Quick response and stable control method for boiler load

Abstract

The invention relates to the technical field of boiler load control, in particular to a boiler load quick response and stable control method which comprises the following steps: step 1, acquiring core operation parameters in real time and constructing a boiler parameter acquisition library; 2, dynamic thermal resistance is calculated based on the soot thickness of the heating surface, and the actual heat exchange efficiency of the boiler is corrected; 3, determining an initial load distribution scheme of the boiler in combination with the corrected actual heat exchange efficiency and the heat demand predicted value; step 4, optimizing an initial load distribution scheme based on the flue gas component parameters, and determining a target load; 5, performing dynamic fine adjustment on the target load based on the real-time heat demand; and 6, boiler fuel input and air supply are controlled according to the finely-adjusted load. According to the method, the traditional mode of estimating the output according to the rated load is abandoned, the influence of ash pollution attenuation on the output is accurately quantified through dynamic thermal resistance correlation heat exchange efficiency correction and actual maximum output load calculation, and equipment damage caused by overload operation and energy idling caused by underload waste are avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of boiler load control, in particular to a boiler load rapid response and stable control method. BACKGROUND

[0002] In the field of industrial production and central heating, as the core heat supply equipment, the rapid response capability and long-term operation stability of the boiler load directly determine the energy utilization efficiency, production continuity and environmental protection indicators.

[0003] A variety of traditional boiler load control methods have been proposed in the prior art. For example, the patent with the patent publication number CN111486428A discloses a boiler load control method, system and storage medium, which mainly predicts the total heat supply in the next time period by querying the user's preset demand or based on the weighted calculation of the actual load of the previous three time periods and the current load; then determines the number of boilers to be started and the initial load according to the principle of starting the maximum heat supply load boiler to the rated value first; finally, the boilers with smaller maximum load are fine-tuned first to adapt to the actual heat supply demand.

[0004] In the actual operation of the boiler, the above-mentioned method has the following problems:

[0005] First, the load distribution only determines the starting priority according to the maximum load size of the boiler, and the dynamic influence of the boiler heating surface ash fouling on the heat exchange efficiency is not considered throughout the process. In the long-term operation of the boiler, the thickness of the heating surface ash fouling will gradually increase with the running time, resulting in a significant decrease in the heat exchange coefficient of the heating surface. Even if the total heat supply in the next time period is predicted and the boiler load is matched according to the method described in the patent, the actual steam production will still be lower than the demand value due to the decrease in heat exchange efficiency, and the number of boilers to be started or the load parameters need to be repeatedly corrected, causing the load response to lag, and frequent adjustment will further exacerbate the load fluctuation, damaging the operation stability.

[0006] Second, the small load boiler with lower thermal efficiency is fine-tuned first, and in order to meet the actual heat supply demand, the fuel input of the boiler needs to be increased by a larger margin, which not only causes fuel waste, but also causes the load regulation sensitivity to decrease due to the low thermal efficiency, and the small fluctuations in the actual load cannot be quickly adapted, and the regulation accuracy is significantly insufficient.

[0007] Therefore, a boiler load rapid response and stable control method is designed to solve the defects of lagging response, insufficient accuracy and low energy efficiency in the prior art, which has significant technical necessity and practical application value. SUMMARY

[0008] The present application is to solve one of the above technical problems, the technical solution adopted is: a boiler load rapid response and stable control method, comprising the following steps: step 1, real-time acquisition of core operating parameters and construction of a boiler parameter acquisition library, the core operating parameters including: the ash fouling thickness of the boiler heating surface; the steam pocket pressure, the steam pocket water level, the main steam flow, the main steam temperature, the flue gas component parameters; the oxygen volume fraction and the carbon monoxide volume fraction in the boiler tail flue; the load demand parameters: real-time heat demand is collected through user-side metering equipment, and based on historical heat demand data and environmental parameters, a prediction model is used to obtain a heat demand prediction value for the next 10-15 minutes, and the prediction update frequency is 1 time / 10 minutes.

[0009] Step 2, calculate the dynamic thermal resistance based on the ash fouling thickness of the heating surface, and correct the actual heat exchange efficiency of the boiler, specifically: calculate the dynamic thermal resistance of the heating surface according to the collected ash fouling thickness and the preset ash fouling heat conduction coefficient; and based on the ratio of the dynamic thermal resistance to the reference thermal resistance, linearly correct the actual heat exchange efficiency of the boiler.

[0010] Step 3, determine the initial load distribution scheme of the boiler in combination with the corrected actual heat exchange efficiency and the heat demand prediction value.

[0011] Step 4, optimize the initial load distribution scheme based on the flue gas component parameters to determine the target load.

