An optimization method and system for denitrification of an orbital rotary kiln

Abstract

The present invention relates to the technical field of data processing. An optimized method and system for denitrification of an orbital rotary kiln provided by the present invention aims to solve the problems that it is difficult to accurately predict the ammonia injection amount and injection points under complex working conditions such as fluctuations in the kiln body rotation 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 and comparing to obtain delay data, establishing zonal 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. Through measures such as multi-faceted data collection, processing, model optimization, and spray gun control, the present invention effectively improves the denitrification efficiency and the utilization rate of the reducing agent, accurately responds to complex working conditions, and realizes stable and efficient denitrification.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to an optimization method and system for denitrification of an orbital rotary kiln. Background Art

[0002] Denitrification of an orbital rotary kiln is an environmental protection technology for controlling the emission of nitrogen oxides (NOx) in the flue gas of industrial kilns, and is mainly applied to the high-temperature production processes in industries such as cement, metallurgy, and chemical industry. Its core principle is to utilize the unique rotating structure and high-temperature environment of the rotary kiln, and inject reducing agents such as ammonia water and urea at specific positions (such as the pre-decomposition zone or the flue) of the kiln body. Through selective non-catalytic reduction (SNCR) or SCR technology combined with a catalyst, the NOx in the flue gas is reduced to harmless nitrogen and water.

[0003] The data processing process of denitrification of an orbital rotary kiln follows a logical chain of monitoring, analysis, and feedback: First, key parameters (such as the temperature distribution in the kiln, NOx concentration, oxygen content, and reducing agent injection amount) are collected in real time through temperature sensors, flue gas composition analyzers, and flow meters on the kiln body, and the data is synchronously transmitted to the central control system; Subsequently, based on the preset denitrification efficiency target, combined with a thermodynamic model or a machine learning algorithm, the efficiency threshold of the reaction between the reducing agent and NOx under the current working conditions is deduced, and the influence of temperature fluctuations or concentration deviations on the reaction path is identified; Then, by comparing the actual emission value with the preset standard, the injection position (moving the spray gun along the track) and injection amount of the reducing agent are dynamically adjusted to keep the reaction zone always in the optimal temperature window (usually 850-1100 °C). At the same time, a prediction model is trained using historical data to optimize the ammonia injection strategy to reduce ammonia slip; Finally, the processed real-time data and process adjustment instructions form a closed-loop feedback to continuously reduce NOx emissions, and the balance relationship between the emission reduction effect and energy consumption is verified through a data visualization platform.

[0004] During the actual data processing of denitrification of an orbital rotary kiln, the high-temperature and dusty environment in the orbital rotary kiln easily causes data drift or distortion of temperature sensors and flue gas composition detectors, and the transient changes in the working conditions when the spray gun moves along the track result in a lag in the real-time monitoring of key parameters (such as local temperature and NOx concentration). Existing thermodynamic models or data-driven algorithms are insufficient in fitting the non-linear and strongly coupled reaction process in the kiln body. Especially in complex working conditions such as fluctuations in the kiln body rotation speed and changes in fuel composition, it is difficult to accurately predict the optimal ammonia injection amount and injection position, resulting in low utilization rate of the reducing agent (increased ammonia slip) or unstable denitrification efficiency. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the present invention provides an optimized method and system for denitrification of an orbital rotary kiln, which solves the problems that it is difficult to accurately predict the optimal ammonia injection amount and injection point under complex working conditions such as the fluctuation of the kiln body speed and the change of fuel composition, resulting in low utilization rate of the reducing agent (increase in ammonia escape) or unstable denitrification efficiency.

[0006] To solve the above technical problems, the specific technical solutions of the present invention are as follows:

[0007] In the first aspect, an optimized method for denitrification of an orbital rotary kiln provided by the present invention includes:

[0008] Step S101: Obtain the denitrification data of the orbital rotary kiln. The denitrification data of the orbital rotary kiln includes the operating parameters of the orbital rotary kiln body, the temperature distribution data inside the orbital rotary kiln, the NOx concentration data inside the orbital rotary kiln, the oxygen content data inside the orbital rotary kiln, the reducing agent injection data inside the orbital rotary kiln, the spray gun state data inside the orbital rotary kiln, and the working condition data inside the orbital rotary kiln;

[0009] Step S102: Based on the denitrification data of the orbital rotary kiln, establish a reference position point inside the orbital rotary kiln, configure coordinates with the reference position point inside the kiln to obtain the coordinate information inside the kiln, substitute the coordinate information inside the kiln into a preset digital twin model to generate a digital twin image inside the kiln, receive the denitrification task of the orbital rotary kiln, substitute the denitrification task of the orbital rotary kiln into a preset denitrification prediction model of the orbital rotary kiln to obtain the denitrification prediction data of the orbital rotary kiln, and substitute the denitrification prediction data of the orbital rotary kiln into the digital twin image inside the kiln to obtain the prediction information of the spray gun on the rail moving mechanism. The prediction information of the spray gun on the rail moving mechanism includes the predicted coordinates of the spray gun on the rail and the predicted time information corresponding to the coordinates;

[0010] Step S103: Collect the real-time data of the spray gun on the rail moving mechanism inside the orbital rotary kiln to obtain the real-time spray gun movement data. The real-time spray gun movement data includes the real-time rail 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 rail moving mechanism to obtain the spray gun delay data;

[0011] Step S104: Establish zones inside the kiln in the digital twin image of the kiln to obtain the zones inside the kiln. Monitor the zones inside the kiln to obtain the gas data of each zone inside the kiln. Retrieve data from the denitrification data of the orbital rotary kiln to obtain the fuel calorific value fluctuation coefficient, sulfur content time-series change value, and ash content ratio dynamic parameters under the optimal operating conditions inside the kiln. Substitute the fuel calorific value fluctuation coefficient, sulfur content time-series change value, and ash content ratio dynamic parameters under the optimal operating conditions inside the kiln into the preset denitrification prediction model of the orbital rotary kiln to obtain the denitrification prediction gas data of the orbital rotary kiln. Compare the denitrification prediction gas data of the orbital rotary kiln with the gas data of each zone inside the kiln in sequence to obtain the gas error data of each zone inside the kiln;

[0012] Step S105: Optimize the preset denitrification prediction model of the orbital rotary kiln based on the gas error data of each zone inside the kiln and the spray gun delay data to obtain the optimized denitrification prediction model of the orbital rotary kiln. Substitute the denitrification task of the orbital rotary kiln into the optimized denitrification prediction model of the orbital rotary kiln to obtain the optimized spray gun rail movement control parameters. The spray gun rail movement control parameters include the spray gun rail movement speed parameter, ammonia injection amount, and injection point position. Use the optimized spray gun rail movement control parameters as the real-time execution data during the movement of the spray gun on the rail movement mechanism.

[0013] Further, for an orbital rotary kiln denitrification optimization method of the present invention, the step S101 includes:

[0014] Collect the rotational speed fluctuation characteristics through a vibration sensor pre-installed in the kiln body drive gearbox and calculate the rotational speed change rate; use the pre-installed distributed optical fiber temperature measurement system to obtain the axial temperature distribution at intervals of 0.5 m, with a sampling frequency not lower than 10 Hz; use the pre-installed laser absorption spectrometer to monitor the NOx concentration gradient in real time; the reductant injection data includes the ammonia water concentration, atomization pressure, and injection flow time-series curve.

