Boiler heating surface soot blowing opportunity judgment method and device, equipment and medium

Abstract

The invention belongs to the technical field of thermal power generation intelligent control, and discloses a boiler heating surface soot blowing opportunity judgment method and device, equipment and a medium, an input signal is constructed according to the load change rate, a compensation coefficient is output through a trained correction model, a boiler heating surface cleaning factor is compensated, and a soot blowing opportunity judgment result is obtained. Identifying that the boiler is in a statistical controlled stable operation interval, collecting historical measurement values of the compensation cleaning factor under various working conditions, and determining a safe operation lower limit; performing sliding window filtering processing on the current measurement value of the compensation cleaning factor to obtain a target cleaning factor; and if the target cleaning factor is continuously lower than the safe operation lower limit and exceeds the set duration, generating a soot blowing trigger instruction, thereby realizing dynamic compensation of the cleaning factor in a rapid variable load state and accurate screening of the stable operation state of the boiler, reducing misjudgment of the soot blowing opportunity, improving the accuracy of determination of the soot blowing opportunity of the heating surface of the boiler, and improving the reliability of the soot blowing opportunity. And intelligent control over soot blowing operation is improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for thermal power generation, specifically to a method, device, equipment, and storage medium for determining the timing of soot blowing on the heating surface of a boiler. Background Technology

[0002] Ash and slag buildup on the heating surfaces of coal-fired boilers is a core issue affecting their safe and economical operation. Cleanliness factors are typically used to assess the degree of surface contamination and guide soot blowing operations. However, conventional cleanliness factors are designed based on steady-state loads. Under rapid load changes, these factors are easily affected by sudden changes, leading to significant deviations. This, in turn, affects the accurate timing of soot blowing operations, impacting boiler heat exchange efficiency and operational safety.

[0003] Traditional methods correct the cleaning factor using a single parameter, but they cannot effectively reflect the true degree of contamination under rapidly changing load conditions, leading to frequent malfunctions in soot blowing. In addition, the widely used fixed threshold soot blowing strategy cannot distinguish between the short-term drop in the cleaning factor caused by load disturbances and the actual ash accumulation trend. Therefore, it is greatly affected by operating condition disturbances, often resulting in over-blowing or under-blowing, causing steam waste or safety hazards.

[0004] In summary, the inaccurate timing of soot blowing in existing technologies is a technical problem that urgently needs to be solved. Summary of the Invention

[0005] To address the aforementioned shortcomings, the present invention aims to provide a method, apparatus, equipment, and storage medium for determining the timing of soot blowing on boiler heating surfaces, which can reduce misjudgments of soot blowing timing and improve the accuracy of determining the timing of soot blowing on boiler heating surfaces.

[0006] The first aspect of this invention discloses a method for determining the timing of soot blowing on a boiler heating surface, comprising: The real-time operating data and static geometric parameters of the boiler's current heating surface are obtained, and the physical properties of the flue gas and steam are calculated based on the real-time operating data and the static geometric parameters. Based on the physical properties of the flue gas and steam, the original cleaning factor is calculated according to the principle of heat exchange balance of the heated surface. An input signal is constructed based on the load change rate. The input signal is then input into a trained correction model to obtain a compensation coefficient. The original cleaning factor is then compensated based on the compensation coefficient to obtain a compensated cleaning factor. Identify that the boiler is in a statistically controlled stable operating range; Historical measurements of the compensating cleaning factor under various operating conditions were collected within the stable operating range. The lower limit of safe operation of the compensation cleaning factor is determined based on the historical measurement values ​​of the compensation cleaning factor; The current measured value of the compensation cleaning factor is subjected to sliding window filtering to obtain the target cleaning factor; If the target cleaning factor remains below the safe operating limit for an extended period of time, a dust removal trigger command is generated.

[0007] In some embodiments, the input signal is constructed based on the load change rate, including: The load change rate is calculated by dividing the difference between the current load and the load at the previous moment by the rated load. Determine whether the fluctuation range of the cleanliness factor of the current heated surface affected by load changes reaches the preset amplitude threshold; If the fluctuation amplitude reaches a preset amplitude threshold, the difference between the current load and the reference load is divided by the rated load to calculate the load deviation change rate; a two-dimensional vector is constructed based on the load change rate and the load deviation change rate as the input signal; If the fluctuation amplitude does not reach the preset amplitude threshold, a one-dimensional vector is constructed based on the load change rate as the input signal.

[0008] In some embodiments, identifying that the boiler is in a statistically controlled stable operating range includes: When the preset starting conditions are met and the fluctuations of each operating parameter are controlled within the preset fluctuation threshold range, and equipment failure, abnormal human operation, and the period of soot blowing are excluded, the boiler is determined to be in a statistically controlled stable operating state. Under continuous and stable operation, the medium load range of 50%-80%, the low load range of 30%-50%, and the high load range of 80%-100% are obtained respectively. The medium load range, low load range, and high load range are divided into sub-ranges, and then combined to obtain the stable operating range of the boiler under statistical control.

[0009] In some embodiments, determining the safe operating lower limit of the compensating cleaning factor based on historical measurements of the compensating cleaning factor includes: After sorting the historical measurements of the compensation cleaning factor in ascending order, the historical measurements below the lower tail threshold are extracted as over-threshold samples; the lower tail threshold is taken from the 85th percentile of the historical measurements after ascending order. The maximum likelihood estimation method is used to fit the generalized Pareto distribution to the over-threshold samples, and the shape and scale parameters of the distribution are solved. Based on the fitted shape parameters and scale parameters, the lower quantile of the tail probability p≤0.003 is calculated, and the lower limit of safe operation is determined according to the lower quantile.

