Glass-melting-furnace waste-gas denitration air-quantity precise regulation method and system

Abstract

The application provides a glass melting furnace waste gas denitration air volume precise regulation and control method and system. Among them, the method is to arrange an ultrasonic Doppler sensor group around the glass melting furnace flue, synchronously emit ultrasonic beams to collect flow velocity fluctuation data of waste gas denitration air volume demand, decompose time sequence waveforms to generate multi-dimensional frequency characteristics, and aggregate into a multi-dimensional flow velocity vector representing the flow field state; correct the vector deviation by spatial coordinate alignment and time sequence synchronization, extract the difference change amount to generate a flow field heat map; establish a space-time correlation model of the multi-dimensional flow velocity vector and the heat map, predict the air volume regulation threshold that meets the denitration demand, convert it into a fan speed control signal to drive the actuator; according to the correlation between the waste gas fluctuation response data caused by the change of the fan speed and the heat map update frequency, dynamically adjust the multi-dimensional flow velocity vector to form a closed-loop control. The application realizes the dynamic demand of denitration air volume and the millisecond closed-loop response of the fan execution, improves the flue gas purification efficiency and reduces the energy consumption.

Description

Technical Field

[0001] This application relates to the field of precise control of denitrification air volume in waste gas, and in particular to a method and system for precise control of denitrification air volume in glass melting furnace waste gas. Background Technology

[0002] Glass melting furnace exhaust gas is characterized by its complex composition and drastic dynamic changes in temperature and flow rate, with nitrogen oxides being one of the main pollutants. To achieve efficient and stable selective catalytic reduction (SCR) denitrification, the key lies in the precise control of ammonia injection, ensuring it matches the transient exhaust gas conditions in real time. The core requirement is to dynamically and in real-time calculate and adjust the necessary ammonia flow rate based on changes in exhaust gas velocity, avoiding both excessive ammonia injection leading to "ammonia escape" and insufficient ammonia injection resulting in decreased denitrification efficiency and excessive emissions. This necessitates a control system with rapid response, high-precision identification of exhaust gas status, and the ability to autonomously execute optimized adjustments.

[0003] One targeted solution is a feedback control system based on a fixed molar ratio. This system installs a continuous emission monitoring system at the inlet of the denitrification system to measure the exhaust gas flow rate and NOx concentration in real time. The controller calculates the theoretically required ammonia flow rate based on a preset fixed ammonia-nitrogen molar ratio, combined with the real-time measured flow and concentration values, and adjusts the ammonia injection valve opening accordingly. However, its drawback is that when the exhaust gas velocity fluctuates rapidly or significantly, the inherent measurement, calculation, and execution lags make it difficult to instantly track the actual dynamic changes. This results in time lags and errors in ammonia injection rate adjustment, failing to achieve optimal matching with sudden flow rate changes, thus affecting the control accuracy and effectiveness. Summary of the Invention

[0004] This application provides a method and system for precise control of denitrification air volume in glass melting furnace exhaust gas, which solves the problems of low purification efficiency and high energy consumption caused by large measurement deviation of denitrification air volume, control lag and lack of feedback mechanism in the prior art.

[0005] Firstly, this application provides a method for precise control of denitrification air volume in glass melting furnace exhaust gas, including:

[0006] An ultrasonic Doppler sensor array installed around the glass melting furnace flue synchronously emits ultrasonic beams to collect flow velocity pulsation data reflecting the air volume requirement for denitrification of waste gas, and obtains the waveform of the time series of the flow velocity pulsation data.

[0007] The waveform of the time series is decomposed to form multi-dimensional frequency features, and the multi-dimensional frequency features collected in the same time window are aggregated to generate a multi-dimensional velocity vector that comprehensively represents the flow field state.

[0008] Spatial coordinate alignment and time series synchronization processing are performed on the flow velocity pulsation data synchronously acquired by the ultrasonic Doppler sensor group to correct the measurement deviation of the multidimensional flow velocity vector, and the difference change of the corrected multidimensional flow velocity vector is extracted.

[0009] The differential change quantities are integrated to generate a flow field thermogram and a spatiotemporal correlation model is established between the multidimensional velocity vector and the dynamic flow field thermogram. The spatiotemporal correlation model is then used to predict the air volume adjustment threshold that meets the requirements of the denitrification reaction.

[0010] The air volume adjustment threshold is converted into a physical control signal that drives the fan speed to change, and the fan actuator is driven to change the fan speed based on the physical control signal;

[0011] The update frequency of the exhaust gas pulsation response data caused by the change of fan speed and the dynamic thermogram of the flow field is obtained, and the multidimensional velocity vector is adjusted based on the correlation between the exhaust gas pulsation response data and the update frequency to form a closed-loop control of data acquisition and execution feedback.

[0012] Optionally, an ultrasonic Doppler sensor array installed around the glass melting furnace flue synchronously emits ultrasonic beams to collect flow velocity pulsation data reflecting the airflow demand for denitrification of the exhaust gas, and acquires the time series waveform of the flow velocity pulsation data, including:

[0013] An ultrasonic Doppler sensor array is arranged at predetermined points on the four walls of the glass melting furnace flue. The measurement positions of each ultrasonic Doppler sensor are evenly distributed according to the cross-section of the glass melting furnace flue to form a closed-loop detection structure.

[0014] In the closed-loop detection structure, a synchronous pulse signal is generated by the trigger controller built into the ultrasonic Doppler sensor group to drive the ultrasonic Doppler sensor group to emit ultrasonic beams at the same time.

[0015] After emitting the ultrasonic beam, each ultrasonic Doppler sensor receives the echo signal reflected by the exhaust gas particles and calculates the instantaneous flow velocity value at each measurement location within the time window of the echo signal.

[0016] The instantaneous flow velocity values ​​at each measurement location are continuously collected and arranged in time sequence to form a flow velocity pulsation data sequence with timestamps, and the waveform of the time series in the flow velocity pulsation data sequence is extracted.

[0017] Optionally, the waveform of the time series is decomposed to form multi-dimensional frequency features, and the multi-dimensional frequency features collected in the same time window are aggregated to generate a multi-dimensional velocity vector that comprehensively represents the flow field state, including:

[0018] The waveform of the time series is divided into independent waveform segments with fixed time intervals, and the average time interval between adjacent waveform inflection points within each independent waveform segment is calculated as a multi-dimensional frequency feature.

[0019] Based on the same time window, multi-dimensional frequency characteristics of independent waveform segments within the time window are collected.

[0020] The multi-dimensional frequency features are aggregated to generate a multi-dimensional velocity vector that comprehensively represents the flow field state.

[0021] Optionally, the flow velocity pulsation data synchronously acquired by the ultrasonic Doppler sensor group is subjected to spatial coordinate alignment and time series synchronization processing to correct the measurement deviation of the multidimensional flow velocity vector, and the difference change of the corrected multidimensional flow velocity vector is extracted, including:

[0022] Obtain the spatial coordinates of the ultrasonic Doppler sensor group, and calculate the delay of ultrasonic wave transmission time between adjacent ultrasonic Doppler sensors based on the spatial coordinates.

[0023] Time series synchronization processing is achieved by adjusting the timestamp of the flow velocity pulsation data of the ultrasonic Doppler sensor based on the delay amount.

[0024] Based on the spatial location coordinates, the time-series synchronized flow velocity fluctuation data is mapped to a preset three-dimensional flow field coordinate grid to generate a set of spatially aligned flow velocity fluctuation data.

[0025] The multidimensional velocity vector is corrected by replacing the velocity fluctuation data with spatially aligned velocity data.

[0026] Multidimensional velocity vectors with the same three-dimensional flow field coordinate grid points are extracted from the corrected multidimensional velocity vectors to calculate the difference in the multidimensional velocity vectors.

[0027] Optionally, the differential changes are fused to generate a flow field thermogram and a spatiotemporal correlation model is established between the multidimensional velocity vector and the dynamic flow field thermogram. The spatiotemporal correlation model is then used to predict the airflow adjustment threshold required to meet the denitrification reaction, including:

[0028] The differential change is superimposed onto the three-dimensional flow field coordinate grid to calculate the density distribution value of the three-dimensional flow field coordinate grid, thereby forming a flow field thermogram that represents the aggregation intensity of the differential change.

[0029] Extract the timestamp and vector element values ​​of the multidimensional velocity vector, and match the vector element values ​​at the same timestamp with the density distribution values ​​of the corresponding three-dimensional flow field coordinate grid points in the flow field thermogram;

[0030] Obtain the changing trend of vector element values ​​and the changing trend of density distribution values ​​within the timestamp. When the directions of the changing trends are consistent, mark the vector element values ​​and the three-dimensional flow field coordinate grid points as a dynamically associated region.

[0031] The proportionality constant between the rate of change of the multidimensional velocity vector and the trend of the density distribution value in the dynamically associated region is statistically analyzed to establish the binding relationship between the proportional constant and the timestamp as a spatiotemporal association model;

[0032] The predicted values ​​of the spatiotemporal correlation model are back-matched with the multidimensional flow velocity vector required for the denitrification reaction to generate the air volume adjustment threshold that drives the fan.

[0033] Optionally, the airflow adjustment threshold is converted into a physical control signal that drives the fan speed change, and the fan actuator is driven to change the fan speed based on the physical control signal, including:

[0034] The airflow adjustment threshold is converted into an electrical pulse modulation signal, and the electrical pulse modulation signal is converted into a physical control signal with a specific pulse width and frequency by a digital signal converter.

