Stable combustion control system based on multi-dimensional furnace combustion measurement method

Abstract

This invention discloses a stable combustion control system based on a multi-dimensional furnace combustion measurement method, specifically relating to the field of furnace combustion measurement. This includes multi-dimensional sensor data acquisition, feature parameter quantification and extraction, combustion stability modeling, intelligent graded stable combustion control, iterative optimization of control effect, and dynamic evaluation of combustion state. Multi-dimensional sensor data acquisition involves synchronously acquiring data through an array of sensors. Feature parameter quantification and extraction involves preprocessing the acquired data to extract key feature parameters. Combustion stability modeling involves constructing a quantitative evaluation model of the combustion stability index based on the key feature parameters. Intelligent graded stable combustion control involves triggering a three-stage direct injection response mechanism for the fuel gun based on the combustion stability index, combined with an adaptive energy matching algorithm to dynamically adjust the stable combustion energy injection. This invention, through a multi-dimensional sensor array and feature extraction algorithm, achieves advanced identification of early combustion deterioration, solving the problem of lag in single flame detector monitoring.

Description

Technical Field

[0001] This invention relates to the field of furnace combustion measurement technology, and more specifically, to a stable combustion control system based on a multi-dimensional furnace combustion measurement method. Background Technology

[0002] With the current high proportion of new energy installed capacity, deep peak shaving of coal-fired units has shifted from unconventional operating conditions to normalized operation mode, with ultra-low load operation time below 30% of rated load accounting for an average of 28% annually. At this time, the sudden drop in furnace heat load leads to a sharp deterioration in combustion stability. The existing single-dimensional flame detection monitoring system has significant technical defects: the ultraviolet flame detection signal response delay is high, and it can only reflect the state after combustion has completely deteriorated; the mechanical atomizing oil gun has a long time from receiving the DCS command to completing atomization and flame establishment, while the critical time window for the deterioration of pulverized coal combustion is very short, resulting in boiler fire extinguishing accidents caused by delayed oil injection.

[0003] More critically, existing technologies suffer from three major technical bottlenecks: First, they lack a multi-physics coupling feature extraction mechanism for early-stage combustion deterioration, making it impossible to provide early warnings in the early stages of signs such as flame morphology variations and abrupt temperature gradient changes; second, the oil gun control strategy uses fixed threshold logic and fails to establish a dynamic mapping relationship between combustion stability and oil gun output, resulting in excessive or insufficient combustion stability during oil injection; third, the correlation analysis between furnace negative pressure fluctuations and combustion status remains at the qualitative level, failing to form a quantitative judgment model based on fluid mechanics, leading to false alarms that interfere with normal operation.

[0004] Therefore, there is an urgent need for a stable combustion control system based on multi-dimensional furnace combustion measurement methods to solve the above problems. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides a stable combustion control system based on a multi-dimensional furnace combustion measurement method, which solves the problems mentioned in the background art through the following scheme.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a stable combustion control system based on a multi-dimensional furnace combustion measurement method, comprising: multi-dimensional sensing data acquisition: synchronously acquiring data through an array of sensors;

[0007] Feature parameter quantization and extraction: Preprocess the collected data to extract key feature parameters;

[0008] Combustion stability modeling: Constructing a quantitative evaluation model for the combustion stability index based on key characteristic parameters;

[0009] Intelligent graded stable combustion control: Triggers a three-stage direct injection response mechanism based on the combustion stability index, and dynamically adjusts the stable combustion energy injection in combination with an adaptive energy matching algorithm;

[0010] Iterative optimization of control effect: The combustion stabilization effect is evaluated through a multi-dimensional constraint verification mechanism, and the control strategy is iteratively adjusted according to the combustion recovery conditions;

[0011] Dynamic assessment of combustion status: Calculates the deviation of the current combustion status from historical benchmark values ​​to achieve real-time early warning and assessment of combustion stability.

[0012] Preferably, the multi-dimensional sensing data acquisition includes flame image acquisition, infrared temperature field monitoring, furnace negative pressure acquisition, and composite signal acquisition; the flame image acquisition uses 12 sets of dual-wavelength high-speed cameras, a sampling frequency of 1000fps, a resolution of 4096×2160, and a dynamic aperture adjustment range of 0.1-10000cd / m². 2 The infrared temperature field monitoring system utilizes a 32×32 elemental mercury cadmium telluride focal plane detector with a temperature range of 300-2200℃ and a spatial resolution of 0.8mrad, equipped with a rotary water-cooled purging device. The furnace negative pressure acquisition system integrates a 16-channel negative pressure sensor with a range of -8000Pa to +8000Pa, a measurement accuracy of 0.2%FS, and a response time ≤5ms. The composite signal acquisition system comprises an 8-channel ultraviolet flame detector and a 4-channel acoustic sensor, using 316L stainless steel armored cables.