[0010] In some embodiments, the current measured value of the compensation cleaning factor is subjected to sliding window filtering to obtain the target cleaning factor, including: A sliding window is used to extract the historical measurement value of the compensation cleaning factor that is closest to the current time as the statistical measurement value; The current measurement value of the compensation cleaning factor is filtered based on the statistical measurement value within the sliding window to obtain the target cleaning factor.

[0011] In some embodiments, filtering the current measurement value of the compensating cleaning factor based on statistical measurements within the sliding window to obtain the target cleaning factor includes: The median and the median absolute deviation of each statistical measurement from the median are calculated based on all statistical measurements within the sliding window. Calculate the target absolute deviation between the current measured value of the compensation cleaning factor and the median, and calculate the product of the preset sensitivity coefficient and the median absolute deviation as the target product. Compare the target absolute deviation with the target product. If the absolute deviation of the target is not greater than the product of the targets, the current measured value of the compensation cleaning factor is directly output as the target cleaning factor; If the absolute deviation of the target is greater than the product of the targets, the current measurement value of the compensation cleaning factor is determined to be an outlier. Adaptive smoothing processing is then performed on the outlier to obtain the target cleaning factor.

[0012] The second aspect of this invention discloses a device for determining the timing of soot blowing on a boiler heating surface, comprising: The acquisition unit is used to acquire real-time operating data and static geometric parameters of the current heating surface of the boiler; The first calculation unit is used to calculate the physical property parameters of flue gas and steam based on the real-time operating data and the static geometric parameters. The second calculation unit is used to calculate the original cleaning factor based on the physical property parameters of the flue gas and steam and the principle of heat exchange balance of the heating surface. Construction unit, used to construct input signal based on load change rate; The compensation unit is used to input the input signal into the trained correction model to obtain compensation coefficients, and to compensate the original cleaning factor according to the compensation coefficients to obtain a compensated cleaning factor. The identification unit is used to identify whether the boiler is in a statistically controlled stable operating range; The data acquisition unit is used to acquire historical measurements of the compensating cleaning factor under various operating conditions within the stable operating range. The determining unit is configured to determine the safe operating limit of the compensating cleaning factor based on the historical measurement values ​​of the compensating cleaning factor. A filtering unit is used to perform sliding window filtering on the current measured value of the compensation cleaning factor to obtain the target cleaning factor; The determination unit is used to generate a dust blowing trigger command when the target cleaning factor is continuously lower than the safe operating limit and exceeds the set time.

[0013] In some embodiments, the building unit includes: The calculation sub-unit is used to divide the difference between the current load and the load at the previous moment by the rated load to calculate the load change rate; The judgment sub-unit is used to determine whether the fluctuation range of the cleaning factor of the current heated surface affected by load changes reaches the preset amplitude threshold. The first construction subunit is used to calculate the load deviation change rate by dividing the difference between the current load and the reference load by the rated load when the judgment subunit determines that the fluctuation amplitude reaches the preset amplitude threshold; and to construct a two-dimensional vector as an input signal based on the load change rate and the load deviation change rate. The second construction subunit is used to construct a one-dimensional vector as an input signal based on the load change rate when the judgment subunit determines that the fluctuation amplitude has not reached the preset amplitude threshold.

[0014] The third aspect of the present invention discloses an electronic device, including a memory storing executable program code and a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the boiler heating surface soot blowing timing method disclosed in the first aspect.

[0015] The fourth aspect of the present invention discloses a computer-readable storage medium storing a computer program, wherein the computer program causes a computer to execute the boiler heating surface soot blowing timing determination method disclosed in the first aspect.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: by constructing an input signal based on the load change rate, outputting a compensation coefficient through a trained correction model to compensate the cleaning factor of the boiler heating surface, and identifying the boiler in a statistically controlled stable operating range, the present invention collects historical measurements of the compensation cleaning factor under various operating conditions based on the stable operating range to determine the safe operating lower limit of the compensation cleaning factor; and performs sliding window filtering on the current measurement value of the compensation cleaning factor to obtain the target cleaning factor; if the target cleaning factor is continuously lower than the safe operating lower limit for more than a set time, a soot blowing trigger command is generated. This realizes dynamic compensation of the cleaning factor under rapid load change and accurate screening of the stable operating state of the boiler, thereby reducing misjudgment of soot blowing timing, improving the accuracy of soot blowing timing determination of the boiler heating surface, and thus improving the intelligent control of soot blowing operation. Attached Figure Description

[0017] Figure 1 This is a flowchart of a method for determining the timing of soot blowing on a boiler heating surface, as disclosed in this invention. Figure 2 This is a schematic diagram of the structure of a boiler heating surface soot blowing timing determination device disclosed in this invention; Figure 3 This is a schematic diagram of the structure of an electronic device disclosed in this invention; Figure 4 This is a schematic diagram of the structure of a computer device disclosed in this invention.

[0018] Explanation of reference numerals in the attached figures: 201. Acquisition unit; 202. First calculation unit; 203. Second calculation unit; 204. Construction unit; 205. Compensation unit; 206. Identification unit; 207. Acquisition unit; 208. Determination unit; 209. Filtering unit; 210. Judgment unit; 301. Memory; 302. Processor. Detailed Implementation

[0019] To enable those skilled in the art to better understand the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0020] In some processes described in the specification, claims, and accompanying drawings of this invention, 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 110, 120, 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 sequential order, nor do they limit "first" and "second" to different types.

[0021] It will be understood by those skilled in the art that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this specification means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “coupled” to another element, it can be directly connected or coupled to the other element, or there may be intermediate elements. Furthermore, “connected” or “coupled” as used herein can include wireless connections or wireless coupling. The term “and / or” as used herein includes all or any units and all combinations of one or more associated listed items.

