A rail-type rotary kiln denitrification optimization method and system

Abstract

The present invention relates to the field of data processing technology. The present invention provides a method and system for optimizing denitrification of a track-type rotary kiln, which aims to solve the problem that it is difficult to accurately predict the amount of ammonia sprayed and the injection point under complex working conditions such as fluctuations in the kiln speed and changes in fuel composition, resulting in low utilization rate of the reducing agent and unstable denitrification efficiency. The present invention includes obtaining denitrification data, establishing reference points and digital twin images in the kiln, collecting real-time data of the spray gun for comparison to obtain delayed data, establishing partition monitoring in the kiln and comparing gas data to obtain error data, and optimizing the denitrification prediction model based on the above data to obtain control parameters. The present invention effectively improves the denitrification efficiency and the utilization rate of the reducing agent through various measures such as data collection, processing, model optimization and spray gun control, accurately responds to complex working conditions, and achieves stable and efficient denitrification.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a denitration optimization method and system for a track-type rotary kiln. Background Art

[0002] Orbital rotary kiln denitrification is an environmental protection technology for controlling nitrogen oxide (NOx) emissions in industrial kiln flue gas, and is mainly used in high-temperature production processes in industries such as cement, metallurgy, and chemicals. Its core principle is to utilize the unique rotating structure and high-temperature environment of the rotary kiln to spray ammonia, urea and other reducing agents at specific locations of the kiln body (such as the pre-decomposition zone or flue), and reduce NOx in the flue gas to harmless nitrogen and water through selective non-catalytic reduction (SNCR) or SCR technology combined with a catalyst.

[0003] The data processing process of orbital rotary kiln denitrification follows the logical chain of monitoring, analysis and feedback: first, key parameters (such as temperature distribution in the kiln, NOx concentration, oxygen content and reducing agent injection amount) are collected in real time through kiln temperature sensors, flue gas composition analyzers and flow meters, and the data are synchronously transmitted to the central control system; then, based on the preset denitrification efficiency target, the system combines thermodynamic models or machine learning algorithms to deduce the efficiency threshold of the reaction between the reducing agent and NOx under the current operating conditions, and identify the impact of temperature fluctuations or concentration deviations on the reaction path; then, by comparing the actual emission values ​​with the preset standards, the reducing agent injection point (moving the spray gun along the track) and the injection amount are dynamically adjusted to keep the reaction zone in the optimal temperature window (usually 850-1100℃), and at the same time, the prediction model is trained using historical data to optimize the ammonia injection strategy to reduce ammonia escape; finally, the processed real-time data and process adjustment instructions form a closed-loop feedback loop to continuously reduce NOx emissions, and verify the balance between emission reduction effect and energy consumption through the data visualization platform.

[0004] In the actual data processing process of orbital rotary kiln denitrification, the high temperature and dusty environment in the orbital rotary kiln can easily cause data drift or distortion of temperature sensors and flue gas composition detectors, and the transient changes in operating conditions when the spray gun moves along the track cause real-time monitoring of key parameters (such as local temperature and NOx concentration) to lag. Existing thermodynamic models or data-driven algorithms are not adequately fitted for the nonlinear and strongly coupled reaction processes in the kiln, especially under complex operating conditions such as kiln speed fluctuations and changes in fuel composition. It is difficult to accurately predict the optimal ammonia injection amount and injection point, resulting in low reductant utilization (increased ammonia escape) or unstable denitrification efficiency. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a track-type rotary kiln denitrification optimization method and system to solve the problem that it is difficult to accurately predict the optimal ammonia injection amount and injection point under complex working conditions such as kiln speed fluctuations and fuel composition changes, resulting in low reducing agent utilization (increased ammonia escape) or unstable denitrification efficiency.

[0006] In order to solve the above technical problems, the specific technical solutions of the present invention are as follows: In a first aspect, the present invention provides a method for optimizing denitration of a track-type rotary kiln, comprising: Step S101, obtaining denitration data of the orbital rotary kiln, the denitration data of the orbital rotary kiln includes the operation parameters of the orbital rotary kiln body, the temperature distribution data in the orbital rotary kiln, the NOx concentration data in the orbital rotary kiln, the oxygen content data in the orbital rotary kiln, the reducing agent injection data in the orbital rotary kiln, the spray gun status data in the orbital rotary kiln and the working condition data in the orbital rotary kiln; Step S102, based on the denitration data of the orbital rotary kiln, establish a reference position point in the kiln in the orbital rotary kiln, configure the coordinates with the reference position point in the kiln, obtain the coordinate information in the kiln, substitute the coordinate information in the kiln into the preset digital twin model, generate the digital twin image in the kiln, receive the denitration task of the orbital rotary kiln, substitute the denitration task of the orbital rotary kiln into the preset denitration prediction model of the orbital rotary kiln, obtain the denitration prediction data of the orbital rotary kiln, substitute the denitration prediction data of the orbital rotary kiln into the digital twin image in the kiln, obtain the prediction information of the spray gun on the orbital moving mechanism, the prediction information of the spray gun on the orbital moving mechanism includes the predicted coordinates of the spray gun on the track and the predicted time information corresponding to the coordinates; Step S103, collecting real-time data of the spray gun on the track moving mechanism in the track type rotary kiln to obtain real-time spray gun movement data, the real-time spray gun movement data includes the real-time track coordinates of the spray gun and the real-time time information corresponding to the coordinates, and comparing the real-time spray gun movement data with the prediction information of the spray gun on the track moving mechanism to obtain spray gun delay data; Step S104, establish kiln partitions in the digital twin image in the kiln, obtain kiln partitions, monitor the kiln partitions, obtain gas data of each kiln partition, retrieve data from the orbital rotary kiln denitration data, obtain the fuel calorific value fluctuation coefficient, sulfur content time series change value and ash content dynamic parameters under the optimal working condition in the kiln, substitute the fuel calorific value fluctuation coefficient, sulfur content time series change value and ash content dynamic parameters under the optimal working condition in the kiln into the preset orbital rotary kiln denitration prediction model, obtain orbital rotary kiln denitration prediction gas data, compare the orbital rotary kiln denitration prediction gas data with the gas data of each kiln partition in turn, and obtain gas error data of each kiln partition; Step S105, based on the gas error data of each kiln partition and the spray gun delay data, the preset orbital rotary kiln denitrification prediction model is optimized to obtain the optimized orbital rotary kiln denitrification prediction model, and the orbital rotary kiln denitrification task is substituted into the optimized orbital rotary kiln denitrification prediction model to obtain the optimized spray gun orbital movement control parameters, the spray gun orbital movement control parameters include the spray gun orbital movement speed parameters, the ammonia spray amount and the injection point, and the optimized spray gun orbital movement control parameters are used as the real-time execution data of the spray gun during the movement of the orbital movement mechanism.

[0007] Furthermore, in the track-type rotary kiln denitrification optimization method described in the present invention, step S101 comprises: The speed fluctuation characteristics are collected by a vibration sensor pre-installed on the kiln drive gearbox to calculate the speed change rate; the axial temperature distribution is obtained at 0.5m intervals using a pre-installed distributed optical fiber temperature measurement system, and the sampling frequency is not less than 10Hz; the NOx concentration gradient is monitored in real time using a pre-installed laser absorption spectrometer; the reducing agent injection data includes ammonia concentration, atomization pressure and injection flow rate timing curves.

[0008] Furthermore, in the track-type rotary kiln denitration optimization method of the present invention, the step S102 of constructing kiln coordinate information based on the track-type rotary kiln denitration data includes: The three-dimensional geometric model of the kiln body is established based on the point cloud scanning data, and the meshing accuracy is ≤2cm; 24 reference positioning RFID tags are set in the axial direction of the kiln body to establish the conversion relationship between the spatial coordinate system inside the kiln and the external polar coordinates; The historical operating condition data is retrieved from the orbital rotary kiln denitrification data, and the historical operating condition data is classified and stored by fuel type to build a denitrification efficiency mapping database including 2000 groups of operating conditions.

[0009] Furthermore, in the track-type rotary kiln denitration optimization method described in the present invention, the step S102 of substituting the track-type rotary kiln denitration task into a preset track-type rotary kiln denitration prediction model includes: collecting the current harmonics of the kiln drive motor, extracting data features of the current harmonics of the kiln drive motor, and obtaining the speed fluctuation characteristics; collecting the axial temperature field at a spatial resolution of 0.3m, monitoring the NOx concentration gradient, and aligning the speed fluctuation characteristics, the axial temperature field, the monitoring of the NOx concentration gradient and the spray gun track encoder position signal in time and space to establish a data matrix with timestamp synchronization; The model is trained based on the data corresponding to the data matrix synchronized with the timestamp to obtain the orbital rotary kiln denitrification prediction model.