[0035] The drive fan actuator receives the physical control signal and adjusts the output frequency of the drive fan actuator according to the physical control signal to change the current intensity of the motor coil;

[0036] The fan speed is adjusted by rotating the shaft bearing of the fan actuator based on the change in the current intensity of the motor coil.

[0037] Optionally, the update frequency of the exhaust gas pulsation response data caused by changes in fan speed and the dynamic thermogram of the flow field is obtained, and the multidimensional velocity vector is adjusted based on the correlation between the exhaust gas pulsation response data and the update frequency to form a closed-loop control of data acquisition and execution feedback, including:

[0038] Collect exhaust gas pulsation response data generated by the ultrasonic Doppler sensor group after the fan speed changes and record the update frequency of the dynamic thermogram of the flow field;

[0039] Calculate the ratio of the number of occurrences of the flow velocity fluctuation peak in the exhaust gas pulsation response data to the update frequency. When the number of occurrences of the fluctuation peak exceeds a preset multiple of the update frequency, it is marked as a high-frequency pulsation event.

[0040] The thermal density distribution of the flow field dynamic thermogram is extracted according to the time window of the occurrence of the high-frequency pulsating event, and the gradient change direction of the thermal density distribution within the time window is extracted.

[0041] The gradient change direction is compared with the change direction of the multidimensional velocity vector in the same time window. If the deviation of the change direction exceeds a preset allowable angle, the spatial sampling point density of the multidimensional velocity vector is adjusted.

[0042] The time series of data acquisition from the ultrasonic Doppler sensor array is reconfigured based on the adjusted spatial sampling point density to form a closed-loop control signal, and the acquisition of the exhaust gas pulsation response data is driven according to the closed-loop control signal.

[0043] Secondly, this application provides a precise control system for denitrification airflow in glass melting furnace exhaust gas, comprising:

[0044] The acquisition module is used to synchronously emit ultrasonic beams through an ultrasonic Doppler sensor group installed around the glass melting furnace flue to acquire flow velocity pulsation data reflecting the air volume requirement for denitrification of exhaust gas, and to obtain the waveform of the time series of the flow velocity pulsation data.

[0045] The decomposition module is used to decompose the waveform of the time series to form multi-dimensional frequency features, and to aggregate the multi-dimensional frequency features collected in the same time window to generate a multi-dimensional velocity vector that comprehensively represents the flow field state.

[0046] The calibration module is used to perform spatial coordinate alignment and time series synchronization processing on the flow velocity pulsation data synchronously acquired by the ultrasonic Doppler sensor group to correct the measurement deviation of the multidimensional flow velocity vector, and extract the difference change of the corrected multidimensional flow velocity vector.

[0047] The fusion module is used to fuse the differential changes to generate a flow field heat map and establish a spatiotemporal correlation model between the multidimensional velocity vector and the dynamic flow field heat map, and predict the air volume adjustment threshold that meets the requirements of the denitrification reaction through the spatiotemporal correlation model.

[0048] The conversion module is used to convert the air volume adjustment threshold into a physical control signal that drives the fan speed change, and drive the fan actuator to change the fan speed based on the physical control signal;

[0049] The adjustment module is used to acquire the update frequency of the exhaust gas pulsation response data and the dynamic thermal map of the flow field caused by the change of fan speed, and adjust the multidimensional velocity vector based on the correlation between the exhaust gas pulsation response data and the update frequency to form a closed-loop control of data acquisition and execution feedback.

[0050] Thirdly, embodiments of this application provide a computing device, including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are to be invoked and executed by the processing component to realize a method for precise control of denitrification air volume in glass melting furnace exhaust gas as described in the first aspect above.

[0051] Fourthly, embodiments of this application provide a computer storage medium storing a computer program, which, when executed by a computer, implements a method for precise control of denitrification airflow in glass melting furnace exhaust gas as described in the first aspect.

[0052] In this application's technical solution, real-time monitoring and precise control of the waste gas flow state are achieved through an ultrasonic Doppler sensor array. A comprehensive flow velocity vector constructed based on multi-dimensional frequency characteristics can fully reflect the dynamic characteristics of the flow field. A flow field thermogram established through a spatiotemporal correlation model enables intelligent prediction of denitrification airflow demand. The airflow adjustment threshold is converted into a physical control signal to drive fan speed adjustment, significantly improving the response speed of the denitrification system. A closed-loop control mechanism based on waste gas pulsation response data continuously optimizes the accuracy of flow velocity monitoring. This method overcomes the lag defect of traditional airflow control, achieving dynamic balance adjustment of the glass melting furnace waste gas denitrification system. It reduces energy consumption while ensuring denitrification efficiency, providing an intelligent solution for industrial waste gas treatment.

[0053] Furthermore, a closed-loop detection structure using an ultrasonic Doppler sensor array enables comprehensive monitoring of the flow of exhaust gas in the glass melting furnace flue. A synchronous pulse triggering mechanism ensures the spatiotemporal consistency of data acquisition from multiple sensors, eliminating measurement errors. Instantaneous velocity calculation based on echo signal analysis achieves high-precision capture of the dynamic characteristics of the flow field. Timestamp-marked velocity pulsation data sequences provide high-resolution temporal characteristics for flow field state analysis. This method overcomes the limitations of traditional single-point velocity measurement, achieving real-time synchronous detection of the velocity distribution across the flue cross-section, providing reliable data support for the precise control of the exhaust gas treatment system, and significantly improving the intelligence level of industrial flue gas treatment.

[0054] These or other aspects of this application will become more apparent in the following description of the embodiments. Attached Figure Description

[0055] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 A flowchart of a method for precise control of denitrification air volume in glass melting furnace exhaust gas provided in this application is shown;

[0057] Figure 2 A scenario diagram is shown for a method for precise control of denitrification air volume in glass melting furnace exhaust gas provided in this application;

[0058] Figure 3This invention provides a schematic diagram of the structure of a glass melting furnace exhaust gas denitrification air volume precision control system.

[0059] Figure 4 A schematic diagram of the structure of a computing device provided in this application is shown. Detailed Implementation

[0060] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0061] In some of the processes described in the specification, claims, and accompanying drawings of this application, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as 101, 102, etc., are merely used to distinguish different operations and do not themselves represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a chronological order, nor do they limit "first" and "second" to different types.

[0062] Researchers have discovered a core bottleneck in the denitrification control of glass melting furnace exhaust gas: while feedback control schemes based on a fixed molar ratio can calculate the basic ammonia injection rate, they cannot respond instantaneously to drastic changes in exhaust gas velocity. Specifically, the measurement and execution lag of the continuous emission monitoring system leads to a significant time difference in ammonia injection adjustment. When the exhaust gas velocity fluctuates rapidly, the preset molar ratio cannot dynamically match the actual air volume demand, causing ammonia escape or a sharp drop in denitrification efficiency. This deficiency stems from the traditional technology's lack of awareness of flow velocity pulsation characteristics and the rigidity of the open-loop control mechanism, necessitating the construction of a closed-loop control system with millisecond-level dynamic prediction capabilities.

[0063] To address the aforementioned challenges, this invention proposes a method for precise control of denitrification airflow in glass melting furnace exhaust gas. Its innovation lies in achieving advanced prediction and closed-loop execution of airflow demand through ultrasonic pulsation feature interpretation and spatiotemporal correlation modeling. Specifically, an ultrasonic Doppler sensor array surrounding the flue gas duct synchronously captures flow velocity pulsation waveforms, and generates a multidimensional velocity vector characterizing the flow field dynamics through multidimensional frequency feature aggregation. Measurement deviations are corrected using spatial coordinate alignment and time series synchronization techniques, and a flow field thermogram is constructed by fusing differential changes. Based on the spatiotemporal correlation model, the required airflow adjustment threshold for denitrification is predicted, and the fan speed is adjusted in real time. This method breaks through the limitations of traditional static molar ratios: multidimensional velocity vector technology achieves millisecond-level feature capture of sudden changes in exhaust gas velocity for the first time, solving the problem of inaccurate adjustment caused by measurement lag; the flow field thermogram-driven prediction model transforms velocity pulsations into air volume demand thresholds, enabling precise pre-matching of ammonia injection adjustment with exhaust gas dynamics; and the closed-loop control mechanism continuously optimizes the acquisition strategy through fan response data, forming an adaptive cycle of "pulse perception - threshold prediction - speed adjustment - feedback calibration", fundamentally eliminating the risks of ammonia escape and denitrification efficiency fluctuations.

[0064] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0065] Figure 1 This application provides a flowchart of a method for precise control of denitrification air volume in glass melting furnace exhaust gas, as shown in the embodiments of this application. Figure 1 As shown, the method includes:

[0066] 101. By synchronously emitting ultrasonic beams through an ultrasonic Doppler sensor array installed around the glass melting furnace flue, flow velocity pulsation data reflecting the air volume requirement for denitrification of exhaust gas is collected, and the waveform of the time series of the flow velocity pulsation data is obtained.

[0067] Optionally, step 101 may specifically include the following steps:

[0068] 1011. An ultrasonic Doppler sensor group is arranged at predetermined points on the four walls of the glass melting furnace flue. The measurement positions of each ultrasonic Doppler sensor are evenly distributed according to the cross-section of the glass melting furnace flue to form a closed-loop detection structure.

[0069] 1012. In the closed-loop detection structure, a synchronous pulse signal is generated by the trigger controller built into the ultrasonic Doppler sensor group to drive the ultrasonic Doppler sensor group to emit ultrasonic beams at the same time.