[0015] Further, for an orbital rotary kiln denitrification optimization method of the present invention, the construction of the coordinate information inside the kiln in the step S102 includes:

[0016] Establish a three-dimensional geometric model of the kiln body based on the point cloud scan data, with a mesh division accuracy ≤ 2 cm;

[0017] Set 24 reference positioning RFID tags axially on the kiln body to establish the conversion relationship between the spatial coordinate system inside the kiln and the external polar coordinate system;

[0018] Retrieve the historical operating condition data from the denitrification data of the orbital rotary kiln, classify and store the historical operating condition data according to the fuel type, and construct a denitrification efficiency mapping database including 2000 groups of operating conditions.

[0019] Further, in an optimized method for denitrification of an orbital rotary kiln according to the present invention, in step S102, substituting the denitrification task of the orbital rotary kiln into a preset denitrification prediction model of the orbital rotary kiln includes: collecting current harmonics of the kiln body drive motor, extracting data features from the current harmonics of the kiln body drive motor to obtain rotational speed fluctuation features; collecting the axial temperature field at a spatial resolution of 0.3 m, monitoring the NOx concentration gradient, aligning the rotational speed fluctuation features, the axial temperature field, the monitored NOx concentration gradient with the position signal of the spray gun track encoder in space and time, and establishing a data matrix with timestamp synchronization.

[0020] Training the model based on the data corresponding to the data matrix with timestamp synchronization to obtain a denitrification prediction model of the orbital rotary kiln.

[0021] Further, in an optimized method for denitrification of an orbital rotary kiln according to the present invention, in step S104, partitioning the inside of the kiln in the digital twin image of the kiln includes:

[0022] Dividing a monitoring section every 0.5 m along the axial direction of the kiln body, and radially dividing it according to the temperature gradient into:

[0023] A high-temperature core area (T≥1050 °C), a transition reaction area (850 °C≤T<1050 °C), and an edge low-temperature area (T<850 °C); the calculation of the gas error data uses: . Where is the comprehensive gas error index of the i-th partition, is the contribution weight of the NOx concentration error, is the contribution weight of the oxygen content error, is the predicted NOx volume concentration, is the measured NOx volume concentration, is the NOx concentration normalization reference value, is the predicted oxygen volume fraction, is the measured oxygen volume fraction, is the oxygen content normalization reference value.

[0024] In a second aspect, the present invention provides an optimized system for denitrification of an orbital rotary kiln, which applies an optimized method for denitrification of an orbital rotary kiln as described above, and includes: a server side, a data acquisition side, and an orbital rotary kiln denitrification equipment side, and the server side establishes communication connections with the data acquisition side and the orbital rotary kiln denitrification equipment side respectively;

[0025] A data acquisition unit for acquiring denitrification data of an orbital rotary kiln. The denitrification data of the orbital rotary kiln includes the operating parameters of the orbital rotary kiln body, the temperature distribution data inside the orbital rotary kiln, the NOx concentration data inside the orbital rotary kiln, the oxygen content data inside the orbital rotary kiln, the reductant injection data inside the orbital rotary kiln, the spray gun status data inside the orbital rotary kiln, and the operating condition data inside the orbital rotary kiln;

[0026] A data analysis unit for establishing a reference position point inside the orbital rotary kiln, configuring coordinates based on the reference position point inside the kiln to obtain the coordinate information inside the kiln, substituting the coordinate information inside the kiln into a preset digital twin model to generate a digital twin image inside the kiln, receiving the denitrification task of the orbital rotary kiln, substituting the denitrification task of the orbital rotary kiln into a preset denitrification prediction model of the orbital rotary kiln to obtain denitrification prediction data of the orbital rotary kiln, and substituting the denitrification prediction data of the orbital rotary kiln into the digital twin image inside the kiln to obtain the prediction information of the spray gun on the rail moving mechanism. The prediction information of the spray gun on the rail moving mechanism includes the predicted coordinates of the spray gun on the rail and the predicted time information corresponding to the coordinates;

[0027] A data comparison unit for collecting the real-time data of the spray gun on the rail moving mechanism inside the orbital rotary kiln to obtain real-time spray gun movement data. The real-time spray gun movement data includes the real-time rail 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 rail moving mechanism to obtain spray gun delay data;

[0028] An error analysis unit for partitioning the inside of the kiln in the digital twin image inside the kiln to obtain the partitions inside the kiln, monitoring the partitions inside the kiln to obtain the gas data of each partition inside the kiln, retrieving data from the denitrification data of the orbital rotary kiln to obtain the fuel calorific value fluctuation coefficient, sulfur content time series change value, and ash content ratio dynamic parameters under the optimal operating condition inside the kiln, substituting the fuel calorific value fluctuation coefficient, sulfur content time series change value, and ash content ratio dynamic parameters under the optimal operating condition inside the kiln into a preset denitrification prediction model of the orbital rotary kiln to obtain denitrification prediction gas data of the orbital rotary kiln, and comparing the denitrification prediction gas data of the orbital rotary kiln with the gas data of each partition inside the kiln in sequence to obtain the gas error data of each partition inside the kiln;

[0029] A control optimization unit for optimizing a preset denitrification prediction model of the orbital rotary kiln based on the gas error data of each partition inside the kiln and the spray gun delay data to obtain an optimized denitrification prediction model of the orbital rotary kiln, substituting the denitrification task of the orbital rotary kiln into the optimized denitrification prediction model of the orbital rotary kiln to obtain optimized spray gun rail movement control parameters. The spray gun rail movement control parameters include the spray gun rail movement speed parameter, ammonia injection amount, and injection point, and using the optimized spray gun rail movement control parameters as the real-time execution data during the movement of the spray gun on the rail moving mechanism.

[0030] Further, for an optimized denitration system of an orbital rotary kiln according to the present invention, the data acquisition unit is further configured to:

[0031] Collect the rotational speed fluctuation characteristics through vibration sensors pre-installed in the kiln body drive gearbox, and calculate the rotational speed change rate; use the pre-installed distributed optical fiber temperature measurement system to obtain the axial temperature distribution at intervals of 0.5 m, and the sampling frequency is not less than 10 Hz; use the pre-installed laser absorption spectrometer to monitor the NOx concentration gradient in real time; the reductant injection data includes the ammonia concentration, atomization pressure, and injection flow time series curve.

[0032] Further, for an optimized denitration system of an orbital rotary kiln according to the present invention, the data analysis unit is further configured to:

[0033] Establish a three-dimensional geometric model of the kiln body based on the point cloud scan data, and the mesh division accuracy ≤ 2 cm;

[0034] Set 24 reference positioning RFID tags axially on the kiln body, and establish the conversion relationship between the internal space coordinate system of the kiln and the external polar coordinates;

[0035] Retrieve historical operating condition data from the denitration data of the orbital rotary kiln, classify and store the historical operating condition data according to the fuel type, and construct a denitration efficiency mapping database including 2000 groups of operating conditions.

[0036] Further, for an optimized denitration system of an orbital rotary kiln according to the present invention, the data analysis unit is further configured to: collect the current harmonics of the kiln body drive motor, extract the data characteristics of the current harmonics of the kiln body drive motor to obtain the rotational speed fluctuation characteristics; collect the axial temperature field with a spatial resolution of 0.3 m, monitor the NOx concentration gradient, and align the rotational speed fluctuation characteristics, axial temperature field, and monitored NOx concentration gradient with the position signal of the spray gun track encoder in space-time to establish a data matrix with time stamp synchronization;

[0037] Train the model based on the data pairs corresponding to the data matrix with time stamp synchronization to obtain an optimized denitration prediction model for the orbital rotary kiln.