[0010] Furthermore, in the track-type rotary kiln denitrification optimization method described in the present invention, the step S104 of establishing kiln partitions in the kiln using the digital twin image in the kiln includes: A monitoring section is divided every 0.5m along the axial direction of the kiln body, and the radial direction is divided into the following sections according to the temperature gradient: High temperature core area (T≥1050℃), transition reaction area (850℃≤T<1050℃), edge low temperature area (T<850℃); the gas error data is calculated using: .in is the comprehensive gas error index of the ith partition, is the contribution weight of NOx concentration error, is the contribution weight of oxygen content error, is the predicted NOx volume concentration, is the measured NOx volume concentration, is the normalized reference value of NOx concentration, is the predicted oxygen volume fraction, is the measured oxygen volume fraction, Oxygen content normalized to baseline value.

[0011] In a second aspect, the present invention provides a track-type rotary kiln denitration optimization system, which uses a track-type rotary kiln denitration optimization method as described above, including: a server end, a data acquisition end, and a track-type rotary kiln denitration equipment end, wherein the server end establishes communication connections with the data acquisition end and the track-type rotary kiln denitration equipment end respectively; A data acquisition unit is used to acquire denitration data of the orbital rotary kiln, wherein the denitration data of the orbital rotary kiln includes operation parameters of the orbital rotary kiln body, temperature distribution data in the orbital rotary kiln, NOx concentration data in the orbital rotary kiln, oxygen content data in the orbital rotary kiln, reducing agent injection data in the orbital rotary kiln, spray gun status data in the orbital rotary kiln, and working condition data in the orbital rotary kiln; A data analysis unit is used to establish a reference position point in the kiln in the track-type rotary kiln, perform coordinate configuration based on the reference position point in the kiln, obtain coordinate information in the kiln, substitute the coordinate information in the kiln into a preset digital twin model, generate a digital twin image in the kiln, receive a denitration task of the track-type rotary kiln, substitute the denitration task of the track-type rotary kiln into a preset denitration prediction model of the track-type rotary kiln, obtain denitration prediction data of the track-type rotary kiln, substitute the denitration prediction data of the track-type rotary kiln into the digital twin image in the kiln, obtain prediction information of the spray gun on the track moving mechanism, and the prediction information of the spray gun on the track moving mechanism includes the predicted coordinates of the spray gun on the track and the predicted time information corresponding to the coordinates; A data comparison unit is used to collect real-time data of the spray gun on the track moving mechanism in the track type rotary kiln to obtain real-time spray gun movement data, the real-time spray gun movement data includes the real-time track coordinates of the spray gun and the real-time time information corresponding to the coordinates, and compare the real-time spray gun movement data with the prediction information of the spray gun on the track moving mechanism to obtain the spray gun delay data; An error analysis unit is used to establish kiln partitions in the digital twin image in the kiln, obtain kiln partitions, monitor the kiln partitions, obtain gas data of each kiln partition, retrieve data from the orbital rotary kiln denitrification data, obtain the fuel calorific value fluctuation coefficient, sulfur content time series change value and ash content dynamic parameters under the optimal operating condition in the kiln, substitute the fuel calorific value fluctuation coefficient, sulfur content time series change value and ash content dynamic parameters under the optimal operating condition in the kiln into a preset orbital rotary kiln denitrification prediction model, obtain orbital rotary kiln denitrification prediction gas data, compare the orbital rotary kiln denitrification prediction gas data with the gas data of each kiln partition in turn, and obtain gas error data of each kiln partition; The control optimization unit is used to optimize the preset orbital rotary kiln denitrification prediction model based on the gas error data of each kiln partition and the spray gun delay data to obtain the optimized orbital rotary kiln denitrification prediction model, substitute the orbital rotary kiln denitrification task into the optimized orbital rotary kiln denitrification prediction model to obtain the optimized spray gun orbital movement control parameters, the spray gun orbital movement control parameters include the spray gun orbital movement speed parameters, the ammonia spray amount and the injection point, and use the optimized spray gun orbital movement control parameters as real-time execution data of the spray gun during the movement of the orbital movement mechanism.

[0012] Furthermore, in the track-type rotary kiln denitration optimization system described in the present invention, the data acquisition unit is also used for: The speed fluctuation characteristics are collected by a vibration sensor pre-installed on the kiln drive gearbox to calculate the speed change rate; the axial temperature distribution is obtained at 0.5m intervals using a pre-installed distributed optical fiber temperature measurement system, and the sampling frequency is not less than 10Hz; the NOx concentration gradient is monitored in real time using a pre-installed laser absorption spectrometer; the reducing agent injection data includes ammonia concentration, atomization pressure and injection flow rate timing curves.

[0013] Furthermore, in the track-type rotary kiln denitrification optimization system described in the present invention, the data analysis unit is also used for: The three-dimensional geometric model of the kiln body is established based on the point cloud scanning data, and the meshing accuracy is ≤2cm; 24 reference positioning RFID tags are set in the axial direction of the kiln body to establish the conversion relationship between the spatial coordinate system inside the kiln and the external polar coordinates; The historical operating condition data is retrieved from the orbital rotary kiln denitrification data, and the historical operating condition data is classified and stored by fuel type to build a denitrification efficiency mapping database including 2000 groups of operating conditions.

[0014] Furthermore, in the track-type rotary kiln denitrification optimization system described in the present invention, the data analysis unit is also used to: collect the current harmonics of the kiln drive motor, extract data features of the current harmonics of the kiln drive motor, and obtain the speed fluctuation characteristics; collect the axial temperature field with a spatial resolution of 0.3m, monitor the NOx concentration gradient, align the speed fluctuation characteristics, the axial temperature field, the NOx concentration gradient monitoring and the spray gun track encoder position signal in time and space, and establish a data matrix with time stamp synchronization; The model is trained based on the data corresponding to the data matrix synchronized with the timestamp to obtain the orbital rotary kiln denitrification prediction model.

[0015] Furthermore, in the track-type rotary kiln denitrification optimization system described in the present invention, the error analysis unit is also used for: A monitoring section is divided every 0.5m along the axial direction of the kiln body, and the radial direction is divided into the following sections according to the temperature gradient: High temperature core area (T≥1050℃), transition reaction area (850℃≤T<1050℃), edge low temperature area (T<850℃); the gas error data is calculated using: .in is the comprehensive gas error index of the ith partition, is the contribution weight of NOx concentration error, is the contribution weight of oxygen content error, is the predicted NOx volume concentration, is the measured NOx volume concentration, is the normalized reference value of NOx concentration, is the predicted oxygen volume fraction, is the measured oxygen volume fraction, Oxygen content normalized to baseline value.

[0016] Beneficial effects of the present invention: Improve the prediction accuracy of ammonia injection amount and injection point: Through comprehensive and refined data collection methods, such as enhancing the speed fluctuation feature collection, optimizing temperature distribution and NOx concentration monitoring, monitoring the full parameters of reducing agent injection, etc., it is possible to more accurately grasp the various working conditions in the kiln. Combined with the precise construction and analysis of the digital twin model, the model training optimization based on current harmonics and temperature fields, and the kiln partition monitoring and data processing optimization measures, the optimal ammonia injection amount and injection point can still be accurately predicted under complex working conditions such as kiln speed fluctuations and fuel composition changes, effectively solving the problem of inaccurate prediction of traditional methods under these complex working conditions.

[0017] Optimize spray gun control: With the help of real-time data monitoring of the spray gun and the establishment of a spray gun delay model, the spray gun delay data can be accurately calculated, and a more accurate spray gun delay model can be established by considering the influence of the airflow field inside the kiln body, thereby achieving more precise control of the spray gun movement trajectory and injection time, ensuring that the reducing agent is injected at the appropriate time and position, and further improving the denitrification effect.

[0018] Improve denitrification efficiency: By accurately predicting the optimal ammonia injection amount and injection point, the reducing agent can react with NOx more fully and effectively, reducing the NOx in the flue gas to harmless nitrogen and water, thereby improving the denitrification efficiency and making the final emission NOx concentration meet more stringent environmental protection standards, effectively reducing the emission of nitrogen oxides in the flue gas of industrial kilns, which is of great significance to environmental protection.