[0070] 1013. After emitting the ultrasonic beam, each ultrasonic Doppler sensor receives the echo signal reflected by the exhaust gas particles and calculates the instantaneous flow velocity value at each measurement location within the time window of the echo signal.

[0071] 1014. Continuously collect the instantaneous flow velocity values ​​corresponding to each measurement location and arrange them in time sequence to form a flow velocity pulsation data sequence with timestamps, and extract the waveform of the time sequence in the flow velocity pulsation data sequence.

[0072] In the above scheme, the ultrasonic Doppler sensor group is a collection of devices that measures fluid velocity by emitting and receiving ultrasonic waves. A glass melting furnace flue refers to a dedicated channel used to discharge high-temperature waste gas during the glass production process. A cross-section refers to the planar shape of an object after it has been transversely cut. A closed-loop detection structure refers to the arrangement of sensors forming a closed loop in space. A trigger controller is a control unit used to generate synchronization signals to coordinate the simultaneous operation of multiple devices. A synchronization pulse signal is an electrical signal used to ensure the simultaneous operation of multiple devices. An ultrasonic beam refers to a region of concentrated ultrasonic energy with a specific directionality. Waste gas particles refer to tiny solid or liquid particles suspended in waste gas generated during combustion. An echo signal refers to the sound wave signal reflected back after an ultrasonic wave encounters an obstacle. A time window refers to a specific time period used for signal analysis. An instantaneous flow velocity value refers to the fluid velocity measured at a certain moment. A timestamp is an identifier used to record the time of data acquisition. A flow velocity pulsation data sequence is a collection of flow velocity change data arranged in chronological order. A time series waveform refers to a graphical representation of data changes over time.

[0073] In this embodiment, firstly, through step 1011, an ultrasonic Doppler sensor array is precisely installed at predetermined points on the four walls of the glass melting furnace flue. A cross-sectional equal-division positioning technique is used to evenly distribute each sensor on the circumference of the flue, forming a spatially continuous closed-loop detection structure through positioning calibration. This structure ensures that the sensor array covers the entire radial area of ​​the flue, establishing a physical basis for full-section flow velocity monitoring.

[0074] Subsequently, in step 1012, the trigger controller built into the ultrasonic Doppler sensor group is activated in the deployed closed-loop detection structure. This controller generates a high-precision synchronization pulse signal to drive all sensors to emit ultrasonic beams at strictly identical times, achieving temporal alignment of multi-point ultrasonic emission. This process eliminates measurement time differences through nanosecond-level time synchronization technology, ensuring spatiotemporal consistency of data acquisition.

[0075] Next, in step 1013, each ultrasonic Doppler sensor receives the echo signal reflected by exhaust gas particles in real time. The echo signal is processed by the Doppler frequency shift analysis algorithm: the frequency shift between the emitted and reflected waves is detected, and combined with sound velocity compensation calculation, the instantaneous flow velocity value of each measurement location within a fixed time window is finally output. This process completes the conversion from physical echo to flow velocity parameters.

[0076] Finally, in step 1014, the system continuously collects instantaneous flow velocity values ​​at each measurement location and arranges them in chronological order using a timestamp engine to form a complete flow velocity pulsation data sequence. Based on this sequence, a waveform extraction operation is performed: the discrete flow velocity values ​​are connected into a continuous waveform curve using a time-domain waveform reconstruction algorithm to capture the pulsation characteristics of the gas flow, and finally output a time-series waveform with complete time attributes.

[0077] In practical applications, in the flue gas control stage of the glass melting furnace exhaust gas denitrification system, the system first arranges ultrasonic Doppler sensor groups on the four walls (including the top, bottom, and side walls) of the glass melting furnace flue. Each sensor is evenly distributed along the cross-section of the flue, with the central axis of the flue as the reference, forming a closed-loop detection structure covering the entire flue area (step 1011). A synchronous pulse signal is generated by the trigger controller built into the sensor group in the closed-loop detection structure, driving all sensors to emit ultrasonic beams at the same time, ensuring the spatiotemporal consistency of multi-point data acquisition (step 1012). When the ultrasonic beam penetrates the exhaust gas flowing in the flue, the particulate matter in the exhaust gas (such as unburned fuel ash and dust) reflects the sound waves. Each sensor receives the echo signal reflected by the exhaust gas particles and calculates the instantaneous flow velocity value of each measurement position within the time window of the echo signal based on the Doppler frequency shift principle (step 1013). The system continuously collects instantaneous flow velocity values ​​at each measurement location, dynamically arranges them in millisecond time sequences, forming a flow velocity pulsation data sequence with timestamps, and extracts the waveform of the time sequence reflecting the periodic fluctuations in exhaust gas flow from this sequence (step 1014). For example, during the stage of a sudden increase in flue gas flow caused by kiln fire switching operation, the waveform of the flow velocity pulsation data sequence shows a steep rising edge, while during the stable combustion period it shows a smooth oscillating waveform. This waveform characteristic provides core input for the subsequent dynamic control of ammonia injection.

[0078] The overall scheme in step 101 above achieves high-precision synchronous acquisition of dynamic flow velocity across the entire cross-section of the flue gas flow field. By combining a closed-loop detection structure design with synchronous pulse triggering technology, the emission timing of the ultrasonic Doppler sensor array is ensured to be strictly uniform, overcoming the spatial limitations of traditional single-point velocity detection. Based on the instantaneous calculation and timestamp marking mechanism of the echo signal, a full-waveform dynamic capture capability for exhaust gas pulsation is constructed, providing a spatiotemporally continuous raw data foundation for flow field state analysis.

[0079] 102. Decompose the waveform of the time series to form multi-dimensional frequency features, and aggregate the multi-dimensional frequency features collected in the same time window to generate a multi-dimensional velocity vector that comprehensively represents the flow field state.

[0080] Optionally, step 102 may specifically include the following steps:

[0081] 1021. Divide the waveform of the time series into independent waveform segments with fixed time intervals, and calculate the average time interval between adjacent waveform inflection points within each independent waveform segment as a multi-dimensional frequency feature.

[0082] 1022. Using the same time window as a reference, collect multi-dimensional frequency characteristics of independent waveform segments within the time window;

[0083] 1023. Aggregate the multi-dimensional frequency features to generate a multi-dimensional velocity vector that comprehensively represents the flow field state.

[0084] In the above scheme, the waveform of a time series refers to the graphical representation of data changing over time. Multidimensional frequency characteristics refer to the set of parameters extracted from the waveform that reflect different frequency domain characteristics. Independent waveform segments refer to discrete waveform units obtained by dividing a continuous waveform into segments of fixed duration. Waveform inflection points refer to characteristic points on the waveform curve where the slope changes significantly. The average time interval refers to the arithmetic mean of the time spans between adjacent waveform inflection points. A time window refers to a specific time period used for signal analysis. Flow field state refers to the distribution of parameters such as velocity and pressure during fluid flow. A multidimensional velocity vector refers to a mathematical vector characterizing the fluid motion state, composed of multiple velocity characteristic parameters.

[0085] In this embodiment, firstly, through step 1021, the system divides the time series waveform into independent waveform segments according to a preset fixed duration (e.g., each segment can be 0.5 seconds). For each segment, a waveform inflection point detection algorithm is executed: the peak and trough positions are located through local curvature analysis, the time interval between adjacent inflection points is calculated, and the arithmetic mean is obtained. This average value serves as the core parameter of the multi-dimensional frequency characteristics, quantifying the fluctuation frequency characteristics of the waveform within the time segment, thus completing the transformation from waveform segment to frequency characteristics.

[0086] Subsequently, in step 1022, multi-dimensional frequency features of all independent waveform segments within the same time window are collected. A time alignment algorithm is used to sort and integrate the feature values ​​of different segments by timestamp, ensuring strict synchronization of feature data at each measurement location within each time window, thus forming a spatially and temporally unified multi-source dataset.

[0087] Finally, in step 1023, the multiple multi-dimensional frequency features collected within the current time window are aggregated: the frequency features of each measurement point are encoded into high-dimensional vector components according to their spatial location using a feature fusion algorithm (for example, the frequency value of point A is used as the first dimension of the vector, and the frequency value of point B is used as the second dimension). This ultimately generates a multi-dimensional velocity vector (e.g., [3.2Hz, 4.1Hz, 2.7Hz...]), with each dimension corresponding to the frequency features of a specific location. The overall data comprehensively characterizes the spatial distribution characteristics of the flue gas flow field state within this time window.

[0088] In practical applications, during the flue gas monitoring stage of a glass melting furnace exhaust gas denitrification system, the system first divides the time-series waveform (such as the velocity change waveform caused by the fire switching operation) into independent waveform segments at fixed time intervals (such as 10 seconds). Each segment contains multiple waveform inflection points (such as the transition point from wave crest to wave trough). The average time interval between all adjacent waveform inflection points within a segment is calculated as a multi-dimensional frequency feature reflecting the local turbulence intensity (step 1021). Using a 1-minute time window, the system collects the multi-dimensional frequency features of all independent waveform segments within this time window (such as 6 segments corresponding to 6 sets of feature values). These feature values ​​respectively characterize the axial pulsation frequency and radial eddy intensity of the flue gas at different times (step 1022). Subsequently, the system aggregates the multi-dimensional frequency features within the same time window: by weighted integration of feature dimensions such as axial pulsation frequency and radial eddy intensity, a multi-dimensional velocity vector comprehensively characterizing the flow field state is generated. This vector includes core flow field state indicators such as the overall flow direction of the flue gas, the uniformity of turbulence distribution, and the energy concentration zone (step 1023). For example, during furnace fire switching, the multidimensional velocity vector can simultaneously reflect the high-speed axial flow in the central area of ​​the flue and the low-frequency eddies in the edge area, accurately indicating the flow field mixing state at the inlet of the ammonia injection reaction chamber, providing key input for the dynamic control of the subsequent ammonia injection position and timing.