[0022] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by those skilled in the art to which this invention pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Throughout the description, the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0024] Please see Figure 1 This invention discloses a method for determining the timing of soot blowing on a boiler heating surface. The method can be executed by electronic devices such as industrial controllers, computers, laptops, tablets, or boiler heating surface soot blowing timing determination devices embedded in electronic devices; this invention does not limit the specific device. In this embodiment, an electronic device is used as an example for illustration.

[0025] like Figure 1 As shown, the method includes the following steps 110-180: 110. Obtain the real-time operating data and static geometric parameters of the boiler's current heating surface, and calculate the physical properties of the flue gas and steam based on the real-time operating data and static geometric parameters.

[0026] The distributed boiler system can acquire real-time operating data for each heating surface of the boiler. This real-time operating data is stored in a Redis database in key-value pairs. The real-time operating data includes flue gas temperature, flue gas pressure, steam temperature, steam pressure, steam flow rate, fuel quantity, flue gas oxygen content, and feedwater flow rate. In addition, the static geometric parameters of each heating surface are also collected and stored in files for real-time calculation. These static geometric parameters include heating area, flue gas flow area, steam flow area, pipe inner diameter, and pipe outer diameter.

[0027] In this invention, the heating surface that needs to be calculated is defined as the current heating surface. Based on the current heating surface identification information, a batch read request is initiated to the Redis database through a predefined key name list. According to the measurement point data of the current heating surface, all real-time operating data related to the current heating surface is read from the Redis database. The real-time operating data is updated every five seconds, and the data reading and calculation frequency is also synchronized at once every five seconds. Then, based on the read real-time operating data and static geometric parameters of the current heating surface, the physical properties of the flue gas and steam are calculated.

[0028] 120. Based on the physical properties of flue gas and steam, the original cleaning factor is calculated according to the principle of heat exchange balance of the heating surface.

[0029] Based on the principle of heat exchange balance of the heating surface, the heat balance equation is solved. For example, based on the principle that the heat released by the flue gas side is equal to the heat absorbed by the working fluid side, and combined with the definition of the cleaning factor (calculated based on the thermal resistance of ash and dirt, the heat transfer coefficient of the flue gas side and the heat transfer coefficient of the steam side), the original cleaning factor is calculated. This original cleaning factor reflects the cleanliness or contamination level of the boiler's current heating surface under the current operating conditions.

[0030] 130. Construct an input signal based on the load change rate, input the input signal into the trained correction model to obtain the compensation coefficient, and compensate the original cleaning factor based on the compensation coefficient to obtain the compensated cleaning factor.

[0031] Considering the differences in structure, working fluid, and thermodynamic characteristics of different heating surfaces in a boiler, and the fact that traditional methods mostly use the same model for boiler heating surfaces without classifying and modeling the fouling formation mechanism and dynamic response characteristics of heating surfaces in different locations, such as low-temperature superheaters, low-temperature reheaters, and economizers, the accuracy of monitoring local ash accumulation risk is insufficient.

[0032] Preferably, in this invention, differentiated compensation strategies are adopted for different types of heated surfaces. The cleaning factor of high-temperature heated surfaces is greatly affected by the load; load fluctuations can significantly interfere with it through instantaneous changes in flue gas velocity and temperature. Therefore, dual-path coupling compensation is required, combining short-term fluctuation responses with long-term trend responses for calculation to obtain a more stable compensation coefficient to offset the interference. Conversely, the cleaning factor of low-temperature heated surfaces is less affected by the load and only requires single-path compensation, calculated solely based on short-term fluctuations. This categorized approach achieves refined and differentiated control strategies, improving the accuracy of localized ash accumulation risk monitoring.

[0033] In practical applications, for heat transfer surfaces where the cleanliness factor is more sensitive to load changes, such as high-temperature superheaters, screen superheaters, and furnaces, the load change rate of the first path and the load deviation change rate of the second path are used for dual-path coupling calculation. For heat transfer surfaces where the cleanliness factor is less sensitive to load changes, such as economizers, low-temperature reheaters, low-temperature superheaters, and high-temperature reheaters, only the load change rate of the first path is used for calculation.

[0034] Specifically, the methods for constructing input signals include: The load change rate is calculated by dividing the difference between the current load and the previous load by the rated load. The fluctuation amplitude of the cleaning factor of the current heated surface affected by the load change is then determined to be within a preset amplitude threshold. If the fluctuation amplitude reaches the preset amplitude threshold, the load deviation change rate is calculated by dividing the difference between the current load and the reference load by the rated load. A two-dimensional vector is constructed based on the load change rate and the load deviation change rate as the input signal. If the fluctuation amplitude does not reach the preset amplitude threshold, a one-dimensional vector is constructed based on the load change rate as the input signal.

[0035] The load change rate of the first path is used to characterize the instantaneous rate of change of boiler load, which is calculated from the real-time load within a preset time window; the load deviation change rate of the second path is used to characterize the long-term deviation of the operating conditions from the design benchmark, which is calculated by subtracting the real-time load from the preset benchmark load.

[0036] In this embodiment of the invention, the modified model is trained using a lightweight neural network to adapt to rapid changes in the cleaning factor under varying loads, and is used to dynamically compensate for the original cleaning factor. The modified model includes a diagonal weight matrix and an improved nonlinear activation function, and is configured to simultaneously satisfy input signals of either a one-dimensional or two-dimensional vector. The input features of the two-dimensional vector are the load change rate of the first path and the load deviation change rate of the second path, respectively.