[0019] Improve the utilization rate of reducing agent: Accurate control of ammonia injection amount and optimization of injection point avoid ammonia escape caused by excessive ammonia injection, so that the reducing agent can fully participate in the denitrification reaction, improve the utilization rate of the reducing agent, reduce production costs, and also reduce the adverse effects of ammonia escape on the environment and equipment.

[0020] Adapting to kiln speed fluctuations: From multi-dimensional monitoring and compensation of speed fluctuations in the data collection stage, to building a prediction model that can adapt to speed changes based on data such as current harmonics, to kiln zone monitoring and adjusting temperature zone boundaries according to speed fluctuations, the entire denitrification system can adapt well to kiln speed fluctuations, ensuring the stability and efficiency of the denitrification process under different speed conditions.

[0021] Responding to changes in fuel composition: By establishing a detailed operating condition database optimization strategy, classifying different fuel types in detail and associating rich feature vectors, and screening and preprocessing historical data according to fuel characteristics during data processing, the system can quickly and accurately adapt to the impact of changes in fuel composition, ensuring accurate denitrification control under different fuel conditions.

[0022] In summary, the present invention has significant beneficial effects in precise denitrification control, coping with complex working conditions, improving denitrification efficiency and reducing agent utilization, optimizing data processing and models, and system operation and maintenance, and has important application value and promotion significance for the denitrification treatment of industrial kilns. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solution of the present invention, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, for ordinary technicians in this field, other drawings can be obtained based on the drawings without paying any creative labor.

[0024] Figure 1A schematic diagram of the process flow of the orbital rotary kiln denitrification optimization method provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described in conjunction with the specific embodiments of the present invention and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. The technical solutions provided by the embodiments of the present invention are described in detail below in conjunction with the drawings.

[0026] In order to better understand the purpose of the present invention, the present invention is described in further detail below.

[0027] In a first aspect, the present invention provides a method for optimizing denitration of a track-type rotary kiln, comprising: Step S101, obtaining denitration data of the orbital rotary kiln, the denitration data of the orbital rotary kiln includes the operation parameters of the orbital rotary kiln body, the temperature distribution data in the orbital rotary kiln, the NOx concentration data in the orbital rotary kiln, the oxygen content data in the orbital rotary kiln, the reducing agent injection data in the orbital rotary kiln, the spray gun status data in the orbital rotary kiln and the working condition data in the orbital rotary kiln; The acquisition of denitration data of the orbital rotary kiln in step S101 includes the following data collection methods: The operating parameters of the orbital rotary kiln include: Speed ​​data: monitor the real-time speed of the kiln and analyze the impact of speed fluctuations on material residence time, temperature distribution and NOx generation.

[0028] Fuel composition data: including fuel type (such as pulverized coal, natural gas), calorific value, sulfur content, ash content, etc., used to predict NOx generation and temperature change trends during combustion.

[0029] Temperature distribution data in the orbital rotary kiln: The temperature gradient of key areas such as the pre-decomposition zone and the firing zone is obtained through multi-point temperature sensors to identify the optimal reaction temperature window (850-1100℃).

[0030] NOx concentration in orbital rotary kiln: Real-time monitoring of NOx concentration (including NO and NO2) in flue gas as the core indicator of denitrification efficiency.

[0031] Oxygen content in the orbital rotary kiln: judge the combustion sufficiency and reduction reaction conditions, and avoid the interference of excessive oxygen on the denitrification reaction.

[0032] Other gas components are also obtained, such as CO, CO2, and SO2 concentrations, to assist in analyzing combustion efficiency and potential impacts on denitrification reactions.

[0033] The data of reducing agent injection in orbital rotary kiln include: Injection amount: Real-time record of the flow rate of ammonia or urea, and calculation of the actual reducing agent amount involved in the reaction based on the concentration data.

[0034] Injection point: The position information of the spray gun moving along the track is used to match the dynamic coverage of the reaction zone when the kiln body is rotating.

[0035] Injection pressure / atomization effect: affects the mixing uniformity and reaction efficiency of the reducing agent and flue gas.

[0036] The status data of the spray gun in the orbital rotary kiln include: Movement trajectory and speed: Monitor the movement path and speed of the spray gun along the track to synchronize the injection point with the kiln rotation.

[0037] Start and stop status: record the spray gun working cycle to avoid invalid spraying or redundant operation.

[0038] The working condition data of the orbital rotary kiln include: Pressure inside the kiln: affects the gas flow direction and the distribution of reducing agents, and needs to be modeled together with temperature and rotation speed.

[0039] Material flow and composition: Monitor the input amount and composition changes of raw materials or fuels to predict NOx generation fluctuations.

[0040] Ambient temperature and humidity: the potential impact of the external environment on kiln heat dissipation and sensor data stability.

[0041] Track type rotary kiln denitrification data include: Historical operation records: including historical data such as speed, fuel ratio, and ammonia injection amount, which are used to train machine learning models.

[0042] Ammonia slip concentration: The concentration of unreacted NH3 is detected by an additional sensor to evaluate the reductant utilization.

[0043] Emissions compliance data: Final flue gas emission NOx concentration for closed-loop feedback control strategy optimization.

[0044] Step S102, based on the denitration data of the orbital rotary kiln, establish a reference position point in the kiln in the orbital rotary kiln, configure the coordinates with the reference position point in the kiln, obtain the coordinate information in the kiln, substitute the coordinate information in the kiln into the preset digital twin model, generate the digital twin image in the kiln, receive the denitration task of the orbital rotary kiln, substitute the denitration task of the orbital rotary kiln into the preset denitration prediction model of the orbital rotary kiln, obtain the denitration prediction data of the orbital rotary kiln, substitute the denitration prediction data of the orbital rotary kiln into the digital twin image in the kiln, obtain the prediction information of the spray gun on the orbital moving mechanism, the prediction information of the spray gun on the orbital moving mechanism includes the predicted coordinates of the spray gun on the track and the predicted time information corresponding to the coordinates; The step S102 of constructing kiln coordinate information based on the orbital rotary kiln denitration data includes the following specific implementations: When building a 3D geometric model of the kiln body based on point cloud scanning data, in addition to the meshing accuracy of ≤2cm, the point cloud scanning data must be updated regularly according to the actual operation of the kiln body. For example, a comprehensive point cloud scan is performed every other week to capture the structural deformation and lining wear that may occur in the long-term operation of the kiln body. The updated point cloud scanning data will be used to correct the 3D geometric model so that the model more accurately reflects the actual state of the kiln body, thereby improving the accuracy of subsequent analysis and prediction based on the model.

[0045] 24 reference positioning RFID tags are set in the axial direction of the kiln body to establish the conversion relationship between the spatial coordinate system in the kiln and the external polar coordinates. The reading accuracy and signal strength of the RFID tags are checked regularly. If any tag reading is abnormal or the signal strength is lower than the set threshold, it is replaced or adjusted in time to ensure that the accuracy of the spatial coordinate system conversion relationship is not affected.

[0046] In terms of retrieving historical operating condition data and storing them by fuel type to build a denitrification efficiency mapping database, the historical operating condition data is further mined in depth. In addition to classification by fuel type, it is also subdivided according to factors such as different seasons and different production batches. Because the ambient temperature, humidity and other conditions are different in different seasons, the quality of raw materials in different production batches may also be different, and these factors will have an impact on the denitrification process. Through this more detailed classification storage, when the database is used for subsequent analysis and prediction, similar operating conditions can be matched more accurately, thereby improving the accuracy of the denitrification prediction model. The preset orbital rotary kiln denitrification prediction model is trained based on historical operating condition data and real-time monitoring data.

[0047] Substituting the orbital rotary kiln denitration task into a preset orbital rotary kiln denitration prediction model in step S102 includes the following model training optimization methods: In terms of collecting the current harmonics of the kiln drive motor and extracting data features to obtain the speed fluctuation characteristics and related data processing, the filtering processing of the current harmonic data is added. The adaptive filter is used to automatically adjust the filtering parameters according to the real-time status of the kiln operation, remove the noise interference in the current harmonic data, and improve the accuracy of the speed fluctuation feature extraction. At the same time, when the speed fluctuation characteristics, axial temperature field, monitoring NOx concentration gradient and spray gun track encoder position signal are aligned in time and space to establish a data matrix with timestamp synchronization, a data interpolation algorithm is introduced. For inconsistent time intervals or missing data that may occur during the acquisition process, the interpolation algorithm is used to complete the data and regularize the time series to ensure the integrity and accuracy of the data matrix, so as to better train the model.

