An AI algorithm-based method and system for accurately regulating oxygen content in a cement clinker kiln

Abstract

The application discloses a cement clinker kiln oxygen content accurate regulation method and system based on an AI algorithm, relates to the cement sintering technical field, and separates the time when the coal quality information changes from the time when the new coal powder actually enters the coal powder bin discharge area, avoids adjusting the current preheater outlet oxygen content target value in advance when the new coal powder has not participated in combustion, simultaneously, confirms the combustion end oxygen supply demand change in combination with the coal quality switching oxygen consumption influence grade, the preheater outlet oxygen content offset trend and the kiln tail carbon monoxide rising state within the bin influence window, and then judges whether the candidate preheater outlet oxygen content target value is executed through the current target retention state, so that the problems of the preheater outlet oxygen content fluctuation, the kiln tail carbon monoxide rising and the clinker sintering stability reduction caused by the oxygen advance adjustment, the oxygen lag adjustment or the frequent oxygen adjustment during the coal source switching are reduced.

Description

Technical Field

[0001] This invention relates to the field of cement firing technology, specifically to a method and system for precise control of oxygen content in cement clinker kilns based on AI algorithms. Background Technology

[0002] In the production process of new dry-process cement clinker, the combustion state of cement clinker kiln needs to be coordinated among fuel supply, ventilation conditions, material decomposition, kiln thermal regime and pollutant emissions. The oxygen content at the preheater outlet can usually reflect the changes in the combustion air margin and reduction risk in the calcination system. As production line operation data, pulverized coal silo data, coal quality data and environmental monitoring data are gradually connected to the central control system, using AI algorithms to accurately control the oxygen content of cement clinker kiln has become an important technical direction to improve calcination stability, reduce reliance on human experience and reduce abnormal fluctuations.

[0003] In actual production, new and old pulverized coal in the pulverized coal silo do not immediately and synchronously enter the kiln head burner or decomposer after coal quality switching. Instead, they undergo stratification replacement during the pulverized coal silo entry, storage, and exit processes. Existing oxygen content control methods typically focus more on the current preheater outlet oxygen content feedback value, kiln tail carbon monoxide feedback value, decomposer outlet temperature feedback value, and coal feed rate feedback value. They lack the ability to determine the actual silo exit impact window of new pulverized coal entering the combustion system based on the silo stratification replacement progress. They also lack the process of selecting candidate preheater outlet oxygen content target values ​​within the silo exit impact window by combining the oxygen consumption impact level of coal quality switching, the preheater outlet oxygen content shift trend, and the kiln tail carbon monoxide rise status. This leads to the tendency for premature misjudgment or delayed response to the impact of coal quality changes on oxygen content. Summary of the Invention

[0004] This invention provides a method and system for precise control of oxygen content in cement clinker kilns based on AI algorithms, which solves the problems mentioned in the background art.

[0005] A method for precise control of oxygen content in cement clinker kilns based on AI algorithms includes the following steps:

[0006] S1. Collect coal powder silo operation information, coal quality information, preheater outlet oxygen content information, kiln tail carbon monoxide information, current preheater outlet oxygen content target value and preheater outlet oxygen content target update record. Define the coal powder entering the coal powder silo after the coal quality information changes as new coal powder. Determine the coal powder silo layer replacement progress based on the coal powder silo operation information, and determine the discharge impact window after the new coal powder enters the coal powder silo discharge area based on the coal powder silo layer replacement progress.

[0007] S2. Within the outflow impact window, the oxygen consumption impact level of coal quality switching is determined based on coal quality information. When the outflow impact window is not determined based on the coal powder bin layer replacement progress, the oxygen consumption impact level of coal quality switching is prohibited from participating in the adjustment of the current preheater outlet oxygen content target value.

[0008] S3. Within the influence window of the outlet, determine the oxygen content deviation trend of the preheater outlet based on the oxygen content information of the preheater outlet and the current target value of the oxygen content of the preheater outlet, determine the carbon monoxide rising state of the kiln tail based on the carbon monoxide information of the kiln tail, and determine the combustion response confirmation state based on the oxygen content deviation trend of the preheater outlet and the carbon monoxide rising state of the kiln tail.

[0009] S4. Based on the oxygen consumption impact level of coal quality switching, the trend of oxygen content deviation at the preheater outlet, the rising state of carbon monoxide at the kiln tail, and the combustion response confirmation state, select a target adjustment strategy from the preset target adjustment strategy table through the AI ​​target selection rule library, and determine the target value of candidate oxygen content at the preheater outlet according to the target adjustment strategy.

[0010] S5. Determine the current target retention status based on the candidate preheater outlet oxygen content target value, the current preheater outlet oxygen content target value, the preheater outlet oxygen content target update record, and the kiln tail carbon monoxide rise status, and generate a preheater outlet oxygen content target update instruction based on the current target retention status.

[0011] Preferably, step S1 includes: S11, reading the coal powder silo inbound metering record, coal powder silo level detection record, and coal powder silo outbound metering record from the coal powder silo inbound metering record, the coal powder silo level detection record, and the coal powder silo outbound metering record, respectively, and transmitting the coal powder silo inbound record, coal powder silo level record, and coal powder silo outbound record to the AI ​​algorithm processing unit; S12, the AI ​​algorithm processing unit establishes a virtual coal powder layer set based on the coal powder silo inbound record, updates the remaining coal powder quantity of each virtual coal powder layer in the virtual coal powder layer set based on the coal powder silo outbound record, and obtains a virtual coal powder layer remaining state set; S13, the AI ​​algorithm processing unit determines the start time and end time of the outbound influence window of the virtual coal powder layer where the new coal powder is located entering the coal powder silo outbound area based on the virtual coal powder layer remaining state set, defines the sampling time period formed from the start time to the end time of the outbound influence window as the outbound influence window, and defines the layer status of the virtual coal powder layer where the new coal powder is located in the coal powder silo as the coal powder silo layer replacement progress.

[0012] Preferably, step S2 includes: S21, reading new coal powder low-calorific value data, new coal powder moisture data, new coal powder volatile matter data, new coal powder ash content data, old coal powder low-calorific value data, old coal powder moisture data, old coal powder volatile matter data, and old coal powder ash content data from coal batch test records; S22, the AI ​​algorithm processing unit performs grading processing on the data read in step S21 according to preset low-calorific value grading range, preset moisture grading range, preset volatile matter grading range, and preset ash content grading range to obtain low-calorific value data grading comparison results, moisture data grading comparison results, volatile matter data grading comparison results, and ash content data grading comparison results; S23, the AI ​​algorithm processing unit combines and judges the low-calorific value data grading comparison results, moisture data grading comparison results, volatile matter data grading comparison results, and ash content data grading comparison results through the AI ​​coal quality status recognition rule library to obtain the oxygen consumption impact level of coal quality switching.

[0013] Preferably, step S3 includes: S31, reading the preheater outlet oxygen content sampling sequence from the online oxygen content detection record, and reading the current preheater outlet oxygen content target value, the preset oxygen content allowable deviation range, and the preset oxygen content judgment sampling number; S32, the AI ​​algorithm processing unit determines the preheater outlet oxygen content target range based on the current preheater outlet oxygen content target value and the preset oxygen content allowable deviation range, and forms the preheater outlet oxygen content judgment sequence from the preheater outlet oxygen content sampling sequence based on the preset oxygen content judgment sampling number; S33, the AI ​​algorithm processing unit determines the preheater outlet oxygen content offset trend as a state within the target range, a state below the target offset, or a state above the target offset based on the comparison result between the preheater outlet oxygen content judgment sequence and the preheater outlet oxygen content target range.

[0014] Preferably, step S3 further includes: S34, reading the kiln tail carbon monoxide sampling sequence from the online carbon monoxide detection record, and reading the preset carbon monoxide upper limit value, the preset carbon monoxide forced update threshold, and the preset carbon monoxide determination sampling number; S35, the AI ​​algorithm processing unit forms a kiln tail carbon monoxide determination sequence from the kiln tail carbon monoxide sampling sequence according to the preset carbon monoxide determination sampling number, and determines the kiln tail carbon monoxide rising state as a normal carbon monoxide state, a carbon monoxide exceeding the upper limit state, or a carbon monoxide forced update state according to the kiln tail carbon monoxide determination sequence, the preset carbon monoxide upper limit value, and the preset carbon monoxide forced update threshold; S36, the AI ​​algorithm processing unit determines the combustion response confirmation state as a combustion response confirmed state or a combustion response unconfirmed state according to the preheater outlet oxygen content shift trend and the kiln tail carbon monoxide rising state.

[0015] Preferably, step S4 includes: S41, assembling a target strategy retrieval data package by combining the coal quality switching oxygen consumption impact level, the preheater outlet oxygen content shift trend, the kiln tail carbon monoxide rise status, and the combustion response confirmation status; S42, calling a preset target adjustment strategy table from the AI ​​target selection rule base, wherein the preset target adjustment strategy table includes multiple target adjustment strategy records, each target adjustment strategy record including a status combination and a target adjustment strategy; S43, the AI ​​target selection rule base retrieves matching target adjustment strategy records from the multiple target adjustment strategy records according to the target strategy retrieval data package, and reads the target adjustment strategy from the matching target adjustment strategy records.

[0016] Preferably, step S4 further includes: S44, calling a preset target value level table, wherein the preset target value level table includes multiple target value level records, each target value level record includes a target adjustment strategy and a candidate target value for the next control cycle; S45, according to the target adjustment strategy read in step S43, retrieving a matching target value level record from the multiple target value level records; S46, reading the candidate target value for the next control cycle from the matching target value level record, and determining the candidate target value for the next control cycle as the candidate preheater outlet oxygen content target value.

