Intelligent temperature control system for glass melting furnace based on multi-parameter collaborative optimization

Abstract

This invention discloses an intelligent temperature control system for glass melting furnaces based on multi-parameter collaborative optimization, relating to the field of furnace temperature control technology. The system includes: acquiring multi-source operating parameters of the glass melting furnace and images of the furnace charge surface to construct a unified time-scaled operating condition sequence; extracting charge coverage information and combining it with the feed mass flow rate, glass melt level changes, and drawing speed changes to construct material endothermic state parameters; sequentially executing feedforward endothermic trigger judgment, back-end thermal diversion judgment, and composite coupling judgment to output the current control mode; and generating joint control quantities according to the control mode and correcting the temperature control parameter set. This invention achieves early identification of material-side endothermic drag, hierarchical differentiation of temperature deviation sources, and collaborative control and continuous adaptive correction of multiple channels, improving the feedforward nature, targeting, and stability of furnace temperature control.

Description

Technical Field

[0001] This invention relates to the field of furnace temperature control technology, and in particular to an intelligent temperature control system for glass melting furnaces based on multi-parameter collaborative optimization. Background Technology

[0002] Conventional glass melting furnace temperature control typically employs methods such as temperature measurement point acquisition, fuel flow regulation, combustion air ratio adjustment, and coordinated furnace pressure control to maintain stable thermal conditions in the melting zone, refining zone, and working end. As glass melting furnaces evolve towards continuous, efficient, and refined operation, temperature control is no longer limited to a single closed-loop temperature control system. Instead, it increasingly incorporates operating parameters such as flue gas oxygen content, flue gas temperature, feeding cycle time, molten glass level, and drawing speed to comprehensively regulate combustion heating, heat exchange organization, and material handling, thereby improving the coordination of heat distribution within the furnace and the stability of the production process.

[0003] However, the above-mentioned conventional methods still have two main limitations in practical applications: First, they mainly rely on temperature results for feedback regulation, and lack feedforward characterization of material-side heat absorption drag caused by changes in furnace charge coverage, feeding load and liquid level, and pulling, thus limiting their ability to identify temperature deviations in the early stages. Second, they mostly use parallel processing for heating-side imbalances, heat exchange-side leakage, and material-side disturbances, making it difficult to differentiate the sources of temperature deviations hierarchically and form differentiated control paths, thereby affecting the targeted nature of multi-channel coordinated control. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a glass melting furnace temperature intelligent control system based on multi-parameter collaborative optimization to solve the problems of insufficient feedforward identification of material-side heat absorption disturbance and insufficient layered identification of temperature deviation sources in the prior art.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides an intelligent temperature control system for a glass melting furnace based on multi-parameter collaborative optimization, comprising: a condition acquisition module, which acquires multi-source operating parameters of the glass melting furnace and images of the furnace charge surface to construct a unified time-scaled condition sequence; a heat absorption construction module, which extracts furnace charge coverage information based on the unified time-scaled condition sequence and constructs a material heat absorption state quantity by combining the feed mass flow rate, glass melt level change, and drawing speed change; a mode determination module, which, based on the unified time-scaled condition sequence and the material heat absorption state quantity, sequentially executes feedforward heat absorption trigger determination, back-end thermal diversion determination, and composite coupling determination, and outputs the current control mode; a joint control module, which, based on the control mode, determines the calling order and adjustment direction of fuel flow rate, combustion air flow rate, flue gas extraction force, and feeding cycle, calculates the joint control quantity, and issues it for execution; and a feedback correction module, which acquires the actual response data after the joint control quantity is executed, constructs a unified response normalized deviation quantity, and corrects the temperature control parameter set for the next control cycle.

[0007] As a preferred embodiment of the intelligent temperature control system for glass melting furnace based on multi-parameter collaborative optimization described in this invention, the construction of a unified time-scaled operating condition sequence includes: collecting corresponding parameters at the melting zone, refining zone, working end, fuel main pipe, combustion air main pipe, main exhaust channel, furnace stable pressure zone, and fixed field of view position at the furnace top; synchronously reading continuous quantities, feeding cycle, and furnace charge surface images at a fixed control cycle using a unified master clock; identifying abnormal sampling values ​​based on range checks and mutation thresholds, and supplementing short-term abnormal values; and stacking all valid parameters in chronological order to form a unified time-scaled operating condition sequence.

[0008] As a preferred embodiment of the intelligent temperature control system for glass melting furnace based on multi-parameter collaborative optimization described in this invention, the construction of the material endothermic state quantity includes: reading the furnace charge surface image, feed mass flow rate, glass melt level, drawing speed, and melting zone temperature according to a historical window based on a unified time-scaled operating condition sequence; extracting a fixed observation area, obtaining the furnace charge coverage area based on the furnace charge coverage identification grayscale threshold, and calculating the coverage rate; identifying continuous coverage segments based on the coverage rate according to the lowest coverage level, and taking the intensity of the maximum segment as the coverage duration characterization quantity.

[0009] As a preferred embodiment of the intelligent temperature control system for a glass melting furnace based on multi-parameter collaborative optimization described in this invention, the construction of the material endothermic state quantity further includes: calculating the absolute value of the glass melt level change in adjacent control cycles and taking the maximum value as the level disturbance characterization data; calculating the absolute value of the pull speed change in adjacent control cycles and taking the maximum value as the pull disturbance characterization data; taking the median level of the feed mass flow rate within the historical window as the feed load characterization data; and constructing the material endothermic state quantity based on the coverage duration characterization quantity, feed load characterization data, level disturbance characterization data, pull disturbance characterization data, average temperature of the melting zone, and temperature slow-release reference coefficient.

[0010] As a preferred embodiment of the intelligent temperature control system for a glass melting furnace based on multi-parameter collaborative optimization described in this invention, the output current control mode includes: constructing a chain-like temperature deviation based on the melting zone temperature, the refining zone temperature, and the working end temperature; constructing a heating-side deviation based on the fuel flow rate, the combustion air flow rate, and the oxygen content of the flue gas, and constructing a heat exchange-side deviation based on the furnace pressure and the flue gas temperature; and sequentially performing feedforward heat absorption triggering determination, back-end thermal diversion determination, and composite coupling determination based on the material heat absorption state quantity, the heating-side deviation, and the heat exchange-side deviation.