[0048] Step S103, collecting real-time data of the spray gun on the track moving mechanism in the track type rotary kiln to obtain real-time spray gun movement data, the real-time spray gun movement data includes the real-time track coordinates of the spray gun and the real-time time information corresponding to the coordinates, and comparing the real-time spray gun movement data with the prediction information of the spray gun on the track moving mechanism to obtain spray gun delay data; In the process of collecting real-time data of the spray gun on the track moving mechanism in the track type rotary kiln, the real-time track coordinates of the spray gun and the real-time time information corresponding to the coordinates are obtained to monitor the vibration of the spray gun. By installing a micro acceleration sensor on the spray gun, the vibration acceleration data of the spray gun is collected in real time, with a sampling frequency of 100 times per second. The vibration of the spray gun will cause the offset of the injection point and the instability of the injection amount. These vibration data will be used together with other real-time data to analyze the actual working status of the spray gun and calculate the spray gun delay data more accurately.

[0049] In order to calculate the spray gun delay data more accurately, in addition to the existing method of comparing the real-time spray gun movement data with the predicted information of the spray gun on the track movement mechanism, the influence of the airflow field inside the kiln on the movement of the spray gun is also considered. By installing wind speed sensors at different positions in the kiln body, real-time data of the airflow speed and direction in the kiln are obtained, and a spray gun delay model considering the influence of airflow is established by combining the movement speed, injection direction and airflow field data of the spray gun. The spray gun delay model will be able to more accurately reflect the delay of the spray gun actually reaching the preset injection point under complex working conditions, providing a more reliable basis for subsequent optimization control. The spray gun delay model is established through the following steps: collecting the airflow speed and direction data in the kiln, combining the movement speed and injection direction of the spray gun, calculating the offset of the airflow to the spray gun trajectory, and generating delay compensation parameters based on the offset.

[0050] Step S104, establish kiln partitions in the digital twin image in the kiln, obtain kiln partitions, monitor the kiln partitions, obtain gas data of each kiln partition, retrieve data from the orbital rotary kiln denitration data, obtain the fuel calorific value fluctuation coefficient, sulfur content time series change value and ash content dynamic parameters under the optimal working condition in the kiln, substitute the fuel calorific value fluctuation coefficient, sulfur content time series change value and ash content dynamic parameters under the optimal working condition in the kiln into the preset orbital rotary kiln denitration prediction model, obtain orbital rotary kiln denitration prediction gas data, compare the orbital rotary kiln denitration prediction gas data with the gas data of each kiln partition in turn, and obtain gas error data of each kiln partition; After dividing a monitoring section every 0.5m along the axial direction of the kiln body and dividing different areas (high-temperature core area, transition reaction area, and edge low-temperature area) according to the temperature gradient, different monitoring parameter weights are further set for each area. For example, in the high-temperature core area, due to its key influence on the denitrification reaction, the monitoring weight of NOx concentration data is set to 0.6, the monitoring weight of oxygen content data is set to 0.3, and the monitoring weight of other related gas data is set to 0.1. In the edge low-temperature area, the monitoring weight of each gas data is adjusted accordingly to better meet the actual situation of the influence of the area on the denitrification reaction.

[0051] In terms of extracting the fuel calorific value fluctuation coefficient, sulfur content time series change value and ash content dynamic parameters under the optimal working condition in the kiln from the orbital rotary kiln denitrification data and substituting them into the preset orbital rotary kiln denitrification prediction model, the historical data screening and preprocessing are increased. According to the current operating parameters of the kiln body (such as speed, temperature, etc.) and the basic characteristics of the fuel (such as calorific value range, sulfur content range, etc.), a set of data that is most similar to the current working conditions is screened from the historical data. The screened data is normalized to make it more in line with the input requirements of the current model in terms of numerical range and data distribution, thereby improving the accuracy of model prediction.

[0052] Step S105, based on the gas error data of each kiln partition and the spray gun delay data, the preset orbital rotary kiln denitrification prediction model is optimized to obtain the optimized orbital rotary kiln denitrification prediction model, and the orbital rotary kiln denitrification task is substituted into the optimized orbital rotary kiln denitrification prediction model to obtain the optimized spray gun orbital movement control parameters, the spray gun orbital movement control parameters include the spray gun orbital movement speed parameters, the ammonia spray amount and the injection point, and the optimized spray gun orbital movement control parameters are used as the real-time execution data of the spray gun during the movement of the orbital movement mechanism.

[0053] When optimizing the preset orbital rotary kiln denitrification prediction model based on the gas error data and spray gun delay data of each kiln partition, in addition to the conventional model parameter adjustment method, an optimization strategy based on reinforcement learning is introduced. According to the actual denitrification effect of the spray gun at different injection points (assessed by real-time monitoring of the NOx concentration of the emission) and the ammonia escape situation, the model is given corresponding reward or penalty signals. The orbital rotary kiln denitrification prediction model continuously adjusts its own prediction strategy based on these feedback signals to more accurately determine the optimal ammonia injection amount and injection point.

[0054] In order to make the optimized spray gun track movement control parameters better adapt to the changes in complex working conditions, after substituting the orbital rotary kiln denitrification task into the optimized orbital rotary kiln denitrification prediction model to obtain the optimized spray gun track movement control parameters, these parameters are verified online in real time. By setting additional monitoring points on the kiln body, the actual denitrification efficiency, ammonia escape rate, and actual movement trajectory of the spray gun are monitored in real time. These actual monitoring data are compared with the results predicted based on the optimized parameters. If the deviation is found to exceed the set threshold, the model is optimized and adjusted again in time to improve the effectiveness and accuracy of the control parameters.

[0055] Specifically, the track-type rotary kiln denitrification optimization method described in the present invention, step S101 comprises: The speed fluctuation characteristics are collected by a vibration sensor pre-installed on the kiln drive gearbox to calculate the speed change rate; the axial temperature distribution is obtained at 0.5m intervals using a pre-installed distributed optical fiber temperature measurement system, and the sampling frequency is not less than 10Hz; the NOx concentration gradient is monitored in real time using a pre-installed laser absorption spectrometer; the reducing agent injection data includes ammonia concentration, atomization pressure and injection flow rate timing curves.

[0056] Enhanced collection of speed fluctuation characteristics: A triaxial MEMS vibration sensor group (X / Y / Z directions) is symmetrically installed on the drive gearbox, and the 1X~5X speed harmonic components are extracted through order analysis; Establish a speed abnormality detection mechanism: when the fluctuation rate of three consecutive sampling cycles is greater than 2%, the Hall sensor auxiliary calibration mode is triggered; Increase the kiln body inertia compensation parameters to eliminate the instantaneous speed jitter interference caused by uneven material distribution.

[0057] Improved reliability of temperature distribution data: The distributed optical fiber adopts a double-layer armored structure (inner layer Φ0.25mm sensing optical fiber, outer layer Φ1mm reference optical fiber); Set temperature mutation verification rules: when the temperature difference between adjacent temperature measurement points is greater than 80°C, activate the infrared thermal imager for regional review; Redundant thermocouple arrays (K type, response time <0.5s) are added at the kiln head / kiln tail to integrate with the optical fiber data in time and space.

[0058] NOx concentration dynamic monitoring optimization: The laser absorption spectrometer is integrated with a self-cleaning optical window (pulse backwash cycle ≤ 30s) to prevent dust adhesion interference; Establish a concentration gradient verification mechanism: when the concentration difference between adjacent monitoring points in the axial direction is greater than 150ppm, start the mobile sampling probe for retesting; Deploy a multi-spectral fusion algorithm to simultaneously analyze the NO / NO2 concentration ratio (accuracy ±3%).

[0059] Reductant injection full parameter monitoring: Install an ultrasonic concentration meter (accuracy ±0.5%) and a Coriolis mass flow meter (accuracy ±0.2%) on the ammonia delivery pipeline; Set the injection pulse width modulation control: when the atomization pressure fluctuation is greater than 0.05MPa, the opening of the pressure regulating valve is automatically adjusted; Monitor the spray gun atomization cone angle in real time (target range 30°-45°), and adjust the gas-liquid ratio through feedback from the high-speed camera system.