[0012] Step 5, dynamically fine-tune the target load based on the real-time heat demand, select the first two boilers with actual heat exchange efficiency, if the load deviation after fine-tuning is >10% of the real-time heat demand corresponding to the steam volume, re-execute steps 3-4 to update the opened boiler group and the target load.

[0013] Step 6, control the boiler fuel input and air supply according to the load after fine-tuning.

[0014] On the basis of any one of the above technical solutions, further optimization is: the ash fouling thickness measurement accuracy of each sensing node of the distributed optical fiber sensor in step 1 is ±0.05mm; the sampling point of the laser gas analyzer is set at 1-2m before the boiler tail flue outlet, the oxygen volume fraction measurement accuracy is ±0.1%, and the carbon monoxide volume fraction measurement accuracy is ±5×10 -6 .

[0015] On the basis of any one of the above technical solutions, further optimization is: the prediction model in step 1 is an existing LSTM neural network model, the input layer of the LSTM neural network model includes historical heat demand data, outdoor temperature, and wind speed, the hidden layer is set to 2-3 layers, the output layer is the heat demand prediction value for the next 10-15 minutes, and the prediction error is controlled within ±5%.

[0016] On the basis of any of the preceding technical solutions, further optimization is that: the calculation steps of the dynamic heat resistance of the heating surface are: step a: collecting original data of the ash fouling thickness.

[0017] Step b: screening effective data of the ash fouling thickness: checking the ash fouling thickness value of each node in the original data list one by one to determine whether it is out of the normal range: if the ash fouling thickness of a node is less than or equal to 3 mm, the data of the node is retained and included in the effective data set; if the ash fouling thickness of a node is greater than 3 mm, the data of the node is excluded and not included in the effective data set.

[0018] Step c: calculating the average value of the ash fouling thickness; the number of data in the effective data set is counted, the ash fouling thickness values of all effective nodes are added to obtain the total ash fouling thickness, and then the average value is calculated through the formula δ=(S*0.001) / m.

[0019] Step d: calculating the dynamic heat resistance; the ash fouling thermal conductivity is obtained, the specific value is determined according to the actual boiler ash fouling type, the average value of the ash fouling thickness calculated in step 3 is combined, and the dynamic heat resistance is calculated through the formula Rd=δ / λ.

[0020] On the basis of any of the preceding technical solutions, further optimization is that: the specific formula of the linear correction of the actual heat exchange efficiency of the boiler in step 2 is: actual heat exchange efficiency= rated heat exchange efficiency* [1-k*(dynamic heat resistance / reference heat resistance)]; wherein the correction coefficient k is determined according to the fuel type.

[0021] On the basis of any of the preceding technical solutions, further optimization is that: the calculation method of the current actual maximum output load of a single boiler in step 3 is: actual maximum output load= boiler rated load*(actual heat exchange efficiency / rated heat exchange efficiency), wherein the boiler rated load is the rated evaporation capacity marked on the boiler nameplate, and the actual heat exchange efficiency is the corrected value in step 2.

[0022] On the basis of any of the preceding technical solutions, further optimization is that: the specific logic of distributing the initial load of a single boiler according to the proportion of the actual maximum output load of each boiler in step 3 is: single boiler initial load= actual maximum output load of the boiler*(steam demand / sum of actual maximum output loads of the started boiler group), and the single boiler initial load cannot exceed 90% of the actual maximum output load, reserving 10% of the load adjustment margin; wherein the actual maximum output load is obtained in the above steps.

[0023] On the basis of any of the preceding technical solutions, further optimization is that: the calculation logic of the current actual steam production of the started boiler group in step 5 is: actual steam production=Σ(single boiler main steam flow*collection duration), wherein the collection duration is 1 min, and the measurement accuracy of the main steam flow is plus or minus 0.1 t / h.

[0024] On the basis of any one of the technical solutions, further optimization is that in the deviation calculation in step 5, if the steam quantity fluctuation frequency corresponding to the real-time heat demand is > 1 / 3 min, the average value of the steam quantity corresponding to the real-time heat demand in the last 3 min is taken to participate in the deviation calculation.

[0025] Compared with the prior art, the application has the following beneficial effects:

[0026] 1. Significantly improve fuel and energy utilization efficiency, reduce operating energy consumption: the application optimizes the load by the flue gas components, reduces the waste of incomplete combustion of fuel, and improves the fuel utilization rate; at the same time, dynamic air supply control avoids invalid air supply, and the energy consumption of the fan is reduced by 15%-20%.