[0013] Preferably, the feature parameter quantization extraction includes flame fractal dimension calculation, temperature field gradient tensor extraction, negative pressure high-frequency component extraction, and acoustic wave energy entropy calculation; the flame fractal dimension calculation: Where r represents the measurement scale, N(r) is the number of squares covering the edge of the flame, and the sampling interval is 10ms; the temperature field gradient tensor is calculated using the finite difference method with a spatial step size of 0.5m; the negative pressure high-frequency component extraction is performed by obtaining the 50-200Hz frequency band components through empirical mode decomposition. Among them, IMF i Represents the i-th order intrinsic mode function, filtering components in the frequency range of 50-200Hz; the acoustic energy entropy is calculated as follows: Where p i This represents the energy percentage of the i-th frequency band.

[0014] Preferably, the combustion stability index in σ(T) represents the coefficient of variation of the temperature field, and σ(T) represents the standard deviation of temperature. This represents the average temperature; when CSI < 0.65, stable combustion intervention is triggered. This threshold is determined by the ROC curve of 3000 sets of historical data, with a false alarm rate ≤ 3% and a false negative rate ≤ 5%.

[0015] Preferably, the intelligent hierarchical stable combustion control strategy includes a three - level response mechanism, an energy matching algorithm, and a combustion recovery determination; the three - level response mechanism: the first - level response CSI = 0.65 - 0.55: 2 oil guns are pulsed and put into operation, the pulse width is 200 ms, the interval is 150 ms, the propulsion time ≤ 300 ms, and the atomizing air pressure is 0.8 - 1.0 MPa; the second - level response CSI = 0.55 - 0.45: 4 oil guns are continuously injected, the propulsion time ≤ 200 ms, and the atomizing air pressure is 1.2 - 1.5 MPa; the third - level response CSI < 0.45: 6 oil guns are fully put into operation, the propulsion time ≤ 150 ms, the atomizing air pressure is 1.8 - 2.0 MPa, and the plasma ignition is started synchronously; the energy matching algorithm where Q r represents the real - time stable combustion energy demand, and the adjustment period is 100 ms; the combustion recovery determination: when CSI ≥ 0.75 for 30 s continuously and Df ≥ 1.7, Cv ≤ 0.15, 1 oil gun is withdrawn every 5 s.

[0016] Preferably, the iterative optimization of the control effect includes sensor fault diagnosis, data filtering processing, and control effect evaluation; the sensor fault diagnosis: a triple - check mechanism, an alarm for a redundant comparison deviation > 15%, interpolation substitution for a temperature over - range of 2000 °C, and activation of a standby sensor for a feature mutation rate > 50%; the data filtering processing: wavelet packet 5 - layer decomposition to remove < 3% energy noise, combined with Kalman filtering, and the error after filtering ≤ 3%; the control effect evaluation: when the increase in CSI within 1.8 s after oil injection is < 0.15, oil guns are added, and the iteration is terminated when the change rate of CSI in 5 consecutive control cycles < 5%.

[0017] Preferably, the dynamic evaluation method of the combustion state is: calculating the deviation degree where CSI ref represents the historical reference value at the same load. When CD ≤ 0.15, the combustion is stable; when 0.15 < CD ≤ 0.3, an early warning is given; when CD > 0.3, a three - level response is started.

[0018] The technical effects and advantages of the present invention are as follows:

[0019] 1. Through the multi - dimensional sensor array and the feature extraction algorithm, the present invention realizes the early identification of the deterioration in the initial stage of combustion, solves the problem of the lag of single - flame detector monitoring, advances the combustion anomaly identification time, and uses a dual - wavelength high - speed camera and an infrared focal - plane detector, combined with the calculation of the flame fractal dimension and the analysis of the temperature - field gradient tensor, to capture the micro - scale pulsation of the flame and the mesoscopic change of the temperature field in real time. Compared with the high response delay of the traditional ultraviolet flame detector, this method can trigger an early warning within the critical window of combustion deterioration, improve the accuracy of combustion anomaly identification, reduce the false alarm rate, and effectively avoid missing the oil - injection opportunity caused by monitoring lag;

[0020] 2. This invention reduces the time required for oil injection and fire start-up by using a three-stage direct oil gun response mechanism and pneumatic drive design, solving the problem of excessively long operation time of mechanical atomizing oil guns. It covers the critical window for combustion deterioration, reducing the risk of boiler fire extinguishing. Through the combustion stability index (CSI) quantitative model and adaptive energy matching algorithm, it achieves precise control of stable combustion energy, solving the problem of fuel waste caused by fixed threshold oil injection and reducing fuel consumption per unit.