[0060] Specifically, in the track-type rotary kiln denitration optimization method of the present invention, the step S102 of constructing kiln coordinate information based on the track-type rotary kiln denitration data includes: The three-dimensional geometric model of the kiln body is established based on the point cloud scanning data, and the meshing accuracy is ≤2cm; 24 reference positioning RFID tags are set in the axial direction of the kiln body to establish the conversion relationship between the spatial coordinate system inside the kiln and the external polar coordinates; The historical operating condition data is retrieved from the orbital rotary kiln denitrification data, and the historical operating condition data is classified and stored by fuel type to build a denitrification efficiency mapping database including 2000 groups of operating conditions.

[0061] Dynamic update mechanism of point cloud data: deploy a mobile laser scanning robot to perform a full kiln scan every 72 hours along the kiln track; Establish a point cloud change rate monitoring model: when the local point cloud offset is greater than 1.5cm for three consecutive scans, a kiln deformation warning is triggered; The ICP algorithm is used to realize the registration of new and old point cloud data and automatically correct the geometric deviation of the 3D model.

[0062] Intelligent calibration of spatial coordinate system: Each RFID tag integrates a temperature compensation module (working range 0-150°C) to eliminate positioning errors caused by thermal expansion; A dual-frequency RFID reader (865MHz / 2.4GHz) is set up to automatically switch the anti-interference frequency band when the kiln rotates; Construct a six-degree-of-freedom coordinate transformation matrix, including the kiln body inclination compensation parameters (±5° range and 0.01° accuracy).

[0063] Working condition database optimization strategy: Establish a five-level fuel classification system: coal / waste-derived fuel / biomass co-firing / petroleum coke / emergency fuel; Deploy data cleaning pipeline: remove sensor outliers (3σ criterion) and fill in missing data (KNN interpolation method); Create a working condition feature fingerprint library: associate a 128-dimensional feature vector (including calorific value fluctuation pattern, sulfur release curve, etc.) with each type of fuel.

[0064] Specifically, the track-type rotary kiln denitrification optimization method described in the present invention, the step S102 of substituting the track-type rotary kiln denitrification task into a preset track-type rotary kiln denitrification prediction model, includes: collecting the current harmonics of the kiln drive motor, extracting data features of the current harmonics of the kiln drive motor, and obtaining the speed fluctuation characteristics; collecting the axial temperature field with a spatial resolution of 0.3m, monitoring the NOx concentration gradient, and aligning the speed fluctuation characteristics, the axial temperature field, the NOx concentration gradient and the spray gun track encoder position signal in time and space, using the IEEE 1588 precision clock protocol to unify the time base, and establishing a data matrix with timestamp synchronization; The model is trained based on the data corresponding to the data matrix synchronized with the timestamp to obtain the orbital rotary kiln denitrification prediction model.

[0065] In-depth analysis of current harmonics: A wide-band current sensor (0-5kHz) is installed on the power supply side of the drive motor to capture the complete spectrum including 23rd harmonics. The wavelet packet decomposition technology is used to separate the fundamental wave and characteristic harmonics. The amplitudes of the 2nd and 5th harmonics are extracted as the core indicators of speed fluctuation. A harmonic-speed mapping database is established, and the harmonic characteristic patterns corresponding to abnormal operating conditions are identified through K-means clustering.

[0066] Dynamic calibration mechanism of temperature field: deploy a movable temperature measurement robot to cruise along the kiln track, perform contact calibration (accuracy ±3°C) on fixed temperature measurement points every 30 minutes, set a 10% overlapping area in adjacent axial temperature measurement intervals, and use cubic spline interpolation to eliminate boundary mutations. When the axial temperature gradient is greater than 150°C / m, activate the redundant infrared thermal imager for full-field scanning verification.

[0067] Multi-source data spatiotemporal fusion: IEEE 1588 precision clock protocol is used to unify the time base of each subsystem (deviation <1ms), and a six-dimensional data cube is constructed: timestamp-axial position-radial depth-speed fluctuation-temperature value-NOx concentration. The data alignment engine is used to automatically compensate for the spatiotemporal coordinate offset caused by the movement of the spray gun (maximum compensation amount ±15cm / 0.5s).

[0068] The prediction model is adaptively trained and a hybrid neural network architecture is designed: 1D-CNN processes the current harmonic sequence, LSTM captures the time series characteristics of the temperature field, the fully connected layer fuses the NOx concentration gradient, and a sliding window mechanism (window length 60s, step length 5s) is deployed to dynamically generate training sample sets.

[0069] Establish an online verification channel: When the deviation between the predicted ammonia injection amount and the actual denitrification efficiency is greater than 8%, incremental learning is triggered to update the model parameters.

[0070] Through the spatial correlation analysis of harmonic characteristics and temperature fields, the speed-temperature coupling fluctuation pattern caused by sudden changes in the calorific value of the fuel (such as a change of ±200kcal / kg) can be identified; the multidimensional characteristics of the data cube enable the model to dynamically adapt to the complex interactive relationship between the movement trajectory of the spray gun and the rotation phase of the kiln body, effectively improving the prediction robustness under variable working conditions.

[0071] Specifically, in the track-type rotary kiln denitrification optimization method described in the present invention, the step S104 of establishing kiln partitions in the kiln using the digital twin image in the kiln includes: A monitoring section is divided every 0.5m along the axial direction of the kiln body, and the radial direction is divided into the following sections according to the temperature gradient: High temperature core area (T≥1050℃), transition reaction area (850℃≤T<1050℃), edge low temperature area (T<850℃); the gas error data is calculated using: .in is the comprehensive gas error index of the ith partition, is the contribution weight of NOx concentration error, is the contribution weight of oxygen content error, is the predicted NOx volume concentration, is the measured NOx volume concentration, is the normalized reference value of NOx concentration, which is the maximum NOx concentration allowed under the current operating conditions. is the predicted oxygen volume fraction, is the measured oxygen volume fraction, The normalized reference value of oxygen content is the maximum volume fraction of oxygen required for theoretical combustion.

[0072] Strengthened axial partition monitoring: three sets of redundant thermocouples (type K / J / T) are arranged in each 0.5m monitoring section to form a triangular temperature measurement array; Develop an axial temperature field self-correction algorithm: when the temperature difference between adjacent sections is greater than 100°C, activate the mobile infrared thermometer for re-measurement and calibration; Establish a dynamic partition boundary mechanism: automatically adjust the high temperature zone boundary according to the real-time thermal image (±0.2m floating range).

[0073] Radial temperature gradient optimization: The high-temperature core area is equipped with a water-cooled high-temperature resistant sensor (working upper limit 1300°C) and a self-cleaning nitrogen purge system; A double-layer monitoring ring is set up in the transition reaction zone: the upper layer monitors the gas composition (spacing 15 cm), and the lower layer monitors the wall temperature; An anti-condensation sensor is installed in the low-temperature edge area, with an integrated heating and dehumidification module (maintaining the probe temperature > 60°C).

[0074] Partition collaborative control strategy: The high-temperature core area implements "temperature priority" control: when T>1100℃, the amount of ammonia spraying is automatically reduced (maximum reduction of 30%); The "optimal efficiency" mode is enabled in the transition reaction zone: the spray gun residence time is dynamically adjusted according to the NOx gradient (base value ± 2s); The "anti-escape" strategy is implemented in the marginal low-temperature area: the ammonia concentration is automatically switched to a low-concentration formula (15%→10%).

[0075] When the sensor data difference within a partition is greater than 5%, the laser scanning temperature field verification is started. Abnormal fluctuations in the high-temperature area trigger an increase in the monitoring frequency of adjacent partitions (10Hz→50Hz). A partition with a calibration deviation greater than 8% for three consecutive times is automatically marked as a fault state.