[0011] As a preferred embodiment of the intelligent temperature control system for glass melting furnace based on multi-parameter collaborative optimization described in this invention, the output current control mode further includes: when the current feed-in heat absorption trigger is established and the composite coupling is not established, outputting a heat absorption suppression priority mode; when the current feed-in heat absorption trigger is not established and the back-end thermal diversion is determined to prioritize heat supply recovery, outputting a heat supply recovery priority mode; when the current feed-in heat absorption trigger is not established and the back-end thermal diversion is determined to prioritize heat exchange closure, outputting a heat exchange closure priority mode; and when the current feed-in heat absorption trigger is established and the composite coupling is established, outputting a composite vibration suppression mode.

[0012] As a preferred embodiment of the intelligent temperature control system for glass melting furnace based on multi-parameter collaborative optimization described in this invention, the calculation and execution of the joint control quantity includes: determining the calling order of the fuel flow channel, combustion air flow channel, exhaust draft channel, and feeding cycle channel according to the current control mode; and generating the joint control base quantity based on the chain temperature deviation, feedforward heat absorption trigger criterion, back-end thermal diversion criterion, composite coupling criterion, and melting zone temperature.

[0013] As a preferred embodiment of the intelligent temperature control system for glass melting furnace based on multi-parameter collaborative optimization described in this invention, the calculation and execution of the joint control quantity further includes: when the heat absorption suppression priority mode, heat supply recovery priority mode, or heat exchange closure priority mode is used, determining the actual increase or decrease of the corresponding channel respectively; when the composite vibration suppression mode is used, generating the final correction quantity of fuel flow rate based on the joint control base quantity, composite coupling criterion, and final correction quantity of feeding cycle time; assembling the final correction quantity of fuel flow rate, final correction quantity of combustion air flow rate, final correction quantity of flue gas extraction force, and final correction quantity of feeding cycle time into a joint control quantity, and issuing it to the corresponding execution mechanism in the order of call.

[0014] As a preferred embodiment of the intelligent temperature control system for a glass melting furnace based on multi-parameter collaborative optimization described in this invention, the construction of a unified response normalized deviation includes: establishing a response observation window after the joint control quantity is issued, collecting the melting zone temperature, clarifying zone temperature, working end temperature, and the actual execution value of each execution channel; comparing the three temperature zones with the target temperature to form a temperature response deviation; comparing the actual execution value of each execution channel with the target correction quantity to form an execution deviation, and constructing a unified response normalized deviation.

[0015] As a preferred embodiment of the intelligent temperature control system for glass melting furnace based on multi-parameter collaborative optimization described in this invention, the specific steps for correcting the temperature control parameter set for the next control cycle are as follows: comparing the normalized deviation of the unified response with the response correction threshold; when the normalized deviation of the unified response is greater than the response correction threshold, correcting the furnace charge coverage identification grayscale threshold, temperature slow-release reference coefficient, feedforward trigger judgment threshold, back-end diversion judgment threshold, and composite coupling threshold; and writing the corrected parameters into the temperature control parameter set for the next control cycle.

[0016] The beneficial effects of this invention are as follows: by constructing the material endothermic state quantity, the early identification of the material endothermic drag is realized, thereby improving the feedforward nature of furnace temperature control; by outputting the current control mode, the hierarchical differentiation of the sources of temperature deviation is realized, thereby improving the pertinence of control decisions; by generating joint control quantities according to the control mode and correcting the temperature control parameter set, multi-channel collaborative control and continuous adaptive correction are realized, thereby improving the stability of temperature control. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of a glass melting furnace temperature intelligent control system based on multi-parameter collaborative optimization.

[0019] Figure 2 A flowchart for constructing a unified time-scaled operating condition sequence.

[0020] Figure 3 A flowchart for constructing the endothermic state parameters of a material.

[0021] Figure 4 This is a flowchart for control mode determination and joint regulation. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figures 1-4 This is one embodiment of the present invention, which provides a glass melting furnace temperature intelligent control system based on multi-parameter collaborative optimization, including the following steps: The operating condition acquisition module collects multi-source operating parameters of the glass melting furnace and images of the furnace charge surface to construct a unified time-scaled operating condition sequence.

[0026] The data collection targets are fixed on the glass melting furnace, and a unique collection location and data source are determined for each type of data collection target. Specifically, a first temperature collection point is set in the melting zone, a second temperature collection point is set in the refining zone, and a third temperature collection point is set at the working end; a fuel flow collection point is set in the fuel main pipe, a combustion air flow collection point is set in the combustion air main pipe, a flue gas oxygen content collection point and a flue temperature collection point are set in the main exhaust channel, and a furnace pressure collection point is set in the furnace stabilization pressure zone; the feeding mass flow rate and feeding cycle are obtained from the feeding actuator, the glass melt level is obtained from the liquid level detection device, the pulling speed is obtained from the pulling actuator, and an imaging device is set at a fixed field of view position on the furnace top to obtain images of the furnace charge surface.

[0027] A unified master clock is established in the controller, and each fixed time (e.g., 2 seconds) is set as a control cycle. At the end of each control cycle, the unified master clock simultaneously triggers each acquisition channel to perform a synchronous read. For continuous quantities such as temperature, flow rate, oxygen content, pressure, liquid level, and pulling speed, the stable sampled value at the end of the current control cycle is read. For event quantities such as feeding cycle time, the number of feeding actions in the current control cycle is counted, and the number is converted into the cycle time value corresponding to the current cycle. For images of the furnace charge surface, a snapshot is taken at the end of the current control cycle, and the image is bound to the current control cycle.