[0037] When a two-dimensional vector is input into the modified model, the two features are first independently linearly scaled using the diagonal weight matrix in the modified model to obtain two-dimensional feature vectors. By setting the diagonal weight matrix for linear transformation, this diagonal weight matrix structure allows the model to assign independent weight coefficients to the two features, learning the different contributions of these two features to the perturbation of the cleaning factor, while greatly reducing the number of parameters and meeting the requirements of lightweight design.

[0038] Then, the two-dimensional feature vector, which is linearly scaled by the diagonal weight matrix and contains the load change rate feature and the load deviation change rate feature, is input into the improved nonlinear activation function in the modified model for transformation, generating the compensation coefficient of the dual-path coupling. This step realizes the nonlinear fusion of the dual-path features.

[0039] In this invention, parameterized activation functions are constructed for two paths. The activation function for the first path amplifies the input response value by multiplying the load change rate by a preset amplification factor λ, aiming to quickly respond to short-term dynamic fluctuations in the load and suppress instantaneous noise. The activation function for the second path amplifies the input response value by multiplying the load deviation change rate by a preset amplification factor λ, aiming to track long-term load trends and correct systematic drift. The inputs to both path activation functions are amplified by the preset amplification factor before input, which improves sensitivity. The output values ​​of the two path activation functions are directly multiplied to obtain the compensation coefficient γ.

[0040] Coupling is achieved through a product operation, rather than a simple weighted summation. This product operation creates a nonlinear interaction between the two features. The product result exhibits a nonlinear amplification effect if and only if both the load change rate feature and the load deviation change rate feature are significant, thereby compensating for rapid load adjustments under conditions deviating from design baselines. After activation function mapping, the coupling modulation factor is greater than one, resulting in a larger compensation coefficient to cope with strong composite interference. When either input is insignificant, the modulation factor approaches one, and the model degenerates into an approximate single-path compensation state.

[0041] The calculation process of the activation function for any path includes: For the i-th path (i=1,2), subtract the preset response threshold of the path from the input response value, and then multiply by the slope parameter. Use this product as the exponential input of the natural exponential function. Add 1 to the result of the natural exponential function and use it as the denominator. Divide the difference between the maximum and minimum output values ​​by the denominator, and add the minimum output value to the result to get the output value of the path.

[0042] Key parameters in the activation function include the preset amplification factor λ, the slope parameter K, and the preset response threshold. The parameters are not fixed. The system can adaptively adjust based on the current load segment, such as low or high load, the operating characteristics of different plants, and the dynamic characteristics of different cleaning factor types. During load increases, if the cleaning factor decreases, the preset amplification factor λ is reduced. The slope parameter K of the activation function is increased during rapid load changes, such as during load increases, when the cleaning factor suddenly rises due to the load change, to obtain a similarly steep curve to offset this and achieve the most suitable compensation curve. The activation function parameters for different heated surfaces are adjusted according to the characteristics of each heated surface.

[0043] 140. Identify that the boiler is in a statistically controlled stable operating range.

[0044] Specifically, step 140 may include the following steps 1401-1402 (not shown): 1401. When the preset starting conditions are met and the fluctuations of each operating parameter are controlled within the preset fluctuation threshold range, and equipment failure, abnormal human operation, and the period of soot blowing are excluded, the boiler is determined to be in a statistically controlled stable operating state.

[0045] The preset starting conditions include: continuous stable operation time of no less than 2 hours and load change rate ≤2% / min (considered as no load step change during the period). Fluctuations in operating parameters are controlled within preset fluctuation thresholds, such as steam temperature ±5℃, steam pressure ±0.2MPa, flue gas flow rate ±3%, and furnace negative pressure ±50Pa. Equipment faults include leaks in the heating surface and abnormal fan operation; human error includes sudden changes in fuel quantity and damper opening adjustments. Through multi-parameter collaborative monitoring and dynamic screening, and by excluding equipment faults, human error, and periods of soot blowing, data can be ensured to remain undisturbed.

[0046] 1402. Under continuous and stable operation, the medium load range of 50%-80%, the low load range of 30%-50%, and the high load range of 80%-100% are obtained respectively. The medium load range, low load range, and high load range are divided into sub-ranges, and then combined to obtain the stable operation range in which the boiler is under statistical control.

[0047] In terms of load segment subdivision principles, the medium load segment with a commonly used boiler load rate of 50%-80% is selected as the core statistical interval. This interval has stable combustion, high heat exchange efficiency, and representative changes in cleanliness factors. At the same time, the low load segment with a load rate of 30%-50% and the high load segment with a load rate of 80%-100% are obtained as supplements. The low load segment and the high load segment correspond to start-up and peak-shaving conditions and full-load operation conditions, respectively. Each load segment is divided into sub-intervals, and the division conditions of each load segment are different. For example, the low load segment needs to meet the requirement of continuous stable operation time ≥3 hours, and the steam temperature fluctuation threshold of the high load segment is tightened to ±3℃ to avoid model deviation caused by data aliasing across load segments.

[0048] Furthermore, the preset starting conditions, preset fluctuation threshold ranges of operating parameters, and data sampling density can be dynamically adjusted in conjunction with seasonal factors. Seasonal factor corrections need to be dynamically adjusted in conjunction with the impact of ambient temperature and humidity on boiler pollution characteristics. For example, in winter, due to increased air density leading to higher furnace heat load and increased risk of slagging on heating surfaces, the preset fluctuation threshold range of operating parameters within the stable operating range is tightened by 20%, and the sliding window statistics need to exclude the slagging warning period. In summer, flue gas temperature is prone to exceed limits, so the preset starting condition needs to be adjusted to "flue gas temperature is stable within the design value ±10℃ range for ≥30 minutes". In spring and autumn, benchmark parameters are used and humidity changes are monitored. When the relative humidity is >75%, the data sampling density is increased.