[0038] Further, for an optimized denitration system of an orbital rotary kiln according to the present invention, the error analysis unit is further configured to:

[0039] Divide a monitoring section every 0.5 m axially along the kiln body, and radially divide it according to the temperature gradient into:

[0040] A high-temperature core area (T ≥ 1050 °C), a transition reaction area (850 °C ≤ T < 1050 °C), and an edge low-temperature area (T < 850 °C); the gas error data is calculated using: . Wherein is the comprehensive gas error index for the i-th partition, is the contribution weight of the NOx concentration error, is the contribution weight of the oxygen content error, is the predicted NOx volume concentration, is the measured NOx volume concentration, is the NOx concentration normalization reference value, is the predicted oxygen volume fraction, is the measured oxygen volume fraction, is the oxygen content normalization reference value.

[0041] Advantages of the present invention:

[0042] Improve the prediction accuracy of ammonia injection quantity and injection points: Through comprehensive and refined data collection means, such as enhancing the collection of rotational speed fluctuation characteristics, optimizing the temperature distribution and NOx concentration monitoring, and monitoring all parameters of the reductant injection, etc., various operating conditions in the kiln can be grasped more accurately. Combining the precise construction and analysis of the digital twin model, the model training optimization based on current harmonics and temperature field, and the optimization of in-kiln partition monitoring and data processing, etc., it is still possible to accurately predict the optimal ammonia injection quantity and injection points under complex operating conditions such as kiln body rotational speed fluctuation and fuel composition change, effectively solving the problem of inaccurate prediction of traditional methods under these complex operating conditions.

[0043] Optimize the spray gun control: By means 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 considering the influence of the airflow field inside the kiln body, so as to achieve more accurate control of the spray gun movement trajectory and injection time, ensure that the reductant is injected at the appropriate time and position, and further improve the denitrification effect.

[0044] Improve the denitrification efficiency: By accurately predicting the optimal ammonia injection quantity and injection points, the reductant can react with NOx more fully and effectively, reducing NOx in the flue gas to harmless nitrogen and water, thereby improving the denitrification efficiency, making the finally emitted NOx concentration meet more stringent environmental protection standards, effectively reducing the emissions of nitrogen oxides in the flue gas of industrial kilns, and being of great significance to environmental protection.

[0045] Improve the utilization rate of the reductant: Precise control of the ammonia injection quantity and optimization of the injection points avoid the phenomenon of ammonia escape caused by excessive ammonia injection, enable the reductant to fully participate in the denitrification reaction, improve the utilization rate of the reductant, reduce the production cost, and at the same time reduce the possible adverse effects of ammonia escape on the environment and equipment.

[0046] Adapt to the rotational speed fluctuation of the kiln body: From the multi-dimensional monitoring and compensation of the rotational speed fluctuation in the data acquisition stage, to constructing a prediction model that can adapt to the rotational speed change based on data such as current harmonics, and then to measures such as monitoring different zones in the kiln and adjusting the temperature zone boundary according to the rotational speed fluctuation, enabling the entire denitration system to well adapt to the rotational speed fluctuation of the kiln body and ensuring the stability and efficiency of the denitration process under different rotational speed conditions.

[0047] Cope with the change of 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 the data processing process, etc., the system can quickly and accurately adapt to the influence brought by the change of fuel composition and ensure accurate denitration control under different fuel conditions.

[0048] In summary, the present invention has significant beneficial effects in aspects such as precise denitration control, coping with complex working conditions, improving denitration efficiency and reductant utilization rate, optimizing data processing and models, and system operation and maintenance, and has important application value and popularization significance for the denitration treatment of industrial kilns. Brief Description of the Drawings

[0049] In order to more clearly illustrate the technical solutions of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, for those of ordinary skill in the art, other drawings can also be obtained according to the drawings without creative efforts.

[0050] Figure 1 It is a schematic diagram of the flow method of the denitration optimization method for the orbital rotary kiln provided by the embodiment of the present invention. Detailed Embodiments

[0051] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the specific embodiments of the present invention and the corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention. The technical solutions provided by each embodiment of the present invention will be described in detail below with reference to the drawings.

[0052] To better understand the objectives of the present invention, the present invention will be further described in detail below.

[0053] In the first aspect, a denitration optimization method for an orbital rotary kiln provided by the present invention includes:

[0054] Step S101: Obtain the denitration data of the orbital rotary kiln. The denitration data of the orbital rotary kiln includes the operating parameters of the orbital rotary kiln body, the temperature distribution data inside the orbital rotary kiln, the NOx concentration data inside the orbital rotary kiln, the oxygen content data inside the orbital rotary kiln, the reductant injection data inside the orbital rotary kiln, the spray gun status data inside the orbital rotary kiln, and the operating condition data inside the orbital rotary kiln.

[0055] The obtaining of the denitration data of the orbital rotary kiln in Step S101 includes the following data collection methods:

[0056] The operating parameters of the orbital rotary kiln body include:

[0057] Rotation speed data: Monitor the real-time rotation speed of the kiln body and analyze the influence of rotation speed fluctuations on the residence time of materials, temperature distribution, and NOx generation.

[0058] Fuel composition data: Include fuel type (such as pulverized coal, natural gas), calorific value, sulfur content, ash content, etc., and are used to predict the NOx generation amount and temperature change trend during the combustion process.

[0059] The temperature distribution data inside the orbital rotary kiln: Obtain the temperature gradient of key areas such as the pre-decomposition zone and the firing zone through multi-point temperature sensors, and identify the optimal reaction temperature window (850 - 1100 °C).

[0060] The NOx concentration inside the orbital rotary kiln: Real-time monitor the NOx concentration in the flue gas (including NO, NO 2 ) as the core index of denitration efficiency.

[0061] The oxygen content inside the orbital rotary kiln: Judge the sufficiency of combustion and the reduction reaction conditions, and avoid the interference of excessive oxygen on the denitration reaction.

[0062] Also obtain other gas components: such as the concentrations of CO, CO 2 , SO 2 to assist in analyzing the combustion efficiency and its potential impact on the denitration reaction.

[0063] The reductant injection data inside the orbital rotary kiln includes:

[0064] Injection volume: Real-time record the flow rate of ammonia water or urea, and calculate the actual amount of reductant participating in the reaction in combination with the concentration data.

[0065] Injection point position: The position information of the spray gun moving along the track, which is used to match the dynamic coverage of the reaction zone under the rotation state of the kiln body.

[0066] Injection pressure / atomization effect: Affect the mixing uniformity of the reductant and the flue gas and the reaction efficiency.

[0067] The spray gun status data inside the orbital rotary kiln includes:

[0068] Moving trajectory and speed: Monitor the moving path and speed of the spray gun along the track to synchronize the spraying points with the rotation of the kiln body.

[0069] Start-stop state: Record the working cycle of the spray gun to avoid ineffective spraying or redundant operations.

[0070] The operating condition data in the orbital rotary kiln includes:

[0071] Pressure inside the kiln: Affects the gas flow direction and the distribution of the reducing agent, and needs to be jointly modeled with temperature and rotation speed.

[0072] Material flow rate and composition: Monitor the input quantity and composition changes of raw meal or fuel to predict fluctuations in NOx generation.

[0073] Ambient temperature and humidity: The potential impact of the external environment on the heat dissipation of the kiln body and the stability of sensor data.

[0074] The denitration data of the orbital rotary kiln includes:

[0075] Historical operation records: Include historical data such as rotation speed, fuel ratio, ammonia injection volume, etc., which are used to train the machine learning model.