[0027] 2. Reduce equipment maintenance cost and prolong service life of core components:

[0028] Relying on the standby boiler priority selection rule, on the one hand, the accumulation of ash pollution caused by the high load operation of the low-efficiency boiler is reduced, the heating surface ash cleaning period is prolonged, and the labor and equipment loss cost is reduced; on the other hand, the wear and tear of the fuel valve and the fan motor caused by frequent load adjustment is reduced, the service life of the fuel valve is prolonged, the probability of fan bearing failure is reduced, and the annual maintenance cost of the overall equipment is reduced.

[0029] 3. Ensure stable supply of heat demand and reduce the risk of interruption on the user side: by updating the heat demand prediction value for 10-15 min in the future once every 10 min to realize demand prediction, the deviation is dynamically adjusted to quickly respond to real-time demand fluctuations, and the heat demand satisfaction rate is improved by combining with the selection of high-efficiency and stable boiler groups.

[0030] 4. Precisely match the actual working condition of the boiler to avoid the risk of unbalanced load distribution:

[0031] The application discards the traditional method of estimating the output according to the rated load, and calculates the actual maximum output load by dynamically correcting the heat exchange efficiency through the dynamic thermal resistance correlation, accurately quantifies the influence of ash pollution attenuation on the output, avoids equipment damage caused by overloading operation and energy idling caused by underloading waste; at the same time, the combustion optimization further ensures the matching of load and combustion state, and avoids the safety and efficiency problems caused by unbalanced load distribution.

[0032] 5. Adapt to multi-scene industrial boilers and have wide industry promotion value: the application considers universality and pertinence in technical design: on the fuel type, the correction coefficient k and the ash pollution thermal conductivity coefficient are subdivided according to coal, gas and biomass, covering mainstream fuel boilers, and having large-scale application prospects. DETAILED DESCRIPTION

[0033] The embodiments of the technical solutions of the application will be described in detail below. The following embodiments are only used to more clearly illustrate the technical solutions of the application.

[0034] Embodiment 1: A boiler load rapid response and stable control method, comprising the following steps: Step 1: Real-time acquisition of core operating parameters and construction of a boiler parameter acquisition library, the core operating parameters including: heating surface state parameters: collecting the ash deposition thickness of the boiler heating surface (water wall, superheater) through a distributed optical fiber sensor, with a collection frequency of 1 time / 3-5 min; conventional operating parameters: collecting the drum pressure, drum water level, main steam flow, and main steam temperature through pressure, water level, flow, and temperature sensing devices, with a collection frequency of 1 time / 1 min; flue gas component parameters: collecting the oxygen volume fraction and carbon monoxide volume fraction in the boiler tail flue through a laser gas analyzer, with a collection frequency of 1 time / 2 min; load demand parameters: collecting real-time heat demand through user-side metering equipment, and based on historical heat demand data and environmental parameters (outdoor temperature, wind speed), obtaining a heat demand prediction value for the next 10-15 min through a prediction model, with a prediction update frequency of 1 time / 10 min.

[0035] The core function of Step 1 is to comprehensively capture, accurately collect, and structurally store the core data required for boiler load regulation and control, specifically including three aspects: first, targeted collection function, matching distributed optical fiber sensors, special pressure sensing devices, and other equipment for the detection needs of different parameters such as ash deposition thickness and drum pressure, to ensure that each parameter can be accurately detected through the appropriate equipment, avoiding data deviation caused by improper equipment selection; second, differentiated frequency control function, setting different collection frequencies from 1 time / 1 min to 1 time / 10 min according to the parameter change rate (such as slow change of ash deposition thickness and real-time fluctuation of main steam flow), ensuring the real-time nature of rapidly changing parameters and avoiding excessive collection of slowly changing parameters; third, data integration and storage function, integrating the four types of parameters collected in a unified dimension (collection time, detection point, etc.), constructing a structured parameter acquisition library, so that the required data can be directly called in subsequent steps without additional data processing.

[0036] Traditional boiler parameter acquisition does not match special detection for different parameters' impact on load regulation, which may lead to insufficient detection accuracy of key parameters (such as ash deposition thickness using conventional visual detection which is easily disturbed by flue gas, and flue gas components using electrochemical sensors with response lag).