[0028] Each parameter undergoes a range validity check, eliminating out-of-bounds values ​​that clearly exceed the equipment's physical range and the process's allowable range. The relative change amplitude and abrupt change threshold between the current control cycle and the previous control cycle are compared; if the relative change amplitude exceeds the abrupt change threshold, the current sampled value is identified as an abrupt anomaly. If a parameter passes both the range check and the abrupt change check within the current control cycle, it is recorded as a valid sampled value; otherwise, it is recorded as an abrupt sampled value. For sampled values ​​determined to be abrupt within the current control cycle, the duration of the abrupt change is further determined. If the duration does not exceed two control cycles, data from adjacent valid cycles before and after the parameter are used to supplement the current abrupt value. If the duration exceeds two control cycles, the current parameter is not forcibly inferred, but rather marked as temporarily excluded from operational assembly within the current control cycle.

[0029] It should be noted that the mutation threshold is determined by the maximum normal fluctuation limit of the relative change amplitude between adjacent periods in the continuous sampling data during the stable operation phase of the equipment, and the value range is [0.15, 0.60].

[0030] All valid parameters within the current control cycle are assembled to form the operating condition record for the current control cycle. The operating condition record includes at least the following: melting zone temperature, refining zone temperature, working end temperature, fuel flow rate, combustion air flow rate, flue gas oxygen content, furnace pressure, flue temperature, feed mass flow rate, feed cycle time, molten glass level, drawing speed, and furnace charge surface image record. The operating condition records of multiple consecutive control cycles (e.g., 15) are stacked in chronological order to form a unified time-stamped operating condition sequence.

[0031] The heat absorption construction module extracts furnace charge coverage information based on a unified time-scaled operating condition sequence, and constructs the material heat absorption state quantity by combining the feed mass flow rate, glass melt level change, and drawing speed change.

[0032] The system continuously reads historical images (e.g., the most recent fifteen control cycles) of the furnace charge surface, charge mass flow rate, molten glass level, drawing speed, and melting zone temperature from the historical window forward from the current control cycle. For each control cycle's corresponding furnace charge surface image, a fixed observation area corresponding to the molten pool surface is first extracted. Specifically, during the equipment installation and commissioning phase, the effective molten pool field of view boundary corresponding to the furnace top observation port is calibrated. During operation, only this effective molten pool field of view area is retained for each frame of image, while the background areas corresponding to the furnace wall, observation port edge, and fixed support structure are removed. The images within the fixed observation area are then processed to grayscale, and the data is analyzed based on the furnace charge coverage area and the bare metal surface. The difference in radiance between exposed high-temperature molten glass is used to extract the coverage area. Since the temperature of the furnace charge coverage area is relatively low, the surface radiance is lower than that of the exposed high-temperature molten glass. Therefore, under fixed observation conditions, the furnace charge coverage area appears as a low-grayscale area in the grayscale image, while the exposed high-temperature molten glass appears as a high-grayscale area. Based on this feature, the furnace charge coverage recognition grayscale threshold separation method is used for the grayscale image of each control cycle to obtain the furnace charge coverage area of ​​the current control cycle. Connectivity repair processing is then performed to fill the small gaps inside the same coverage block and remove isolated noise points. Each control cycle obtains a furnace charge coverage area that corresponds one-to-one with the current cycle.

[0033] It should be noted that the grayscale threshold for identifying furnace charge coverage is determined by the grayscale statistical results of the exposed high-temperature molten glass area and the furnace charge coverage area in the furnace charge surface image collected during the stable operation phase of the equipment. Preferably, the boundary value between the mean grayscale values ​​of the two types of areas is taken, with a value range of [80, 180].

[0034] After obtaining the charge coverage area for each control cycle, the area ratio of the charge coverage area in the fixed observation area is calculated to obtain the coverage rate corresponding to the control cycle. Specifically, the pixel area occupied by the charge coverage area in each control cycle is ratio-processed to the total pixel area of ​​the fixed observation area to obtain the coverage rate value of the current control cycle. The coverage rates of each control cycle within the historical window are obtained sequentially to form a coverage rate sequence. After forming the coverage rate sequence, the coverage rate sequence within the historical window is continuously characterized to distinguish between short-term transient coverage and continuous maintenance coverage. Specifically, the coverage rate of multiple consecutive control cycles is continuously analyzed. If the coverage rate of adjacent control cycles remains at the minimum coverage level, then... If the coverage rate is above the minimum level, the current coverage is considered to be in a continuous state. If the coverage rate only increases briefly within a single control cycle and then quickly falls back, the current coverage is considered to be non-continuous. The coverage rates of each control cycle within the historical window are arranged in chronological order, and coverage segments that are continuously greater than or equal to the minimum coverage level are identified. The number of consecutive cycles of the coverage segment is recorded, and the average coverage rate within the segment is calculated. The number of consecutive cycles and the average coverage rate within the segment are used as the segment intensity of the coverage segment. After traversing all coverage segments, the maximum segment intensity is taken as the coverage continuity characteristic. The larger the coverage continuity characteristic, the longer the furnace charge coverage lasts within the current historical window, and the higher the probability of subsequent endothermic drag formation.

[0035] It should be noted that the minimum coverage level is determined by historical data of furnace charge coverage rate collected during the stable operation phase of the equipment. The preferred value is the boundary between the upper limit of coverage rate under normal non-continuous coverage conditions and the lower limit of coverage rate at the start of continuous coverage, with a value range of [0.20, 0.45].

[0036] After establishing a continuous representation of the coverage, the changes in the molten glass level data within the historical window are analyzed to characterize the degree of disturbance in the current material balance. Specifically, the changes in molten glass level between adjacent control cycles are calculated cycle by cycle. The molten glass level values ​​for multiple consecutive control cycles within the historical window are read. First, the absolute value of the level change between every two adjacent control cycles is calculated. Then, the maximum value is extracted from all the absolute values ​​of the level change as the level disturbance representation data. If the level change is relatively gradual within the historical window, it indicates that the current molten glass material balance is relatively stable. If the level change is frequent and significant, it indicates that the balance between the current cold material entry, melting, and molten glass flow is changing, and this change will enhance the influence of material heat absorption on temperature.