[0076] In a second aspect, the present invention provides a track-type rotary kiln denitration optimization system, which uses a track-type rotary kiln denitration optimization method as described above, including: a server end, a data acquisition end, and a track-type rotary kiln denitration equipment end, wherein the server end establishes communication connections with the data acquisition end and the track-type rotary kiln denitration equipment end respectively; A data acquisition unit is used to acquire denitration data of the orbital rotary kiln, wherein the denitration data of the orbital rotary kiln includes operation parameters of the orbital rotary kiln body, temperature distribution data in the orbital rotary kiln, NOx concentration data in the orbital rotary kiln, oxygen content data in the orbital rotary kiln, reducing agent injection data in the orbital rotary kiln, spray gun status data in the orbital rotary kiln, and working condition data in the orbital rotary kiln; A data analysis unit is used to establish a reference position point in the kiln in the track-type rotary kiln, perform coordinate configuration based on the reference position point in the kiln, obtain coordinate information in the kiln, substitute the coordinate information in the kiln into a preset digital twin model, generate a digital twin image in the kiln, receive a denitration task of the track-type rotary kiln, substitute the denitration task of the track-type rotary kiln into a preset denitration prediction model of the track-type rotary kiln, obtain denitration prediction data of the track-type rotary kiln, substitute the denitration prediction data of the track-type rotary kiln into the digital twin image in the kiln, obtain prediction information of the spray gun on the track moving mechanism, and the prediction information of the spray gun on the track moving mechanism includes the predicted coordinates of the spray gun on the track and the predicted time information corresponding to the coordinates; A data comparison unit is used to collect real-time data of the spray gun on the track moving mechanism in the track type rotary kiln to obtain real-time spray gun movement data, the real-time spray gun movement data includes the real-time track coordinates of the spray gun and the real-time time information corresponding to the coordinates, and compare the real-time spray gun movement data with the prediction information of the spray gun on the track moving mechanism to obtain the spray gun delay data; An error analysis unit is used to establish kiln partitions in the digital twin image in the kiln, obtain kiln partitions, monitor the kiln partitions, obtain gas data of each kiln partition, retrieve data from the orbital rotary kiln denitrification data, obtain the fuel calorific value fluctuation coefficient, sulfur content time series change value and ash content dynamic parameters under the optimal operating condition in the kiln, substitute the fuel calorific value fluctuation coefficient, sulfur content time series change value and ash content dynamic parameters under the optimal operating condition in the kiln into a preset orbital rotary kiln denitrification prediction model, obtain orbital rotary kiln denitrification prediction gas data, compare the orbital rotary kiln denitrification prediction gas data with the gas data of each kiln partition in turn, and obtain gas error data of each kiln partition; The control optimization unit is used to optimize the preset orbital rotary kiln denitrification prediction model based on the gas error data of each kiln partition and the spray gun delay data to obtain the optimized orbital rotary kiln denitrification prediction model, substitute the orbital rotary kiln denitrification task into the optimized orbital rotary kiln denitrification prediction model to obtain the optimized spray gun orbital movement control parameters, the spray gun orbital movement control parameters include the spray gun orbital movement speed parameters, the ammonia spray amount and the injection point, and use the optimized spray gun orbital movement control parameters as real-time execution data of the spray gun during the movement of the orbital movement mechanism.

[0077] Specifically, in the rail-type rotary kiln denitrification optimization system described in the present invention, the data acquisition unit is also used for: The speed fluctuation characteristics are collected by a vibration sensor pre-installed on the kiln drive gearbox to calculate the speed change rate; the axial temperature distribution is obtained at 0.5m intervals using a pre-installed distributed optical fiber temperature measurement system, and the sampling frequency is not less than 10Hz; the NOx concentration gradient is monitored in real time using a pre-installed laser absorption spectrometer; the reducing agent injection data includes ammonia concentration, atomization pressure and injection flow rate timing curves.

[0078] Specifically, in the track-type rotary kiln denitrification optimization system of the present invention, the data analysis unit is also used for: The three-dimensional geometric model of the kiln body is established based on the point cloud scanning data, and the meshing accuracy is ≤2cm; 24 reference positioning RFID tags are set in the axial direction of the kiln body to establish the conversion relationship between the spatial coordinate system inside the kiln and the external polar coordinates; The historical operating condition data is retrieved from the orbital rotary kiln denitrification data, and the historical operating condition data is classified and stored by fuel type to build a denitrification efficiency mapping database including 2000 groups of operating conditions.

[0079] Specifically, in the track-type rotary kiln denitrification optimization system described in the present invention, the data analysis unit is also used to: collect the current harmonics of the kiln drive motor, extract data features of the current harmonics of the kiln drive motor, and obtain the speed fluctuation characteristics; collect the axial temperature field with a spatial resolution of 0.3m, monitor the NOx concentration gradient, align the speed fluctuation characteristics, the axial temperature field, the NOx concentration gradient monitoring and the spray gun track encoder position signal in time and space, and establish a data matrix with time stamp synchronization; The model is trained based on the data corresponding to the data matrix synchronized with the timestamp to obtain the orbital rotary kiln denitrification prediction model.

[0080] Specifically, in the track-type rotary kiln denitrification optimization system described in the present invention, the error analysis unit is also used for: A monitoring section is divided every 0.5m along the axial direction of the kiln body, and the radial direction is divided into the following sections according to the temperature gradient: High temperature core area (T≥1050℃), transition reaction area (850℃≤T<1050℃), edge low temperature area (T<850℃); the gas error data is calculated using: .in is the comprehensive gas error index of the ith partition, is the contribution weight of NOx concentration error, is the contribution weight of oxygen content error, is the predicted NOx volume concentration, is the measured NOx volume concentration, is the normalized reference value of NOx concentration, is the predicted oxygen volume fraction, is the measured oxygen volume fraction, Oxygen content normalized to baseline value.

[0081] The present invention solves the problem of low reductant utilization (increased ammonia escape) or unstable denitrification efficiency due to difficulty in accurately predicting the optimal ammonia injection amount and injection point under complex working conditions such as kiln speed fluctuations and fuel composition changes through the following measures: Comprehensive and sophisticated data collection and processing: Enhanced speed fluctuation feature acquisition: A three-axis MEMS vibration sensor group is symmetrically installed on the drive gearbox. The 1X - 5X speed harmonic components are extracted through order analysis, which can capture the speed changes more carefully. A speed abnormality detection mechanism is established. When the fluctuation rate of three consecutive sampling cycles is greater than 2%, the Hall sensor auxiliary calibration mode is triggered, which increases the accuracy of monitoring speed abnormalities. At the same time, the kiln body inertia compensation parameters are added to eliminate the instantaneous speed jitter interference caused by uneven material distribution, so that the acquired speed data can more truly reflect the actual operating status of the kiln body, providing a reliable basis for subsequent accurate predictions.

[0082] Improved reliability of temperature distribution data: Distributed optical fiber adopts a double-layer armored structure, which improves the stability and anti-interference ability of the optical fiber. Set temperature mutation verification rules. When the temperature difference between adjacent temperature measurement points is greater than 80°C, activate the infrared thermal imager for regional review to ensure the accuracy of temperature data. Add redundant thermocouple arrays at the head / tail of the kiln and integrate them with the optical fiber data in time and space. Multi-directional temperature monitoring methods complement each other, which can more comprehensively and accurately grasp the temperature distribution in the kiln, which is crucial for determining the optimal reaction temperature window and analyzing the impact of temperature on the denitrification reaction.

[0083] Optimization of dynamic monitoring of NOx concentration: The laser absorption spectrometer is integrated with a self-cleaning optical window to prevent dust adhesion interference and ensure the stability of monitoring data. A concentration gradient calibration mechanism is established. When the concentration difference between adjacent axial monitoring points is greater than 150ppm, the mobile sampling probe is started for retesting to promptly detect and correct possible concentration monitoring deviations. A multi-spectral fusion algorithm is deployed to simultaneously analyze the NO / NO2 concentration ratio, more accurately grasp the actual situation of NOx, and provide a key basis for accurately predicting the optimal ammonia injection amount, because NOx concentration is the core indicator of denitrification efficiency.

[0084] Monitoring of all parameters of reducing agent injection: ultrasonic concentration meter and Coriolis mass flowmeter are installed on the ammonia delivery pipeline to accurately measure the concentration and flow of ammonia, and comprehensively grasp the reducing agent injection situation by combining the information such as injection point, injection pressure / atomization effect. Set the injection pulse width modulation control. When the atomization pressure fluctuation is greater than 0.05MPa, the opening of the pressure regulating valve is automatically adjusted. The spray gun atomization cone angle is monitored in real time and the gas-liquid ratio is adjusted through feedback from the high-speed camera system to ensure the mixing uniformity and reaction efficiency of the reducing agent and flue gas, which indirectly helps to accurately determine the optimal ammonia injection amount and injection point.

[0085] Analysis and optimization based on digital twin model and database: Dynamic update mechanism of point cloud data and model correction: By deploying a mobile laser scanning robot to scan the entire kiln regularly (every 72 hours), a point cloud change rate monitoring model is established to timely detect kiln structure deformation, lining wear, etc. The ICP algorithm is used to realize the registration of new and old point cloud data, and the three-dimensional geometric model is automatically corrected to make the model more accurately reflect the actual state of the kiln, providing a more realistic basis for subsequent analysis and prediction based on the digital twin model, and helping to more accurately analyze the relationship between the situation in the kiln and the denitrification process under complex working conditions.