[0037] Change parameter collection from single storage to multi-dimensional application: First, when building a structured parameter collection library, store it according to the association of collection time-detection point-parameter type, so that multiple parameter combinations under the same time dimension can be quickly called in subsequent steps (such as step 4 optimization of combustion, main steam flow, flue gas composition, and drum pressure data can be simultaneously called to judge the matching relationship between load and combustion); Second, realize associated verification through multi-parameter synchronous collection (such as when the main steam flow increases, if the oxygen volume fraction of flue gas decreases and the carbon monoxide volume fraction increases, it can be predicted in advance that the combustion is insufficient, and a warning is provided for step 4 load optimization); Third, the stored historical data can support long-term optimization (such as step 6 daily optimization of heat exchange efficiency correction logic, the correlation data of ash fouling thickness and heat exchange efficiency in the past 30 days can be called to optimize the dynamic thermal resistance calculation coefficient). Through the extension of data application, parameter collection not only becomes a basic link, but also becomes a support carrier for load regulation and control precision and iteration.

[0038] Step 2: Calculate dynamic thermal resistance based on the ash fouling thickness of the heating surface, and correct the actual heat exchange efficiency of the boiler, specifically: according to the collected ash fouling thickness and the preset ash fouling thermal conductivity coefficient (0.18-0.22 W / (m.K) for coal-fired boilers, 0.15-0.18 W / (m.K) for gas-fired boilers, and 0.20-0.25 W / (m.K) for biomass boilers), calculate the dynamic thermal resistance of the heating surface; then based on the ratio of dynamic thermal resistance to baseline thermal resistance (0.003-0.005 m 2 .K / W), linearly correct the actual heat exchange efficiency of the boiler. The corrected actual heat exchange efficiency decreases with the increase of dynamic thermal resistance.

[0039] This step calculates dynamic thermal resistance through the linkage of ash fouling thickness and thermal conductivity coefficient, realizes the precision of thermal resistance quantification, and is fundamentally different from traditional experience-based estimation of thermal resistance.

[0040] In traditional boiler heat exchange efficiency calculation, the change of thermal resistance caused by ash fouling on the heating surface is often estimated by experience (such as uniformly corrected by a fixed thermal resistance value), without considering the real-time change of ash fouling thickness and the difference in thermal conductivity characteristics of different fuel ash fouling, resulting in large deviation in thermal resistance calculation (error often exceeds 15%), which further affects the accuracy of heat exchange efficiency correction.

[0041] The advantages of this step are: first, the real-time collected soot thickness of the heating surface (precise monitoring by the distributed optical fiber sensor in step 1, accuracy ± 0.05 mm) is used as the basic data for heat resistance calculation, avoiding the subjectivity of empirical estimation, and making the heat resistance calculation directly related to the actual soot accumulation state; second, different fuel types (coal, gas, biomass) have different soot thermal conductivity characteristics, and different soot thermal conductivity coefficients are preset (coal 0.18-0.22 W / (m·K), gas 0.15-0.18 W / (m·K), biomass 0.20-0.25 W / (m·K)), such as biomass fuel soot containing high alkali metals, which has a higher thermal conductivity coefficient, and the heat resistance calculation is more realistic. Through the linkage calculation of real-time soot thickness + fuel adaptive thermal conductivity coefficient, the quantitative error of dynamic heat resistance can be controlled within 5%, providing accurate heat resistance basis for subsequent heat exchange efficiency correction.

[0042] The heat exchange efficiency is linearly corrected by the ratio of dynamic heat resistance to reference heat resistance, realizing the dynamic efficiency correction and solving the hysteresis problem of traditional fixed efficiency calculation.

[0043] Traditional boilers often use the rated heat exchange efficiency (a fixed value under design conditions) as the basis for load calculation, ignoring the decrease in heat exchange efficiency caused by soot accumulation during operation, resulting in a large deviation between actual output and calculated value (for example, the rated efficiency is 93%, the actual efficiency is reduced to 88% due to soot accumulation, and the load is still calculated according to 93%, which may cause insufficient output).

[0044] The advantages of this step are: a linear correction logic of dynamic heat resistance / reference heat resistance is established, which directly links the real-time heat resistance change of the heating surface to the heat exchange efficiency - the reference heat resistance (0.003-0.005 m 2 ·K / W) is determined based on the rated heat exchange efficiency of the boiler, reflecting the ideal heat resistance state under design conditions; when the dynamic heat resistance increases due to soot accumulation, the ratio of dynamic heat resistance to reference heat resistance increases, and the actual heat exchange efficiency after correction decreases accordingly (for example, the dynamic heat resistance increases from 0.004 m 2 ·K / W to 0.008 m 2 ·K / W, the ratio increases from 1 to 2, combined with the correction coefficient k, the efficiency decreases linearly). This correction method can track the efficiency decay caused by soot in real time, making the actual heat exchange efficiency always match the current operating condition, avoiding the load calculation deviation caused by traditional fixed efficiency, and providing accurate efficiency basis for step 3 to calculate the actual maximum output load.