[0037] The variation of the drawing speed within the historical window is analyzed to characterize the disturbance of the material residence state and heat balance on the discharge side. Specifically, the change in drawing speed between adjacent control cycles is calculated cycle by cycle. The drawing speed values ​​of multiple consecutive control cycles within the historical window are read. First, the absolute value of the drawing speed change between every two adjacent control cycles is calculated. Then, the maximum value is extracted from all the absolute values ​​of the drawing speed change as the drawing disturbance characterization data. If the drawing speed is basically stable within the historical window, it indicates that the discharge side has little impact on the material residence time inside the melting furnace. If the drawing speed changes frequently within the historical window, it indicates that the change in the rhythm of the discharge side will change the glass melt renewal rate and material residence state inside the furnace, thereby enhancing the impact of cold material heat absorption on the temperature chain.

[0038] The material flow rate within the historical window is processed to form material load characterization data. Specifically, the material flow rate within the historical window is continuously read, and the median of the material flow rate within the historical window is taken as the load value of the current cycle's material feeding level.

[0039] First, the basic strength of the cold material heat source is characterized by the combined characteristics of the coverage duration and the feeding load. Then, the slow-release threshold of the melting zone temperature on the heat source is altered through the multiplicative effect of the liquid level disturbance and the pulling disturbance. This results in a lower weakening effect of the melting zone temperature on the cold material's heat absorption when the material equilibrium disturbance is more pronounced. The material heat absorption state quantity is then constructed, expressed as: ; in, For the first The amount of heat absorbed by the material in each control cycle. To cover continuous characterization, This is a normalized characterization of the material feeding load data within the historical window. This is normalized data of the average temperature of the melting zone within the historical window. This is the temperature-controlled release baseline coefficient. Normalized data to characterize liquid level disturbances. To characterize the normalized data for the pull perturbation.

[0040] It should be noted that the temperature-slow-release baseline coefficient is determined by statistically analyzing the slope of the decrease in the material's endothermic state quantity as the average temperature of the melting zone increases over multiple historical periods when the furnace charge coverage and feed mass flow rate are at the same level. The median of this slope is used as the temperature-slow-release baseline coefficient, with a value range of [0.001, 0.020]. The ratio of the average temperature of the melting zone to the target temperature of the melting zone within the historical window is used as the baseline coefficient. The ratio of liquid level disturbance characterization data to the maximum allowable fluctuation value of liquid level during the stable production stage is used as... The ratio of the pulling disturbance characterization data to the maximum allowable fluctuation value of the pulling speed during the stable production stage is used as... ; The data on the feeding load characterization were obtained by performing interval normalization on the upper and lower limits of the feeding mass flow rate obtained statistically under normal operating conditions during the production stage.

[0041] The mode determination module, based on the unified time-scale operating condition sequence and the material heat absorption state quantity, sequentially performs feedforward heat absorption trigger determination, back-end thermal diversion determination, and composite coupling determination, and outputs the current control mode.

[0042] The melting zone temperature, refining zone temperature, and working end temperature are read from the unified time-scale operating condition sequence for the current control cycle and its historical window. The melting zone temperature difference is used as the main characteristic of the current temperature deviation. The refining zone temperature difference and the working end temperature difference are introduced together as chain-transmission constraints to indicate whether the current temperature deviation has expanded from a local thermal fluctuation in the melting zone to a continuous thermal deviation that continues to propagate along the refining zone and the working end. When the temperature difference between the refining zone and the working end is not synchronized with the temperature difference in the melting zone, the chain-transmission temperature deviation mainly represents a local deviation. When the temperature difference between the refining zone and the working end increases synchronously with the temperature difference in the melting zone, the chain-transmission temperature deviation represents a chain-transmission thermal deviation with continuous transmission characteristics between the key temperature zones. The chain-transmission temperature deviation is constructed to reflect the forward and backward transmission relationship between the three key temperature zones of the glass melting furnace. The expression is as follows: ; in, For the first The chain temperature deviation per control cycle For the first The temperature of the melting zone in each control cycle, For the first The clarification zone temperature for each control cycle. For the first Operating end temperature per control cycle , as well as These are the target temperatures for the melting zone, the refining zone, and the working end, respectively. , as well as These are the allowable deviation bandwidths for the melting zone, refining zone, and working end, respectively. This indicates a small positive number that prevents the denominator from being zero.

[0043] It should be noted that the allowable deviation bandwidth of the melting zone, refining zone and working end is determined based on the target temperature fluctuation range of the corresponding temperature zone during the stable production stage. Specifically, the maximum allowable fluctuation range of continuous historical temperature data under qualified product operating conditions is taken as the allowable deviation bandwidth of the corresponding temperature zone.

[0044] True imbalances on the heating side of a glass melting furnace are rarely caused by slight drifts in a single parameter. Rather, they only manifest as heating anomalies when fuel, air, and flue gas oxygen levels all deviate simultaneously. Only when all three are synchronously abnormal does the heating-side deviation amplify significantly. Using the fuel flow rate, combustion air flow rate, and flue gas oxygen content of the current control cycle, a heating-side deviation can be constructed to characterize whether a significant deviation has occurred in the heating chain. The expression is: ; in, For the first The deviation of the heating side in each control cycle For the first Fuel flow rate per control cycle For the first Combustion air flow rate per control cycle For the first Oxygen content in flue gas during each control cycle. For the target value of fuel flow rate, The target value for combustion airflow. This is the target value for the oxygen content in the flue gas. , and These are the allowable deviation bandwidths for fuel flow rate, combustion air flow rate, and flue gas oxygen content, respectively.

[0045] It should be noted that the allowable deviation bandwidth of fuel flow rate, combustion air flow rate and flue gas oxygen content is determined based on the actual fluctuation range of the corresponding parameters during the stable production stage. Specifically, the maximum allowable fluctuation range of each parameter around its set value during continuous production of qualified products is taken as the corresponding allowable deviation bandwidth.

[0046] When abnormal furnace pressure and flue temperature occur simultaneously, it usually indicates that heat leakage and cold air intrusion effects are more pronounced. A heat transfer-side deviation is constructed using the furnace pressure and flue temperature during the current control cycle to indicate whether there has been a significant deviation in the furnace's thermal organization and external heat transfer status. The expression is: ; in, For the first Deviation on the heat exchange side per control cycle For the first Furnace pressure for each control cycle For the first Flue temperature for each control cycle The target value for furnace pressure. The target value for flue gas temperature. and These are the allowable deviation bandwidths for furnace pressure and flue gas temperature, respectively.