[0017] Preferably, step S5 includes: S51, the AI ​​algorithm processing unit determines the target update difference status based on the target value difference between the candidate preheater outlet oxygen content target value and the current preheater outlet oxygen content target value; S52, the AI ​​algorithm processing unit determines the target update count status based on the number of times the preheater outlet oxygen content target update record is updated in the target update statistics window; S53, the AI ​​algorithm processing unit determines the current target retention status as an allowed retention status or an unallowed retention status based on the target update difference status, the target update count status, and the kiln tail carbon monoxide rise status.

[0018] Preferably, step S5 further includes: S54, when the current target retention state is a permitted retention state, generating a maintain current target instruction and using the maintain current target instruction as the preheater outlet oxygen content target update instruction; S55, when the current target retention state is a disallowed retention state and the kiln tail carbon monoxide rise state is not a carbon monoxide forced update state, generating an update to candidate target instruction and using the update to candidate target instruction as the preheater outlet oxygen content target update instruction; S56, when the kiln tail carbon monoxide rise state is a carbon monoxide forced update state, generating a forced target update instruction and using the forced target update instruction as the preheater outlet oxygen content target update instruction.

[0019] A precise oxygen content control system for cement clinker kilns based on AI algorithms includes an AI algorithm processing unit, a pulverized coal silo layer processing unit, a coal oxygen consumption identification unit, a combustion response confirmation unit, a target strategy selection unit, and a target update control unit.

[0020] When connected to the pulverized coal silo layer processing unit, coal quality oxygen consumption identification unit, combustion response confirmation unit, target strategy selection unit, and target update control unit, it is used to receive data transmitted by each unit, call the AI ​​coal quality status identification rule library and AI target selection rule library, and output the pulverized coal silo layer replacement progress, silo exit impact window, coal quality switching oxygen consumption impact level, preheater outlet oxygen content offset trend, kiln tail carbon monoxide rise status, combustion response confirmation status, target adjustment strategy, candidate preheater outlet oxygen content target value, and current target retention status;

[0021] The coal powder silo layer processing unit is used to collect coal powder silo operation information and transmit the coal powder silo operation information to the AI ​​algorithm processing unit;

[0022] The coal oxygen consumption identification unit is used to collect coal quality information and transmit the coal quality information to the AI ​​algorithm processing unit;

[0023] The combustion response confirmation unit is used to collect information on oxygen content at the preheater outlet, carbon monoxide information at the kiln tail, and the current target value of oxygen content at the preheater outlet, and transmits the information on oxygen content at the preheater outlet, carbon monoxide information at the kiln tail, and the current target value of oxygen content at the preheater outlet to the AI ​​algorithm processing unit.

[0024] The target strategy selection unit is used to store a preset target adjustment strategy table and make the preset target adjustment strategy table available for the AI ​​algorithm processing unit to call.

[0025] The target update control unit is used to collect the target update records of oxygen content at the preheater outlet and generate a target update instruction for oxygen content at the preheater outlet based on the current target retention status output by the AI ​​algorithm processing unit.

[0026] This invention provides a method and system for precise control of oxygen content in cement clinker kilns based on AI algorithms, which has the following beneficial effects:

[0027] 1. By separating the time when coal quality information changes from the time when new pulverized coal actually enters the pulverized coal silo discharge area, the target value of the current preheater outlet oxygen content is avoided from being adjusted in advance before the new pulverized coal has participated in combustion. At the same time, within the silo discharge impact window, the changes in oxygen demand at the combustion end are confirmed by combining the oxygen consumption impact level of coal quality switching, the trend of preheater outlet oxygen content deviation, and the state of carbon monoxide rise at the kiln tail. Then, the current target retention status is used to determine whether to implement the candidate preheater outlet oxygen content target value. This reduces the problems of preheater outlet oxygen content fluctuation, kiln tail carbon monoxide rise, and clinker burning stability decline caused by early, delayed, or frequent oxygen adjustment during coal source switching.

[0028] 2. The combustion response confirmation status is used to confirm whether the new pulverized coal has formed an identifiable combustion feedback in the oxygen content information at the preheater outlet and the carbon monoxide information at the kiln tail after entering the outlet window, so as to avoid directly changing the current target value of oxygen content at the preheater outlet based solely on the oxygen consumption impact level of coal quality switching.

[0029] 3. By continuously processing data packets retrieved through target strategies, preset target adjustment strategy tables, preset target value range tables, and current target retention status, the preheater outlet oxygen content target update command can respond to changes in oxygen supply demand caused by coal quality switching, while avoiding frequent updates of the current preheater outlet oxygen content target value due to small candidate changes. Attached Figure Description

[0030] Figure 1 A schematic diagram of a method for precise control of oxygen content in cement clinker kilns based on AI algorithms;

[0031] Figure 2 This is a schematic diagram of a cement clinker kiln oxygen content precision control system based on AI algorithms.

[0032] Figure 3 A schematic diagram illustrating the logic for generating candidate target values ​​and updating targets. Detailed Implementation

[0033] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0034] Example 1

[0035] This embodiment can be applied to scenarios such as switching coal sources in cement clinker kilns, changing coal supply in pulverized coal silos, adjusting the coal blending ratio in coal mill systems, feeding high-moisture coal into silos during the rainy season, changing the blending ratio of alternative fuels, and changing the combustion air demand caused by adjustments in clinker production.

[0036] Taking the coal source switching scenario as an example, when the pulverized coal bin is gradually switched from old pulverized coal to new pulverized coal, although the new pulverized coal has entered the pulverized coal bin, it will not immediately enter the pulverized coal bin discharge area and participate in the combustion of the kiln head burner or decomposition furnace. If the control is made only based on the change of coal quality information or the current oxygen content information at the preheater outlet, it is easy to have problems such as premature adjustment, delayed adjustment or frequent adjustment of the current oxygen content target value at the preheater outlet. Therefore, this embodiment determines the discharge impact window after the new pulverized coal enters the pulverized coal bin discharge area by the pulverized coal bin layer replacement progress, and determines the candidate preheater outlet oxygen content target value by combining the oxygen consumption impact level of coal quality switching, the preheater outlet oxygen content offset trend and the kiln tail carbon monoxide rise state within the discharge impact window. Then, it determines whether to generate a preheater outlet oxygen content target update instruction by the current target retention status.

[0037] This invention provides a method for precise control of oxygen content in cement clinker kilns based on AI algorithms. Please refer to [link to relevant documentation]. Figure 1 This includes the following steps:

[0038] S1. Collect coal powder silo operation information, coal quality information, preheater outlet oxygen content information, kiln tail carbon monoxide information, current preheater outlet oxygen content target value and preheater outlet oxygen content target update record. Define the coal powder entering the coal powder silo after the coal quality information changes as new coal powder. Determine the coal powder silo layer replacement progress based on the coal powder silo operation information, and determine the discharge impact window after the new coal powder enters the coal powder silo discharge area based on the coal powder silo layer replacement progress.

[0039] S2. Within the outflow impact window, the oxygen consumption impact level of coal quality switching is determined based on coal quality information. When the outflow impact window is not determined based on the coal powder bin layer replacement progress, the oxygen consumption impact level of coal quality switching is prohibited from participating in the adjustment of the current preheater outlet oxygen content target value.

[0040] S3. Within the influence window of the outlet, determine the oxygen content deviation trend of the preheater outlet based on the oxygen content information of the preheater outlet and the current target value of the oxygen content of the preheater outlet, determine the carbon monoxide rising state of the kiln tail based on the carbon monoxide information of the kiln tail, and determine the combustion response confirmation state based on the oxygen content deviation trend of the preheater outlet and the carbon monoxide rising state of the kiln tail.

[0041] S4. Based on the oxygen consumption impact level of coal quality switching, the trend of oxygen content deviation at the preheater outlet, the rising state of carbon monoxide at the kiln tail, and the combustion response confirmation state, select a target adjustment strategy from the preset target adjustment strategy table through the AI ​​target selection rule library, and determine the target value of candidate oxygen content at the preheater outlet according to the target adjustment strategy.

[0042] S5. Determine the current target retention status based on the candidate preheater outlet oxygen content target value, the current preheater outlet oxygen content target value, the preheater outlet oxygen content target update record, and the kiln tail carbon monoxide rise status, and generate a preheater outlet oxygen content target update instruction based on the current target retention status.

[0043] In this embodiment, the above method can distinguish between changes in coal quality information and the actual time when new pulverized coal enters the pulverized coal silo discharge area during coal source switching, avoiding premature adjustment of the current preheater outlet oxygen content target value before the new pulverized coal participates in combustion. Simultaneously, within the silo discharge impact window, the oxygen consumption impact level of coal quality switching, the preheater outlet oxygen content deviation trend, and the kiln tail carbon monoxide rise status are combined to confirm whether there has been a change in oxygen supply demand at the combustion end. This ensures that the candidate preheater outlet oxygen content target value is not simply generated based on coal quality test results, but corresponds to both the preheater outlet oxygen content feedback and the kiln tail carbon monoxide feedback. Furthermore, by judging the candidate preheater outlet oxygen content target value before updating it based on the current target retention status, the current preheater outlet oxygen content target value can be maintained if the kiln tail carbon monoxide has not reached a risk state and the target change is insufficient to generate effective control benefits. This reduces the problems of preheater outlet oxygen content fluctuations, kiln tail carbon monoxide rises, and clinker burning stability decreases caused by premature, delayed, or frequent oxygen adjustments during coal source switching.

[0044] Example 2

[0045] Based on the application scenario of Example 1, this example illustrates the scenario of coal source switching in a cement clinker kiln. The coal powder silo originally stored old coal powder. After the coal quality information changes, new coal powder is sent into the coal powder silo by the coal mill system. The AI ​​algorithm processing unit receives the coal powder silo operation data, coal quality batch test data, and subsequent combustion feedback data, and performs time alignment processing on each data according to the same sampling time, so that the process of new coal powder entering the coal powder silo, staying in the coal powder silo, entering the coal powder silo discharge area, and participating in combustion can be continuously identified.