[0021] 3. This invention improves the reliability of the system under complex operating conditions through multi-dimensional constraint verification and data fault tolerance technology, solves the problem of false alarms caused by sensor failure, and improves the efficiency of data; through combustion state quantitative assessment and dynamic response mechanism, the minimum stable combustion load of the unit is reduced, solving the problem of insufficient deep peak shaving capability and improving the peak shaving depth. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the overall structure of the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] As attached Figure 1 The combustion stability control system shown is based on a multi-dimensional furnace combustion measurement method and includes:

[0025] Multidimensional sensing data acquisition: Data is acquired synchronously through an array of sensors;

[0026] In this embodiment, it should be specifically noted that the multi-dimensional sensing data acquisition includes flame image acquisition, infrared temperature field monitoring, furnace negative pressure acquisition, and composite signal acquisition; the flame image acquisition uses 12 sets of dual-wavelength high-speed cameras, with a sampling frequency of 1000fps, a resolution of 4096×2160, and a dynamic aperture adjustment range of 0.1-10000cd / m². 2 The infrared temperature field monitoring system utilizes a 32×32 elemental mercury cadmium telluride focal plane detector with a temperature range of 300-2200℃ and a spatial resolution of 0.8mrad, equipped with a rotary water-cooled purging device. The furnace negative pressure acquisition system integrates a 16-channel negative pressure sensor with a range of -8000Pa to +8000Pa, a measurement accuracy of 0.2%FS, and a response time ≤5ms. The composite signal acquisition system comprises an 8-channel ultraviolet flame detector and a 4-channel acoustic sensor, using 316L stainless steel armored cables.

[0027] Feature parameter quantization and extraction: Preprocess the collected data to extract key feature parameters;

[0028] In this embodiment, it should be specifically noted that: the feature parameter quantization extraction includes flame fractal dimension calculation, temperature field gradient tensor, negative pressure high-frequency component extraction, and acoustic wave energy entropy calculation; the flame fractal dimension calculation: Where r represents the measurement scale, N(r) is the number of squares covering the edge of the flame, and the sampling interval is 10ms; the temperature field gradient tensor is calculated using the finite difference method with a spatial step size of 0.5m; the negative pressure high-frequency component extraction is performed by obtaining the 50-200Hz frequency band components through empirical mode decomposition. Among them, IMF i Represents the i-th order intrinsic mode function, filtering components in the frequency range of 50-200Hz; the acoustic energy entropy is calculated as follows: Where p i This represents the energy percentage of the i-th frequency band.

[0029] Combustion stability modeling: Constructing a quantitative evaluation model for the combustion stability index based on key characteristic parameters;

[0030] In this embodiment, it should be specifically noted that the combustion stability index in σ(T) represents the coefficient of variation of the temperature field, and σ(T) represents the standard deviation of temperature. This represents the average temperature; when CSI < 0.65, stable combustion intervention is triggered. This threshold is determined by the ROC curve of 3000 sets of historical data, with a false alarm rate ≤ 3% and a false negative rate ≤ 5%.

[0031] Intelligent graded stable combustion control: Triggers a three-stage direct injection response mechanism based on the combustion stability index, and dynamically adjusts the stable combustion energy injection in combination with an adaptive energy matching algorithm;

[0032] In this embodiment, it is specifically noted that the intelligent graded stable combustion control strategy includes a three-level response mechanism, an energy matching algorithm, and combustion recovery determination. The three-level response mechanism is as follows: Level 1 response CSI = 0.65-0.55: two fuel guns are pulsed, with a pulse width of 200ms, an interval of 150ms, a propulsion time ≤ 300ms, and an atomized gas pressure of 0.8-1.0MPa; Level 2 response CSI = 0.55-0.45: four fuel guns continuously spray, with a propulsion time ≤ 200ms and an atomized gas pressure of 1.2-1.5MPa; Level 3 response CSI < 0.45: all six fuel guns are engaged, with a propulsion time ≤ 150ms, an atomized gas pressure of 1.8-2.0MPa, and simultaneous plasma ignition. The energy matching algorithm... Q rIt represents the real-time stable combustion energy demand, and the adjustment period is 100 ms; for the combustion recovery determination: when CSI≥0.75 for 30 consecutive seconds and Df≥1.7, Cv≤0.15, one oil gun is withdrawn every 5 s.