[0086] Intelligent calibration of spatial coordinate system: Each RFID tag integrates a temperature compensation module to eliminate positioning errors caused by thermal expansion, sets a dual-frequency RFID reader and automatically switches the anti-interference frequency band, and constructs a six-degree-of-freedom coordinate conversion matrix including the kiln body inclination compensation parameters, which ensures the accuracy of the conversion relationship between the spatial coordinate system in the kiln and the external polar coordinates. This is very important for accurately determining the position of the spray gun in the kiln and analyzing its relationship with various parameters, and provides a reliable coordinate basis for accurately predicting the injection point.

[0087] Working condition database optimization strategy: Establish a five-level fuel classification system, classify different types of fuels in detail, deploy data cleaning pipelines to remove outliers, fill in missing data, and create a working condition feature fingerprint library to associate 128-dimensional feature vectors for each type of fuel. Through this more detailed classification storage and data processing, when the database is used for subsequent analysis and prediction, similar working conditions can be matched more accurately, and the accuracy of the denitrification prediction model can be improved, so as to better cope with complex working conditions such as changes in fuel composition.

[0088] Model training optimization based on current harmonics and temperature fields: The current harmonics of the kiln drive motor are collected and deeply analyzed. The complete spectrum is captured by installing a wide-band current sensor. The key harmonic amplitude is extracted using wavelet packet decomposition technology as the core indicator of speed fluctuation. A harmonic-speed mapping database is established and abnormal working conditions are identified through K-means clustering. At the same time, the axial temperature field is collected with a spatial resolution of 0.3m and dynamically calibrated, such as deploying a mobile temperature measurement robot for cruise calibration and setting up overlapping area interpolation processing. These multi-source data are fused in time and space to construct a six-dimensional data cube, and a hybrid neural network architecture is designed based on this for model training. The training sample set is generated through a sliding window mechanism, and an online verification channel is established to trigger incremental learning and update model parameters according to the prediction deviation. This processing method enables the model to better adapt to complex working conditions such as kiln speed fluctuations and temperature changes, and improves the accuracy and robustness of the prediction.

[0089] Real-time data monitoring of spray gun and establishment of spray gun delay model: In the process of collecting real-time data of the spray gun on the track moving mechanism in the track-type rotary kiln, not only the position and time information of the spray gun is obtained, but also the vibration of the spray gun is monitored by installing a micro accelerometer, with a sampling frequency of 100 times per second. The vibration of the spray gun will cause the offset of the injection point and the instability of the injection amount. Combining these vibration data with other real-time data to analyze the actual working state of the spray gun can more accurately calculate the spray gun delay data. In addition, considering the influence of the airflow field inside the kiln body on the movement of the spray gun, the airflow speed and direction data are obtained by installing wind speed sensors at different positions in the kiln body, and the spray gun delay model considering the influence of airflow is established by combining the movement speed and injection direction of the spray gun. This model can more accurately reflect the delay of the spray gun actually reaching the preset injection point under complex working conditions, and provide a reliable basis for the subsequent precise control of the injection point and time of the spray gun, thereby helping to solve the problem of difficulty in accurately predicting the optimal ammonia injection amount and injection point due to spray gun delay.

[0090] Kiln partition monitoring and data processing optimization: Strengthening and dynamic adjustment of axial partition monitoring: a monitoring section is divided every 0.5m along the axial direction of the kiln body and monitoring is strengthened. Three groups of redundant thermocouples are arranged in each section to form a triangular temperature measurement array. An axial temperature field self-correction algorithm is developed. When the temperature difference between adjacent sections is greater than 100°C, the mobile infrared thermometer is activated for re-measurement and calibration, and a dynamic partition boundary mechanism is established to automatically adjust the high-temperature zone boundary according to the real-time thermal image. These measures make the monitoring of the temperature distribution in the kiln more accurate and flexible, and can better adapt to the temperature field changes caused by changes in working conditions such as kiln speed fluctuations, and provide more accurate temperature-related data support for the accurate prediction of the optimal ammonia injection amount and injection point.

[0091] Radial temperature gradient optimization and zoning collaborative control strategy: Optimization measures are taken for different radial temperature gradient areas (high-temperature core area, transition reaction area, and edge low-temperature area), such as deploying special sensors in the high-temperature core area and equipping them with a self-cleaning system, setting up a double-layer monitoring ring in the transition reaction area, and installing anti-condensation sensors in the edge low-temperature area and integrating heating and dehumidification modules. At the same time, different control strategies are formulated for each area, such as "temperature priority" control in the high-temperature core area, "optimal efficiency" mode in the transition reaction area, and "anti-escape" strategy in the edge low-temperature area. In addition, when the difference in sensor data within the partition is greater than 5%, the laser scanning temperature field verification is started, and abnormal fluctuations in the high-temperature area trigger the increase in the monitoring frequency of adjacent partitions. Through these partition monitoring optimization and collaborative control strategies, the amount of ammonia spray and the injection point can be adjusted more accurately according to the actual situation of different areas, effectively responding to the denitrification needs under complex working conditions such as kiln speed fluctuations and changes in fuel composition, and improving the denitrification efficiency and reducing agent utilization rate.

[0092] Optimization of gas error data calculation: When calculating the gas error data of each kiln partition, different monitoring parameter weights are set according to the degree of influence of different areas on the denitration reaction, such as giving a higher weight to NOx concentration data in the high-temperature core area. At the same time, when retrieving relevant parameters from the orbital rotary kiln denitration data and substituting them into the prediction model, the screening and preprocessing of historical data are increased, and similar operating condition data are screened and normalized according to the current operating parameters of the kiln body and fuel characteristics. These measures make the calculation of gas error data more consistent with actual working conditions, improve the accuracy of optimizing the prediction model based on gas error data, and thus help to more accurately predict the optimal ammonia injection amount and injection point.

[0093] Prediction model optimization and real-time online verification: Prediction model optimization based on reinforcement learning: When optimizing the preset orbital rotary kiln denitrification prediction model based on the gas error data and spray gun delay data of each kiln partition, an optimization strategy based on reinforcement learning is introduced. According to the actual denitrification effect of the spray gun at different injection points (assessed by real-time monitoring of the NOx concentration of the emission) and the ammonia escape situation, the model is given corresponding reward or penalty signals. The model continuously adjusts its prediction strategy based on these feedback signals to more accurately determine the optimal ammonia injection amount and injection point, so that it can better adapt to the denitrification needs under complex working conditions such as kiln speed fluctuations and changes in fuel composition.

[0094] Real-time online verification and model re-optimization: In order to make the optimized spray gun track movement control parameters better adapt to the changes in complex working conditions, after substituting the orbital rotary kiln denitrification task into the optimized orbital rotary kiln denitrification prediction model to obtain the optimized spray gun track movement control parameters, these parameters are verified online in real time. By setting additional monitoring points on the kiln body, the actual denitrification efficiency, ammonia escape rate, and the actual movement trajectory of the spray gun are monitored in real time. These actual monitoring data are compared with the results predicted based on the optimized parameters. If the deviation is found to exceed the set threshold, the model is optimized and adjusted again in time to improve the effectiveness and accuracy of the control parameters, so that accurate ammonia spraying amount and injection point control can always be achieved under complex working conditions, thereby improving the denitrification efficiency and reducing agent utilization.

[0095] Through the above comprehensive measures from data collection, model construction and optimization, spray gun monitoring, kiln zoning processing and real-time verification, the present invention can effectively solve the problem that it is difficult to accurately predict the optimal ammonia injection amount and injection point under complex working conditions such as kiln speed fluctuations and fuel composition changes, resulting in low reducing agent utilization (increased ammonia escape) or unstable denitrification efficiency.