[0046] Specifically: Step S1 includes: S11, reading the coal powder silo inbound metering record, coal powder silo level detection record and coal powder silo outbound metering record from the coal powder silo inbound metering record, coal powder silo level detection record and coal powder silo outbound metering record respectively, and transmitting the coal powder silo inbound record, coal powder silo level record and coal powder silo outbound record to the AI ​​algorithm processing unit;

[0047] Specifically: The coal powder silo entry metering record is generated by the coal mill system's entry metering equipment. The coal powder silo entry record includes at least the entry sampling time, entry batch identifier, and entry coal powder quantity.

[0048] The coal powder silo level detection record is generated by the coal powder silo level gauge. The coal powder silo level record includes at least the sampling time and the coal powder silo inventory.

[0049] The coal powder silo discharge metering record is generated by the coal powder silo discharge metering equipment. The coal powder silo discharge record includes at least the discharge sampling time and the amount of coal powder discharged.

[0050] During data transmission, the coal powder silo entry record, coal powder silo level record, and coal powder silo exit record are matched according to the sampling time, so that the AI ​​algorithm processing unit can identify the amount of coal powder entering, stored, and leaving the coal powder silo in each sampling period.

[0051] S12. The AI ​​algorithm processing unit establishes a set of virtual coal powder layers based on the coal powder silo entry records, updates the remaining coal powder quantity of each virtual coal powder layer in the set of virtual coal powder layers based on the coal powder silo exit records, and obtains a set of remaining states of the virtual coal powder layers.

[0052] Specifically: The AI ​​algorithm processing unit establishes a set of virtual coal powder layers according to the batch identifier and sampling time in the coal powder silo entry record. Each entry batch corresponds to one virtual coal powder layer in the set of virtual coal powder layers.

[0053] Each virtual coal powder layer records the corresponding batch identifier, the initial amount of coal powder in the virtual coal powder layer, and the remaining amount of coal powder in the virtual coal powder layer.

[0054] It should be noted that the coal powder discharge area is the area at the bottom of the coal powder silo that is connected to the discharge metering equipment and can be deducted from the coal powder during the current sampling period.

[0055] The AI ​​algorithm processing unit prioritizes the discharge of virtual coal powder layers closest to the coal powder silo discharge area, and deducts the remaining amount of virtual coal powder layer based on the coal powder silo discharge records in each sampling period.

[0056] During data processing, the coal powder silo level record is used to verify the total amount of remaining coal powder in each virtual coal powder layer in the virtual coal powder layer set. The AI ​​algorithm processing unit first calculates the total amount of remaining coal powder in all virtual coal powder layers in the virtual coal powder layer set, and then allocates the coal powder silo storage quantity corresponding to the coal powder silo level record to each virtual coal powder layer according to the proportion of the remaining coal powder in each virtual coal powder layer in the total value, so as to complete the correction of the remaining coal powder in the virtual coal powder layer and obtain the set of remaining states of the virtual coal powder layer.

[0057] It should be noted that during the correction of the remaining coal powder quantity in the virtual coal powder layer, the proportional correction is only used to correct the remaining coal powder quantity in each virtual coal powder layer, and does not change the order in which the virtual coal powder layers in the set of virtual coal powder layers are formed according to the sampling time of entering the warehouse.

[0058] When the deviation between the coal powder silo level record corresponding to the coal powder silo storage quantity and the total remaining coal powder quantity of all virtual coal powder layers in the virtual coal powder layer set does not exceed the preset allowable deviation for level correction, the remaining coal powder quantity of each virtual coal powder layer is proportionally corrected according to the coal powder silo level record.

[0059] The preset material level calibration allowable deviation is preset based on the detection error of the pulverized coal silo level gauge, the measurement error of the pulverized coal silo outlet metering equipment, and the minimum identifiable amount of pulverized coal to be discharged within a single sampling period.

[0060] When the deviation between the coal powder silo level record corresponding to the coal powder silo storage quantity and the total remaining coal powder quantity of all virtual coal powder layers in the virtual coal powder layer set exceeds the preset allowable deviation for level correction, the sampling period is marked as an abnormal level correction period, and the virtual coal powder layer sequence and coal powder silo layer replacement progress of the previous sampling period are maintained.

[0061] When the coal powder silo discharge record shows that there is a discharge volume of coal powder, but the coal powder silo level record does not show a decrease in the inventory corresponding to the discharge volume of coal powder within the preset discharge verification time, the sampling period is marked as an abnormal discharge period, and the update of the start time and end time of the discharge impact window is suspended.

[0062] When the coal powder silo level record shows a decrease in the silo level, but there is no corresponding amount of coal powder discharged from the silo in the coal powder discharge record, the sampling period is marked as a metering abnormal period, and the virtual coal powder layer remaining state set of the previous sampling period is maintained.

[0063] When the abnormal period of material level correction, abnormal period of discharge, or abnormal period of metering within the continuous sampling period ends, the virtual coal powder layer remaining state set is recalibrated using the coal powder silo entry record, coal powder silo material level record, and coal powder silo discharge record after the recovery to normal, and the virtual coal powder layer remaining coal powder quantity is updated according to the principle of prioritizing the discharge of virtual coal powder layers closer to the coal powder silo discharge area.

[0064] The above processing is used to cover data inconsistencies caused by coal powder silo deviation, material arching, mixing, backflow, or abnormal level gauge errors, so that abnormal data will not directly change the layer order of the virtual coal powder layer where the new coal powder is located, nor will it prematurely trigger the silo exit impact window.

[0065] S13. The AI ​​algorithm processing unit determines the start and end times of the discharge influence window of the virtual coal powder layer where the new coal powder is located entering the coal powder silo discharge area based on the set of remaining states of the virtual coal powder layer. The sampling time period formed from the start to the end of the discharge influence window is defined as the discharge influence window. The layer status of the virtual coal powder layer where the new coal powder is located in the coal powder silo is defined as the coal powder silo layer replacement progress.

[0066] Specifically: The virtual coal powder layer where the new coal powder is located is the virtual coal powder layer corresponding to the first batch of coal powder entering the coal powder silo after the coal quality information changes;

[0067] The AI ​​algorithm processing unit determines the layer status of the new pulverized coal in the virtual pulverized coal layer from the set of remaining virtual pulverized coal layer states. The layer status includes not entering the pulverized coal silo discharge area, entering the pulverized coal silo discharge area, and completing the discharge from the pulverized coal silo discharge area.

[0068] When the virtual coal powder layer containing the new coal powder is first deducted from the remaining coal powder amount of the virtual coal powder layer, the sampling time is determined as the start time of the outflow impact window;

[0069] When the amount of remaining coal powder in the virtual coal powder layer where the new coal powder is located is not greater than the preset layer emptying threshold, the sampling time is determined as the end time of the outflow influence window.

[0070] It should be noted that the preset layer clearing threshold is set based on the upper limit of the metering error of the coal powder bin discharge metering equipment and the minimum identifiable amount of coal powder discharged within a single sampling period. The larger of the two is taken so that the virtual coal powder layer where the new coal powder is located can be stably identified when it has basically left the coal powder bin discharge area.

[0071] It should be noted that the preset layer emptying threshold is used to determine whether the virtual coal powder layer where the new coal powder is located has been discharged from the coal powder silo discharge area. The preset layer emptying threshold can be taken as one percent to five percent of the initial coal powder amount of the virtual coal powder layer where the new coal powder is located, or as the larger value between the upper limit of the metering error of the coal powder silo discharge metering equipment and the minimum identifiable coal powder amount in a single sampling period.

[0072] In this embodiment, when the metering error of the coal powder bin's discharge metering device is small and the sampling period is short, the preset layer clearing threshold is preferentially taken as two percent of the initial coal powder amount of the virtual coal powder layer where the new coal powder is located, in order to avoid the discharge window end time being uncertain for a long time due to a small amount of residual coal powder affecting the discharge due to metering error.

[0073] It should also be noted that there is a delay in the conveying of pulverized coal from the pulverized coal silo discharge area into the kiln head burner or decomposition furnace. This delay is defined as the pulverized coal combustion arrival delay time.

[0074] The coal powder combustion arrival delay time is predetermined based on the conveying path length of the coal powder conveying equipment, the coal powder conveying air volume, the operating frequency of the coal feeding equipment, and the coal powder exit time and combustion feedback occurrence time recorded in the stable operation test, and is fixedly stored within the same production line.

[0075] When determining the impact window of coal pulverized coal discharge, the start time of the impact window of coal pulverized coal discharge is delayed by the coal pulverized coal combustion arrival delay time to obtain the start time of the combustion impact window;

[0076] The combustion influence window end time is obtained by extending the coal powder combustion arrival delay time forward from the end time of the silo exit influence window.

[0077] The time period from the start time of the combustion influence window to the end time of the combustion influence window is used to subsequently determine the trend of oxygen content shift at the preheater outlet, the state of carbon monoxide rise at the kiln tail, and the state of combustion response confirmation.

[0078] Through the above processing, the transport lag between the discharge of new pulverized coal from the pulverized coal silo and its actual participation in combustion can be incorporated into the judgment process, avoiding premature confirmation of combustion response before the new pulverized coal has reached the kiln head burner or decomposition furnace.

[0079] In this embodiment, the outflow influence window is determined by the start and end times of the outflow influence window, so that the new coal powder corresponding to the coal quality change only participates in the subsequent oxygen content control judgment after it actually enters the coal powder silo discharge area.