[0033] Iterative optimization of control effect: Evaluate the stable combustion effect through a multi-dimensional constraint verification mechanism, and iteratively adjust the control strategy according to the combustion recovery conditions;

[0034] In this embodiment, specifically, it should be noted that the iterative optimization of the control effect includes sensor fault diagnosis, data filtering processing, and control effect evaluation; for the sensor fault diagnosis: a triple verification mechanism, redundant comparison deviation>15% alarm, temperature over-range 2000 °C interpolation replacement, feature mutation rate>50% enable standby sensor; for the data filtering processing, wavelet packet 5-layer decomposition is used to eliminate energy noise <3%, combined with Kalman filtering, and the error after filtering ≤3%; for the control effect evaluation: when the CSI increase amount <0.15 within 1.8 s after oil injection, add oil guns, and terminate the iteration when the CSI change rate <5% for 5 consecutive control cycles.

[0035] Dynamic evaluation of combustion state: Calculate the deviation degree between the current combustion state and the historical reference value to achieve real-time early warning and evaluation of combustion stability.

[0036] In this embodiment, specifically, it should be noted that the method for dynamic evaluation of the combustion state is: calculate the deviation degree where CSI ref represents the historical reference value at the same load. When CD≤0.15, the combustion is stable; when 0.15<CD≤0.3, an early warning is issued; when CD>0.3, a three-level response is started.

[0037] This invention achieves advanced identification of early-stage combustion deterioration through a multi-dimensional sensor array and feature extraction algorithm, solving the problem of lag in single fire detector monitoring and advancing the identification time of combustion anomalies. It uses a dual-wavelength high-speed camera and an infrared focal plane detector, combined with flame fractal dimension calculation and temperature field gradient tensor analysis, to capture real-time microscale pulsations of flame and subtle changes in the temperature field. Compared to the high response delay of traditional ultraviolet flame detectors, this method can trigger early warnings within the critical window of combustion deterioration, improving the accuracy of combustion anomaly identification, reducing false alarm rates, and effectively avoiding missed fuel injection opportunities due to monitoring lag. Through a three-stage direct fuel injection response mechanism and pneumatic drive design, the fuel injection and fire-starting time is compressed, solving the problem of excessively long operation time for mechanical atomized fuel guns. It covers the critical window of combustion deterioration, reducing the risk of boiler flameout. Through the Combustion Stability Index (CSI) quantification model and adaptive energy matching algorithm, precise control of stable combustion energy is achieved, solving the problem of fuel waste caused by fixed threshold fuel injection and reducing fuel consumption per unit. Through multi-dimensional constraint verification and data fault tolerance technology, the reliability of the system under complex operating conditions is improved, solving the problem of false alarms caused by sensor failures and improving data efficiency. Through quantitative assessment of combustion status and a dynamic response mechanism, the minimum stable combustion load of the unit is reduced, solving the problem of insufficient deep peak-shaving capacity and improving peak-shaving depth.