[0089] The overall scheme in step 102 above achieves intelligent aggregation and characterization of multi-dimensional frequency features of the flow field. It employs a waveform inflection point interval averaging algorithm to decompose the time-series waveform into frequency-dimensional features, and innovatively designs a time-window-driven multi-parameter aggregation framework. The generated multi-dimensional velocity vector integrates time-domain waveforms and frequency-domain characteristics, breaking through the characterization bottleneck of a single scalar parameter and forming a structured index that comprehensively reflects the complex state of the flow field.

[0090] 103. Perform spatial coordinate alignment and time series synchronization processing on the flow velocity pulsation data synchronously acquired by the ultrasonic Doppler sensor group to correct the measurement deviation of the multidimensional flow velocity vector, and extract the difference change of the corrected multidimensional flow velocity vector.

[0091] Optionally, step 103 may specifically include the following steps:

[0092] 1031. Obtain the spatial coordinates of the ultrasonic Doppler sensor group, and calculate the delay of ultrasonic wave transmission time between adjacent ultrasonic Doppler sensors based on the spatial coordinates.

[0093] 1032. Time series synchronization processing is achieved by adjusting the timestamp of the flow velocity pulsation data of the ultrasonic Doppler sensor based on the delay amount;

[0094] 1033. Based on the spatial location coordinates, the time-series synchronized flow velocity pulsation data is mapped to a preset three-dimensional flow field coordinate grid to generate a set of spatially aligned flow velocity pulsation data.

[0095] 1034. Replace the multidimensional velocity vector with velocity fluctuation data aligned to spatial coordinates to complete the correction of the multidimensional velocity vector;

[0096] 1035. Extract multidimensional velocity vectors with the same three-dimensional flow field coordinate grid points from the corrected multidimensional velocity vector to calculate the difference in the multidimensional velocity vector.

[0097] In the above scheme, velocity pulsation data refers to a set of velocity change data arranged in chronological order. Spatial coordinate alignment refers to unifying measurement data from different locations into the same coordinate system. Time series synchronization processing refers to adjusting the time reference of data collected by different sensors to keep them consistent. The spatial position coordinates of the ultrasonic Doppler sensor group refer to the specific position information of each sensor in three-dimensional space. The delay in ultrasonic wave transmission time refers to the time difference required for ultrasonic waves to propagate between different sensors. A timestamp is an identifier that records the time of data acquisition. Three-dimensional flow field coordinate grid points refer to pre-defined three-dimensional spatial grid points used to describe the flow field state. A multidimensional velocity vector refers to a mathematical vector characterizing the fluid motion state, composed of multiple velocity characteristic parameters. The corrected multidimensional velocity vector refers to the velocity characteristic vector after spatial and temporal calibration. The difference in change refers to the degree of change between multidimensional velocity vectors at the same location at different times.

[0098] In this embodiment, firstly, through step 1031, the system obtains the precise spatial coordinates (X / Y / Z installation coordinates) of the ultrasonic Doppler sensor group. The straight-line distance between adjacent sensors is measured using a spatial geometric calculation algorithm, and the delay in ultrasonic transmission time (e.g., the time Δt required for sensor A to travel to B) is calculated based on the propagation speed of sound in the gas medium (a fixed value). This delay quantifies the measurement time difference caused by spatial position differences, providing a correction basis for timing synchronization.

[0099] Subsequently, in step 1032, the flow velocity pulsation data is time-series corrected based on the calculated time delay: the timestamps of the data collected by each sensor are compensated in reverse according to the delay (e.g., the timestamp of sensor B's data is advanced by Δt), so that all sensor data are strictly aligned under a unified virtual time reference. This process achieves time series synchronization processing of multi-source data through a time axis offset algorithm, eliminating measurement time differences caused by spatial location.

[0100] Next, in step 1033, the synchronized flow velocity fluctuation data is mapped to a preset three-dimensional flow field coordinate grid: based on the spatial coordinates of the sensor and the geometric distance to the nearest grid point, the data is assigned to the corresponding grid coordinates using a nearest neighbor interpolation algorithm. This operation generates a set of flow velocity fluctuation data with precisely corresponding spatial coordinates, achieving accurate positioning of the measurement data in the three-dimensional flow field space.

[0101] Then, in step 1034, the data in the original multidimensional velocity vector is updated with the spatially aligned set of velocity fluctuation data: the original measurement values ​​at corresponding positions in the vector are replaced with the corrected grid data values. This replacement operation completes the correction of the multidimensional velocity vector through a data overlay algorithm, ensuring that the spatial and temporal attributes of each dimension of the vector are strictly aligned.

[0102] Finally, in step 1035, corrected multidimensional velocity vectors at different time points are extracted for the same three-dimensional flow field coordinate grid point: the vector components are grouped according to the same spatial location, and the component differences at the same spatial location at adjacent time points are calculated (e.g., X value at time t2 - X value at time t1), to obtain the change in difference at each spatial location. This change is output through a vector differential algorithm to measure the dynamic evolution characteristics of the flow field state.

[0103] In practical applications, during the flow field monitoring stage of the glass melting furnace flue gas denitrification system, the system first acquires the spatial coordinates of the ultrasonic Doppler sensor group (e.g., the three-dimensional coordinates of the sensor installation points on the four walls of the flue). Based on the geometric distance between each sensor and the propagation speed of ultrasonic waves in high-temperature exhaust gas, the system calculates the time delay of ultrasonic wave transmission between adjacent ultrasonic Doppler sensors (e.g., the acoustic wave transmission lag time between the top sensor and the side wall sensor) (step 1031). Based on this time delay, the system dynamically adjusts the timestamps of the flow velocity pulsation data of each sensor (e.g., advancing the timestamps of the data collected by the side wall sensors to match the timing of the top sensor), achieving time-series synchronization processing of data across spatial points and eliminating timing misalignment caused by differences in acoustic wave transmission paths (step 1032). Subsequently, based on the spatial coordinates (e.g., the angle between the sensor coordinates and the flue axis), the system maps the time-series synchronized flow velocity pulsation data to preset three-dimensional flow field coordinate grid points (e.g., a three-dimensional grid with the center of the flue as the origin), generating a set of spatially aligned flow velocity pulsation data (step 1033). This set directly replaces the original multidimensional velocity vector (such as the combined vector of axial pulsation frequency and radial vortex intensity) to complete the correction of the multidimensional velocity vector (step 1034). Finally, the system extracts the multidimensional velocity vectors of the same three-dimensional flow field coordinate grid points (such as the axial and radial velocity components at grid point [X,Y,Z]) from the corrected multidimensional velocity vector. By comparing the fluctuation trend differences of the same grid point vectors in adjacent time windows, the system calculates the difference in the multidimensional velocity vector (such as the abrupt change in axial pulsation frequency and the attenuation of radial vortex intensity). This change directly reflects the risk of flow field instability at the inlet of the denitrification reaction chamber (step 1035).

[0104] The overall scheme in step 103 above achieves accurate spatiotemporal field reconstruction capability across sensor data. A spatially aligned velocity dataset correction mechanism is constructed through ultrasonic transmission delay compensation and three-dimensional flow field coordinate grid mapping technology. Based on an algorithm for extracting differences in vectors at the same grid points, the error accumulation problem of traditional independent analysis of multi-sensor data is overcome, establishing a spatially distributed sensing dynamic benchmark model of the flow field.

[0105] 104. Integrate the difference changes to generate a flow field thermogram and establish a spatiotemporal correlation model between the multidimensional velocity vector and the dynamic flow field thermogram, and predict the air volume adjustment threshold that meets the requirements of the denitrification reaction through the spatiotemporal correlation model.

[0106] Optionally, step 104 may specifically include the following steps:

[0107] 1041. The differential change amount is superimposed on the three-dimensional flow field coordinate grid point to calculate the density distribution value of the three-dimensional flow field coordinate grid point to form a flow field thermogram representing the accumulation intensity of the differential change amount;

[0108] 1042. Extract the timestamp and vector element values ​​of the multidimensional velocity vector, and match the vector element values ​​at the same timestamp with the density distribution values ​​of the corresponding three-dimensional flow field coordinate grid points in the flow field thermogram;

[0109] 1043. Obtain the changing trend of the vector element values ​​within the timestamp and the changing trend of the density distribution values. When the direction of the changing trends is consistent, mark the vector element values ​​and the three-dimensional flow field coordinate grid points as a dynamically related region.

[0110] 1044. Statistically calculate the proportionality constant between the rate of change of the multidimensional velocity vector and the trend of the density distribution value in the dynamically associated region to establish the binding relationship between the proportionality constant and the timestamp as a spatiotemporal association model;

[0111] 1045. Based on the multidimensional flow velocity vector required for the denitrification reaction, the predicted value of the spatiotemporal correlation model is back-matched to generate the air volume adjustment threshold for driving the fan.