[0080] Step S2 includes: S21, reading the lower heating value data, moisture data, volatile matter data, ash content data, lower heating value data, moisture data, volatile matter data, and ash content data of new pulverized coal from the coal batch test records;

[0081] Specifically: The coal quality batch test record is generated by the coal quality test system according to the batch identifier of the coal entering the warehouse. The AI ​​algorithm processing unit reads the low calorific value data, moisture data, volatile matter data and ash content data of the new coal powder according to the batch identifier of the virtual coal powder layer where the new coal powder is located.

[0082] Old coal powder refers to the coal powder that has entered the coal powder silo discharge area but has not yet been discharged from the silo before the new coal powder enters the virtual coal powder silo discharge area.

[0083] When there are multiple virtual coal powder layers corresponding to old coal powder in the coal powder silo discharge area, the virtual coal powder layer that is adjacent to the virtual coal powder layer where the new coal powder is located and enters the coal powder silo discharge area first is selected as the virtual coal powder layer corresponding to the old coal powder.

[0084] In the coal source switching scenario set in Example 1, the new coal powder low heating value data, new coal powder moisture data, new coal powder volatile matter data, and new coal powder ash content data are used to represent the coal quality status corresponding to the new coal source, while the old coal powder low heating value data, old coal powder moisture data, old coal powder volatile matter data, and old coal powder ash content data are used to represent the coal quality status corresponding to the coal source before the switching.

[0085] S22. The AI ​​algorithm processing unit performs grading processing on the data read in step S21 according to the preset low calorific value grading range, preset moisture grading range, preset volatile matter grading range and preset ash content grading range, and obtains the low calorific value data grading comparison results, moisture data grading comparison results, volatile matter data grading comparison results and ash content data grading comparison results.

[0086] Specifically: The preset ranges for lower calorific value, moisture content, volatile matter, and ash content are established according to the coal quality acceptance standards of the production line and the batch test records of coal quality during stable production. Each range is stored using the corresponding unit of measurement for the coal quality index. The lower calorific value data, moisture content data, volatile matter data, and ash content data are not added together.

[0087] It should be noted that the preset low calorific value range, preset moisture range, preset volatile matter range, and preset ash content range are all pre-divided into low, medium, and high grades according to the coal quality acceptance standards and historical stable production data of the same production line.

[0088] The preset low calorific value grading range is based on the lower limit of low calorific value acceptance, the target range of stable low calorific value operation, and the upper limit of low calorific value acceptance in the coal quality acceptance standard of the production line as the grading boundaries;

[0089] The preset moisture grading range is based on the upper limit of moisture acceptance in the coal quality acceptance standard of the production line, the commonly used moisture range during stable production, and the control range of high moisture coal in the rainy season as the grading boundary.

[0090] The preset volatile matter grading range is based on the lower limit of volatile matter acceptance, the target range of stable volatile matter operation, and the upper limit of volatile matter acceptance in the coal quality acceptance standard of the production line as the grading boundaries;

[0091] The preset ash content grading range is based on the upper limit of ash content acceptance in the coal quality acceptance standard of the production line, the commonly used range of ash content during stable production, and the control range of high ash coal as the grading boundaries.

[0092] During the grading process, the coal quality data corresponding to the new pulverized coal and the coal quality data corresponding to the old pulverized coal are respectively determined in the grading, and the grading of the new pulverized coal is compared with that of the old pulverized coal to obtain the grading comparison results of the low calorific value data, the moisture data, the volatile matter data, and the ash content data.

[0093] In one feasible example, when the lower heating value of the new pulverized coal is one level lower than that of the old pulverized coal, the moisture content of the new pulverized coal is one level higher than that of the old pulverized coal, the volatile matter content of the new pulverized coal is in the same level as that of the old pulverized coal, and the ash content of the new pulverized coal is one level higher than that of the old pulverized coal, the oxygen consumption impact level of the coal quality switching corresponding to this level combination is configured as a medium oxygen consumption impact level.

[0094] In another feasible example, when the lower heating value of the new pulverized coal is lower than that of the old pulverized coal by at least one grade, and the moisture content of the new pulverized coal is higher than that of the old pulverized coal by at least one grade, and the ash content of the new pulverized coal is higher than that of the old pulverized coal by at least one grade, the oxygen consumption impact level of the coal quality switching corresponding to the grade combination is configured as a high oxygen consumption impact level.

[0095] The above-mentioned grading boundaries are fixed and stored in advance within the same production line. The oxygen consumption impact level of coal quality switching is obtained by grading combination matching, without the need to perform cross-dimensional summation processing on low calorific value data, moisture data, volatile matter data and ash content data.

[0096] Once the lower heating value data of new pulverized coal and the lower heating value data of old pulverized coal fall into the lower heating value grading, a comparison result of the lower heating value data grading is formed.

[0097] Once the moisture data of new pulverized coal and the moisture data of old pulverized coal fall into the moisture grading, a comparison result of the moisture data grading is formed.

[0098] Once the volatile matter data of the new pulverized coal and the volatile matter data of the old pulverized coal fall into the volatile matter category, a comparison result of the volatile matter data category is formed;

[0099] Once the ash content data of the new pulverized coal and the ash content data of the old pulverized coal are determined to fall into the ash content category, a comparison result of the ash content data category is formed.

[0100] It should be noted that the comparison results of low calorific value data, moisture data, volatile matter data, and ash content data are all matching results, and it is not necessary to convert coal quality data of different dimensions into the same value.

[0101] S23. The AI ​​algorithm processing unit combines and judges the results of the comparison of low calorific value data, moisture data, volatile matter data and ash content data through the AI ​​coal quality status recognition rule library to obtain the oxygen consumption impact level of coal quality switching.

[0102] Specifically: The AI ​​coal quality status recognition rule base is pre-configured from the coal quality acceptance standards of the production line, the batch test records of coal quality during historical stable operation, and the experience of manually confirmed coal quality switching.

[0103] The AI ​​coal quality status recognition rule base includes multiple coal quality status recognition rule records. Each coal quality status recognition rule record includes a low calorific value data classification comparison result, a moisture data classification comparison result, a volatile matter data classification comparison result, an ash content data classification comparison result, and a coal quality switching oxygen consumption impact level.

[0104] Each coal quality status recognition rule record in the AI ​​coal quality status recognition rule base corresponds to a classification combination consisting of the classification comparison results of low calorific value data, moisture data, volatile matter data, and ash content data.

[0105] When the current grade combination is completely consistent with the coal quality status identification rule record, the AI ​​algorithm processing unit reads the corresponding coal quality switching oxygen consumption impact level.

[0106] When the current grade combination does not match the coal quality status identification rule record, the AI ​​algorithm processing unit determines the oxygen consumption impact level of coal quality switching as the preset conservative oxygen consumption impact level.

[0107] In this embodiment, the oxygen consumption impact level of coal quality switching is used to represent the classification result of the combustion oxygen supply demand after the new coal powder enters the coal powder silo discharge area. The oxygen consumption impact level of coal quality switching does not directly change the current target value of oxygen content at the preheater outlet, but is used as one of the state conditions for the selection of subsequent target adjustment strategies.

[0108] Example 3

[0109] Based on the application scenarios of Examples 1 and 2, this example is used to illustrate how, after new pulverized coal enters the outlet impact window, it is possible to confirm whether the combustion system has shown a response related to coal quality switching based on the feedback of oxygen content at the preheater outlet and carbon monoxide feedback at the kiln tail, thereby avoiding directly adjusting the current target value of oxygen content at the preheater outlet based solely on coal quality information.

[0110] Specifically: Step S3 includes: S31, reading the oxygen content sampling sequence at the preheater outlet from the online oxygen content detection record, and reading the current target value of oxygen content at the preheater outlet, the preset allowable deviation range of oxygen content, and the preset number of oxygen content determination samplings;

[0111] Specifically: the online oxygen content detection record is generated by the online oxygen content detection device installed at the preheater outlet flue gas detection location, and the preheater outlet oxygen content sampling sequence consists of multiple preheater outlet oxygen content sampling values ​​arranged according to the sampling time;

[0112] During the data reading process, the oxygen content sampling values ​​at the preheater outlet generated when the equipment is in calibration state, maintenance state, or communication abnormal state are removed, and the oxygen content sampling values ​​at the preheater outlet with valid sampling time are retained.

[0113] The current target value of oxygen content at the preheater outlet is obtained by reading the oxygen content control target currently being executed by the central control system;

[0114] The preset allowable deviation range of oxygen content is set in advance based on the detection error of the online oxygen content detection equipment, the normal fluctuation range of oxygen content at the preheater outlet during stable production, and the production line control requirements.

[0115] The preset number of oxygen content determination sampling times is pre-set based on the sampling cycle of the online oxygen content detection equipment and the response time of the cement clinker kiln oxygen content control, so that the oxygen content determination sequence at the preheater outlet can cover one oxygen content feedback determination process. S32, the AI ​​algorithm processing unit determines the target range of oxygen content at the preheater outlet based on the current target value of oxygen content at the preheater outlet and the preset allowable deviation range of oxygen content, and forms the preheater outlet oxygen content determination sequence from the preheater outlet oxygen content sampling sequence according to the preset number of oxygen content determination sampling times.

[0116] Specifically: Taking the current target value of oxygen content at the preheater outlet as the center, the target range of oxygen content at the preheater outlet is determined according to the preset allowable deviation range of oxygen content. The target range of oxygen content at the preheater outlet includes the lower limit and the upper limit of the target range of oxygen content at the preheater outlet.

[0117] When forming the oxygen content determination sequence at the preheater outlet, select the oxygen content sampling values ​​at the preheater outlet that are continuously arranged according to the sampling time and whose number reaches the preset number of oxygen content determination sampling times from the current sampling time backward.