Claims

1. A method for optimizing denitration of a track-type rotary kiln, characterized in that: include: Step S101, obtaining denitration data of a track-type rotary kiln; Step S102, based on the orbital rotary kiln denitration data, construct the kiln coordinate information, generate the kiln digital twin image, receive the orbital rotary kiln denitration task, substitute the orbital rotary kiln denitration task into the preset orbital rotary kiln denitration prediction model, obtain the orbital rotary kiln denitration prediction data, substitute the orbital rotary kiln denitration prediction data into the kiln digital twin image, and obtain the prediction information of the spray gun on the orbital moving mechanism; Step S103, collecting real-time data of the spray gun on the track moving mechanism in the track type rotary kiln to obtain real-time spray gun movement data, and comparing the real-time spray gun movement data with the prediction information of the spray gun on the track moving mechanism to obtain spray gun delay data; Step S104, establish kiln partitions in the digital twin image in the kiln, monitor the kiln partitions, obtain gas data of each kiln partition, substitute the fuel calorific value fluctuation coefficient, sulfur content time series change value and ash content ratio dynamic parameters under the optimal working condition in the orbital rotary kiln denitration data into the preset orbital rotary kiln denitration prediction model, obtain orbital rotary kiln denitration prediction gas data, compare the orbital rotary kiln denitration prediction gas data with the gas data of each kiln partition in turn, and obtain gas error data of each kiln partition; Step S105, based on the gas error data of each kiln partition and the spray gun delay data, the preset orbital rotary kiln denitrification prediction model is optimized, the orbital rotary kiln denitrification task is substituted into the optimized orbital rotary kiln denitrification prediction model, and the optimized spray gun orbital movement control parameters are obtained, and the optimized spray gun orbital movement control parameters are used as real-time execution data of the spray gun during the movement of the orbital movement mechanism.

2. According to the track type rotary kiln denitrification optimization method as claimed in claim 1, it is characterized in that: The step S101 includes: The denitrification data of the orbital rotary kiln include the operating parameters of the orbital rotary kiln, the temperature distribution data in the orbital rotary kiln, the NOx concentration data in the orbital rotary kiln, the oxygen content data in the orbital rotary kiln, the reducing agent injection data in the orbital rotary kiln, the spray gun status data in the orbital rotary kiln and the working condition data in the orbital rotary kiln; The speed fluctuation characteristics are collected by a vibration sensor pre-installed on the kiln drive gearbox to calculate the speed change rate; the axial temperature distribution is obtained at 0.5m intervals using a pre-installed distributed optical fiber temperature measurement system, and the sampling frequency is not less than 10Hz; the NOx concentration gradient is monitored in real time using a pre-installed laser absorption spectrometer; the reducing agent injection data includes ammonia concentration, atomization pressure and injection flow rate timing curves.

3. According to the track type rotary kiln denitrification optimization method as claimed in claim 1, it is characterized in that: The step S102 described in which the kiln coordinate information is constructed based on the orbital rotary kiln denitration data includes: The three-dimensional geometric model of the kiln body is established based on the point cloud scanning data, and the meshing accuracy is ≤2cm; 24 reference positioning RFID tags are set in the axial direction of the kiln body to establish the conversion relationship between the spatial coordinate system inside the kiln and the external polar coordinates; The historical operating condition data is retrieved from the orbital rotary kiln denitrification data, and the historical operating condition data is classified and stored by fuel type to build a denitrification efficiency mapping database including 2000 groups of operating conditions.

4. A method for optimizing denitration of a track-type rotary kiln according to claim 1, characterized in that: Substituting the denitration task of the orbital rotary kiln into the preset orbital rotary kiln denitration prediction model described in step S102 includes: collecting the current harmonics of the kiln drive motor, extracting data features of the current harmonics of the kiln drive motor, and obtaining the speed fluctuation characteristics; collecting the axial temperature field with a spatial resolution of 0.3m, monitoring the NOx concentration gradient, and aligning the speed fluctuation characteristics, the axial temperature field, the monitoring of the NOx concentration gradient and the position signal of the spray gun track encoder in time and space to establish a data matrix with time stamp synchronization; The model is trained based on the data corresponding to the data matrix synchronized with the timestamp to obtain the orbital rotary kiln denitrification prediction model.

5. According to the track type rotary kiln denitrification optimization method as claimed in claim 1, it is characterized in that: The step S104 of establishing kiln partitions in the digital twin image in the kiln includes: A monitoring section is divided every 0.5m along the axial direction of the kiln body, and the radial direction is divided into the following sections according to the temperature gradient: High temperature core zone (T≥1050℃), transition reaction zone (850℃≤T<1050℃), and edge low temperature zone (T<850℃).

6. A track type rotary kiln denitration optimization system, applied to a track type rotary kiln denitration optimization method as claimed in any one of claims 1 to 5, characterized in that: include: The server side, the data collection side and the rail-type rotary kiln denitration equipment side, the server side establishes communication connections with the data collection side and the rail-type rotary kiln denitration equipment side respectively; A data acquisition unit, used to acquire denitrification data of a track-type rotary kiln; A data analysis unit is used to construct kiln coordinate information based on the orbital rotary kiln denitration data, generate a digital twin image in the kiln, receive a denitration task of the orbital rotary kiln, substitute the denitration task of the orbital rotary kiln into a preset denitration prediction model of the orbital rotary kiln, obtain denitration prediction data of the orbital rotary kiln, substitute the denitration prediction data of the orbital rotary kiln into the digital twin image in the kiln, and obtain prediction information of the spray gun on the orbital moving mechanism; A data comparison unit is used to collect real-time data of the spray gun on the track moving mechanism in the track type rotary kiln, obtain real-time spray gun movement data, compare the real-time spray gun movement data with the prediction information of the spray gun on the track moving mechanism, and obtain spray gun delay data; The error analysis unit is used to establish kiln partitions in the digital twin image in the kiln, monitor the kiln partitions, obtain gas data of each kiln partition, substitute the fuel calorific value fluctuation coefficient, sulfur content time series change value and ash content ratio dynamic parameters under the optimal working condition in the orbital rotary kiln denitrification data into the preset orbital rotary kiln denitrification prediction model, obtain orbital rotary kiln denitrification prediction gas data, compare the orbital rotary kiln denitrification prediction gas data with the gas data of each kiln partition in turn, and obtain the gas error data of each kiln partition; The control optimization unit is used to optimize the preset orbital rotary kiln denitrification prediction model based on the gas error data of each kiln partition and the spray gun delay data, substitute the orbital rotary kiln denitrification task into the optimized orbital rotary kiln denitrification prediction model, obtain the optimized spray gun orbital movement control parameters, and use the optimized spray gun orbital movement control parameters as real-time execution data of the spray gun during the movement of the orbital movement mechanism.

7. A track type rotary kiln denitrification optimization system according to claim 6, characterized in that: The data acquisition unit is further used for: The denitrification data of the orbital rotary kiln include the operating parameters of the orbital rotary kiln, the temperature distribution data in the orbital rotary kiln, the NOx concentration data in the orbital rotary kiln, the oxygen content data in the orbital rotary kiln, the reducing agent injection data in the orbital rotary kiln, the spray gun status data in the orbital rotary kiln and the working condition data in the orbital rotary kiln; The speed fluctuation characteristics are collected by a vibration sensor pre-installed on the kiln drive gearbox to calculate the speed change rate; the axial temperature distribution is obtained at 0.5m intervals using a pre-installed distributed optical fiber temperature measurement system, and the sampling frequency is not less than 10Hz; the NOx concentration gradient is monitored in real time using a pre-installed laser absorption spectrometer; the reducing agent injection data includes ammonia concentration, atomization pressure and injection flow rate timing curves.

8. A track type rotary kiln denitrification optimization system according to claim 6, characterized in that: The data analysis unit is further used for: The three-dimensional geometric model of the kiln body is established based on the point cloud scanning data, and the meshing accuracy is ≤2cm; 24 reference positioning RFID tags are set in the axial direction of the kiln body to establish the conversion relationship between the spatial coordinate system inside the kiln and the external polar coordinates; The historical operating condition data is retrieved from the orbital rotary kiln denitrification data, and the historical operating condition data is classified and stored by fuel type to build a denitrification efficiency mapping database including 2000 groups of operating conditions.

9. A track type rotary kiln denitrification optimization system according to claim 6, characterized in that: The data analysis unit is further used to: collect current harmonics of the kiln drive motor, extract data features of the current harmonics of the kiln drive motor, and obtain speed fluctuation features; The axial temperature field is collected at a spatial resolution of 0.3m, the NOx concentration gradient is monitored, the speed fluctuation characteristics, the axial temperature field, the NOx concentration gradient and the position signal of the spray gun track encoder are aligned in time and space to establish a data matrix with time stamp synchronization; The model is trained based on the data corresponding to the data matrix synchronized with the timestamp to obtain the orbital rotary kiln denitrification prediction model.

10. A track type rotary kiln denitrification optimization system according to claim 6, characterized in that: The error analysis unit is further used for: A monitoring section is divided every 0.5m along the axial direction of the kiln body, and the radial direction is divided into the following sections according to the temperature gradient: High temperature core zone (T≥1050℃), transition reaction zone (850℃≤T<1050℃), and edge low temperature zone (T<850℃).

Images