[0112] In the above scheme, the difference in change refers to the degree of change between multidimensional velocity vectors at the same location at different times. A flow field heatmap is a graphic representation of the spatial distribution of flow field parameters using color gradients. Three-dimensional flow field coordinate grid points are pre-defined three-dimensional spatial grid points used to describe the flow field state. A multidimensional velocity vector is a mathematical vector characterizing the fluid motion state, composed of multiple velocity characteristic parameters. A spatiotemporal correlation model is a mathematical model describing the temporal and spatial relationship between velocity characteristics and the flow field heatmap. A density distribution value is a statistical measure of the difference in change per unit volume. Aggregation intensity refers to the degree of spatial concentration of the difference in change. A dynamic correlation region refers to the spatial range where velocity changes and heatmap changes are synchronous. The direction of change trend refers to the increase or decrease characteristics of parameters over time. The rate of change refers to how quickly parameters change over time. A proportionality constant is a fixed ratio between the changes of two correlated parameters. A timestamp is an identifier for recording the time of data acquisition. Denitrification reaction refers to the chemical process of removing nitrogen oxides from exhaust gas. An airflow regulation threshold is a critical value for controlling the operating parameters of the blower.

[0113] In this embodiment, firstly, through step 1041, the system processes the difference change data: based on the baseline density value of each three-dimensional flow field coordinate grid point, the difference change is converted into a density increment through a spatial weighted superposition algorithm. Gradient accumulation calculation is performed on the same-direction difference within continuous regions (e.g., if the difference at grid point A is +0.3 and the difference at adjacent grid point B is +0.5, then an increasing gradient is formed from A to B), and finally, the density distribution value of each grid point is output. When the density values ​​of multiple grid points form a continuous gradient distribution, a flow field heatmap is generated through a thermal rendering engine, where the thermal intensity is proportional to the degree of aggregation of the difference change—high aggregation areas are presented as red gradients, and low aggregation areas are presented as blue gradients, visually representing the local focal areas of flow field state changes.

[0114] Subsequently, in step 1042, the precise timestamps and corresponding vector element values ​​(such as vector components [Vx, Vy, Vz]) contained in the multidimensional velocity vector are extracted. Simultaneously, the density distribution values ​​of the corresponding three-dimensional flow field coordinate grid points at the same timestamp are extracted from the flow field heatmap. A four-dimensional index (X, Y, Z, timestamp) is established using a spatiotemporal matching algorithm, strictly aligning the vector element values ​​and density distribution values ​​according to spatial location and time nodes, forming a paired dataset of "vector value-density value," thus establishing a bridging relationship for subsequent correlation analysis.

[0115] Next, step 1043 analyzes the variation patterns of paired datasets within the same time stamp: the derivative of the vector element value variation trend is calculated (three consecutive positive growth points are marked as "rising"), and the slope analysis of the density distribution value variation trend is performed simultaneously (continuous density increase is marked as "increasing"). When both satisfy the same trend direction at a specific coordinate point (e.g., vector value rises and density value increases), the system automatically marks the vector element at that time stamp and its associated three-dimensional flow field coordinate grid point as a dynamically associated region, indicating that there is a physical-level coordinated response mechanism between the flow field parameters and thermodynamic evolution at that location.

[0116] Then, in step 1044, parameter statistics are performed within the calibrated dynamic correlation region: the rate of change of the multidimensional velocity vector (the change in components per unit time ΔV / Δt) is calculated, and the slope of the density distribution value (ΔD / Δt) is measured simultaneously. The proportional relationship between the two is determined through linear regression analysis, generating a proportionality constant K=(ΔD / Δt) / (ΔV / Δt). The proportionality constant K is dynamically bound to a timestamp to form a time-varying parameter sequence, constructing a spatiotemporal correlation model. This model establishes a quantitative relationship between velocity change and thermal response; for example, K=0.8 indicates that for every 1 m / s increase in velocity, the density increases by 0.8 units.

[0117] Finally, in step 1045, based on the optimal multidimensional flow velocity vector parameter range required for the denitrification reaction (e.g., a flow velocity of 2.5 ± 0.3 m / s is required for ammonia mixing), the thermodynamic density value corresponding to the target flow velocity is retrieved and matched in the spatiotemporal correlation model. This thermodynamic density threshold is converted into a damper opening command, and the airflow adjustment threshold for driving the fan speed is output, thereby achieving precise control of the denitrification reaction gas flow rate as needed.

[0118] In practical applications, during the airflow control stage of the denitrification system in glass melting furnaces, the system first superimposes the differential changes (such as the abrupt change in axial velocity in the central area of ​​the flue and the attenuation of edge eddies) onto the three-dimensional flow field coordinate grid points. A weighted fusion algorithm is then used to calculate the density distribution values ​​of the three-dimensional flow field coordinate grid points, generating a flow field thermogram that uses color gradients to represent the aggregation intensity of the differential changes (step 1041). For example, the red high-density area formed in the middle of the flue due to the impact of fire exchange indicates strong turbulent aggregation. Simultaneously, the system extracts the timestamps and vector element values ​​(such as axial pulsation frequency and radial eddy intensity) of the multi-dimensional velocity vectors. The vector element values ​​at the same timestamp are matched with the density distribution values ​​of the corresponding three-dimensional flow field coordinate grid points in the flow field thermogram (step 1042). For example, the peak value of the axial pulsation frequency at 14:05 matches the peak value of the density at the same coordinate grid point in the thermogram. Based on temporal continuity, the system acquires the changing trends of vector element values ​​and density distribution values ​​within a timestamp (e.g., a continuous increase in axial pulsation frequency and a synchronous increase in density distribution values). When the changing trends are consistent (both increasing or decreasing), the vector element value and the three-dimensional flow field coordinate grid point are marked as a dynamic correlation region (step 1043). This type of region directly reflects the key turbulent coupling points of ammonia injection mixing efficiency. The system further calculates the proportionality constant between the rate of change of multidimensional velocity vectors and the changing trend of density distribution values ​​in the dynamic correlation region (e.g., a fixed ratio of the rate of change of axial velocity to the increase in density), and establishes a binding relationship between the proportionality constant and the timestamp as a spatiotemporal correlation model (step 1044). This model can quantify the dynamic mapping relationship between ammonia injection reaction efficiency and flow field state. Finally, the system back-matches the predicted values ​​of the spatiotemporal correlation model based on the multidimensional velocity vector required for the denitrification reaction (such as the ideal turbulence intensity to maintain an ammonia-nitrogen molar ratio of 1:1) and outputs the air volume adjustment threshold to ensure the sufficiency of the denitrification reaction (step 1045). For example, it is predicted that the fan speed needs to be increased to a specific threshold 5 seconds after the flame change to eliminate the vortex dead zone. This threshold is executed by the DCS system in real time to drive the fan to ensure that the flue gas is evenly diffused in the denitrification reaction chamber.

[0119] The overall scheme in step 104 above achieves collaborative modeling and prediction of the flow field thermodynamic state and fluid motion. It innovatively employs density distribution value overlay technology to generate differential change thermograms, combined with a timestamp-driven vector element and thermodynamic value trend matching mechanism. By extracting the proportional constant of dynamically correlated regions and mapping it inversely to denitrification requirements, it overcomes the lag limitation of empirical parameter adjustment, forming a predictive capability for airflow thresholds oriented towards reaction requirements.

[0120] 105. Convert the air volume adjustment threshold into a physical control signal that drives the fan speed change, and drive the fan actuator to change the fan speed based on the physical control signal.

[0121] Optionally, step 105 may specifically include the following steps:

[0122] 1051. The air volume adjustment threshold is converted into an electrical pulse modulation signal, and the electrical pulse modulation signal is converted into a physical control signal with a specific pulse width and frequency through a digital signal converter;

[0123] 1052. The drive fan actuator receives the physical control signal and adjusts the output frequency of the drive fan actuator according to the physical control signal to change the current intensity of the motor coil;

[0124] 1053. Based on the change in the current intensity of the motor coil, the rotation of the shaft bearing of the fan actuator is driven to achieve the adjustment of the fan speed.

[0125] In the above scheme, the airflow regulation threshold refers to the key critical value for controlling the fan's operating parameters. The physical control signal refers to the electrical signal that can directly drive the actuator. The electrical pulse modulation signal refers to the electrical signal that transmits control information through changes in pulse width or frequency. The digital signal converter is an electronic device that converts digital control signals into analog physical signals. The pulse width refers to the duration of a single electrical pulse. The pulse frequency refers to the number of repetitions of an electrical pulse per unit time. The fan actuator is the device that converts the control signal into mechanical action. The output frequency refers to the periodicity of the control signal output by the fan actuator. The motor coil current intensity refers to the magnitude of the current flowing through the motor windings. The shaft bearing refers to the mechanical component that supports the rotation of the fan impeller. The fan speed refers to the number of rotations of the fan impeller per unit time.

[0126] In this embodiment, firstly, through step 1051, the system converts the airflow adjustment threshold (e.g., the target fan speed can be 1500 rpm) into an electrical pulse modulation signal: the digital signal processor maps the threshold value to waveform parameters of a specific pulse width (e.g., the duty cycle can be 60%) and frequency (can be 1 kHz), driving the digital signal converter to generate the corresponding physical control signal. This signal is a high-precision square wave pulse, and its pulse width ratio directly corresponds to the percentage value of the target fan speed, realizing the accurate conversion of digital commands into physical electrical signals.