[0049] 150. Collect historical measurements of the compensating cleaning factor under various operating conditions within the stable operating range.

[0050] It should be noted that historical measurements of the compensation cleaning factor should be collected within the statistically controlled stable operating range of the boiler. The data collection period should be no less than 6 months to ensure that the sample size is sufficient and representative. The data sampling interval should be consistent with the compensation cleaning factor update interval, with a default setting of 30 seconds.

[0051] 160. Determine the lower limit of safe operation of the cleaning factor based on historical measurements of the cleaning factor.

[0052] Specifically, step 160 may include the following steps 1601-1603 (not shown): 1601. After sorting the historical measurements of the compensation cleaning factor in ascending order, extract the historical measurements below the lower tail threshold as over-threshold samples.

[0053] The lower tail threshold is taken from the 85th percentile of historical measurements sorted in ascending order. The over-threshold sample set is used to reflect the low-state distribution characteristics of the compensation cleaning factor.

[0054] 1602. The maximum likelihood estimation method is used to fit the generalized Pareto distribution to the samples exceeding the threshold, and the shape parameter ξ and scale parameter σ of the distribution are solved.

[0055] Maximum likelihood estimation is achieved by constructing a likelihood function and finding its maximum value. The likelihood function construction needs to be based on the probability density function of the generalized Pareto distribution. During the fitting process, the optimal solution of the parameters is solved by an iterative algorithm to ensure that the goodness of fit R² ≥ 0.9.

[0056] 1603. Based on the shape and scale parameters obtained from the fitting, calculate the lower quantile of the tail probability p≤0.003, and determine the lower limit of safe operation based on the lower quantile.

[0057] The lower quantile can be directly determined as the lower limit of safe operation. Alternatively, it can be further determined whether the compensation cleaning factor has a theoretical minimum value. If the calculated lower quantile is lower than the theoretical minimum value, then the theoretical minimum value is taken as the lower limit of safe operation L, ensuring that the lower limit setting does not violate the physical boundary.

[0058] Specifically, when the shape parameter is less than 0, the theoretical minimum value (i.e., the left endpoint of the distributed support set is σ) is calculated based on the lower tail threshold, the shape parameter ξ, and the scale parameter σ. If the lower quantile is lower than the theoretical minimum, then the theoretical minimum is taken as the lower limit for safe operation.

[0059] 170. Perform sliding window filtering on the current measured value of the compensation cleaning factor to obtain the target cleaning factor.

[0060] Specifically, step 170 may include the following steps 1701-1702 (not shown): 1701. Use a sliding window to capture the historical measurement value of the compensation cleaning factor that is closest to the current time as the statistical measurement value.

[0061] The sliding window is used to initialize a fixed-length data queue. In practical applications, this sliding window can be maintained to record the compensated cleaning factor. When a new compensated cleaning factor arrives, it is added to the tail of the queue. If the queue is full, the old data at the head of the queue is removed. When the sliding window is full, for each new compensated cleaning factor, the sliding window is moved to extract different historical measurements for statistical analysis.

[0062] 1702. Filter the current measurement value of the compensation cleaning factor based on the statistical measurement value in the sliding window to obtain the target cleaning factor.

[0063] Specifically, step 1702 may include the following steps S1 to S3 (not shown): S1. Calculate the median and the median absolute deviation of each statistical measurement from the median based on all statistical measurements within the sliding window.

[0064] The median is calculated by sorting all statistical measurements in the sliding window from largest to smallest and taking the middle value as the median; the absolute deviation of the median is calculated by taking the median of the absolute deviations of each statistical measurement from the median.

[0065] S2. Calculate the target absolute deviation between the current measured value of the compensation cleaning factor and the median, and calculate the product of the preset sensitivity coefficient and the median absolute deviation as the target product. Compare the target absolute deviation with the target product.

[0066] S3. If the absolute deviation of the target is not greater than the target product, the current measured value of the compensation cleaning factor is directly output as the target cleaning factor; if the absolute deviation of the target is greater than the target product, the current measured value of the compensation cleaning factor is determined to be an outlier, and adaptive smoothing processing is performed on the outlier to obtain the target cleaning factor.

[0067] The adaptive smoothing processing method specifically includes: First, the normalized deviation parameter is calculated. This parameter is the ratio of the absolute deviation of the current measurement value and the median of the statistical measurements within the sliding window to the target product, and it does not exceed 1. Then, based on the normalized deviation parameter, a smoothing factor is dynamically calculated, where the smoothing factor is linearly positively correlated with the normalized deviation parameter. Next, the weighted average of the statistical measurements within the sliding window is calculated. Finally, based on the smoothing factor, the weighted average of the outliers and the statistical measurements is weighted and fused to obtain the smoothing result as the target cleanliness factor.

[0068] 180. If the target cleaning factor remains below the safe operating limit for an extended period of time, a soot blowing trigger command will be generated.

[0069] At a preset time interval, such as every 10 seconds, the target cleaning factor is compared with the safe operating limit. If the target cleaning factor is continuously lower than the safe operating limit and exceeds the set time, a soot blowing trigger command is generated to trigger the soot blowing operation.