[0047] It should be noted that the allowable deviation bandwidth of furnace pressure and flue temperature is determined based on the actual fluctuation range of furnace pressure and flue temperature around their respective target values ​​during the stable production stage. Specifically, the maximum allowable fluctuation range corresponding to the two under the continuous production conditions of qualified products is taken as the allowable deviation bandwidth.

[0048] The coupling amount of the material's endothermic state quantity with the deviation of heating supply and heat exchange is compared to calculate the feedforward endothermic trigger criterion, expressed as: ; in, For the first Feedforward heat absorption triggering criteria for each control cycle.

[0049] Then, the feedforward endothermic trigger criterion is compared with the feedforward trigger judgment threshold. If the feedforward endothermic trigger criterion is greater than or equal to the feedforward trigger judgment threshold, the current temperature deviation is determined to be triggered first by the material side endothermic drag. If the feedforward endothermic trigger criterion is less than the feedforward trigger judgment threshold, the current temperature deviation is determined to be triggered first by the back-end thermal anomaly.

[0050] It should be noted that the feedforward triggering threshold is determined by calculating the ratio of the material's endothermic state quantity to the downstream thermal anomaly coupling quantity on a cycle-by-cycle basis, selecting the historical cycles in which the subsequent temperature drop is indeed triggered first by the material side, and taking the median value between the minimum and maximum values ​​of the corresponding ratios in the historical cycles as the feedforward triggering threshold, with a value range of [0.55, 0.80].

[0051] When it is determined that the current temperature deviation is not dominated by feedforward heat absorption, the downstream thermal anomaly is further analyzed to distinguish whether the current temperature deviation is mainly due to imbalance on the heating side or the heat exchange side. The downstream thermal anomaly analysis is performed using the following expression: ; in, For the first Criteria for back-end thermal diversion in each control cycle.

[0052] The back-end thermal diversion criterion is compared with the back-end diversion judgment threshold. If the back-end thermal diversion criterion is greater than or equal to the back-end diversion judgment threshold, the current temperature deviation is determined to be of the heating recovery priority type. If the back-end thermal diversion criterion is less than the back-end diversion judgment threshold, the current temperature deviation is determined to be of the heat exchange closure priority type.

[0053] The back-end diversion judgment threshold is determined by calculating the proportion of the deviation on the heating side to the total deviation on the heating side and the deviation on the heat exchange side on a cycle-by-cycle basis. The historical cycles that correspond to which subsequent regulation requires priority restoration of heating are selected, and the median value between the minimum and maximum values ​​of the corresponding proportions in the historical cycles is taken as the back-end diversion judgment threshold, with a value range of [0.45, 0.70].

[0054] The expression for determining whether the current operating condition belongs to a complex coupled state of material heat absorption dragging and heat supply imbalance is as follows: ; in, For the first Composite coupling criterion for each control cycle.

[0055] It should be noted that the composite coupling criterion is used to characterize whether the material heat absorption drag and the heat supply imbalance work together to affect the current temperature deviation within the same control cycle. Among them, the material heat absorption state quantity is used to characterize the intensity of the material heat absorption drag, and the heat supply side deviation quantity is used to characterize the intensity of the heat supply chain imbalance. When both increase together, it indicates that the current operating condition has both the characteristics of front-end heat absorption drag and back-end insufficient heat supply. The heat exchange side deviation quantity is introduced as a suppression term to avoid misidentifying the operating condition as a composite coupling state when the heat exchange anomaly is dominant. Therefore, when the material heat absorption state quantity and the heat supply side deviation quantity increase synchronously and the heat exchange side deviation quantity is not dominant, the composite coupling criterion increases, indicating that the current operating condition is more in line with the triggering conditions of the composite vibration suppression mode.

[0056] The composite coupling criterion is compared with the composite coupling threshold. If the composite coupling criterion is greater than or equal to the composite coupling threshold, and feedforward heat absorption triggering has been determined, then the current operating condition is determined to enter the composite vibration suppression mode. If the composite coupling criterion is less than the composite coupling threshold, then the single control mode determination result obtained in the previous step is maintained.

[0057] It should be noted that the composite coupling threshold is determined by calculating the coupling results of the material's endothermic state quantity and the deviation of the heating side on a cycle-by-cycle basis. Historical cycles that require simultaneous execution of endothermic suppression and heating recovery are selected, and the average of the minimum and maximum values ​​in the corresponding coupling results of the historical cycles is used as the composite coupling threshold, with a value range of [0.10, 0.80].

[0058] The system performs hierarchical mode mapping on the current temperature deviation source and outputs the current control mode. Specifically, when it is determined to be feedforward heat absorption triggering and not determined to be compound coupling, the system outputs heat absorption suppression priority mode; when it is determined to be back-end thermal anomaly and determined to be heating recovery priority, the system outputs heating recovery priority mode; when it is determined to be back-end thermal anomaly and determined to be heat exchange closure priority, the system outputs heat exchange closure priority mode; and when it is determined to enter a compound coupling state, the system outputs compound vibration suppression mode.

[0059] The joint control module determines the order and direction of adjustment for fuel flow, combustion air flow, exhaust suction, and feeding cycle based on the control mode, calculates the joint control quantity, and issues it for execution.

[0060] If the current control mode is heat absorption suppression priority mode, the feeding cycle channel is called first, followed by the fuel flow channel, while the combustion air flow channel and the exhaust draft channel remain frozen during this cycle. If the current control mode is heat supply recovery priority mode, the fuel flow channel is called first, followed by the combustion air flow channel, while the feeding cycle channel and the exhaust draft channel remain frozen during this cycle. If the current control mode is heat exchange closure priority mode, the exhaust draft channel is called first, followed by the fuel flow channel, while the combustion air flow channel and the feeding cycle channel remain frozen during this cycle. If the current control mode is composite vibration suppression mode, the feeding cycle channel is called first, followed by the fuel flow channel, then the combustion air flow channel, and finally the exhaust draft channel.