[0127] Subsequently, in step 1052, the circuit control unit of the fan actuator receives the physical control signal and analyzes the pulse width parameter in the signal through the pulse width demodulation module (e.g., a 60% duty cycle corresponds to a speed of 1500 rpm). Based on the analysis result, the control unit dynamically adjusts the output frequency of the fan actuator (e.g., increasing the motor drive frequency from 50Hz to 60Hz) and changes the current intensity of the input motor coil through the frequency converter circuit (e.g., increasing the current from 10A to 12A) to match the coil magnetic field strength to the target speed requirement.

[0128] Finally, in step 1053, the change in the current intensity of the motor coil causes a change in the electromagnetic field intensity: the enhanced current generates a stronger magnetic field, driving the shaft bearing to rotate faster under the action of the Lorentz force. The bearing transmits torque to the fan impeller through a mechanical transmission device (such as a gear set or coupling), achieving linear adjustment of the fan speed (e.g., from 1200 rpm to 1500 rpm). The speed change is fed back to the control system in real time, forming a closed-loop control of "command-current-speed" to ensure that the air volume is precisely matched to the denitrification reaction requirements.

[0129] In practical applications, in the airflow control stage of the glass melting furnace denitrification system, the system first converts the airflow adjustment threshold (such as the minimum airflow value required to eliminate the eddy current dead zone during the fire switching period) into an electrical pulse modulation signal. This signal is processed by a digital signal converter into a physical control signal with a specific pulse width and frequency (such as the pulse width corresponding to the airflow increase ratio and the frequency corresponding to the adjustment response speed) (step 1051). After receiving the physical control signal, the drive fan actuator adjusts the output frequency according to the physical control signal through the built-in frequency converter (such as increasing the frequency to match the pulse width increment), dynamically changing the current intensity of the motor coil (step 1052). Based on the change in the current intensity of the motor coil, the electromagnetic field intensity inside the drive fan actuator increases or decreases synchronously, driving the shaft bearing to rotate (such as the bearing speed increasing linearly with the current intensity), ultimately achieving precise adjustment of the fan speed (step 1053). For example, when a sudden drop in flow velocity is detected in the central area of ​​the flue during the initial stage of kiln fire switching, the physical control signal drives the fan speed to increase from the baseline value to the target threshold within 5 seconds, ensuring that the flow field at the inlet of the denitrification reaction chamber quickly recovers to a uniform distribution and avoiding ammonia escape caused by local low velocity.

[0130] The overall scheme in step 105 above achieves a lossless transmission link from control commands to physical drives. Through a two-stage processing architecture of electrical pulse modulation and digital signal conversion, abstract thresholds are transformed into physical control signals that can drive actuators. Based on the precise mapping relationship between pulse width / frequency and motor current intensity, the anti-interference bottleneck of traditional analog signal control is overcome, achieving millisecond-level precise response of the fan speed.

[0131] 106. Obtain the update frequency of the exhaust gas pulsation response data and the dynamic thermal map of the flow field caused by the change of fan speed, and adjust the multidimensional velocity vector based on the correlation between the exhaust gas pulsation response data and the update frequency to form a closed-loop control of data acquisition and execution feedback.

[0132] Optionally, step 106 may specifically include the following steps:

[0133] 1061. Collect exhaust gas pulsation response data generated by the ultrasonic Doppler sensor group after the fan speed changes and record the update frequency of the dynamic thermogram of the flow field;

[0134] 1062. Calculate the ratio of the number of occurrences of the flow velocity fluctuation peak in the exhaust gas pulsation response data to the update frequency. When the number of occurrences of the fluctuation peak exceeds a preset multiple of the update frequency, it is marked as a high-frequency pulsation event.

[0135] 1063. Extract the thermal density distribution of the flow field dynamic thermogram corresponding to the time window of the occurrence of the high-frequency pulsating event, and extract the gradient change direction of the thermal density distribution within the time window;

[0136] 1064. Compare the gradient change direction with the change direction of the multidimensional velocity vector in the same time window in the multidimensional velocity vector. If the deviation of the change direction exceeds a preset allowable angle, adjust the spatial sampling point density of the multidimensional velocity vector.

[0137] 1065. The time series of data acquisition from the ultrasonic Doppler sensor group is reconfigured based on the adjusted spatial sampling point density to form a closed-loop control signal, and the acquisition of the exhaust gas pulsation response data is driven according to the closed-loop control signal.

[0138] In the above scheme, exhaust gas pulsation response data refers to the data on the change in exhaust gas flow state within the flue after a change in fan speed. The dynamic thermal map of the flow field refers to a real-time graphic representation of the spatial distribution of flow field parameters, visually displayed through color gradients. The update frequency refers to the rate at which the dynamic thermal map of the flow field is regenerated and displayed. Closed-loop control refers to a system adjustment method that continuously optimizes the control process through real-time feedback data. High-frequency pulsation events refer to abnormally rapid fluctuations in exhaust gas flow. The time window refers to a specific time period used for signal analysis. The thermal density distribution refers to the spatial distribution of energy intensity at each point in the thermal map of the flow field. The gradient change direction refers to the direction of the greatest spatial rate of change of thermal density. The multidimensional velocity vector refers to a mathematical vector characterizing the fluid motion state, composed of multiple velocity characteristic parameters. The spatial sampling point density refers to the number of ultrasonic Doppler sensor arrays arranged per unit space. The closed-loop control signal refers to the feedback control command used to achieve automatic system adjustment.

[0139] In this embodiment, firstly, through step 1061, the system collects in real time the exhaust gas pulsation response data (including flow velocity fluctuation waveform) generated by the ultrasonic Doppler sensor group after the fan speed changes, and simultaneously monitors the refresh frequency of the dynamic thermal map of the flow field (e.g., twice per second). This process synchronously transmits the exhaust gas velocity time series and the thermal map update timestamp through the data bus, establishing a time-domain correlation between the control action and the flow field response.

[0140] Subsequently, in step 1062, the flow velocity waveform characteristics in the exhaust gas pulsation response data are analyzed: the number of peak fluctuations within a continuous time window is scanned (e.g., 120 peaks per minute), and the ratio of peak occurrence frequency to update frequency is calculated (e.g., 120 times / 30Hz = 4 times). When this ratio exceeds a preset multiple threshold (which can be 3 times), the system automatically marks it as a high-frequency pulsation event, indicating abnormal turbulent oscillations in the flow field.

[0141] Next, in step 1063, based on the specific time window of the high-frequency pulsation event, thermal data for the corresponding period of the flow field dynamic thermal map is extracted: the rate of change of thermal density in three-dimensional space during this period is calculated using a spatial gradient algorithm, and the gradient change direction vector of the direction with the maximum rate of change is extracted (the density gradient is strongest along the 45° direction of the flue axis). This direction reflects the propagation path of thermal anomalies.

[0142] Then, in step 1064, the direction vector of the thermal gradient change is compared with the spatial change direction of the multidimensional velocity vector within the same time window: the cosine of the angle between the two vectors is calculated. If the absolute value of the cosine value deviating from 1 exceeds the preset allowable angle (which can be >15°), it indicates a significant directional deviation between the thermal evolution of the flow field and the velocity change. Based on this, the system initiates a spatial sampling point density adjustment mechanism—increasing the density of ultrasonic sensor placement in the deviation area to improve monitoring resolution.

[0143] Finally, through step 1065, the system reconfigures the data acquisition timing of the ultrasonic Doppler sensor group based on the adjusted spatial sampling point density: in the detected abnormal area, the sampling frequency is increased (from 10Hz to 15Hz), and the activation time points of each sensor are reallocated through the timing optimization algorithm to eliminate signal interference. A closed-loop control signal containing regional coordinates, sampling frequency and sensor activation sequence is generated, and this signal is embedded into the control loop to drive the high-resolution acquisition of exhaust gas pulsation response data in real time. This completes the closed-loop control process of "response analysis → thermal map diagnosis → sampling optimization → data re-acquisition", ensuring that the flow field state monitoring dynamically matches the exhaust gas flow characteristics under the change of fan speed.

[0144] In practical applications, in the closed-loop control of the glass melting furnace denitrification system, the system first collects the flue gas pulsation response data generated by the ultrasonic Doppler sensor group after the fan speed changes (such as the signal of a sudden increase in flow velocity in the central area of ​​the flue caused by the fan speed increase after the fire change), and simultaneously records the update frequency of the dynamic thermal map of the flow field (such as the real-time density distribution of the thermal map refreshed once per second) (step 1061). Based on time series analysis, the system calculates the ratio of the number of occurrences of the flow velocity fluctuation peak in the flue gas pulsation response data to the update frequency (such as detecting 12 flow velocity peaks within 1 minute after the fire change, while the thermal map is only updated 6 times). When the number of occurrences of the fluctuation peak exceeds a preset multiple of the update frequency (such as the number of peaks being more than twice the update frequency), the period is marked as a high-frequency pulsation event (step 1062). Subsequently, the system extracts the thermal density distribution of the dynamic thermal map corresponding to the time window of the high-frequency pulsation event (e.g., the thermal map color block sequence from 0 to 60 seconds after the fire switch), and extracts the gradient change direction of the thermal density distribution within this time window (e.g., the radial gradient of thermal density decreasing from the center of the flue to the edge) (step 1063). This gradient change direction is compared with the change direction of the multidimensional velocity vector in the same time window in the multidimensional velocity vector. If the deviation of the change direction exceeds the preset allowable angle (e.g., the angle between the axial velocity enhancement and the radial density decay > 30°), the spatial sampling point density of the multidimensional velocity vector is adjusted (e.g., the sensor sampling point density in the flue edge region is increased by 50% to capture radial vortex details) (step 1064). Finally, the system reconfigures the time series of data acquisition of the ultrasonic Doppler sensor group based on the adjusted spatial sampling point density, forming a closed-loop control signal to drive the fine acquisition of exhaust gas pulsation response data (step 1065). For example, during the high-frequency pulsation event window after the fire switch, the eddy regeneration process of the flue dead zone is captured by encrypted sampling, providing dynamic flow field mapping for the orientation adjustment of the ammonia injection grid.