[0038] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0039] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A stable combustion control system based on multi-dimensional furnace combustion measurement, characterized in that, include: Multidimensional sensing data acquisition: Data is acquired synchronously through an array of sensors; Feature parameter quantization and extraction: Preprocess the collected data to extract key feature parameters; The feature parameter quantization extraction includes flame fractal dimension calculation, temperature field gradient tensor extraction, negative pressure high-frequency component extraction, and acoustic energy entropy calculation; the flame fractal dimension calculation: Where Df represents the fractal dimension of the flame, r represents the measurement scale, N(r) is the number of squares covering the edge of the flame, and the sampling interval is 10ms; the temperature field gradient tensor is calculated using the finite difference method with a spatial step size of 0.5m; the negative pressure high-frequency component extraction is performed by obtaining the 50-200Hz frequency band components through empirical mode decomposition. Where Hf represents the negative pressure high-frequency component. Represents the i-th order intrinsic mode function, filtering components in the frequency range of 50-200Hz; the acoustic energy entropy is calculated as follows: Where Ee represents the entropy of sound wave energy. This represents the energy percentage of the i-th frequency band; Combustion stability modeling: Constructing a quantitative evaluation model for the combustion stability index based on key characteristic parameters; The combustion stability index , Represents the coefficient of variation of the temperature field. Indicates the standard deviation of temperature. This represents the average temperature; when CSI < 0.65, stable combustion intervention is triggered. This threshold is determined by the ROC curve of 3000 sets of historical data, with a false alarm rate ≤ 3% and a false negative rate ≤ 5%. Intelligent graded stable combustion control: Triggers a three-stage direct injection response mechanism based on the combustion stability index, and dynamically adjusts the stable combustion energy injection in combination with an adaptive energy matching algorithm; The intelligent graded stable combustion control strategy includes a three-level response mechanism, an energy matching algorithm, and combustion recovery determination. The three-level response mechanism is as follows: Level 1 response (CSI = 0.65-0.55): two fuel nozzles are pulsed, with a pulse width of 200ms, an interval of 150ms, a propulsion time ≤ 300ms, and an atomization pressure of 0.8-1.0MPa; Level 2 response (CSI = 0.55-0.45): four fuel nozzles continuously spray, with a propulsion time ≤ 200ms and an atomization pressure of 1.2-1.5MPa; Level 3 response (CSI < 0.45): all six fuel nozzles are engaged, with a propulsion time ≤ 150ms, an atomization pressure of 1.8-2.0MPa, and simultaneous plasma ignition. The energy matching algorithm... ,in This indicates the real-time stable combustion energy demand, with an adjustment cycle of 100ms; the combustion recovery determination is as follows: when CSI≥0.75 for 30 consecutive seconds and Df≥1.7 and Cv≤0.15, one fuel gun is withdrawn every 5 seconds; Iterative optimization of control effect: The combustion stabilization effect is evaluated through a multi-dimensional constraint verification mechanism, and the control strategy is iteratively adjusted according to the combustion recovery conditions; Dynamic assessment of combustion status: Calculates the deviation of the current combustion status from historical benchmark values ​​to achieve real-time early warning and assessment of combustion stability.

2. The stable combustion control system based on multi-dimensional furnace combustion measurement method according to claim 1, characterized in that: The multi-dimensional sensing data acquisition includes flame image acquisition, infrared temperature field monitoring, furnace negative pressure acquisition, and composite signal acquisition; the flame image acquisition uses 12 sets of dual-wavelength high-speed cameras, a sampling frequency of 1000fps, a resolution of 4096×2160, and a dynamic aperture adjustment range of 0.1-10000cd / The infrared temperature field monitoring utilizes a 32×32 element mercury cadmium telluride focal plane array detector with a temperature measurement range of 300-2200°C. The spatial resolution is 0.8 mrad, and it is equipped with a rotary water-cooled purging device; the furnace negative pressure acquisition: integrates a 16-channel negative pressure sensor with a range of -8000Pa to +8000Pa, a measurement accuracy of 0.2%FS, and a response time of ≤5ms; the composite signal acquisition: an 8-channel ultraviolet flame detector and a 4-channel acoustic sensor, using 316L stainless steel armored cables.

3. The stable combustion control system based on multi-dimensional furnace combustion measurement method according to claim 1, characterized in that: The iterative optimization of control effect includes sensor fault diagnosis, data filtering, and control effect evaluation. Sensor fault diagnosis employs a triple verification mechanism: redundancy comparison deviation >15% alarm, temperature over-range interpolation replacement at 2000℃, and activation of a backup sensor if the feature mutation rate >50%. Data filtering involves wavelet packet decomposition at 5 levels to remove <3% energy noise, combined with Kalman filtering, resulting in an error ≤3%. Control effect evaluation involves increasing the oil gun input when the CSI increase is <0.15 within 1.8s after oil injection, and terminating the iteration when the CSI change rate is <5% for 5 consecutive control cycles.

4. The stable combustion control system based on multi-dimensional furnace combustion measurement method according to claim 1, characterized in that: The dynamic combustion state evaluation method is as follows: Calculate the deviation degree where, represents the historical reference value at the same load. When CD ≤ 0.15, the combustion is stable; when 0.15 < CD ≤ 0.3, it is a warning; when CD > 0.3, a three-level response is initiated.

Images