[0061] After the control channel call sequence is determined, a joint control baseline quantity for the current control cycle is generated. The intensity of the current temperature deviation is characterized by a chain-like temperature deviation quantity. Then, the most prominent mode feature of the current control cycle is introduced to demonstrate that the current deviation not only exists but also has a clear dominant mechanism. Simultaneously, the thermal buffering capacity characterized by the melting zone temperature is combined to constrain the adjustment intensity. This ensures that the stronger the thermal buffering within the furnace, the smoother the adjustment action in this cycle. This is used to uniformly characterize the overall adjustment intensity of this cycle, expressed as: ; in, For the first The joint control base quantity of each control cycle This is the temperature release coefficient in the melting zone.

[0062] It should be noted that the melting zone temperature slow-release coefficient is determined by statistically analyzing the change in the reduction of the joint control base quantity when the melting zone temperature increases over multiple historical periods when the control mode remains unchanged and the material endothermic state quantity is at the same level. The median value of the change is then used as the melting zone temperature slow-release coefficient, with a value range of [0.001, 0.020].

[0063] When the current control mode is the endothermic suppression priority mode, the product of the joint control base quantity and the feedforward endothermic trigger criterion is used as the target reduction amount of the feeding cycle in the current control cycle, and the minimum value between the target reduction amount of the feeding cycle and the maximum allowable suppression boundary of the feeding cycle is used as the actual reduction of the feeding cycle. After the feeding cycle is reduced, the absolute value of the ratio of the joint control base quantity and the actual reduction of the feeding cycle is used as the target compensation amount of the fuel flow rate, and the minimum value between the target compensation amount of the fuel flow rate and the maximum allowable increase boundary of the fuel flow rate is used as the actual increase of the fuel flow rate.

[0064] It should be noted that the maximum allowable reduction boundary of the feeding cycle is obtained by ensuring the continuous and stable feeding of the melting furnace and the stability of the glass melt level; the maximum allowable increase boundary of the fuel flow rate is obtained by ensuring the maximum allowable increase under the conditions of not exceeding the burner safe load, the allowable rate of furnace temperature rise and the product quality control requirements.

[0065] When the current control mode is the heating recovery priority mode, the product of the joint control base quantity and the back-end thermal diversion criterion is used as the target increase in fuel flow rate, and the minimum value between the target increase in fuel flow rate and the maximum allowable increase boundary of fuel flow rate is selected as the actual increase in fuel flow rate. After the actual increase in fuel flow rate is determined, the square root of the product of the actual increase in fuel flow rate and the joint control base quantity is used to obtain the target increase in combustion air flow rate, and the minimum value between the target increase in combustion air flow rate and the maximum allowable increase boundary of combustion air flow rate is selected as the actual increase in combustion air flow rate.

[0066] It should be noted that the maximum allowable increase boundary for combustion air flow is obtained by the maximum allowable increase under the conditions of not exceeding the allowable load of the combustion fan, the safe range of combustion air distribution, and the control requirements for the oxygen content of flue gas.

[0067] When the current control mode is the heat exchange closed-loop priority mode, the combined control base quantity is first multiplied by the result of the 1 minus the back-end thermal diversion criterion to obtain the target increase in flue gas extraction force. The minimum value between the target increase in flue gas extraction force and the maximum allowable increase boundary of flue gas extraction force is selected as the actual increase in flue gas extraction force. After the actual increase in flue gas extraction force is determined, the ratio of the combined control base quantity and the actual increase in flue gas extraction force is used as the target compensation quantity for fuel flow rate. The minimum value between the target compensation quantity for fuel flow rate and the maximum allowable increase boundary of fuel flow rate is selected as the actual increase in fuel flow rate. In the heat exchange closed-loop priority mode, the combustion air flow rate and the feeding cycle remain unchanged in this cycle.

[0068] It should be noted that the maximum allowable increase boundary for flue gas extraction force is obtained by the maximum allowable increase under the condition that it does not exceed the allowable load of the induced draft actuator, the safe range of furnace pressure, and the requirements for flue gas temperature control.

[0069] When the current control mode is the composite vibration suppression mode, four actions are executed sequentially: material reduction, heat replenishment, air following, and flue gas extraction fine-tuning. This allows for the simultaneous suppression of material heat absorption drag and heat supply imbalance, generating the final correction amount for fuel flow, expressed as: ; in, For the first The final correction amount of fuel flow rate for each control cycle To maximize the allowable increase in fuel flow rate, For the first The final correction amount for the feeding cycle that has been determined in each control cycle.

[0070] After the final correction values ​​for each channel corresponding to the current control mode are formed, the final correction values ​​for fuel flow, combustion air flow, exhaust suction, and feeding cycle are assembled to form the joint control value for the current control cycle.

[0071] The joint control quantities are sequentially issued to the corresponding actuators according to the calling order corresponding to the current control mode. The final correction quantity for fuel flow is issued to the fuel supply actuator, the final correction quantity for combustion air flow is issued to the combustion air conditioning actuator, the final correction quantity for exhaust draft is issued to the exhaust draft actuator, and the final correction quantity for feeding cycle time is issued to the feeding actuator.

[0072] The feedback correction module collects the actual response data after the joint control quantity is executed, constructs a unified response normalized deviation, and corrects the temperature control parameter set for the next control cycle.

[0073] After the joint control quantity of the current control cycle is issued, a response observation window of fixed length (e.g., ten control cycles) is established starting from the end time of the control cycle. Within the response observation window, the following data are collected for each control cycle: melting zone temperature, refining zone temperature, working end temperature, actual executed value of fuel flow rate, actual executed value of combustion air flow rate, actual executed value of flue gas extraction force, actual executed value of charging cycle, furnace charge coverage status, molten glass level, and drawing speed. Among these, the three temperature zones are used to reflect the actual impact of the current control action on the thermal state; the actual executed values ​​of fuel flow rate, combustion air flow rate, flue gas extraction force, and charging cycle are used to determine whether the actuator is in place as instructed; and the furnace charge coverage status, molten glass level, and drawing speed are used to determine whether material-side disturbances have been suppressed.