[0145] The overall scheme in step 106 above achieves dynamic closed-loop optimization of execution feedback and data acquisition. Through correlation analysis between the peak frequency of pulsations and the update frequency of the heatmap, an intelligent calibration mechanism for high-frequency pulsation events is constructed. An innovative gradient direction deviation-driven spatial sampling density adaptive adjustment algorithm is designed to overcome the response hysteresis defect of static acquisition strategies, forming a closed-loop control system for data acquisition based on real-time flow field feedback.

[0146] The following are specific examples for steps 101 to 106, such as Figure 2 As shown:

[0147] In the SCR denitrification system of a float glass melting furnace, an ultrasonic Doppler airflow closed-loop control method is applied. Ultrasonic Doppler sensor arrays are uniformly arranged on the four walls of the float glass melting furnace flue according to preset three-dimensional coordinates, forming a closed-loop detection structure. The sensors have built-in trigger controllers that generate synchronous pulse signals, driving all sensors to simultaneously emit ultrasonic beams. When the ultrasonic beams penetrate the dust-laden flue gas, they are reflected by the gas particles to form echo signals. Each sensor calculates the instantaneous flow velocity value at its measurement location within a fixed time window, and generates a flow velocity pulsation data sequence sorted by timestamp, capturing the transient waveforms of sudden velocity drops in the central area of ​​the flue and enhanced edge vortices during flame changeover.

[0148] The system segments the time-series waveform of the velocity fluctuation data sequence into independent waveform segments of fixed duration, extracts the average time interval between adjacent waveform inflection points within each segment, and uses this as a multi-dimensional frequency feature characterizing the turbulence intensity. The system then aggregates the multi-dimensional frequency features (such as axial fluctuation frequency and radial eddy intensity) of all independent waveform segments within a unified time window to generate a multi-dimensional velocity vector that comprehensively reflects the mixing uniformity of the flow field.

[0149] The time delay of ultrasonic transmission between adjacent sensors is calculated based on the spatial coordinates of the sensor array. The timestamps of each sensor's data are then adjusted accordingly to achieve time-series synchronization. The synchronized velocity fluctuation data is mapped to a preset three-dimensional flow field coordinate grid, generating a spatially aligned velocity fluctuation data set. This set replaces the original multidimensional velocity vector to complete the correction. The axial and radial components at the same grid point are extracted from the corrected vector, and the differences before and after the ignition switch are calculated (e.g., the coupling effect of axial velocity attenuation and radial vortex enhancement).

[0150] The differential changes are superimposed onto the 3D flow field coordinate grid points. The density distribution values ​​of each grid point are calculated through weighted fusion, generating a flow field heatmap that uses color gradients to represent the intensity of differential changes (e.g., a high-density area in the center of the flue indicates strong turbulence attenuation). The timestamps and vector element values ​​of the multidimensional velocity vectors are extracted and matched with the density distribution values ​​of corresponding grid points in the heatmap at the same timestamp. When the trends of both are consistent (e.g., both continuously decreasing), the grid point is marked as a dynamically correlated region (i.e., the critical area for ammonia mixing). The proportionality constant between the rate of change of the multidimensional velocity vector and the trend of change of the density distribution value in the dynamically correlated region is statistically analyzed, and a spatiotemporal correlation model is constructed by binding the timestamp. Based on the ideal turbulence conditions required for SCR denitrification (e.g., maintaining a 1:1 ammonia-nitrogen molar ratio for flow field uniformity), the predicted values ​​of the reverse matching model are used to generate airflow adjustment thresholds (e.g., the fan needs to be accelerated to eliminate low-speed areas at the edges).

[0151] The airflow adjustment threshold is converted into a physical control signal with a specific pulse width and frequency by a digital signal converter (pulse width corresponds to the speed increase ratio, and frequency corresponds to the response speed). After receiving the signal, the drive fan actuator adjusts the output frequency to increase the current intensity of the motor coil, driving the shaft bearing to rotate and thus increasing the fan speed. After the speed increase, the vortex intensity in the flue edge area is significantly reduced, and the flow field uniformity is improved.

[0152] The closed-loop system collects exhaust gas pulsation response data after fan speed changes (e.g., changes in the number of vortex peaks after speed increase) and compares it with the update frequency of the dynamic thermal map of the flow field. When the number of fluctuation peaks exceeds a preset multiple of the update frequency, the time period is marked as a high-frequency pulsation event. The thermal map sequence within the event time window is extracted, and the gradient change direction of the thermal density distribution is extracted (e.g., the radial decreasing gradient from the center to the edge of the flue). This gradient is compared with the change direction of the multidimensional velocity vector in the same time window. If the deviation exceeds the allowable angle (e.g., the angle between a sudden increase in axial velocity and the radial diffusion direction of thermal density is too large), the spatial sampling point density of the multidimensional velocity vector is adjusted (e.g., increasing the sensor density at the bottom of the flue). Based on the adjusted density, the data acquisition time series of the ultrasonic Doppler sensor group is reconfigured to generate a closed-loop control signal to drive refined data acquisition, thereby optimizing the airflow pre-adjustment strategy for the next fire-changing cycle.

[0153] Figure 3 This application provides a schematic diagram of a glass melting furnace exhaust gas denitrification airflow precision control system, as shown in the embodiment. Figure 3 As shown, the system includes:

[0154] The acquisition module 31 is used to synchronously emit ultrasonic beams through an ultrasonic Doppler sensor group installed around the glass melting furnace flue to acquire flow velocity pulsation data reflecting the air volume requirement for denitrification of waste gas, and to obtain the waveform of the time series of the flow velocity pulsation data.

[0155] The decomposition module 32 is used to decompose the waveform of the time series to form multi-dimensional frequency features, and to aggregate the multi-dimensional frequency features collected in the same time window to generate a multi-dimensional velocity vector that comprehensively represents the flow field state.

[0156] The calibration module 33 is used to perform spatial coordinate alignment and time series synchronization processing on the flow velocity pulsation data synchronously acquired by the ultrasonic Doppler sensor group to correct the measurement deviation of the multidimensional flow velocity vector, and extract the difference change of the corrected multidimensional flow velocity vector.

[0157] The fusion module 34 is used to fuse the differential changes to generate a flow field heat map and establish a spatiotemporal correlation model between the multidimensional velocity vector and the dynamic heat map of the flow field, and predict the air volume adjustment threshold that meets the requirements of the denitrification reaction through the spatiotemporal correlation model.

[0158] The conversion module 35 is used to convert the air volume adjustment threshold into a physical control signal that drives the fan speed to change, and to drive the fan actuator to change the fan speed based on the physical control signal.

[0159] The adjustment module 36 is used to acquire the update frequency of the exhaust gas pulsation response data and the dynamic thermal map of the flow field caused by the change of fan speed, and adjust the multidimensional velocity vector based on the correlation between the exhaust gas pulsation response data and the update frequency to form a closed-loop control of data acquisition and execution feedback.

[0160] Figure 3 The aforementioned glass melting furnace exhaust gas denitrification air volume precision control system can perform... Figure 1 The implementation principle and technical effects of the precise control method for denitrification airflow in glass melting furnace exhaust gas described in the illustrated embodiment will not be repeated here. The specific operation methods of each module and unit in the precise control system for denitrification airflow in glass melting furnace exhaust gas described in the above embodiments have been described in detail in the embodiments related to this method, and will not be elaborated upon here.

[0161] In one possible design, Figure 3 The glass melting furnace exhaust gas denitrification air volume precision control system of the embodiment shown can be implemented as a computing device, such as... Figure 4 As shown, the computing device may include a storage component 41 and a processing component 42;

[0162] The storage component 41 stores one or more computer instructions, wherein the one or more computer instructions are invoked and executed by the processing component 42.

[0163] The processing component 42 is used for the above Figure 1 The embodiment describes a method for precise control of denitrification air volume in glass melting furnace exhaust gas.

[0164] The processing component 42 may include one or more processors to execute computer instructions to complete all or part of the steps in the above-described method. Alternatively, the processing component may be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-described method.

[0165] Storage component 41 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0166] Of course, computing devices may also include other components, such as input / output interfaces, display components, communication components, etc.

[0167] Input / output interfaces provide interfaces between processing components and peripheral interface modules, which can be output devices, input devices, etc.

[0168] The communication components are configured to facilitate wired or wireless communication between computing devices and other devices.

[0169] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform. In this case, the computing device can refer to a cloud server, and the aforementioned processing components, storage components, etc., can be basic server resources rented or purchased from the cloud computing platform.

[0170] This application also provides a computer storage medium storing a computer program, which, when executed by a computer, can perform the above-described functions. Figure 1 The embodiment shown illustrates a method for precise control of denitrification air volume in glass melting furnace exhaust gas.