[0070] In this embodiment of the invention, whether to generate a soot blowing trigger command can be determined by a three-level state judgment based on the relationship between the target cleaning factor and the lower limit of safe operation. Specifically, the lower limit of safe operation is added to the warning interval threshold δ to obtain the upper limit of warning; wherein, the warning interval threshold δ is 5%-8% of L, and the specific value is determined according to the type of heating surface, for example, 8% for furnace heating surface and 5% for flue heating surface; if the target cleaning factor is greater than the upper limit of warning (i.e., target cleaning factor > L + δ), it is determined to be in a state where soot blowing is not required; if the target cleaning factor is greater than or equal to the lower limit of safe operation and less than or equal to the upper limit of warning (i.e., L ≤ target cleaning factor ≤ L + δ), it is determined to be in a warning state, and soot blowing is temporarily suspended; if the target cleaning factor is less than the lower limit of safe operation (i.e., target cleaning factor < L), it is determined to be in a state where soot blowing is required, and the count value of the counter used to continuously record the state where soot blowing is required is incremented by one. When the continuous count of the state where soot blowing is required reaches the first preset number, it is considered that the state where soot blowing is required has been maintained for more than the set time, and the soot blowing trigger condition is confirmed to be met, and a soot blowing trigger command is generated. This avoids misjudgment caused by instantaneous fluctuations and can increase the accuracy of judgment.

[0071] Understandably, when continuously recording the number of times soot blowing is required, the count is reset to zero if any instance is determined to be in another state. For furnace heating surfaces, a continuous count of 6 times requiring soot blowing is considered sufficient; for flue heating surfaces, a continuous count of 8 times requiring soot blowing is considered sufficient to maintain the soot blowing state for more than the set duration, corresponding to 1-1.3 minutes.

[0072] Preferably, the validity of the target cleaning factor can be verified through a data validity check mechanism. If invalid, the state corresponding to the target cleaning factor at the previous moment is maintained. Invalid data scenarios include the target cleaning factor being consistent with the target cleaning factor at the previous moment, and the data value exceeding the physically reasonable range. For example, the cleaning factor is usually 0.6-1.0. If it exceeds this range, it is marked as invalid data. During the invalid data period, the dust removal judgment is not performed, and the previous state is maintained.

[0073] Furthermore, when the target cleanliness factor is < L, the system determines that a soot blowing state is required. After incrementing the counter used to continuously record the soot blowing state, and before generating the soot blowing trigger command, the soot blowing determination result can be adjusted based on the load change trend. Specifically, if the load increase rate is > 1% / min, the soot blowing trigger condition is confirmed and a soot blowing trigger command is generated only when the continuous count of the soot blowing state reaches a second preset number, where the second preset number is greater than the first preset number. By extending the continuous count threshold (i.e., using a larger second preset number), the set time is also extended accordingly, thereby delaying the soot blowing determination and avoiding the aggravation of thermal deviation during high-load periods.

[0074] Alternatively, if the load change rate is ≤0.5% / min, the standard continuous counting threshold (i.e., using a smaller first preset number of times) will be used; or, if the load decrease rate is >1% / min and the target cleanliness factor is <L-0.05, a soot blowing trigger command will be generated immediately to trigger the soot blowing operation and prevent the accumulation of contaminants on the heated surface before low load.

[0075] In some possible implementations, alarms can be dynamically triggered and displayed in real time based on status counts. Specifically, when the count value of the counter used to continuously record the alarm status is greater than 1, the alarm is activated; when the count value of the counter used to continuously record the status requiring soot blowing is greater than 1, the alarm is activated; when the count value of the counter used to continuously record the status not requiring soot blowing is greater than 1, the alarm is deactivated. Various status results are written to a Redis database in real time and dynamically displayed using color and numerical values ​​through a visual interface, forming a plant-wide monitoring system.

[0076] like Figure 2 As shown in the figure, this embodiment of the invention also discloses a device for determining the timing of soot blowing on a boiler heating surface, comprising an acquisition unit 201, a first calculation unit 202, a second calculation unit 203, a construction unit 204, a compensation unit 205, an identification unit 206, a collection unit 207, a determination unit 208, a filtering unit 209, and a determination unit 210, wherein, The acquisition unit 201 is used to acquire the real-time operating data and static geometric parameters of the current heating surface of the boiler; The first calculation unit 202 is used to calculate the physical properties of flue gas and steam based on real-time operating data and static geometric parameters. The second calculation unit 203 is used to calculate the original cleaning factor based on the physical properties of flue gas and steam and the principle of heat exchange balance of the heating surface. Construction unit 204 is used to construct the input signal based on the load change rate; The compensation unit 205 is used to input the input signal into the trained correction model to obtain the compensation coefficient, and to compensate the original cleaning factor according to the compensation coefficient to obtain the compensated cleaning factor. The identification unit 206 is used to identify that the boiler is in a statistically controlled stable operating range; The acquisition unit 207 is used to acquire historical measurements of the compensating cleaning factor under various operating conditions within a stable operating range; The determining unit 208 is used to determine the safe operating limit of the compensating cleaning factor based on the historical measurement values ​​of the compensating cleaning factor; The filtering unit 209 is used to perform sliding window filtering on the current measured value of the compensation cleaning factor to obtain the target cleaning factor; The determination unit 210 is used to generate a soot blowing trigger command when the target cleaning factor is continuously lower than the safe operating limit and exceeds the set time.

[0077] As an optional implementation, building unit 204 includes the following sub-units (not shown): The calculation sub-unit is used to divide the difference between the current load and the load at the previous moment by the rated load to calculate the load change rate; The judgment sub-unit is used to determine whether the fluctuation range of the cleaning factor of the current heated surface affected by load changes reaches the preset amplitude threshold. The first construction subunit is used to calculate the load deviation change rate by dividing the difference between the current load and the reference load by the rated load when the judgment subunit determines that the fluctuation amplitude has reached the preset amplitude threshold; and to construct a two-dimensional vector as the input signal based on the load change rate and the load deviation change rate. The second construction subunit is used to construct a one-dimensional vector as an input signal based on the load change rate when the judgment subunit determines that the fluctuation amplitude has not reached the preset amplitude threshold.