[0074] After the data collection in the response observation window is completed, a temperature response result is first generated. Specifically, the temperatures of the three temperature zones within the response observation window are compared with the corresponding target temperatures on a cycle basis to determine whether the temperatures of the three temperature zones converge toward the target value, remain unchanged, or continue to deviate after the current joint control is implemented. If the deviation of at least two of the three temperature zones continues to decrease, the control in this cycle is considered effective in terms of temperature results. If only one temperature zone improves or all three temperature zones continue to deteriorate simultaneously, the control in this cycle is considered insufficient in terms of temperature results, and an execution result is generated. Specifically, the actual execution values ​​of fuel flow rate, combustion air flow rate, exhaust suction, and feeding cycle in the response observation window are compared with the corresponding target correction values ​​in the joint control of this cycle. If the actual execution values ​​can follow the target correction values ​​within the allowable deviation, the channel is considered to have executed well. If the actual execution values ​​are consistently less than or lag behind the target correction values, the channel is considered to have not executed well.

[0075] It should be noted that the allowable deviation is obtained by taking the maximum allowable fluctuation range of the corresponding execution channel following error under the continuous production conditions of qualified products.

[0076] The unified response normalized bias is constructed as follows: ; in, For the first The normalized deviation of the unified response for each control cycle For the first Normalized deviation of temperature response per control cycle For the first The normalized deviation of the execution of each control cycle.

[0077] It should be noted that, The actual temperature of each temperature zone within the response observation window is compared with the corresponding target temperature and then normalized according to the allowable temperature deviation bandwidth. The allowable temperature deviation bandwidth is obtained by comparing the actual execution value of each execution channel in the response observation window with the corresponding target correction amount and then normalizing it according to the allowable execution deviation bandwidth. The allowable temperature deviation bandwidth is obtained by statistically analyzing the historical temperature fluctuation range of each temperature zone in the stable production stage and combining it with the upper limit of the allowable fluctuation of the corresponding temperature zone process control. The allowable execution deviation bandwidth is obtained by statistically analyzing the historical execution deviation range of each execution channel in the stable following stage and combining it with the upper limit of the allowable following error of the corresponding actuator.

[0078] The normalized deviation of the unified response is compared with the response correction threshold. When the normalized deviation of the unified response is greater than the response correction threshold, it is determined that there is an effective deviation between the joint control result of the current control cycle and the target response, and the parameters are corrected for the next control cycle. When the normalized deviation of the unified response is not greater than the response correction threshold, the temperature control parameter set for the next control cycle remains unchanged.

[0079] It should be noted that the response correction threshold is calculated by periodically calculating the normalized deviation of the unified response after the joint control quantity is executed, screening the historical periods for which subsequent parameters do not need to be corrected, and taking the maximum value of the normalized deviation of the unified response corresponding to the historical period as the response correction threshold, with a value range of [0.05, 0.30].

[0080] Perform parameter correction for the next control cycle.

[0081] Furthermore, when the current control mode is the heat absorption suppression priority mode or the composite vibration suppression mode, and three consecutive control cycles (e.g., 3) within the response observation window show that the charge coverage continues, the glass melt level fluctuation does not converge, and the temperature deviation of at least two temperature zones continues to expand, the charge coverage identification grayscale threshold and the temperature slow-release reference coefficient are modified so that more low-heat state charge areas are identified as effective coverage areas in the next control cycle, and the characterization result of the material heat absorption state quantity on the material-side heat absorption drag is enhanced; when the current control mode is the heat absorption suppression priority mode or the composite vibration suppression mode, and multiple consecutive control cycles within the response observation window show that the charge coverage weakens rapidly, the temperature deviation of at least two temperature zones continues to converge, and the temperature recovery process is faster than the target recovery trend corresponding to the current control mode, the charge coverage identification grayscale threshold and the temperature slow-release reference coefficient are adjusted in reverse so that the boundary low-heat state area is no longer identified as an effective coverage area in the next control cycle, and the characterization result of the material heat absorption state quantity on the material-side heat absorption drag is weakened.

[0082] When the current control mode is the heat absorption suppression priority mode, and multiple consecutive control cycles within the response observation window show that the material reduction action has been executed, the temperature deviation of at least two temperature zones has not converged, and the deviation on the heating side continues to increase, the feedforward heat absorption trigger judgment parameter is modified so that the condition accompanied by continuous imbalance on the heating side will no longer be classified into the heat absorption suppression priority mode. When the current control mode is the heating recovery priority mode, and multiple consecutive control cycles within the response observation window show that the fuel recovery action has been executed, the temperature deviation has not converged according to the target recovery trend corresponding to the current control mode, and the flue temperature continues to remain at a high level, the back-end thermal diversion judgment parameter is modified so that the condition with both heating side deviation and heat exchange side abnormality is transferred to the judgment range of the heat exchange closure priority mode. When the control mode alternates between the heat absorption suppression priority mode and the heating recovery priority mode in multiple adjacent control cycles, and the response observation window shows both material heat absorption dragging characteristics and heating imbalance characteristics, the composite coupling judgment parameter is modified so that this type of condition is transferred to the judgment range of the composite vibration suppression mode.

[0083] The corrected parameters are uniformly written into the temperature control parameter set of the next control cycle. The corrected parameters include the furnace charge coverage identification grayscale threshold, temperature slow release reference coefficient, feedforward trigger judgment threshold, back-end diversion judgment threshold, and composite coupling threshold for the next control cycle.

[0084] In summary, this invention achieves early identification of material-side heat absorption drag by constructing material endothermic state variables, thereby improving the feedforward nature of furnace temperature control; it achieves hierarchical differentiation of temperature deviation sources by outputting the current control mode, thereby improving the pertinence of control decisions; and it achieves multi-channel collaborative control and continuous adaptive correction by generating joint control variables according to the control mode and correcting the temperature control parameter set, thereby improving the stability of temperature control.