[0076] Ammonia slip concentration: Detect the unreacted NH 3 concentration through an additional sensor to evaluate the utilization rate of the reducing agent.

[0077] Emission compliance data: The NOx concentration of the final flue gas emission, which is used to optimize the closed-loop feedback control strategy.

[0078] In step S102, based on the denitration data of the orbital rotary kiln, establish a reference position point inside the kiln, configure coordinates with the reference position point inside the kiln to obtain the coordinate information inside the kiln, substitute the coordinate information inside the kiln into the preset digital twin model to generate the digital twin image inside 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 to obtain the denitration prediction data of the orbital rotary kiln, and substitute the denitration prediction data of the orbital rotary kiln into the digital twin image inside the kiln to obtain the prediction information of the spray gun on the track moving mechanism. 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;

[0079] In step S102, the construction of the coordinate information inside the kiln based on the denitration data of the orbital rotary kiln includes the following specific implementation methods:

[0080] When establishing a three-dimensional geometric model of the kiln body based on point cloud scanning data, in addition to the grid division accuracy ≤ 2 cm, it is also necessary to regularly update the point cloud scanning data according to the actual operation conditions of the kiln body. For example, a comprehensive point cloud scan is carried out once a week to capture possible structural deformations, lining wear, etc. that may occur during the long-term operation of the kiln body. The updated point cloud scanning data will be used to correct the three-dimensional geometric model, making the model more accurately reflect the actual state of the kiln body, and thus improving the accuracy of subsequent analysis and prediction based on this model.

[0081] For setting 24 reference positioning RFID tags axially on the kiln body to establish the conversion relationship between the internal space coordinate system of the kiln and the external polar coordinates, regularly check the reading accuracy and signal strength of the RFID tags. If it is found that a tag reads abnormally or the signal strength is lower than the set threshold, replace it or adjust its position in time so that the accuracy of the space coordinate system conversion relationship is not affected.

[0082] In terms of retrieving historical operating condition data, classifying and storing it according to fuel types to construct a denitration efficiency mapping database, further deeply mine the historical operating condition data. In addition to classifying according to fuel types, it is also subdivided according to factors such as different seasons and different production batches. Because environmental conditions such as temperature and humidity are different in different seasons, and the raw material quality of different production batches may also vary, these factors will all affect the denitration process. Through this more detailed classification and storage, when using the database for analysis and prediction subsequently, similar operating conditions can be more accurately matched, improving the accuracy of the denitration prediction model. The preset orbital rotary kiln denitration prediction model is trained based on historical operating condition data and real-time monitoring data.

[0083] Substituting the orbital rotary kiln denitration task into the preset orbital rotary kiln denitration prediction model described in step S102 includes the following model training and optimization methods:

[0084] In terms of collecting the current harmonics of the kiln body drive motor, extracting data features to obtain rotational speed fluctuation features and related data processing, increase the filtering processing of the current harmonic data. Use an adaptive filter to automatically adjust the filtering parameters according to the real-time state of the kiln body operation, remove the noise interference in the current harmonic data, and improve the accuracy of rotational speed fluctuation feature extraction. At the same time, when establishing a data matrix with time stamp synchronization by spatio-temporally aligning the rotational speed fluctuation features, axial temperature field, monitored NOx concentration gradient and the position signal of the spray gun track encoder, introduce a data interpolation algorithm. For situations where the time intervals may be inconsistent or data is missing during the collection process, use the interpolation algorithm 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.

[0085] Step S103: Collect the real-time data of the spray gun on the rail moving mechanism in the orbital rotary kiln to obtain the real-time spray gun movement data. The real-time spray gun movement data includes the real-time rail coordinates of the spray gun and the corresponding real-time time information. Compare the real-time spray gun movement data with the predicted information of the spray gun on the rail moving mechanism to obtain the spray gun delay data;

[0086] During the process of collecting the real-time data of the spray gun on the rail moving mechanism in the orbital rotary kiln, obtain the real-time rail coordinates of the spray gun and the corresponding real-time time information, and monitor the vibration condition of the spray gun. By installing a micro acceleration sensor on the spray gun, collect the vibration acceleration data of the spray gun in real time, and the sampling frequency is 100 times per second. The vibration of the spray gun will cause the deviation of the spraying point and the instability of the spraying amount. These vibration data will be used together with other real-time data to analyze the actual working state of the spray gun and calculate the spray gun delay data more accurately.

[0087] 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 rail moving mechanism, also consider the influence of the internal gas flow field of the kiln body on the movement of the spray gun. By installing wind speed sensors at different positions in the kiln, obtain the real-time data of the gas flow velocity and direction in the kiln, and combine the movement speed, spraying direction of the spray gun and the gas flow field data to establish a spray gun delay model considering the influence of the gas flow. The spray gun delay model will be able to more accurately reflect the delay situation of the spray gun actually reaching the preset spraying point under complex working conditions, and provide a more reliable basis for subsequent optimization control. The spray gun delay model is established through the following steps: collect the gas flow velocity and direction data in the kiln, combine the movement speed and spraying direction of the spray gun, calculate the offset of the gas flow on the spray gun trajectory, and generate delay compensation parameters based on the offset.

[0088] Step S104: Establish partitions in the kiln in the digital twin image of the kiln to obtain the kiln partitions. Monitor the kiln partitions to obtain the gas data of each kiln partition. Retrieve data in the denitration data of the orbital rotary kiln to obtain 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 kiln. 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 kiln into the preset denitration prediction model of the orbital rotary kiln to obtain the denitration prediction gas data of the orbital rotary kiln. Compare the denitration prediction gas data of the orbital rotary kiln with the gas data of each kiln partition in turn to obtain the gas error data of each kiln partition;

[0089] After dividing each 0.5 m along the axial direction of the kiln body into a monitoring section and dividing different regions according to the temperature gradient (high-temperature core region, transition reaction region, edge low-temperature region), different monitoring parameter weights are further set for each region. For example, in the high-temperature core region, due to its key influence on the denitration 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 relevant gas data is set to 0.1. In the edge low-temperature region, the monitoring weights of each gas data are adjusted accordingly to better conform to the actual situation of the influence degree of this region on the denitration reaction.

[0090] In terms of retrieving the fuel calorific value fluctuation coefficient, sulfur content time-series change value, and ash content ratio dynamic parameters under the optimal operating conditions in the kiln from the denitration data of the orbital rotary kiln and substituting them into the preset orbital rotary kiln denitration prediction model, the screening and preprocessing of historical data are increased. According to the current operating parameters of the kiln body (such as rotation speed, temperature, etc.) and the basic characteristics of the fuel (such as calorific value range, sulfur content range, etc.), a set of data most similar to the current operating conditions is screened out from the historical data. The screened data is normalized to make it more conform to the input requirements of the current model in terms of numerical range and data distribution, improving the accuracy of model prediction.

[0091] Step S105, optimize the preset orbital rotary kiln denitration prediction model based on the gas error data and spray gun delay data in each kiln inner partition to obtain the optimized orbital rotary kiln denitration prediction model. Substitute the orbital rotary kiln denitration task into the optimized orbital rotary kiln denitration prediction model to obtain the optimized spray gun rail movement control parameters. The spray gun rail movement control parameters include the spray gun rail movement speed parameter, ammonia injection amount, and injection point. Use the optimized spray gun rail movement control parameters as the real-time execution data during the movement of the spray gun in the rail movement mechanism.