[0118] If the number of valid preheater outlet oxygen content samples before the current sampling time has not reached the preset number of oxygen content determination samples, the preheater outlet oxygen content offset trend will not be updated temporarily, and the preheater outlet oxygen content samples will be replenished at the next sampling time. S33, the AI ​​algorithm processing unit determines the preheater outlet oxygen content offset trend as being within the target range, below the target offset, or above the target offset state based on the comparison result between the preheater outlet oxygen content determination sequence and the target range of preheater outlet oxygen content.

[0119] Specifically: when every sampled value of oxygen content at the preheater outlet in the preheater outlet oxygen content determination sequence is within the target range of oxygen content at the preheater outlet, the preheater outlet oxygen content offset trend is determined to be within the target range.

[0120] When each sampled value of preheater outlet oxygen content in the preheater outlet oxygen content determination sequence is less than the lower limit of the target range of preheater outlet oxygen content, the preheater outlet oxygen content offset trend is determined to be below the target offset state.

[0121] When every sampled value of preheater outlet oxygen content in the preheater outlet oxygen content determination sequence is greater than the upper limit of the target range of preheater outlet oxygen content, the preheater outlet oxygen content offset trend is determined to be above the target offset state.

[0122] When the sampled values ​​in the preheater outlet oxygen content determination sequence simultaneously include those within the target range of preheater outlet oxygen content, those below the lower limit of the target range of preheater outlet oxygen content, or those above the upper limit of the target range of preheater outlet oxygen content, the preheater outlet oxygen content offset trend already determined in the previous sampling period will not be changed.

[0123] If there is no preheater outlet oxygen content offset trend that has been determined in the previous sampling period, then the preheater outlet oxygen content offset trend is determined as the state within the target range.

[0124] Step S3 further includes: S34, reading the carbon monoxide sampling sequence at the kiln tail from the online carbon monoxide detection record, and reading the preset carbon monoxide upper limit value, the preset carbon monoxide forced update threshold and the preset number of carbon monoxide determination samplings;

[0125] Specifically: The carbon monoxide online detection record is generated by the carbon monoxide online detection device set at the kiln tail flue gas detection location, and the kiln tail carbon monoxide sampling sequence consists of multiple kiln tail carbon monoxide sampling values ​​arranged according to the sampling time;

[0126] During the data reading process, the carbon monoxide sampling values ​​at the kiln tail generated when the equipment is in calibration state, maintenance state, or communication abnormal state are removed, and the carbon monoxide sampling values ​​at the kiln tail with valid sampling time are retained.

[0127] The preset upper limit for carbon monoxide is set in advance based on the normal combustion control requirements of the production line and the allowable range of carbon monoxide at the kiln tail during stable production.

[0128] The preset carbon monoxide forced replacement threshold is set in advance according to the safety control requirements of the production line, and the preset carbon monoxide forced replacement threshold is greater than the preset carbon monoxide upper limit value;

[0129] The preset number of carbon monoxide determination sampling times is pre-set based on the sampling cycle of the online carbon monoxide detection equipment and the response time of carbon monoxide feedback at the kiln tail, so that the carbon monoxide determination sequence at the kiln tail can cover one carbon monoxide status determination process.

[0130] S35, the AI ​​algorithm processing unit forms a kiln tail carbon monoxide determination sequence from the kiln tail carbon monoxide sampling sequence according to the preset number of carbon monoxide determination samplings, and determines the kiln tail carbon monoxide rising state as a normal carbon monoxide state, a carbon monoxide exceeding the upper limit state, or a carbon monoxide forced update state according to the kiln tail carbon monoxide determination sequence, the preset carbon monoxide upper limit value, and the preset carbon monoxide forced update threshold.

[0131] Specifically: When forming the carbon monoxide determination sequence at the kiln tail, select the carbon monoxide sampling values ​​at the kiln tail that are consecutively arranged according to the sampling time and whose number reaches the preset number of carbon monoxide determination sampling times from the current sampling time backward.

[0132] If the number of valid carbon monoxide samples at the kiln tail before the current sampling time has not reached the preset number of carbon monoxide determination samples, the carbon monoxide rise status at the kiln tail will not be updated temporarily, and the kiln tail carbon monoxide sample value will be supplemented at the next sampling time.

[0133] When the carbon monoxide sampling value at the kiln tail in the kiln tail carbon monoxide determination sequence is not greater than the preset upper limit value of carbon monoxide, the rising state of carbon monoxide at the kiln tail is determined as the normal state of carbon monoxide.

[0134] When every sampled value of carbon monoxide at the kiln tail in the carbon monoxide determination sequence is greater than the preset upper limit of carbon monoxide and less than the preset forced update threshold of carbon monoxide, the rising state of carbon monoxide at the kiln tail is determined as the state of carbon monoxide exceeding the upper limit.

[0135] When the carbon monoxide sample value at the kiln tail in the kiln tail determination sequence is not less than the preset carbon monoxide forced update threshold, the rising state of carbon monoxide at the kiln tail is determined as the carbon monoxide forced update state.

[0136] When the sampled value in the carbon monoxide determination sequence at the kiln tail does not meet the determination conditions of normal carbon monoxide state, carbon monoxide exceeding the upper limit state, or carbon monoxide forced update state, the rising state of carbon monoxide at the kiln tail that was determined in the previous sampling period will not be changed.

[0137] If there is no kiln tail carbon monoxide rising state that has been determined in the previous sampling period, then the kiln tail carbon monoxide rising state is determined as the normal carbon monoxide state.

[0138] S36. The AI ​​algorithm processing unit determines the combustion response confirmation status as either a confirmed combustion response status or an unconfirmed combustion response status based on the trend of oxygen content shift at the preheater outlet and the rising state of carbon monoxide at the kiln tail.

[0139] Specifically: When the oxygen content at the preheater outlet deviates below the target deviance state, it indicates that the oxygen content at the preheater outlet within the outlet influence window is already below the allowable range corresponding to the current target value of oxygen content at the preheater outlet, and the combustion response confirmation state is determined to be the combustion response confirmed state.

[0140] When the carbon monoxide rises at the kiln tail to the state of carbon monoxide exceeding the upper limit or the state of carbon monoxide forced renewal, it indicates that incomplete combustion feedback has occurred within the outflow impact window, and the combustion response confirmation state is determined to be the combustion response confirmed state.

[0141] When the oxygen content at the preheater outlet is within the target range or higher than the target deviation, and the carbon monoxide rise at the kiln tail is in a normal carbon monoxide state, the combustion response confirmation state is determined to be the combustion response unconfirmed state.

[0142] In this embodiment, the combustion response confirmation status is used to confirm whether the new pulverized coal has formed an identifiable combustion feedback in the preheater outlet oxygen content information and kiln tail carbon monoxide information after entering the outlet impact window, so as to avoid directly changing the current preheater outlet oxygen content target value based solely on the oxygen consumption impact level of coal quality switching.

[0143] Example 4

[0144] Based on the application scenarios of Examples 1 to 3, this example is used to explain how to select a target adjustment strategy, generate candidate target values ​​for oxygen content at the preheater outlet, and determine whether to retain the current target value for oxygen content at the preheater outlet before generating the target update instruction for oxygen content at the preheater outlet.

[0145] Please see Figure 3 Specifically, step S4 includes: S41, assembling a target strategy retrieval data package by combining the coal quality switching oxygen consumption impact level, the preheater outlet oxygen content shift trend, the kiln tail carbon monoxide rise status, and the combustion response confirmation status.

[0146] Specifically: The target strategy retrieval data package includes four fields, namely the coal quality switching oxygen consumption impact level field, the preheater outlet oxygen content offset trend field, the kiln tail carbon monoxide rise status field, and the combustion response confirmation status field.

[0147] Before generating the target strategy retrieval data package, the effectiveness of the oxygen consumption impact level of coal quality switching, the oxygen content shift trend at the preheater outlet, the carbon monoxide rise status at the kiln tail, and the combustion response confirmation status are checked respectively.

[0148] When any field is empty, no new target policy retrieval data packet is generated, and the target policy retrieval data packet generated in the previous control cycle is retained.

[0149] When all four fields have a definite status value, the four fields are written into the same target strategy retrieval data packet in the preset field order.

[0150] S42. The AI ​​target selection rule base calls the preset target adjustment strategy table, which includes multiple target adjustment strategy records. Each target adjustment strategy record includes a state combination and a target adjustment strategy.

[0151] Specifically: The preset target adjustment strategy table is pre-configured from the operation records, coal quality switching operation records, preheater outlet oxygen content target update records, kiln tail carbon monoxide sampling records, and clinker burning stability records during the stable production period of the same production line;

[0152] It should be noted that: pre-configuration refers to creating a data record table based on historical data records of the same production line before the system runs, and calling the data record table according to field matching during operation, rather than making a manual judgment on each adjustment.

[0153] Manual confirmation is only used to annotate the target adjustment results in historical data records. The annotation content includes maintaining the target strategy, improving the target strategy, reducing the target strategy, and forcibly improving the target strategy. After the annotation is completed, a target adjustment strategy record is formed.

[0154] Each state combination includes a coal quality switching oxygen consumption impact level, a preheater outlet oxygen content shift trend, a kiln tail carbon monoxide rise state, and a combustion response confirmation state.

[0155] Each target adjustment strategy record saves a state combination corresponding to a target adjustment strategy;

[0156] The AI ​​target selection rule base retrieves the preset target adjustment strategy table in a way that the fields are completely consistent. When the coal quality switching oxygen consumption impact level, preheater outlet oxygen content offset trend, kiln tail carbon monoxide rise status and combustion response confirmation status in the target strategy retrieval data packet are all consistent with the corresponding fields in a target adjustment strategy record, the target adjustment strategy in that target adjustment strategy record is read.