[0045] Step 3: Combine the corrected actual heat exchange efficiency with the heat demand prediction value to determine the initial load distribution scheme of the boiler. Specifically, according to the ratio of the actual heat exchange efficiency to the rated heat exchange efficiency of the boiler, the current actual maximum output load of a single boiler is calculated. Compare the steam demand corresponding to the heat demand prediction value with the actual maximum output load of a single boiler. If the steam demand is less than or equal to the actual maximum output load of a single boiler, start the current boiler. If the steam demand is greater than the actual maximum output load of a single boiler, select standby boilers according to the principle of actual maximum output load from large to small, and form a starting boiler group, so that the sum of the actual maximum output loads of the starting boiler group is greater than or equal to the steam demand. Then, according to the proportion of the actual maximum output load of each boiler, the initial load of a single boiler is distributed.

[0046] Traditional boiler load distribution often directly uses rated load (design output marked on nameplate) as the basis for actual callable output, ignoring the impact of heat exchange efficiency decay (such as efficiency reduction due to ash pollution) on actual output, resulting in a theoretical demand that can be met but actual output that is insufficient (such as rated load 40t / h, actual maximum output only 38t / h due to efficiency reduction, still distributing load according to 40t / h, which is easy to cause steam supply gap).

[0047] The advantage of this step is that the corrected actual heat exchange efficiency of step 2 is associated with the rated heat exchange efficiency, and the calculation logic of actual maximum output load = boiler rated load x (actual heat exchange efficiency / rated heat exchange efficiency) is constructed - actual heat exchange efficiency reflects the true heat exchange capacity of the boiler under current working conditions, and the ratio directly reflects the impact of efficiency decay on output (such as actual heat exchange efficiency 89.2%, rated efficiency 93.5%, ratio ≈0.954, actual maximum output of boiler with rated load 40t / h ≈38.16t / h). Through this calculation method, the quantitative error of actual maximum output can be controlled within 3%, avoiding the estimation bias of traditional rated load direct application, and providing accurate capacity reference for subsequent boiler start-stop and load distribution.

[0048] In traditional boiler group operation, boiler start-stop often relies on operator experience (such as feeling that demand is large to open more boilers), which is easy to cause two problems: one is under-provision (total output of starting boiler group is insufficient, which cannot meet the demand), the other is over-provision (too many boilers are started, single boiler runs at low load for a long time, energy consumption increases, equipment wear and tear intensifies).

[0049] The advantage of this step is to establish a precise matching logic of demand-capacity: first, convert the heat demand prediction value into steam demand, and then compare it with the actual maximum output load of a single boiler: if the demand is less than or equal to the actual maximum output of a single boiler, only the current boiler needs to be started to meet the demand, avoiding energy waste caused by multiple boiler start-stop; if the demand is greater than the actual maximum output of a single boiler, select standby boilers according to the principle of actual maximum output from large to small (preferentially call boilers with strong output to reduce the number of start-ups), and ensure that the total output of the started boiler group is greater than or equal to the steam demand (for example, if the demand is 65 t / h, the actual maximum output of a single boiler is 38 t / h, and a 35 t / h boiler needs to be added, the total output is 73 t / h, which is greater than 65 t / h). This configuration method not only avoids the demand gap caused by under-provisioning, but also prevents resource waste caused by over-provisioning, achieving economic and efficient operation of the boiler group.

[0050] Traditional initial load distribution often uses average distribution (such as two boilers each taking 50% of the load) or empirical distribution (operating personnel subjectively set the load ratio), ignoring the differences in actual maximum output of different boilers, resulting in overloading of boilers with weak output and low load of boilers with strong output (for example, the actual maximum output of A boiler is 38 t / h, and the actual maximum output of B boiler is 35 t / h. If they are evenly distributed, each will bear 32.5 t / h, and the load ratio of B boiler will be approximately 92.9%, close to full-load operation, while the load ratio of A boiler will be approximately 85.5%, still with redundancy).

[0051] The advantage of this step is to use the actual maximum output ratio of a single boiler in the started boiler group as the basis for distribution to build a distribution logic of single boiler initial load = actual maximum output load of the boiler × (steam demand / sum of actual maximum output load of started boiler group) - the higher the ratio of a boiler (the stronger the actual output), the more initial load it will bear, avoiding unbalanced load distribution.

[0052] Step 4: Optimize the initial load distribution scheme based on flue gas component parameters to determine the target load. Specifically, determine whether the collected oxygen volume fraction (reasonable range 3%-6%) and carbon monoxide volume fraction (reasonable range ≤80×10 -6 ) are within the preset reasonable range: if oxygen is <3% and carbon monoxide is >80×10 -6 , reduce the initial load of the corresponding boiler by 5%-8% of the current load and increase the speed of the induced draft fan by 10%-15%; if oxygen is >6% and carbon monoxide is ≤80×10 -6 , increase the initial load of the corresponding boiler by 3%-5% of the current load and reduce the speed of the induced draft fan by 5%-10%; if both are within the reasonable range, maintain the initial load; the optimized load is used as the target load of the started boiler group.