[0092] When optimizing the preset orbital rotary kiln denitration prediction model based on the gas error data and spray gun delay data in each kiln inner partition, in addition to the conventional model parameter adjustment methods, an optimization strategy based on reinforcement learning is introduced. According to the actual denitration effect (evaluated by real-time monitoring of the NOx concentration emitted) and ammonia slip situation of the spray gun at different injection points, the model is given corresponding reward or punishment signals. The orbital rotary kiln denitration prediction model continuously adjusts its prediction strategy according to these feedback signals to more accurately determine the optimal ammonia injection amount and injection point.

[0093] In order to enable the optimized control parameters of the spray gun track movement to better adapt to the changes in complex working conditions, after substituting the denitration task of the orbital rotary kiln into the optimized denitration prediction model of the orbital rotary kiln to obtain the optimized control parameters of the spray gun track movement, these parameters are verified in real time online. By setting additional monitoring points on the kiln body, information such as the actual denitration efficiency, ammonia escape rate, and the actual movement trajectory of the spray gun is monitored in real time. The actual monitoring data is compared with the predicted results based on the optimized parameters. If the deviation is found to exceed the set threshold, the model is optimized and adjusted again in a timely manner to improve the effectiveness and accuracy of the control parameters.

[0094] Specifically, for the denitration optimization method of the orbital rotary kiln described in the present invention, the step S101 includes:

[0095] Collect the rotational speed fluctuation characteristics through the vibration sensors pre-installed in the kiln body drive gearbox, and calculate the rotational speed change rate; use the pre-installed distributed optical fiber temperature measurement system to obtain the axial temperature distribution at intervals of 0.5 m, and the sampling frequency is not less than 10 Hz; use the pre-installed laser absorption spectrometer to monitor the NOx concentration gradient in real time; the reducing agent injection data includes the ammonia water concentration, atomization pressure, and injection flow time series curve.

[0096] Enhanced acquisition of rotational speed fluctuation characteristics:

[0097] Install a three-axis MEMS vibration sensor group (X / Y / Z directions) symmetrically on the drive gearbox, and extract the rotational speed harmonic components of 1X to 5X through order analysis;

[0098] Establish a rotational speed anomaly detection mechanism: when the volatility in three consecutive sampling periods > 2%, trigger the Hall sensor auxiliary verification mode;

[0099] Increase the kiln body inertia compensation parameter to eliminate the instantaneous rotational speed jitter interference caused by uneven material distribution.

[0100] Improve the reliability of temperature distribution data:

[0101] The distributed optical fiber adopts a double-layer armored structure (inner layer Φ0.25 mm sensing optical fiber, outer layer Φ1 mm reference optical fiber);

[0102] Set the temperature mutation verification rule: when the temperature difference between adjacent temperature measurement points > 80 °C, activate the infrared thermal imager for regional recheck;

[0103] Add redundant thermocouple arrays (K type, response time < 0.5 s) at the kiln head / kiln tail, and perform spatio-temporal fusion with the optical fiber data.

[0104] Optimization of dynamic monitoring of NOx concentration:

[0105] The laser absorption spectrometer integrates a self-cleaning optical window (pulse back-blowing period ≤ 30 s) to prevent dust adhesion interference;

[0106] Establish a concentration gradient verification mechanism: when the concentration difference between axially adjacent monitoring points > 150 ppm, start the retest of the mobile sampling probe;

[0107] Deploy a multi-spectral fusion algorithm to synchronously analyze the NO / NO 2 Concentration ratio (accuracy ± 3%).

[0108] Full-parameter monitoring of the reductant injection:

[0109] Install an ultrasonic concentration meter (accuracy ± 0.5%) and a Coriolis mass flow meter (accuracy ± 0.2%) on the ammonia water delivery pipeline;

[0110] Set the injection pulse width modulation control: when the atomization pressure fluctuation > 0.05 MPa, automatically adjust the opening of the pressure stabilizing valve;

[0111] Real-time monitor the atomization cone angle of the spray gun (target range 30° - 45°), and feedback to adjust the gas-liquid ratio through the high-speed camera system.

[0112] Specifically, for an optimized method for denitrification of an orbital rotary kiln according to the present invention, in step S102, the construction of the in-kiln coordinate information based on the denitrification data of the orbital rotary kiln includes:

[0113] Establish a three-dimensional geometric model of the kiln body based on the point cloud scanning data, with a mesh division accuracy ≤ 2 cm;

[0114] Set 24 reference positioning RFID tags axially on the kiln body to establish the conversion relationship between the in-kiln space coordinate system and the external polar coordinate system;

[0115] Retrieve historical operating condition data from the denitrification data of the orbital rotary kiln, classify and store the historical operating condition data according to the fuel type, and construct a denitrification efficiency mapping database including 2000 groups of operating conditions.

[0116] Point cloud data dynamic update mechanism: Deploy a mobile laser scanning robot to perform a full-kiln scan along the kiln body track every 72 hours;

[0117] Establish a point cloud change rate monitoring model: when the local point cloud offset > 1.5 cm in three consecutive scans, trigger the kiln body deformation warning;

[0118] Adopt the ICP algorithm to realize the registration of new and old point cloud data and automatically correct the geometric deviation of the three-dimensional model.

[0119] Intelligent calibration of the space coordinate system:

[0120] Each RFID tag integrates a temperature compensation module (operating range 0 - 150 °C) to eliminate positioning errors caused by thermal expansion;

[0121] A dual - frequency RFID reader - writer (865 MHz / 2.4 GHz) is set up to automatically switch to an anti - interference frequency band when the kiln body rotates;

[0122] A six - degree - of - freedom coordinate transformation matrix is constructed, including kiln body tilt compensation parameters (accuracy 0.01 ° in the range of ±5 °).

[0123] Optimization strategy for the operating condition database:

[0124] Establish a five - level fuel classification system: coal / refuse - derived fuel / biomass co - firing / petroleum coke / emergency fuel;

[0125] Deploy a data cleaning pipeline: eliminate sensor outliers (3σ criterion), fill in missing data (KNN interpolation method);

[0126] Create an operating condition feature fingerprint library: each type of fuel is associated with a 128 - dimensional feature vector (including calorific value fluctuation pattern, sulfur release curve, etc.).

[0127] Specifically, for the orbital rotary kiln denitration optimization method described in the present invention, substituting the orbital rotary kiln denitration task into the preset orbital rotary kiln denitration prediction model in step S102 includes: collecting the current harmonics of the kiln body drive motor, extracting data features from the current harmonics of the kiln body drive motor to obtain rotational speed fluctuation features; collecting the axial temperature field with a spatial resolution of 0.3 m, monitoring the NOx concentration gradient, aligning the rotational speed fluctuation features, axial temperature field, monitored NOx concentration gradient with the position signal of the spray gun rail encoder in space - time, using the IEEE 1588 precision clock protocol to unify the time reference, and establishing a data matrix with timestamp synchronization;

[0128] Training the model based on the data corresponding to the data matrix with timestamp synchronization to obtain an orbital rotary kiln denitration prediction model.

[0129] In - depth analysis of current harmonics:

[0130] Install a broadband current sensor (0 - 5 kHz) on the power supply side of the drive motor to capture the complete spectrum including the 23rd harmonic, use wavelet packet decomposition technology to separate the fundamental wave and characteristic harmonics, extract the amplitudes of the 2nd and 5th harmonics as the core indicators of rotational speed fluctuation, establish a harmonic - rotational speed mapping database, and identify the harmonic feature patterns corresponding to abnormal operating conditions through K - means clustering.