[0085] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A glass melting furnace temperature intelligent regulation system based on multi-parameter collaborative optimization, characterized in that, include: The operating condition acquisition module collects multi-source operating parameters of the glass melting furnace and images of the furnace charge surface to construct a unified time-scaled operating condition sequence. The heat absorption construction module extracts furnace charge coverage information based on a unified time-scaled operating condition sequence, and combines it with the feed mass flow rate, glass melt level change and drawing speed change to construct the material heat absorption state quantity; The mode determination module, based on the unified time-scale operating condition sequence and the material heat absorption state quantity, sequentially performs feedforward heat absorption trigger determination, back-end thermal diversion determination, and composite coupling determination, and outputs the current control mode. The joint control module determines the order and direction of adjustment of fuel flow, combustion air flow, exhaust suction, and feeding cycle according to the control mode, calculates the joint control quantity, and issues it for execution. The feedback correction module collects the actual response data after the joint control quantity is executed, constructs a unified response normalized deviation, and corrects the temperature control parameter set for the next control cycle.

2. The multi-parameter synergy-optimization based intelligent temperature regulation system for glass melting furnace of claim 1, wherein, The construction of the unified time-scaled operating condition sequence includes: Parameters were collected at the melting zone, clarification zone, working end, fuel main pipe, combustion air main pipe, main flue gas duct, furnace stabilization pressure zone, and fixed field of view position on the furnace top. The continuous quantity, feeding cycle, and furnace charge surface image are read synchronously using a unified master clock at a fixed control cycle. Abnormal sampled values ​​are identified based on range checks and mutation thresholds, and short-term abnormal values ​​are filled in. All valid parameters are stacked in chronological order to form a unified time-scaled operating condition sequence.

3. The multi-parameter synergy-optimization-based intelligent temperature regulating system for glass melting furnace according to claim 1 or 2, wherein, The endothermic state of the building material includes: Based on a unified time-scale operating condition sequence, the furnace charge surface image, charge mass flow rate, glass melt level, drawing speed, and melting zone temperature are read according to the historical window. Extract a fixed observation area, obtain the furnace material coverage area based on the grayscale threshold of furnace material coverage identification, and calculate the coverage rate; Based on coverage rate, continuous coverage segments are identified according to the lowest coverage level, and the intensity of the segment with the highest intensity is taken as the characteristic of coverage persistence.

4. The multi-parameter synergy-optimized based intelligent temperature regulating system for glass melting furnace according to claim 3, wherein, The endothermic state of the building material also includes: Calculate the absolute value of the glass melt level change in adjacent control cycles and take the maximum value as the liquid level disturbance characterization data; Calculate the absolute value of the pull speed change between adjacent control cycles and take the maximum value as the pull disturbance characterization data; The median level of the feed mass flow rate within the historical window is taken as the feed load characterization data; Based on the continuous coverage characterization data, feeding load characterization data, liquid level disturbance characterization data, pulling disturbance characterization data, average temperature of the melting zone and temperature slow release reference coefficient, the endothermic state parameters of the material are constructed.

5. The multi-parameter synergy-optimized based intelligent temperature regulating system for glass melting furnace of claim 1 or 4, wherein, The current control mode of the output includes: A chain-like temperature deviation is constructed based on the melting zone temperature, the refining zone temperature, and the working end temperature. The deviation on the heating side is constructed based on the fuel flow rate, combustion air flow rate, and flue gas oxygen content, and the deviation on the heat exchange side is constructed based on the furnace pressure and flue temperature. Based on the material's heat absorption state, the deviation on the heating side, and the deviation on the heat exchange side, the feedforward heat absorption triggering judgment, the back-end thermal diversion judgment, and the composite coupling judgment are performed sequentially.

6. The multi-parameter synergy-optimized based intelligent temperature regulating system for glass melting furnace of claim 5, wherein, The current output control mode also includes: When the feed-in heat absorption trigger is valid and the composite coupling is invalid, the output heat absorption suppression priority mode is used. When the current feed-in heat absorption trigger is not established and the back-end thermal diversion is determined to prioritize heating recovery, output the heating recovery priority mode; When the current feed-in heat absorption trigger is not established and the back-end thermal shunting is determined to prioritize heat exchange closure, output the heat exchange closure priority mode. When the current feed-in heat absorption trigger is established and the composite coupling is established, the output composite vibration suppression mode is activated.

7. The multi-parameter synergy-optimized based intelligent temperature regulating system for glass melting furnace of claim 6, wherein, The calculation and execution of the joint control quantity includes: The order of calling the fuel flow channel, combustion air flow channel, exhaust suction channel, and feeding cycle channel is determined based on the current control mode. Based on the chain temperature deviation, the feedforward endothermic trigger criterion, the back-end thermal diversion criterion, the composite coupling criterion, and the melting zone temperature, a joint control basis quantity is generated.

8. The multi-parameter synergy-optimization based intelligent temperature regulation system for glass melting furnace of claim 7, wherein, The calculation and execution of the joint control quantity also includes: When the heat absorption suppression priority mode, the heat supply recovery priority mode, or the heat exchange closure priority mode is selected, the actual increase or decrease of the corresponding channel is determined respectively. When it is a composite vibration suppression mode, the final correction amount of fuel flow rate is generated based on the joint control base amount, composite coupling criterion and final correction amount of feeding cycle time. The final correction values ​​for fuel flow, combustion air flow, exhaust draft, and feeding cycle are combined to form a joint control quantity, which is then issued to the corresponding actuators in the order of call.

9. The multi-parameter synergy-optimized based intelligent temperature regulating system for glass melting furnace of claim 8, wherein, The constructed unified response normalization bias includes: After the joint control quantity is issued, a response observation window is established to collect the melting zone temperature, clarifying zone temperature, working end temperature, and actual execution values ​​of each execution channel; The temperature response deviation is calculated by comparing the temperatures of the three temperature zones with the target temperature. By comparing the actual execution value of each execution channel with the target correction amount, the execution deviation is formed, and a unified response normalized deviation is constructed.

10. The multi-parameter synergy-optimized based intelligent temperature regulating system for glass melting furnace of claim 9, wherein, The specific steps for modifying the temperature control parameter set for the next control cycle are as follows: The normalized deviation of the unified response is compared with the response correction threshold. When the normalized deviation of the unified response is greater than the response correction threshold, the gray scale threshold for furnace charge coverage identification, the temperature slow release reference coefficient, the feedforward trigger judgment threshold, the back-end diversion judgment threshold, and the composite coupling threshold are corrected. Write the corrected parameters into the temperature control parameter set for the next control cycle.

Images