[0157] In an implementable example, when the oxygen consumption impact level of coal quality switching is medium, the oxygen content deviation trend at the preheater outlet is below the target deviation state, the carbon monoxide rise state at the kiln tail is carbon monoxide exceeding the upper limit state, and the combustion response confirmation state is combustion response confirmed state, the target adjustment strategy is determined to be the target improvement strategy.

[0158] It should be noted that the AI ​​coal quality status recognition rule base includes multiple coal quality status recognition rule records. Each coal quality status recognition rule record includes the comparison results of low calorific value data, moisture data, volatile matter data, ash content data, and the oxygen consumption impact level of coal quality switching.

[0159] In an implementable example, when the comparison results of the low calorific value data show that the new coal powder is one grade lower than the old coal powder, the comparison results of the moisture data show that the new coal powder is one grade higher than the old coal powder, the comparison results of the volatile matter data show that the new coal powder and the old coal powder are in the same grade, and the comparison results of the ash content data show that the new coal powder is one grade higher than the old coal powder, the oxygen consumption impact level of the coal quality switching is determined to be a medium oxygen consumption impact level.

[0160] The preset target value level table includes multiple target value level records. Each target value level record includes a target adjustment strategy and an alternative target value for the next control cycle.

[0161] In an implementable example, when the target adjustment strategy is to improve the target strategy, the candidate target value for the next control cycle corresponding to the improved target strategy is read from the preset target value level table, and the candidate target value for the next control cycle is determined as the candidate target value for oxygen content at the preheater outlet.

[0162] The above-mentioned grading range, oxygen consumption impact level of coal quality switching, target adjustment strategy and alternative target value for the next control cycle are pre-configured by the coal quality acceptance standard of the production line, the operation record during stable production, the oxygen content control parameter table and the safety control parameter table, and are kept consistent within the same production line;

[0163] It should be noted that the preset target adjustment strategy table does not add the values ​​of coal quality data, oxygen content data, and carbon monoxide data, but instead performs field matching processing on the already determined state results.

[0164] S43. The AI ​​target selection rule base retrieves data packets based on target strategies, searches for matching target adjustment strategy records in multiple target adjustment strategy records, and reads the target adjustment strategy from the matching target adjustment strategy records.

[0165] Specifically: retrieval matching means that the coal quality switching oxygen consumption impact level, preheater outlet oxygen content shift trend, kiln tail carbon monoxide rise status, and combustion response confirmation status in the target strategy retrieval data package are completely consistent with the corresponding fields in the target adjustment strategy record.

[0166] When a completely identical target adjustment strategy record exists, read the target adjustment strategy from that target adjustment strategy record;

[0167] When there is no completely identical target adjustment strategy record, read the target adjustment strategy from the preset default target adjustment strategy record;

[0168] The preset default target adjustment strategy record is pre-configured according to the production line safety control requirements, and the preset default target adjustment strategy record prioritizes the target adjustment strategy that does not reduce the current target value of oxygen content at the preheater outlet.

[0169] Step S4 further includes: S44, calling the preset target value level table, wherein the preset target value level table includes multiple target value level records, and each target value level record includes a target adjustment strategy and a candidate target value for the next control cycle;

[0170] Specifically: the preset target value range table is obtained by pre-configuring the oxygen content control parameter table of the production line;

[0171] It should be noted that: the preset target value range table is pre-configured according to the oxygen content control parameter table of the production line, and the alternative target value for the next control cycle is stored in the same oxygen content unit as the current preheater outlet oxygen content target value;

[0172] The preset target value level table includes at least the target value level corresponding to the target strategy, the target value level corresponding to the target strategy, the target value level corresponding to the target strategy, the target value level corresponding to the target strategy, and the target value level corresponding to the target strategy that is forcibly increased.

[0173] In one feasible example, when the current target oxygen content at the preheater outlet is 3.0%, the alternative target value for the next control cycle corresponding to the target strategy is maintained at 3.0%, the alternative target value for the next control cycle corresponding to the target strategy is increased to 3.2%, the alternative target value for the next control cycle corresponding to the target strategy is decreased to 2.8%, and the alternative target value for the next control cycle corresponding to the target strategy is forcibly increased to 3.4%.

[0174] In another configuration, the target adjustment amount corresponding to the target strategy is kept at 0, the target adjustment amount corresponding to the target strategy is increased by 0.1 to 0.3 percentage points, the target adjustment amount corresponding to the target strategy is decreased by 0.1 to 0.3 percentage points, the target adjustment amount corresponding to the target strategy is forcibly increased by 0.3 to 0.5 percentage points, and the target value formed by the current target value of oxygen content at the preheater outlet and the corresponding target adjustment amount is determined as the alternative target value for the next control cycle.

[0175] The target value range and target adjustment amount mentioned above are predetermined based on the oxygen content control parameter table of the production line, the detection error of the online oxygen content detection equipment, the stable operation record of clinker burning, and the safety control parameter table, and are fixedly stored within the same production line.

[0176] Once the target adjustment strategy is determined, the candidate target value for the next control cycle corresponding to the target adjustment strategy is read from the preset target value level table, and the candidate target value for the next control cycle is determined as the candidate target value for oxygen content at the preheater outlet.

[0177] Each target value level record saves a target adjustment strategy and a candidate target value for the next control cycle.

[0178] The alternative target value for the next control cycle is stored using the same unit of measurement as the current target value for oxygen content at the preheater outlet.

[0179] In this embodiment, the candidate target value for the next control cycle corresponding to the target strategy is kept at the current target value of oxygen content at the preheater outlet. The candidate target value for the next control cycle corresponding to the target strategy is increased to the target value of the increased gear allowed by the production line. The candidate target value for the next control cycle corresponding to the target strategy is decreased to the target value of the decreased gear allowed by the production line. The candidate target value for the next control cycle corresponding to the target strategy is forcibly increased to the target value of the increased gear under the safety control requirements.

[0180] S45. Based on the target adjustment strategy read in step S43, retrieve the matching target value level record from multiple target value level records;

[0181] Specifically: the retrieval of target value level records uses the target adjustment strategy as the retrieval field;

[0182] When the target adjustment strategy read in step S43 is consistent with the target adjustment strategy in the target value level record, the target value level record is determined as the matching target value level record.

[0183] When multiple target value records are consistent with the same target adjustment strategy, the target value record corresponding to the current production line operation mode is selected.

[0184] The current production line operation mode is determined by the clinker output level currently being executed by the central control system, the kiln condition stability indicator, or the operation mode confirmed manually.

[0185] S46. Read the candidate target value for the next control cycle from the matched target value range record, and determine the candidate target value for the next control cycle as the candidate target value for oxygen content at the preheater outlet.

[0186] Specifically: the target value of the oxygen content at the outlet of the candidate preheater is the target value of the oxygen content at the outlet of the preheater that can be adopted in the next control cycle;

[0187] The target value of oxygen content at the outlet of the preheater is only saved as an alternative target value and does not directly overwrite the current target value of oxygen content at the outlet of the preheater.

[0188] After generating the candidate preheater outlet oxygen content target value, record the generation time of the candidate preheater outlet oxygen content target value, the corresponding target adjustment strategy, and the corresponding target strategy retrieval data packet, so as to determine whether the current preheater outlet oxygen content target value needs to be updated.

[0189] Step S5 includes: S51, the AI ​​algorithm processing unit determines the target update difference status based on the target value difference between the candidate preheater outlet oxygen content target value and the current preheater outlet oxygen content target value;

[0190] The target value gap is the absolute value of the difference between the candidate preheater outlet oxygen content target value and the current preheater outlet oxygen content target value, and the target value gap is stored in the unit of measurement of the preheater outlet oxygen content target value itself.

[0191] The preset target update difference threshold is pre-set based on the detection error of the online oxygen content detection equipment, the minimum executable target adjustment amount of the control system, and the target value adjustment records during stable production.

[0192] When the target value difference is less than the preset target update difference threshold, the target update difference status is determined as the target update difference threshold not being reached.

[0193] When the difference between the target values ​​is not less than the preset target update difference threshold, the target update difference state is determined to be the target update difference threshold state.

[0194] S52, the AI ​​algorithm processing unit determines the target update count status based on the number of updates recorded in the target update statistics window for the oxygen content target update record at the preheater outlet;

[0195] Specifically: The preheater outlet oxygen content target update record includes the execution time of each update of the current preheater outlet oxygen content target value and the updated current preheater outlet oxygen content target value;

[0196] The target update statistics window is preset based on the response time of the oxygen content control in the cement clinker kiln and the allowable frequency of the actuator.

[0197] The target update statistics window displays the number of times an update command to a candidate target or a forced target update command has been executed.

[0198] When the number of updates in the target update statistics window is less than the preset target update count limit, the target update count status is determined to be "not reached target update count limit".

[0199] When the number of updates in the target update statistics window is not less than the preset target update count limit, the target update count status is determined to be the target update count limit status.

[0200] S53. The AI ​​algorithm processing unit determines the current target retention status as either an allowable retention status or an unallowable retention status based on the target update difference status, the target update count status, and the kiln tail carbon monoxide rise status.

[0201] Specifically: when the carbon monoxide rises at the kiln tail and the carbon monoxide is in a forced renewal state, the current target retention state is determined to be a state where retention is not allowed.

[0202] When the carbon monoxide level at the kiln tail rises to the state where the carbon monoxide exceeds the upper limit and the target update difference state reaches the target update difference threshold, the current target retention state is determined to be a state where retention is not allowed.

[0203] When the carbon monoxide rise at the kiln tail is in a normal carbon monoxide state and the target update difference state is in a state where the target update difference threshold has not been reached, the current target retention state is determined as an allowable retention state.

[0204] When the carbon monoxide rise status at the kiln tail is normal, the target update difference status is at the target update difference threshold, and the target update count status is at the target update count limit threshold, the current target retention status is determined as the allowable retention status.