[0131] Temperature Field Dynamic Calibration Mechanism: Deploy a movable temperature measurement robot to cruise along the kiln body track. Conduct contact calibration (accuracy ±3°C) on fixed temperature measurement points every 30 minutes. Set a 10% overlapping area in adjacent axial temperature measurement intervals. Use cubic spline interpolation to eliminate boundary mutations. When the axial temperature gradient > 150°C / m, activate redundant infrared thermal imagers for full-field scanning verification.

[0132] Multi-source Data Spatiotemporal Fusion: Use the IEEE 1588 Precision Clock Protocol to unify the time reference of each subsystem (deviation < 1ms). Construct a six-dimensional data cube: timestamp - axial position - radial depth - rotational speed fluctuation - temperature value - NOx concentration. Use a data alignment engine to automatically compensate for spatiotemporal coordinate offsets caused by the movement of the spray gun (maximum compensation amount ±15cm / 0.5s).

[0133] Adaptive Training of the Prediction Model, Design a Hybrid Neural Network Architecture: 1D-CNN processes current harmonic sequences, LSTM captures the temporal characteristics of the temperature field, and the fully connected layer fuses the NOx concentration gradient. Deploy a sliding window mechanism (window length 60s, step size 5s) to dynamically generate a training sample set.

[0134] Establish an Online Verification Channel: When the deviation between the predicted ammonia injection amount and the actual denitration efficiency > 8%, trigger incremental learning to update the model parameters.

[0135] Through the spatial correlation analysis of harmonic characteristics and the temperature field, the rotational speed-temperature coupling fluctuation mode caused by sudden changes in fuel calorific value (such as a change of ±200 kcal / kg) can be identified; the multi-dimensional characteristics of the data cube enable the model to dynamically adapt to the complex interaction relationship between the spray gun movement trajectory and the kiln body rotation phase, effectively improving the prediction robustness under variable working conditions.

[0136] Specifically, for the orbital rotary kiln denitration optimization method described in the present invention, the establishment of kiln internal zoning in step S104 includes:

[0137] Divide a monitoring section every 0.5m along the axial direction of the kiln body, and radially divide it according to the temperature gradient into:

[0138] High-temperature core area (T ≥ 1050°C), transition reaction area (850°C ≤ T < 1050°C), edge low-temperature area (T < 850°C); the calculation of the gas error data adopts: . Where is the comprehensive gas error index of the i-th zone, is the contribution weight of the NOx concentration error, is the contribution weight of the oxygen content error, is the predicted NOx volume concentration, is the measured NOx volume concentration, is the NOx concentration normalization reference value, and its value is the maximum NOx concentration allowed under the current working conditions. is the predicted oxygen volume fraction. is the measured oxygen volume fraction. The oxygen content normalization reference value, and its value is the maximum volume fraction of oxygen required for theoretical combustion.

[0139] Axial partition monitoring enhancement: Three groups of redundant thermocouples (type K / J / T) are arranged in each 0.5m monitoring section to form a triangular temperature measurement array.

[0140] Develop an axial temperature field self-correction algorithm: When the temperature difference between adjacent sections > 100°C, activate the mobile infrared thermometer for re-measurement and calibration.

[0141] Establish a dynamic partition boundary mechanism: Automatically adjust the boundary of the high-temperature area (±0.2m floating range) according to the real-time thermal image.

[0142] Radial temperature gradient optimization:

[0143] Deploy water-cooled high-temperature-resistant sensors (working upper limit 1300°C) in the high-temperature core area, and equip with a self-cleaning nitrogen purge system.

[0144] Set up a double-layer monitoring ring in the transition reaction area: The upper layer monitors the gas composition (spacing 15cm), and the lower layer monitors the wall temperature.

[0145] Install anti-condensation sensors in the edge low-temperature area, and integrate a heating and dehumidification module (maintain the probe temperature > 60°C).

[0146] Partition collaborative control strategy:

[0147] Implement "temperature priority" control in the high-temperature core area: Automatically reduce the ammonia injection amount (maximum reduction 30%) when T > 1100°C.

[0148] Enable the "efficiency optimal" mode in the transition reaction area: Dynamically adjust the spray gun residence time according to the NOx gradient (±2s from the reference value).

[0149] Implement the "anti-escape" strategy in the edge low-temperature area: Automatically switch the ammonia water concentration to a low-concentration formula (15% → 10%).

[0150] When the difference in sensor data within the partition > 5%, start laser scanning temperature field verification. Abnormal fluctuations in the high-temperature area trigger an increase in the monitoring frequency of adjacent partitions (10Hz → 50Hz). Partitions with a calibration deviation > 8% for 3 consecutive times are automatically marked as faulty states.

[0151] In a second aspect, the present invention provides an optimized system for denitrification of an orbital rotary kiln, which applies an optimized method for denitrification of an orbital rotary kiln as described above, and includes: a server end, a data acquisition end, and an orbital rotary kiln denitrification equipment end. The server end establishes communication connections with the data acquisition end and the orbital rotary kiln denitrification equipment end respectively;

[0152] A data acquisition unit for acquiring denitrification data of the orbital rotary kiln. The denitrification data of the orbital rotary kiln includes operating parameters of the orbital rotary kiln body, temperature distribution data inside the orbital rotary kiln, NOx concentration data inside the orbital rotary kiln, oxygen content data inside the orbital rotary kiln, reductant injection data inside the orbital rotary kiln, spray gun state data inside the orbital rotary kiln, and operating condition data inside the orbital rotary kiln;

[0153] A data analysis unit for establishing a reference position point inside the rotary kiln, configuring coordinates based on the reference position point inside the kiln to obtain coordinate information inside the kiln, substituting the coordinate information inside the kiln into a preset digital twin model to generate a digital twin image inside the kiln, receiving a denitrification task of the orbital rotary kiln, substituting the denitrification task of the orbital rotary kiln into a preset denitrification prediction model of the orbital rotary kiln to obtain denitrification prediction data of the orbital rotary kiln, and substituting the denitrification prediction data of the orbital rotary kiln into the digital twin image inside the kiln to obtain prediction information of the spray gun on the rail moving mechanism. The prediction information of the spray gun on the rail moving mechanism includes the predicted coordinates of the spray gun on the rail and the predicted time information corresponding to the coordinates;

[0154] A data comparison unit for collecting real-time data of the spray gun on the rail moving mechanism inside the orbital rotary kiln to obtain real-time spray gun movement data. The real-time spray gun movement data includes the real-time coordinates of the spray gun on the rail 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 rail moving mechanism to obtain spray gun delay data;

[0155] An error analysis unit for partitioning the inside of the kiln in the digital twin image of the kiln to obtain partitions inside the kiln, monitoring the partitions inside the kiln to obtain gas data of each partition inside the kiln, retrieving data from the denitrification data of the orbital rotary kiln to obtain the fuel calorific value fluctuation coefficient, sulfur content time series change value, and ash content ratio dynamic parameters under the optimal operating condition inside the kiln, substituting the fuel calorific value fluctuation coefficient, sulfur content time series change value, and ash content ratio dynamic parameters under the optimal operating condition inside the kiln into a preset denitrification prediction model of the orbital rotary kiln to obtain denitrification prediction gas data of the orbital rotary kiln, and comparing the denitrification prediction gas data of the orbital rotary kiln with the gas data of each partition inside the kiln in sequence to obtain gas error data of each partition inside the kiln;

[0156] A control optimization unit is used to optimize a preset denitration prediction model for an orbital rotary kiln based on the gas error data and lance delay data of each partition in the kiln, obtaining an optimized denitration prediction model for the orbital rotary kiln. Substituting the denitration task of the orbital rotary kiln into the optimized denitration prediction model for the orbital rotary kiln, optimized lance rail movement control parameters are obtained. The lance rail movement control parameters include lance rail movement speed parameters, ammonia injection amount, and injection points. The optimized lance rail movement control parameters are used as real-time execution data during the movement of the lance in the rail movement mechanism.