[0053] Traditional boiler load adjustment is usually only around to meet the steam demand, without relating to the combustion state (such as blindly increasing fuel to improve load, without synchronously adjusting the air supply, leading to insufficient oxygen and increased carbon monoxide), which is easy to cause load to meet the demand but combustion to be inefficient (such as load meeting the demand but carbon monoxide exceeding the standard, fuel waste increasing) or combustion safety risk (such as too low oxygen, causing incomplete combustion coking).

[0054] The advantage of this step is that the oxygen volume fraction in flue gas (reflecting the matching degree of air supply and fuel) and carbon monoxide volume fraction (reflecting the combustion sufficiency) are taken as core judgment indexes to build the linkage logic of combustion state-load adjustment. When oxygen < 3% and carbon monoxide > 80x10 -6 , it indicates that the fuel is excessive and the air supply is insufficient, so the load needs to be reduced (fuel input is reduced) and the induced draft fan speed needs to be increased (flue gas volume is increased to improve air supply); when oxygen > 6% and carbon monoxide ≤ 80x10 -6 , it indicates that the air supply is excessive and the fuel is relatively insufficient, so the load needs to be increased (fuel input is increased) and the induced draft fan speed needs to be reduced (invalid air supply is reduced to reduce fan energy consumption). Through this linkage optimization, the load adjustment not only meets the demand, but also synchronously ensures that the combustion is in the efficient interval (oxygen 3%-6%, carbon monoxide ≤ 80x10 -6 ), the fuel utilization rate is improved, and the traditional load and combustion disconnection inefficiency problem is avoided.

[0055] This step synchronously adjusts the induced draft fan speed and the load to realize the dynamic matching of wind resistance and load. When the load is increased, if the oxygen is reasonable, the induced draft fan speed is simultaneously increased to ensure that the air supply matches the increase of fuel and avoids insufficient air supply; when the load is reduced, the induced draft fan speed is simultaneously reduced to reduce invalid air supply and reduce fan energy consumption.

[0056] Step 5: dynamically fine-tune the target load based on real-time heat demand, specifically: calculate the deviation of the real-time heat demand corresponding steam quantity and the actual steam production of the opened boiler group (the sum of the main steam flow of each boiler); if the deviation ≤ 5% of the real-time heat demand corresponding steam quantity, select the boiler with the highest actual heat exchange efficiency in the opened boiler group and fine-tune the current load by ±2%-±3%; if 5% of the real-time heat demand corresponding steam quantity < deviation ≤ 10% of the real-time heat demand corresponding steam quantity, select the top two boilers in terms of actual heat exchange efficiency and fine-tune the current load of each by ±3%-±5%; if the deviation > 10% of the real-time heat demand corresponding steam quantity, re-execute steps 3-4 to update the opened boiler group and the target load.

[0057] The advantage of this step is that: first, the deviation of real-time heat demand corresponding to steam volume and the current actual steam output of the opened boiler group is classified as ≤5%, 5%-10% and >10%, and then the differentiated fine-tuning strategy is matched - small deviation (≤5%) only needs to fine-tune a single high-efficiency boiler with a small amplitude (±2%-±3%) to avoid excessive intervention; medium deviation (5%-10%) needs to fine-tune two high-efficiency boilers with a medium amplitude (±3%-±5%) to quickly reduce the deviation; large deviation (>10%) is determined as the current boiler group cannot meet the demand, and the scheme is directly reconstructed.

[0058] Step 6: According to the load control after fine-tuning, the boiler fuel input and air supply are controlled, and a parameter feedback updating mechanism is established, specifically:

[0059] According to the ratio of the load after fine-tuning and the actual maximum output load, the fuel valve opening degree is determined (the ratio is increased by 10%, and the opening degree is increased by 8%-12%), and the air supply fan speed is determined in combination with the oxygen volume fraction of flue gas (the load is increased by 10% and the oxygen is reasonable, and the speed is increased by 6%-9%); every 1h, the heating surface ash thickness is re-measured, if the growth exceeds 0.3mm, steps 2-6 are re-executed to update the parameters; every day, based on the load adjustment data of the day, the heat exchange efficiency correction logic and the flue gas component adjustment amplitude are optimized to improve the subsequent control precision.