[0171] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0172] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0173] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0174] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for precise control of denitrification air volume in glass melting furnace exhaust gas, characterized in that, include: An ultrasonic Doppler sensor array installed around the glass melting furnace flue synchronously emits ultrasonic beams to collect flow velocity pulsation data reflecting the air volume requirement for denitrification of waste gas, and obtains the waveform of the time series of the flow velocity pulsation data. The waveform of the time series is decomposed to form multi-dimensional frequency features, and the multi-dimensional frequency features collected in the same time window are aggregated to generate a multi-dimensional velocity vector that comprehensively represents the flow field state. Spatial coordinate alignment and time series synchronization processing are performed on the flow velocity pulsation data synchronously acquired by the ultrasonic Doppler sensor group to correct the measurement deviation of the multidimensional flow velocity vector, and the difference change of the corrected multidimensional flow velocity vector is extracted. The differential change quantities are integrated to generate a flow field thermogram and a spatiotemporal correlation model is established between the multidimensional velocity vector and the dynamic flow field thermogram. The spatiotemporal correlation model is then used to predict the air volume adjustment threshold that meets the requirements of the denitrification reaction. The air volume adjustment threshold is converted into a physical control signal that drives the fan speed to change, and the fan actuator is driven to change the fan speed based on the physical control signal; The update frequency of the exhaust gas pulsation response data caused by the change of fan speed and the dynamic thermogram of the flow field is obtained, and the multidimensional velocity vector is adjusted based on the correlation between the exhaust gas pulsation response data and the update frequency to form a closed-loop control of data acquisition and execution feedback.

2. The method according to claim 1, characterized in that, An ultrasonic Doppler sensor array installed around the glass melting furnace flue synchronously emits ultrasonic beams to collect flow velocity pulsation data reflecting the airflow demand for denitrification of the exhaust gas, and acquires the time series waveform of the flow velocity pulsation data, including: An ultrasonic Doppler sensor array is arranged at predetermined points on the four walls of the glass melting furnace flue. The measurement positions of each ultrasonic Doppler sensor are evenly distributed according to the cross-section of the glass melting furnace flue to form a closed-loop detection structure. In the closed-loop detection structure, a synchronous pulse signal is generated by the trigger controller built into the ultrasonic Doppler sensor group to drive the ultrasonic Doppler sensor group to emit ultrasonic beams at the same time. After emitting the ultrasonic beam, each ultrasonic Doppler sensor receives the echo signal reflected by the exhaust gas particles and calculates the instantaneous flow velocity value at each measurement location within the time window of the echo signal. The instantaneous flow velocity values ​​at each measurement location are continuously collected and arranged in time sequence to form a flow velocity pulsation data sequence with timestamps, and the waveform of the time series in the flow velocity pulsation data sequence is extracted.

3. The method according to claim 1, characterized in that, The waveform of the time series is decomposed to form multi-dimensional frequency features, and the multi-dimensional frequency features collected in the same time window are aggregated to generate a multi-dimensional velocity vector that comprehensively represents the flow field state, including: The waveform of the time series is divided into independent waveform segments with fixed time intervals, and the average time interval between adjacent waveform inflection points within each independent waveform segment is calculated as a multi-dimensional frequency feature. Based on the same time window, multi-dimensional frequency characteristics of independent waveform segments within the time window are collected. The multi-dimensional frequency features are aggregated to generate a multi-dimensional velocity vector that comprehensively represents the flow field state.

4. The method according to claim 1, characterized in that, The flow velocity pulsation data synchronously acquired by the ultrasonic Doppler sensor array are subjected to spatial coordinate alignment and time series synchronization processing to correct the measurement deviation of the multidimensional flow velocity vector, and the difference change of the corrected multidimensional flow velocity vector is extracted, including: Obtain the spatial coordinates of the ultrasonic Doppler sensor group, and calculate the delay of ultrasonic wave transmission time between adjacent ultrasonic Doppler sensors based on the spatial coordinates. Time series synchronization processing is achieved by adjusting the timestamp of the flow velocity pulsation data of the ultrasonic Doppler sensor based on the delay amount. Based on the spatial location coordinates, the time-series synchronized flow velocity fluctuation data is mapped to a preset three-dimensional flow field coordinate grid to generate a set of spatially aligned flow velocity fluctuation data. The multidimensional velocity vector is corrected by replacing the velocity fluctuation data with spatially aligned velocity data. Multidimensional velocity vectors with the same three-dimensional flow field coordinate grid points are extracted from the corrected multidimensional velocity vectors to calculate the difference in the multidimensional velocity vectors.

5. The method according to claim 4, characterized in that, The differential changes are fused to generate a flow field thermogram, and a spatiotemporal correlation model is established between the multidimensional velocity vector and the dynamic flow field thermogram. The spatiotemporal correlation model is then used to predict the airflow adjustment threshold required to meet the denitrification reaction, including: The differential change is superimposed onto the three-dimensional flow field coordinate grid to calculate the density distribution value of the three-dimensional flow field coordinate grid, thereby forming a flow field thermogram that represents the aggregation intensity of the differential change. Extract the timestamp and vector element values ​​of the multidimensional velocity vector, and match the vector element values ​​at the same timestamp with the density distribution values ​​of the corresponding three-dimensional flow field coordinate grid points in the flow field thermogram; Obtain the changing trend of vector element values ​​and the changing trend of density distribution values ​​within the timestamp. When the directions of the changing trends are consistent, mark the vector element values ​​and the three-dimensional flow field coordinate grid points as a dynamically associated region. The proportionality constant between the rate of change of the multidimensional velocity vector and the trend of the density distribution value in the dynamically associated region is statistically analyzed to establish the binding relationship between the proportional constant and the timestamp as a spatiotemporal association model; The predicted values ​​of the spatiotemporal correlation model are back-matched with the multidimensional flow velocity vector required for the denitrification reaction to generate the air volume adjustment threshold that drives the fan.

6. The method according to claim 1, characterized in that, Converting the airflow adjustment threshold into a physical control signal that drives the fan speed change, and driving the fan actuator to change the fan speed based on the physical control signal, includes: The airflow adjustment threshold is converted into an electrical pulse modulation signal, and the electrical pulse modulation signal is converted into a physical control signal with a specific pulse width and frequency by a digital signal converter. The drive fan actuator receives the physical control signal and adjusts the output frequency of the drive fan actuator according to the physical control signal to change the current intensity of the motor coil; The fan speed is adjusted by rotating the shaft bearing of the fan actuator based on the change in the current intensity of the motor coil.

7. The method according to claim 1, characterized in that, The update frequency of the exhaust gas pulsation response data caused by changes in fan speed and the dynamic thermogram of the flow field is obtained. Based on the correlation between the exhaust gas pulsation response data and the update frequency, the multidimensional velocity vector is adjusted to form a closed-loop control of data acquisition and execution feedback, including: Collect exhaust gas pulsation response data generated by the ultrasonic Doppler sensor group after the fan speed changes and record the update frequency of the dynamic thermogram of the flow field; Calculate the ratio of the number of occurrences of the flow velocity fluctuation peak in the exhaust gas pulsation response data to the update frequency. When the number of occurrences of the fluctuation peak exceeds a preset multiple of the update frequency, it is marked as a high-frequency pulsation event. The thermal density distribution of the flow field dynamic thermogram is extracted according to the time window of the occurrence of the high-frequency pulsating event, and the gradient change direction of the thermal density distribution within the time window is extracted. The gradient change direction is compared with the change direction of the multidimensional velocity vector in the same time window. If the deviation of the change direction exceeds a preset allowable angle, the spatial sampling point density of the multidimensional velocity vector is adjusted. The time series of data acquisition from the ultrasonic Doppler sensor array is reconfigured based on the adjusted spatial sampling point density to form a closed-loop control signal, and the acquisition of the exhaust gas pulsation response data is driven according to the closed-loop control signal.

8. A precise control system for denitrification air volume in glass melting furnace exhaust gas, characterized in that, include: The acquisition module is used to synchronously emit ultrasonic beams through an ultrasonic Doppler sensor group installed around the glass melting furnace flue to acquire flow velocity pulsation data reflecting the air volume requirement for denitrification of exhaust gas, and to obtain the waveform of the time series of the flow velocity pulsation data. The decomposition module is used to decompose the waveform of the time series to form multi-dimensional frequency features, and to aggregate the multi-dimensional frequency features collected in the same time window to generate a multi-dimensional velocity vector that comprehensively represents the flow field state. The calibration module is used to perform spatial coordinate alignment and time series synchronization processing on the flow velocity pulsation data synchronously acquired by the ultrasonic Doppler sensor group to correct the measurement deviation of the multidimensional flow velocity vector, and extract the difference change of the corrected multidimensional flow velocity vector. The fusion module is used to fuse the differential changes to generate a flow field heat map and establish a spatiotemporal correlation model between the multidimensional velocity vector and the dynamic flow field heat map, and predict the air volume adjustment threshold that meets the requirements of the denitrification reaction through the spatiotemporal correlation model. The conversion module is used to convert the air volume adjustment threshold into a physical control signal that drives the fan speed change, and drive the fan actuator to change the fan speed based on the physical control signal; The adjustment module is used to acquire the update frequency of the exhaust gas pulsation response data and the dynamic thermal map of the flow field caused by the change of fan speed, and adjust the multidimensional velocity vector based on the correlation between the exhaust gas pulsation response data and the update frequency to form a closed-loop control of data acquisition and execution feedback.

9. A computing device, characterized in that, It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are invoked and executed by the processing component to realize the method for precise control of denitrification air volume of glass melting furnace exhaust gas as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that, The device contains a computer program that, when executed by a computer, implements a method for precise control of denitrification air volume in glass melting furnace exhaust gas as described in any one of claims 1 to 7.

Images