[0157] Specifically, for an orbital rotary kiln denitration optimization system according to the present invention, the data acquisition unit is further used for:

[0158] Collecting the rotational speed fluctuation characteristics through a vibration sensor pre-installed in the kiln body drive gearbox and calculating the rotational speed change rate; obtaining the axial temperature distribution at 0.5 m intervals using a pre-installed distributed optical fiber temperature measurement system with a sampling frequency not lower than 10 Hz; real-time monitoring of the NOx concentration gradient using a pre-installed laser absorption spectrometer; the reductant injection data includes the ammonia water concentration, atomization pressure, and injection flow time series curve.

[0159] Specifically, for an orbital rotary kiln denitration optimization system according to the present invention, the data analysis unit is further used for:

[0160] Establishing a three-dimensional geometric model of the kiln body based on the point cloud scan data with a mesh division accuracy ≤ 2 cm;

[0161] Setting 24 reference positioning RFID tags axially on the kiln body and establishing the conversion relationship between the internal space coordinate system of the kiln and the external polar coordinates;

[0162] Retrieving historical operating condition data from the orbital rotary kiln denitration data, classifying and storing the historical operating condition data according to the fuel type, and constructing a denitration efficiency mapping database including 2000 groups of operating conditions.

[0163] Specifically, for an orbital rotary kiln denitration optimization system according to the present invention, the data analysis unit is further used for: collecting the current harmonics of the kiln body drive motor, extracting data characteristics from the current harmonics of the kiln body drive motor to obtain the rotational speed fluctuation characteristics; collecting the axial temperature field with a spatial resolution of 0.3 m, monitoring the NOx concentration gradient, and performing spatio-temporal alignment of the rotational speed fluctuation characteristics, axial temperature field, and monitored NOx concentration gradient with the lance rail encoder position signal to establish a data matrix with time stamp synchronization;

[0164] Training the model based on the data corresponding to the data matrix with time stamp synchronization to obtain a denitration prediction model for the orbital rotary kiln.

[0165] Specifically, for an optimized denitration system of an orbital rotary kiln according to the present invention, the error analysis unit is further configured to:

[0166] Divide a monitoring section every 0.5 m along the axial direction of the kiln body, and divide it radially according to the temperature gradient into:

[0167] A high-temperature core area (T≥1050 °C), a transition reaction area (850 °C≤T<1050 °C), and an edge low-temperature area (T<850 °C); the calculation of the gas error data adopts: . Where is the comprehensive gas error index of the i-th partition, is the contribution weight of the NOx concentration error, is the contribution weight of the oxygen content error, is the predicted NOx volume concentration, is the measured NOx volume concentration, is the NOx concentration normalization reference value, is the predicted oxygen volume fraction, is the measured oxygen volume fraction, is the oxygen content normalization reference value.

[0168] The present invention solves the problems that it is difficult to accurately predict the optimal ammonia injection amount and injection point under complex working conditions such as the fluctuation of the kiln body rotation speed and the change of fuel composition, which in turn leads to low reductant utilization rate (increased ammonia escape) or unstable denitration efficiency through the following measures in multiple aspects:

[0169] Comprehensive and refined data collection and processing:

[0170] Enhanced collection of rotational speed fluctuation characteristics: Install a three-axis MEMS vibration sensor group symmetrically on the drive gearbox, extract the 1X - 5X rotational speed harmonic components through order analysis, and can capture the rotational speed change more carefully. Establish a rotational speed anomaly detection mechanism. When the volatility of three consecutive sampling periods > 2%, trigger the Hall sensor auxiliary verification mode, which increases the monitoring accuracy of rotational speed anomalies. At the same time, increase the kiln body inertia compensation parameter, which can eliminate the instantaneous rotational speed jitter interference caused by uneven material distribution, so that the obtained rotational speed data can more truly reflect the actual operating state of the kiln body and provide a reliable basis for subsequent accurate prediction.

[0171] Improving the reliability of temperature distribution data: The distributed optical fiber adopts a double-layer armored structure, which improves the stability and anti-interference ability of the optical fiber. A temperature mutation verification rule is set. When the temperature difference between adjacent temperature measurement points > 80 °C, the infrared thermal imager is activated for regional verification to ensure the accuracy of temperature data. Redundant thermocouple arrays are added at the kiln head / kiln tail and spatially and temporally fused with the optical fiber data. The multi-faceted temperature monitoring means complement each other, enabling a more comprehensive and accurate understanding of the temperature distribution in the kiln. This is crucial for determining the optimal reaction temperature window and analyzing the impact of temperature on the denitrification reaction.

[0172] Optimizing the dynamic monitoring of NOx concentration: The laser absorption spectrometer integrates a self-cleaning optical window to prevent dust adhesion interference and ensure the stability of monitoring data. A concentration gradient verification mechanism is established. When the concentration difference between adjacent axial monitoring points > 150 ppm, the mobile sampling probe is activated for remeasurement to promptly detect and correct possible concentration monitoring deviations. A multi-spectral fusion algorithm is deployed to simultaneously analyze the NO / NO 2 concentration ratio to more precisely understand the actual situation of NOx, providing a key basis for accurately predicting the optimal ammonia injection volume, as the NOx concentration is the core indicator of denitrification efficiency.

[0173] Full-parameter monitoring of reductant injection: An ultrasonic concentration meter and a Coriolis mass flowmeter are installed on the ammonia water delivery pipeline to accurately measure the ammonia water concentration and flow rate. Combining information such as injection points, injection pressure / atomization effect, etc., comprehensively understand the reductant injection situation. A pulse width modulation control for injection is set. When the atomization pressure fluctuation > 0.05 MPa, the opening of the pressure stabilizing valve is automatically adjusted, and the atomization cone angle of the spray gun is monitored in real time and the gas-liquid ratio is feedback-adjusted through a high-speed camera system to improve the mixing uniformity and reaction efficiency of the reductant and flue gas, thereby indirectly helping to accurately determine the optimal ammonia injection volume and injection points.

[0174] Analysis and optimization based on the digital twin model and database:

[0175] Dynamic update mechanism of point cloud data and model correction: By deploying a mobile laser scanning robot to perform a full-kiln scan of the kiln body regularly (every 72 hours), a point cloud change rate monitoring model is established to promptly detect situations such as kiln body structure deformation and lining wear. The ICP algorithm is used to register the new and old point cloud data and automatically correct the three-dimensional geometric model, making the model more accurately reflect the actual state of the kiln body, providing a more practical 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.

[0176] Intelligent calibration of the spatial coordinate system: Each RFID tag integrates a temperature compensation module to eliminate the positioning error caused by thermal expansion. A dual-frequency RFID reader is set up and the anti-interference frequency band is automatically switched. A six-degree-of-freedom coordinate transformation matrix containing the kiln body inclination compensation parameter is constructed to ensure the accuracy of the conversion relationship between the spatial coordinate system inside 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, providing a reliable coordinate basis for accurately predicting the spraying point.