[0078] As an optional implementation, the identification unit 206 includes the following sub-units (not shown): The status recognition subunit is used to determine that the boiler is in a statistically controlled stable operating state when it detects that the preset start conditions are met and the fluctuations of each operating parameter are controlled within the preset fluctuation threshold range, while excluding equipment failure, abnormal human operation and soot blowing. The interval division sub-unit is used to obtain the medium load segment with a load rate of 50%-80%, the low load segment with a load rate of 30%-50%, and the high load segment with a load rate of 80%-100% under continuous and stable operation conditions. The medium load segment, low load segment, and high load segment are divided into sub-intervals, and then combined to obtain the stable operation range of the boiler under statistical control.

[0079] As an optional implementation, the determining unit 208 includes the following sub-units (not shown): The extraction sub-unit is used to sort the historical measurements of the compensation cleaning factor in ascending order and extract the historical measurements below the lower tail threshold as over-threshold samples; the lower tail threshold is taken from the 85th percentile of the historical measurements after ascending order. The distribution fitting subunit is used to fit the generalized Pareto distribution to the over-threshold samples using the maximum likelihood estimation method, and to solve for the shape and scale parameters of the distribution. The lower limit setting subunit is used to calculate the lower quantile of the tail probability p≤0.003 based on the fitted shape and scale parameters, and to determine the lower limit of safe operation based on the lower quantile.

[0080] As an optional implementation, the filter unit 209 includes the following sub-units (not shown): The truncation sub-unit is used to use a sliding window to truncate the historical measurement value of the compensation cleaning factor that is closest to the current time as the statistical measurement value; The filtering subunit is used to filter the current measurement value of the compensation cleaning factor based on the statistical measurement value within the sliding window to obtain the target cleaning factor.

[0081] Further optionally, the above-mentioned filtering subunit includes the following modules not shown: The first calculation module is used to calculate the median and the median absolute deviation of each statistical measurement from the median based on all statistical measurements within the sliding window; The second calculation module is used to calculate the target absolute deviation between the current measured value of the compensation cleaning factor and the median, and to calculate the product of the preset sensitivity coefficient and the median absolute deviation as the target product, and compare the target absolute deviation with the target product. The output module is used to directly output the current measured value of the compensation cleaning factor as the target cleaning factor when the absolute deviation of the target is not greater than the target product. The anomaly detection module is used to determine that the current measurement value of the compensation cleaning factor is an anomaly when the absolute deviation of the target is greater than the target product. The smoothing module is used to perform adaptive smoothing on outliers to obtain the target cleanliness factor.

[0082] Furthermore, the smoothing module is specifically used to calculate the normalized deviation parameter, which is the ratio of the target product as the numerator and the absolute deviation of the current measurement value and the median of the statistical measurements within the sliding window as the denominator, and does not exceed 1. Subsequently, based on the normalized deviation parameter, a smoothing factor is dynamically calculated, wherein the smoothing factor is linearly positively correlated with the normalized deviation parameter. Then, the weighted average of the statistical measurements within the sliding window is calculated. Finally, based on the smoothing factor, the weighted average of the outliers and the statistical measurements is weighted and fused to obtain the smoothing result as the target cleanliness factor.

[0083] like Figure 3 As shown, this embodiment of the invention also discloses an electronic device, including a memory 301 storing executable program code and a processor 302 coupled to the memory 301; The processor 302 calls the executable program code stored in the memory 301 to execute the boiler heating surface soot blowing timing determination method described in the above embodiments.

[0084] like Figure 4 As shown in the figure, this invention also discloses a computer device. The computer device includes a processor, a memory, a network interface, a display screen, and an input device connected via a system bus. The processor in this computer design provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer device is used to store data related to the boiler heating surface soot blowing timing determination method. The network interface of the computer device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the boiler heating surface soot blowing timing determination method described in the above embodiments.

[0085] This invention also discloses a computer-readable storage medium storing a computer program that causes a computer to execute the boiler heating surface soot blowing timing determination method described in the above embodiments.

[0086] The storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, or optical data storage device, etc.

[0087] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This computer program can be stored in a storage medium, and when executed, it can include the processes of the embodiments of the methods described above. The aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, optical disk, or read-only memory (ROM), or random access memory (RAM).

[0088] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0089] The embodiments described above are merely examples of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. A method for determining the timing of soot blowing on a boiler heating surface, characterized in that, include: The real-time operating data and static geometric parameters of the boiler's current heating surface are obtained, and the physical properties of the flue gas and steam are calculated based on the real-time operating data and the static geometric parameters. Based on the physical properties of the flue gas and steam, the original cleaning factor is calculated according to the principle of heat exchange balance of the heated surface. An input signal is constructed based on the load change rate. The input signal is then input into a trained correction model to obtain a compensation coefficient. The original cleaning factor is then compensated based on the compensation coefficient to obtain a compensated cleaning factor. Identify that the boiler is in a statistically controlled stable operating range; Historical measurements of the compensating cleaning factor under various operating conditions were collected within the stable operating range. The lower limit of safe operation of the compensation cleaning factor is determined based on the historical measurement values ​​of the compensation cleaning factor; The current measured value of the compensation cleaning factor is subjected to sliding window filtering to obtain the target cleaning factor; If the target cleaning factor remains below the safe operating limit for an extended period of time, a dust removal trigger command is generated.