[0060] The traditional boiler fuel air supply control often uses a fixed ratio (such as the fuel valve opening degree and the load are adjusted according to 1:1, and the air supply fan speed is constant), without considering the actual maximum output load difference and combustion state feedback, which is easy to cause two problems: one is fuel excess; the other is air supply mismatch.

[0061] The advantage of this step is that: a double linkage control logic is constructed, on the one hand, the fuel valve opening degree is determined according to the ratio of the load after fine-tuning and the actual maximum output load (the ratio is increased by 10%, and the opening degree is increased by 8%-12%), to ensure that the fuel input matches the actual carrying capacity of the boiler; on the other hand, the air supply fan speed is adjusted in combination with the oxygen volume fraction of flue gas to ensure that the air supply changes synchronously with the fuel and load.

[0062] On the basis of any one of the above technical solutions, further optimization is: the ash thickness measurement accuracy of each sensing node of the distributed optical fiber sensor in step 1 is ±0.05mm; the sampling point of the laser gas analyzer is set at 1-2m before the boiler tail flue outlet, the oxygen volume fraction measurement accuracy is ±0.1%, and the carbon monoxide volume fraction measurement accuracy is ±5×10 -6 .

[0063] The sampling point of the laser gas analyzer is arranged at a position 1-2 m before the outlet of the tail flue, which is in a stable region of the flue flow field, and the flue gas is fully mixed to be uniform in composition (to avoid measurement deviation caused by local high / low oxygen), so that the collected flue gas component data can truly reflect the overall combustion state of the boiler.

[0064] On the basis of any one of the technical solutions above, further optimization is that: the prediction model in step 1 is an LSTM neural network model, the input layer of the LSTM neural network model comprises historical heat demand data (heat demand in the same time period in the past 30 days), outdoor temperature (measurement accuracy ±0.5℃), and wind speed (measurement accuracy ±0.2 m / s), the hidden layer is set to 2-3 layers, the output layer is a heat demand prediction value in the future 10-15 min, and the prediction error is controlled within ±5%.

[0065] It should be noted that: the core invention point of the patent is a boiler load rapid response and stable control method, rather than the LSTM prediction model itself. Details of model training (such as iteration number and activation function) belong to conventional parameters in the application of the LSTM model (a person skilled in the art can set the parameters according to the data volume and accuracy requirements, such as selecting ReLU as the activation function and setting the iteration number to 100-200 times), and the core architecture (input layer, hidden layer, and output layer) and performance index (±5% error) of the model are clearly defined in the present application. A person skilled in the art can complete model construction and training based on the existing LSTM model without technical obstacles, which belongs to the prior art.

[0066] Specific operation examples: I. Data preparation. Data collection:

[0067] Historical heat demand: user-side steam meter (±0.1 t / h) collects data in the same time period in the past 30 days (such as 9:00-9:15 every day), a total of 30 groups (58-65 t / h);

[0068] Environmental parameters: platinum resistance sensor (±0.5℃) collects outdoor temperature in the same period, and cup-type anemometer (±0.2 m / s) collects wind speed in the same period.

[0069] Preprocessing: missing values: linear interpolation (such as using the average values of the previous and next days to supplement the missing temperature).

[0070] Normalization: Min-Max method mapping to [0, 1] (formula: normalized value = (original value-minimum value) / (maximum value-minimum value)).

[0071] Set division: 7:3 division of training set (21 groups) and test set (9 groups).

[0072] II. Model construction (according to conventional parameters in the industry).

[0073] Based on TensorFlow / Keras:

[0074] Architecture: Input layer: dimension 3 (heat demand + temperature + wind speed), time step 5, shape (None, 5, 3); Hidden layer: 2 layers (optionally 3 layers), 64 / 32 units, ReLU activation (to prevent gradient vanishing); Output layer: 1 fully connected layer, Linear activation (regression task).

[0075] Training parameters: Optimizer: Adam (learning rate 0.001); Loss function: Mean Squared Error (MSE).

[0076] Number of iterations: 150 times (regular range of 100-200 times), batch_size = 32.

[0077] On the basis of any one of the technical solutions described above, further optimization is as follows: the calculation steps of the dynamic heat resistance of the heated surface are as follows: step a: collect the original data of the thickness of the ash deposition; collect the thickness data of the ash deposition of each monitoring node of the heated surface through the distributed optical fiber sensor, record the specific value of each node (unit: mm), form an original data list, for example: node 1 (2.1 mm), node 2 (3.2 mm), node 3 (2.8 mm) …… node n (xmm).