[0205] When the carbon monoxide rise status at the kiln tail is normal, the target update difference status is the target update difference threshold status, and the target update count status is the target update count limit threshold status, the current target retention status is determined to be a retention-not-allowed status.

[0206] It should be noted that the current target retention status is used to suppress unnecessary updates to the current preheater outlet oxygen content target value when no safety risk is triggered, and to abandon the retention of the current preheater outlet oxygen content target value when the carbon monoxide rise at the kiln tail reaches the safety control condition.

[0207] Step S5 further includes: S54, when the current target retention status is the allowed retention status, generating a retain current target instruction, and using the retain current target instruction as the preheater outlet oxygen content target update instruction;

[0208] Specifically: The "Maintain Current Target" command is used to ensure that the central control system continues to execute the current target value for oxygen content at the preheater outlet;

[0209] After generating the command to maintain the current target, the target value of the oxygen content at the preheater outlet is not changed, and the generation time of the command to maintain the current target is recorded.

[0210] S55. When the current target retention status is not allowed to be retained and the kiln tail carbon monoxide rise status is not the carbon monoxide forced update status, generate an update to candidate target instruction and use the update to candidate target instruction as the preheater outlet oxygen content target update instruction.

[0211] Specifically: The update to candidate target instruction is used to update the current preheater outlet oxygen content target value to a candidate preheater outlet oxygen content target value;

[0212] After generating the update command to the candidate target, the candidate preheater outlet oxygen content target value is written into the oxygen content target setting position of the central control system, and the execution time and the updated current preheater outlet oxygen content target value are written into the preheater outlet oxygen content target update record.

[0213] S56. When the carbon monoxide rises at the kiln tail and the carbon monoxide is in a forced carbon monoxide renewal state, a forced target renewal command is generated and the forced target renewal command is used as the target renewal command for the oxygen content at the preheater outlet.

[0214] Specifically: The forced target update instruction is used to prioritize the execution of the target value of oxygen content at the outlet of the candidate preheater when the carbon monoxide information at the kiln tail reaches the forced update condition;

[0215] After generating the forced target update instruction, the candidate preheater outlet oxygen content target value is written into the oxygen content target setting position of the central control system, and the execution time, the updated current preheater outlet oxygen content target value, and the carbon monoxide forced update status are written into the preheater outlet oxygen content target update record.

[0216] In this embodiment, by continuously processing the target strategy retrieval data packet, the preset target adjustment strategy table, the preset target value level table, and the current target retention status, the preheater outlet oxygen content target update command can not only respond to changes in oxygen supply demand caused by coal quality switching, but also avoid frequent updates of the current preheater outlet oxygen content target value due to small candidate changes.

[0217] Example 5

[0218] A precise oxygen content control system for cement clinker kilns based on AI algorithms, please refer to... Figure 2 Specifically, it includes an AI algorithm processing unit, a pulverized coal silo layer processing unit, a coal quality oxygen consumption identification unit, a combustion response confirmation unit, a target strategy selection unit, and a target update control unit;

[0219] Based on the application scenarios of Examples 1 to 4, this example illustrates the implementable structure and data processing method of the AI ​​algorithm processing unit. The AI ​​algorithm processing unit can be deployed in the industrial computer, edge computing controller or central control server of the cement clinker production line, and receive coal powder silo operation information, coal quality information, preheater outlet oxygen content information, kiln tail carbon monoxide information, current preheater outlet oxygen content target value and preheater outlet oxygen content target update record through the existing data interface of the production line.

[0220] It should be noted that the AI ​​algorithm in this embodiment is a knowledge rule reasoning algorithm based on a rule base, and is not limited to a neural network model, a training model, or a parameter weight model.

[0221] The AI ​​algorithm performs time alignment, validity checks, tiered processing, status identification, rule matching, and target value tier mapping on coal powder silo operation information, coal quality information, preheater outlet oxygen content information, kiln tail carbon monoxide information, current preheater outlet oxygen content target value, and preheater outlet oxygen content target update records. This results in coal powder silo layer replacement progress, silo impact window, coal quality switching oxygen consumption impact level, preheater outlet oxygen content offset trend, kiln tail carbon monoxide rise status, combustion response confirmation status, target adjustment strategy, candidate preheater outlet oxygen content target values, and current target retention status.

[0222] Among them, the AI ​​coal quality status recognition rule base is used to infer the oxygen consumption impact level of coal quality switching based on coal quality grade combination, and the AI ​​target selection rule base is used to infer the target adjustment strategy based on the oxygen consumption impact level of coal quality switching, the oxygen content deviation trend at the preheater outlet, the carbon monoxide rise status at the kiln tail, and the combustion response confirmation status.

[0223] Therefore, the AI ​​algorithm processing unit in this embodiment does not rely on training samples, model structure, loss function or model update mechanism, but completes state reasoning and target selection through a pre-configured rule base and data table that can be verified by the production line, so as to ensure that the same combination of states can obtain a definite processing result.

[0224] When connected to the pulverized coal silo layer processing unit, coal quality oxygen consumption identification unit, combustion response confirmation unit, target strategy selection unit, and target update control unit, it is used to receive data transmitted by each unit, call the AI ​​coal quality status identification rule library and AI target selection rule library, and output the pulverized coal silo layer replacement progress, silo exit impact window, coal quality switching oxygen consumption impact level, preheater outlet oxygen content offset trend, kiln tail carbon monoxide rise status, combustion response confirmation status, target adjustment strategy, candidate preheater outlet oxygen content target value, and current target retention status;

[0225] Specifically: The AI ​​algorithm processing unit includes data receiving function, data time alignment function, data validity check function, rule base calling function, status result generation function, and target update judgment function;

[0226] During the data reception process, the AI ​​algorithm processing unit receives coal powder silo operation information transmitted by the coal powder silo layer processing unit, coal quality information transmitted by the coal quality oxygen consumption identification unit, preheater outlet oxygen content information, kiln tail carbon monoxide information and current preheater outlet oxygen content target value transmitted by the combustion response confirmation unit, as well as preheater outlet oxygen content target update record transmitted by the target update control unit.

[0227] During the data time alignment process, the AI ​​algorithm processing unit matches the coal powder silo operation information, preheater outlet oxygen content information, kiln tail carbon monoxide information and preheater outlet oxygen content target update record according to the sampling time, and matches the coal quality information with the coal powder silo operation information according to the batch identification of the coal powder entering the silo, so that the coal powder entering the silo process, exiting the silo process, coal quality data and combustion feedback data of the same batch of coal powder can be processed continuously.

[0228] During the data validity check, the AI ​​algorithm processing unit removes data that lacks sampling time, lacks batch identification, is in the calibration state of the detection equipment, is in the maintenance state of the detection equipment, or is in a communication abnormal state, and uses the data that passes the check as the data for subsequent processing.

[0229] During the rule base call process, the AI ​​algorithm processing unit calls the AI ​​coal quality status recognition rule base to match the comparison results of low calorific value data, moisture data, volatile matter data, and ash content data to obtain the oxygen consumption impact level of coal quality switching. It then calls the AI ​​target selection rule base and the preset target adjustment strategy table to match the oxygen consumption impact level of coal quality switching, the oxygen content deviation trend at the preheater outlet, the carbon monoxide rise status at the kiln tail, and the combustion response confirmation status to obtain the target adjustment strategy.

[0230] During the state result generation process, the AI ​​algorithm processing unit generates the coal powder bin layer replacement progress and bin exit impact window according to the processing method described in Example 2, generates the preheater outlet oxygen content offset trend, kiln tail carbon monoxide rise status and combustion response confirmation status according to the processing method described in Example 3, and generates the candidate preheater outlet oxygen content target value and current target retention status according to the processing method described in Example 4.

[0231] During the target update judgment process, the AI ​​algorithm processing unit does not directly change the actions of the field actuators. Instead, it transmits the current target retention status and the candidate preheater outlet oxygen content target value to the target update control unit. The target update control unit generates a preheater outlet oxygen content target update command and writes it into the oxygen content target setting position of the central control system.

[0232] The coal powder silo layer processing unit is used to collect coal powder silo operation information and transmit the coal powder silo operation information to the AI ​​algorithm processing unit;

[0233] The coal oxygen consumption identification unit is used to collect coal quality information and transmit the coal quality information to the AI ​​algorithm processing unit;

[0234] The combustion response confirmation unit is used to collect information on oxygen content at the preheater outlet, carbon monoxide information at the kiln tail, and the current target value of oxygen content at the preheater outlet, and transmits the information on oxygen content at the preheater outlet, carbon monoxide information at the kiln tail, and the current target value of oxygen content at the preheater outlet to the AI ​​algorithm processing unit.

[0235] The target strategy selection unit is used to store a preset target adjustment strategy table and make the preset target adjustment strategy table available for the AI ​​algorithm processing unit to call.

[0236] The target update control unit is used to collect the target update records of oxygen content at the preheater outlet and generate a target update instruction for oxygen content at the preheater outlet based on the current target retention status output by the AI ​​algorithm processing unit.

[0237] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended technical solutions and their equivalents.