2. The method for determining the timing of soot blowing on the boiler heating surface according to claim 1, characterized in that, The input signal is constructed based on the load change rate, including: The load change rate is calculated by dividing the difference between the current load and the load at the previous moment by the rated load. Determine whether the fluctuation range of the cleanliness factor of the current heated surface affected by load changes reaches the preset amplitude threshold; If the fluctuation amplitude reaches a preset amplitude threshold, the difference between the current load and the reference load is divided by the rated load to calculate the load deviation change rate; a two-dimensional vector is constructed based on the load change rate and the load deviation change rate as the input signal; If the fluctuation amplitude does not reach the preset amplitude threshold, a one-dimensional vector is constructed based on the load change rate as the input signal.

3. The method for determining the timing of soot blowing on the boiler heating surface according to claim 1, characterized in that, Identifying that the boiler is in a statistically controlled stable operating range includes: When the preset starting conditions are met and the fluctuations of each operating parameter are controlled within the preset fluctuation threshold range, and equipment failure, abnormal human operation, and the period of soot blowing are excluded, the boiler is determined to be in a statistically controlled stable operating state. Under continuous and stable operation, the medium load range of 50%-80%, the low load range of 30%-50%, and the high load range of 80%-100% are obtained respectively. The medium load range, low load range, and high load range are divided into sub-ranges, and then combined to obtain the stable operating range of the boiler under statistical control.

4. The method for determining the timing of soot blowing on the boiler heating surface according to claim 1, characterized in that, Based on historical measurements of the compensating cleaning factor, the safe operating lower limit of the compensating cleaning factor is determined, including: After sorting the historical measurements of the compensation cleaning factor in ascending order, the historical measurements below the lower tail threshold are extracted as over-threshold samples; the lower tail threshold is taken from the 85th percentile of the historical measurements after ascending order. The maximum likelihood estimation method is used to fit the generalized Pareto distribution to the over-threshold samples, and the shape and scale parameters of the distribution are solved. Based on the fitted shape parameters and scale parameters, the lower quantile of the tail probability p≤0.003 is calculated, and the lower limit of safe operation is determined according to the lower quantile.

5. The method for determining the timing of soot blowing on a boiler heating surface according to any one of claims 1 to 4, characterized in that, The target cleaning factor is obtained by performing sliding window filtering on the current measured value of the compensation cleaning factor, including: A sliding window is used to extract the historical measurement value of the compensation cleaning factor that is closest to the current time as the statistical measurement value; The current measurement value of the compensation cleaning factor is filtered based on the statistical measurement value within the sliding window to obtain the target cleaning factor.

6. The method for determining the timing of soot blowing on the boiler heating surface according to claim 5, characterized in that, The current measurement value of the compensation cleaning factor is filtered based on the statistical measurement value within the sliding window to obtain the target cleaning factor, including: The median and the median absolute deviation of each statistical measurement from the median are calculated based on all statistical measurements within the sliding window. Calculate the target absolute deviation between the current measured value of the compensation cleaning factor and the median, and calculate the product of the preset sensitivity coefficient and the median absolute deviation as the target product. Compare the target absolute deviation with the target product. If the absolute deviation of the target is not greater than the product of the targets, the current measured value of the compensation cleaning factor is directly output as the target cleaning factor; If the absolute deviation of the target is greater than the product of the targets, the current measurement value of the compensation cleaning factor is determined to be an outlier. Adaptive smoothing processing is then performed on the outlier to obtain the target cleaning factor.

7. A device for determining the timing of soot blowing on a boiler heating surface, characterized in that, include: The acquisition unit is used to acquire real-time operating data and static geometric parameters of the current heating surface of the boiler; The first calculation unit is used to calculate the physical property parameters of flue gas and steam based on the real-time operating data and the static geometric parameters. The second calculation unit is used to calculate the original cleaning factor based on the physical property parameters of the flue gas and steam and the principle of heat exchange balance of the heating surface. Construction unit, used to construct input signal based on load change rate; The compensation unit is used to input the input signal into the trained correction model to obtain compensation coefficients, and to compensate the original cleaning factor according to the compensation coefficients to obtain a compensated cleaning factor. The identification unit is used to identify whether the boiler is in a statistically controlled stable operating range; The data acquisition unit is used to acquire historical measurements of the compensating cleaning factor under various operating conditions within the stable operating range. The determining unit is configured to determine the safe operating limit of the compensating cleaning factor based on the historical measurement values ​​of the compensating cleaning factor. A filtering unit is used to perform sliding window filtering on the current measured value of the compensation cleaning factor to obtain the target cleaning factor; The determination unit is used to generate a dust blowing trigger command when the target cleaning factor is continuously lower than the safe operating limit and exceeds the set time.

8. The boiler heating surface soot blowing timing determination device according to claim 7, characterized in that, The building unit includes: The calculation sub-unit is used to divide the difference between the current load and the load at the previous moment by the rated load to calculate the load change rate; The judgment sub-unit is used to determine whether the fluctuation range of the cleaning factor of the current heated surface affected by load changes reaches the preset amplitude threshold. The first construction subunit is used to calculate the load deviation change rate by dividing the difference between the current load and the reference load by the rated load when the judgment subunit determines that the fluctuation amplitude reaches the preset amplitude threshold; and to construct a two-dimensional vector as an input signal based on the load change rate and the load deviation change rate. The second construction subunit is used to construct a one-dimensional vector as an input signal based on the load change rate when the judgment subunit determines that the fluctuation amplitude has not reached the preset amplitude threshold.

9. An electronic device, characterized in that, It includes a memory storing executable program code and a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the boiler heating surface soot blowing timing determination method according to any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program causes a computer to execute the boiler heating surface soot blowing timing determination method according to any one of claims 1 to 6.

Images