[0177] Optimization strategy for the operating condition database: Establish a five-level fuel classification system to classify different types of fuels in detail, and deploy a data cleaning pipeline to eliminate outliers and fill in missing data. Create an operating condition feature fingerprint library and associate a 128-dimensional feature vector with each type of fuel. Through this more detailed classification storage and data processing, when using the database for analysis and prediction later, similar operating conditions can be more accurately matched, improving the accuracy of the denitration prediction model and thus better coping with complex operating conditions such as fuel composition changes.

[0178] Model training optimization based on current harmonics and temperature field: Collect and deeply analyze the current harmonics of the kiln body drive motor. Capture the complete spectrum by installing a broadband current sensor, and use the wavelet packet decomposition technique to extract the key harmonic amplitude as the core index of rotational speed fluctuation. Establish a harmonic-rotational speed mapping database and identify abnormal operating condition patterns through K-means clustering. At the same time, collect the axial temperature field with a spatial resolution of 0.3m and perform dynamic calibration, such as deploying a movable temperature measurement robot for cruise calibration and setting interpolation processing for overlapping areas. Fuse these multi-source data in space and time to construct a six-dimensional data cube, and design a hybrid neural network architecture based on this for model training. Generate a training sample set through a sliding window mechanism, and also establish an online verification channel to trigger incremental learning to update model parameters according to the prediction deviation. Such a processing method enables the model to better adapt to complex operating conditions such as kiln body rotational speed fluctuation and temperature change, improving the accuracy and robustness of the prediction.

[0179] Real-time data monitoring of the spray gun and establishment of the spray gun delay model:

[0180] In the process of collecting real-time data of the spray gun on the rail moving mechanism in the collecting orbital rotary kiln, not only the position and time information of the spray gun are obtained, but also the vibration condition of the spray gun is monitored by installing a micro acceleration sensor, with a sampling frequency of 100 times per second. The vibration of the spray gun will cause the deviation of the spraying point and the instability of the spraying amount. By analyzing the actual working state of the spray gun by combining these vibration data with other real-time data, the spray gun delay data can be calculated more accurately. In addition, considering the influence of the internal airflow field of the kiln body on the movement of the spray gun, the air velocity and direction data are obtained by installing wind speed sensors at different positions in the kiln body, and a spray gun delay model considering the influence of airflow is established by combining the moving speed and spraying direction of the spray gun. This model can more accurately reflect the delay situation of the spray gun actually reaching the preset spraying point under complex working conditions, providing a reliable basis for accurately controlling the spraying point and time of the spray gun in the future, and thus helping to solve the problem of difficult to accurately predict the optimal ammonia spraying amount and spraying point caused by the spray gun delay.

[0181] In-kiln Zonal Monitoring and Data Processing Optimization:

[0182] Axial Zonal Monitoring Enhancement and Dynamic Adjustment: Divide a monitoring section every 0.5 m along the axial direction of the kiln body and strengthen the monitoring. Arrange three groups of redundant thermocouples in each section to form a triangular temperature measurement array. Develop an axial temperature field self-correction algorithm. When the temperature difference between adjacent sections > 100 °C, activate the mobile infrared thermometer for re-measurement and calibration, and establish a dynamic zonal boundary mechanism to automatically adjust the boundary of the high-temperature zone according to the real-time thermal image. These measures make the monitoring of the temperature distribution in the kiln more accurate and flexible, can better adapt to the temperature field changes caused by working condition changes such as the fluctuation of the kiln body rotation speed, and provide more accurate temperature-related data support for accurately predicting the optimal ammonia spraying amount and spraying point.

[0183] Radial Temperature Gradient Optimization and Zonal Cooperative Control Strategy: Take optimization measures for different radial temperature gradient regions (high-temperature core region, transition reaction region, edge low-temperature region) respectively. For example, deploy special sensors and equip with a self-cleaning system in the high-temperature core region, set up a double-layer monitoring ring in the transition reaction region, and install dew-proof sensors and integrate heating and dehumidification modules in the edge low-temperature region. At the same time, formulate different control strategies for each region, such as the "temperature priority" control in the high-temperature core region, the "efficiency optimal" mode in the transition reaction region, and the "anti-escape" strategy in the edge low-temperature region. And when the difference in sensor data within the zone > 5%, start the laser scanning temperature field verification, and the abnormal fluctuation in the high-temperature zone triggers an increase in the monitoring frequency of the adjacent zone. Through these zonal monitoring optimizations and cooperative control strategies, the ammonia spraying amount and spraying point can be adjusted more accurately according to the actual conditions of different regions, effectively meeting the denitration requirements under complex working conditions such as the fluctuation of the kiln body rotation speed and the change of fuel composition, and improving the denitration efficiency and the utilization rate of the reducing agent.

[0184] Gas error data calculation optimization: When calculating the gas error data for each partition in the kiln, different monitoring parameter weights are set according to the influence degree of different regions on the denitration reaction. For example, a higher weight is assigned to the NOx concentration data in the high-temperature core area. At the same time, when retrieving relevant parameters from the denitration data of the orbital rotary kiln and substituting them into the prediction model, the screening and preprocessing of historical data are increased. Similar operating condition data are screened according to the current operating parameters of the kiln body and the fuel characteristics and normalized. These measures make the calculation of gas error data more in line with the actual operating 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 points.

[0185] Prediction model optimization and real-time online verification:

[0186] Optimization of the prediction model based on reinforcement learning: When optimizing the preset denitration prediction model of the orbital rotary kiln based on the gas error data for each partition in the kiln and the gun delay data, an optimization strategy based on reinforcement learning is introduced. According to the actual denitration effect of the gun at different injection points (evaluated by real-time monitoring of the emitted NOx concentration) and the ammonia slip situation, the model is given corresponding reward or punishment signals. The model continuously adjusts its prediction strategy according to these feedback signals to more accurately determine the optimal ammonia injection amount and injection points, enabling it to better adapt to the denitration requirements under complex operating conditions such as the fluctuation of the kiln body rotation speed and the change of fuel composition.

[0187] Real-time online verification and model re-optimization: In order to make the optimized gun rail movement control parameters better adapt to the changes in complex operating conditions, after substituting the denitration task of the orbital rotary kiln into the optimized denitration prediction model of the orbital rotary kiln to obtain the optimized gun rail movement control parameters, these parameters are verified in real time online. By setting additional monitoring points on the kiln body, information such as the actual denitration efficiency, ammonia slip rate, and the actual movement trajectory of the gun is monitored in real time. The 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 promptly re-optimized and adjusted to improve the effectiveness and accuracy of the control parameters, enabling precise control of the ammonia injection amount and injection points under complex operating conditions all the time, thereby improving the denitration efficiency and the utilization rate of the reducing agent.

[0188] Through the above comprehensive measures in aspects such as data collection, model construction and optimization, gun monitoring, in-kiln partition processing, and real-time verification, the present invention can effectively solve the problems that it is difficult to accurately predict the optimal ammonia injection amount and injection points under complex operating conditions such as the fluctuation of the kiln body rotation speed and the change of fuel composition, resulting in low utilization rate of the reducing agent (increased ammonia slip) or unstable denitration 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: The temperature range of the high temperature core zone is T≥1050℃, the temperature range of the transition reaction zone is 850℃≤T<1050℃, and the temperature range of the edge low temperature zone is 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: The temperature range of the high temperature core zone is T≥1050℃, the temperature range of the transition reaction zone is 850℃≤T<1050℃, and the temperature range of the edge low temperature zone is T<850℃.

Images