Claims

1. A method for precise control of oxygen content in cement clinker kilns based on AI algorithms, characterized in that, Includes the following steps: S1. Collect coal powder silo operation information, coal quality information, preheater outlet oxygen content information, kiln tail carbon monoxide information, current preheater outlet oxygen content target value and preheater outlet oxygen content target update record. Define the coal powder entering the coal powder silo after the coal quality information changes as new coal powder. Determine the coal powder silo layer replacement progress based on the coal powder silo operation information, and determine the discharge impact window after the new coal powder enters the coal powder silo discharge area based on the coal powder silo layer replacement progress. S2. Within the outflow impact window, the oxygen consumption impact level of coal quality switching is determined based on coal quality information. When the outflow impact window is not determined based on the coal powder bin layer replacement progress, the oxygen consumption impact level of coal quality switching is prohibited from participating in the adjustment of the current preheater outlet oxygen content target value. S3. Within the influence window of the outlet, determine the oxygen content deviation trend of the preheater outlet based on the oxygen content information of the preheater outlet and the current target value of the oxygen content of the preheater outlet, determine the carbon monoxide rising state of the kiln tail based on the carbon monoxide information of the kiln tail, and determine the combustion response confirmation state based on the oxygen content deviation trend of the preheater outlet and the carbon monoxide rising state of the kiln tail. S4. Based on the oxygen consumption impact level of coal quality switching, the trend of oxygen content deviation at the preheater outlet, the rising state of carbon monoxide at the kiln tail, and the combustion response confirmation state, select a target adjustment strategy from the preset target adjustment strategy table through the AI ​​target selection rule library, and determine the target value of candidate oxygen content at the preheater outlet according to the target adjustment strategy. S5. Determine the current target retention status based on the candidate preheater outlet oxygen content target value, the current preheater outlet oxygen content target value, the preheater outlet oxygen content target update record, and the kiln tail carbon monoxide rise status, and generate a preheater outlet oxygen content target update instruction based on the current target retention status.

2. The method for precise control of oxygen content in cement clinker kilns based on AI algorithms according to claim 1, characterized in that, Step S1 includes: S11, reading the coal powder silo inbound metering record, coal powder silo level detection record, and coal powder silo outbound metering record from the coal powder silo inbound metering record, coal powder silo level detection record, and coal powder silo outbound metering record respectively, and transmitting the coal powder silo inbound record, coal powder silo level record, and coal powder silo outbound record to the AI ​​algorithm processing unit; S12, the AI ​​algorithm processing unit establishes a virtual coal powder layer set based on the coal powder silo inbound record, updates the remaining coal powder quantity of each virtual coal powder layer in the virtual coal powder layer set based on the coal powder silo outbound record, and obtains a virtual coal powder layer remaining state set; S13, the AI ​​algorithm processing unit determines the start time and end time of the outbound influence window of the virtual coal powder layer where the new coal powder is located entering the coal powder silo outbound area based on the virtual coal powder layer remaining state set, defines the sampling time period from the start time to the end time of the outbound influence window as the outbound influence window, and defines the layer status of the virtual coal powder layer where the new coal powder is located in the coal powder silo as the coal powder silo layer replacement progress.

3. The method for precise control of oxygen content in cement clinker kilns based on AI algorithms according to claim 2, characterized in that, Step S2 includes: S21, reading the lower heating value data, moisture data, volatile matter data, and ash content data of new pulverized coal from the coal quality batch test records; S22, the AI ​​algorithm processing unit performs grading processing on the data read in step S21 according to the preset grading ranges for lower heating value, moisture, volatile matter, and ash content, obtaining the grading comparison results for lower heating value, moisture, volatile matter, and ash content; S23, the AI ​​algorithm processing unit combines and judges the grading comparison results for lower heating value, moisture, volatile matter, and ash content using the AI ​​coal quality status recognition rule library to obtain the oxygen consumption impact level of coal quality switching.

4. The method for precise control of oxygen content in cement clinker kilns based on AI algorithms according to claim 3, characterized in that, Step S3 includes: S31, reading the oxygen content sampling sequence at the preheater outlet from the online oxygen content detection record, and reading the current target value of oxygen content at the preheater outlet, the preset allowable deviation range of oxygen content, and the preset number of oxygen content determination samplings; S32, the AI ​​algorithm processing unit determines the target range of oxygen content at the preheater outlet based on the current target value of oxygen content at the preheater outlet and the preset allowable deviation range of oxygen content, and forms a preheater outlet oxygen content determination sequence from the sampling sequence of oxygen content at the preheater outlet based on the preset number of oxygen content determination samplings; S33, the AI ​​algorithm processing unit determines the preheater outlet oxygen content offset trend as being within the target range, below the target offset, or above the target offset state based on the comparison result between the preheater outlet oxygen content determination sequence and the target range of oxygen content at the preheater outlet.

5. The method for precise control of oxygen content in cement clinker kilns based on AI algorithms according to claim 4, characterized in that, Step S3 further includes: S34, reading the kiln tail carbon monoxide sampling sequence from the online carbon monoxide detection record, and reading the preset carbon monoxide upper limit, the preset carbon monoxide forced update threshold, and the preset carbon monoxide judgment sampling number; S35, the AI ​​algorithm processing unit forms a kiln tail carbon monoxide judgment sequence from the kiln tail carbon monoxide sampling sequence according to the preset carbon monoxide judgment sampling number, and determines the kiln tail carbon monoxide rising state as a normal carbon monoxide state, a carbon monoxide exceeding the upper limit state, or a carbon monoxide forced update state according to the kiln tail carbon monoxide judgment sequence, the preset carbon monoxide upper limit, and the preset carbon monoxide forced update threshold; S36, the AI ​​algorithm processing unit determines the combustion response confirmation state as a combustion response confirmed state or a combustion response unconfirmed state according to the preheater outlet oxygen content shift trend and the kiln tail carbon monoxide rising state.

6. The method for precise control of oxygen content in cement clinker kilns based on AI algorithms according to claim 5, characterized in that, Step S4 includes: S41, assembling a target strategy retrieval data package by combining the coal quality switching oxygen consumption impact level, the preheater outlet oxygen content shift trend, the kiln tail carbon monoxide rise status, and the combustion response confirmation status; S42, calling a preset target adjustment strategy table from the AI ​​target selection rule base, wherein the preset target adjustment strategy table includes multiple target adjustment strategy records, each target adjustment strategy record including a status combination and a target adjustment strategy; S43, the AI ​​target selection rule base retrieves matching target adjustment strategy records from the multiple target adjustment strategy records according to the target strategy retrieval data package, and reads the target adjustment strategy from the matching target adjustment strategy records.

7. The method for precise control of oxygen content in cement clinker kilns based on AI algorithms according to claim 6, characterized in that, Step S4 further includes: S44, calling a preset target value level table, wherein the preset target value level table includes multiple target value level records, each target value level record includes a target adjustment strategy and a candidate target value for the next control cycle; S45, according to the target adjustment strategy read in step S43, searching for a matching target value level record among the multiple target value level records; S46, reading the candidate target value for the next control cycle from the matching target value level record, and determining the candidate target value for the next control cycle as the candidate preheater outlet oxygen content target value.

8. The method for precise control of oxygen content in cement clinker kilns based on AI algorithms according to claim 7, characterized in that, Step S5 includes: S51, the AI ​​algorithm processing unit determines the target update difference status based on the target value difference between the candidate preheater outlet oxygen content target value and the current preheater outlet oxygen content target value; S52, the AI ​​algorithm processing unit determines the target update count status based on the number of times the preheater outlet oxygen content target update record is updated in the target update statistics window; S53, the AI ​​algorithm processing unit determines the current target retention status as either an allowed retention status or an unallowed retention status based on the target update difference status, the target update count status, and the kiln tail carbon monoxide rise status.

9. A method for precise control of oxygen content in cement clinker kilns based on AI algorithms according to claim 8, characterized in that, Step S5 further includes: S54, when the current target retention state is a permitted retention state, generating a maintain current target instruction and using the maintain current target instruction as the preheater outlet oxygen content target update instruction; S55, when the current target retention state is a disallowed retention state and the kiln tail carbon monoxide rise state is not a carbon monoxide forced update state, generating an update to candidate target instruction and using the update to candidate target instruction as the preheater outlet oxygen content target update instruction; S56, when the kiln tail carbon monoxide rise state is a carbon monoxide forced update state, generating a forced target update instruction and using the forced target update instruction as the preheater outlet oxygen content target update instruction.

10. A precise oxygen content control system for cement clinker kilns based on AI algorithms, based on the precise oxygen content control method for cement clinker kilns based on AI algorithms as described in any one of claims 1 to 9, characterized in that, It includes an AI algorithm processing unit, a pulverized coal bin layer processing unit, a coal quality oxygen consumption identification unit, a combustion response confirmation unit, a target strategy selection unit, and a target update control unit; When connected to the pulverized coal silo layer processing unit, coal quality oxygen consumption identification unit, combustion response confirmation unit, target strategy selection unit, and target update control unit, it is used to receive data transmitted by each unit, call the AI ​​coal quality status identification rule library and AI target selection rule library, and output the pulverized coal silo layer replacement progress, silo exit impact window, coal quality switching oxygen consumption impact level, preheater outlet oxygen content offset trend, kiln tail carbon monoxide rise status, combustion response confirmation status, target adjustment strategy, candidate preheater outlet oxygen content target value, and current target retention status; The coal powder silo layer processing unit is used to collect coal powder silo operation information and transmit the coal powder silo operation information to the AI ​​algorithm processing unit; The coal oxygen consumption identification unit is used to collect coal quality information and transmit the coal quality information to the AI ​​algorithm processing unit; The combustion response confirmation unit is used to collect information on oxygen content at the preheater outlet, carbon monoxide information at the kiln tail, and the current target value of oxygen content at the preheater outlet, and transmits the information on oxygen content at the preheater outlet, carbon monoxide information at the kiln tail, and the current target value of oxygen content at the preheater outlet to the AI ​​algorithm processing unit. The target strategy selection unit is used to store a preset target adjustment strategy table and make the preset target adjustment strategy table available for the AI ​​algorithm processing unit to call. The target update control unit is used to collect the target update records of oxygen content at the preheater outlet and generate a target update instruction for oxygen content at the preheater outlet based on the current target retention status output by the AI ​​algorithm processing unit.

Images