[0078] Step b: screen the effective data of the thickness of the ash deposition; check the thickness value of the ash deposition of each node in the original data list one by one, and judge whether it is out of the normal range (the normal range is ≤3 mm): if the thickness of the ash deposition of a node is ≤3 mm, the data of the node is retained and included in the effective data set; if the thickness of the ash deposition of a node is >3 mm, the data of the node is excluded and not included in the effective data set.

[0079] Example: node 2 (3.2 mm) in the original data is >3 mm, and the data is excluded, and the effective data set is node 1 (2.1 mm), node 3 (2.8 mm) …… node n (xmm, x≤3).

[0080] Step c: calculate the average value of the thickness of the ash deposition; count the number of data (denoted as m) in the effective data set, add the thickness values of all the effective nodes of the ash deposition to obtain the total thickness of the ash deposition (denoted as S, unit: mm), and then calculate the average value (denoted as δ, unit: m) through the formula (mm needs to be converted to m, and the conversion coefficient is 1 mm = 0.001 m); formula: δ = (S×0.001) / m.

[0081] Example: the effective data is 2.1 mm and 2.8 mm, S = 4.9 mm, m = 2, and δ = (4.9×0.001) / 2 = 0.00245 m.

[0082] Step d: Calculate the dynamic thermal resistance; Obtain the ash fouling thermal conductivity (denoted as λ, unit: W / (m.K), the specific value needs to be determined according to the actual boiler ash fouling type, such as the ash fouling thermal conductivity of common coal-fired boilers is about 0.1-0.2 W / (m.K)), combined with the average value of ash fouling thickness (δ) calculated in step 3, the dynamic thermal resistance (denoted as Rd, unit: m 2 .K / W) is calculated by the formula: Rd=δ / λ.

[0083] Example: δ=0.00245m, λ=0.15W / (m.K), Rd=0.00245 / 0.15≈0.0163m 2 .K / W.

[0084] The reference thermal resistance determination step is:

[0085] Step 1: Obtain the rated heat exchange efficiency of the boiler; From the boiler design file, product specification or equipment parameter account book, query and confirm the rated heat exchange efficiency (denoted as η, unit: %) of the boiler, ensure the accuracy of the value, such as the rated heat exchange efficiency of a certain boiler is marked as 93.5%.

[0086] Step 2: Match the reference thermal resistance range; According to the obtained rated heat exchange efficiency (η), compare the preset efficiency and reference thermal resistance corresponding relationship, determine the value range of reference thermal resistance (denoted as R_b, unit: m 2 .K / W).

[0087] If 92%≤η≤95%, the value range of R_b is 0.003-0.004m 2 .K / W, the specific value can be selected within the range according to the actual operation condition of the boiler (such as fuel type, load fluctuation), such as 0.0035m 2 .K / W is taken under normal working condition.

[0088] If 88%≤η≤91%, the value range of R_b is 0.004-0.005m 2 .K / W, the specific value can be selected according to the actual working condition, such as 0.0048m 2 .K / W is taken under high load working condition.

[0089] If η is out of the above two intervals (such as η<88% or η>95%), the accuracy of the rated heat exchange efficiency data needs to be rechecked, or the reference thermal resistance value rule is supplemented according to the design standard of similar boilers.

[0090] This optimization step is not isolated, but forms a deep cooperation with step 1 equipment precision optimization and step 2 heat exchange efficiency correction: the distributed optical fiber sensor (±0.05mm) of step 1 provides high-precision input for step a original data collection.

[0091] The steps a-d of the optimization provide low error dynamic thermal resistance and reference thermal resistance for step 2.

[0092] Step 2 corrects the heat exchange efficiency based on the accurate thermal resistance parameter, and step 3 calculates the actual maximum output load, and finally step 6 realizes accurate control of fuel air supply.

[0093] Embodiment 2: Compared with embodiment 1, the difference lies in that it further comprises the following technical features:

[0094] On the basis of any one of the above technical solutions, further optimization is that the specific formula for linearly correcting the actual heat exchange efficiency of the boiler in step 2 is: actual heat exchange efficiency = rated heat exchange efficiency x [1-k x (dynamic thermal resistance / reference thermal resistance)]; wherein the correction coefficient k is determined according to the type of fuel: for coal-fired boilers, k is preferably 0.08-0.10, for gas-fired boilers, k is preferably 0.06-0.08, and for biomass boilers, k is preferably 0.09-0.11.

[0095] The value range of the correction coefficient k is strictly based on the current boiler industry standard in China, and all values are based on the published industry standard. Technical personnel can ensure that the corrected heat efficiency calculation is operable and verifiable according to the test method of GB / T10184, which belongs to the existing technical field. The value logic, range definition and application principle are derived from published boiler industry standards and technical literature, which will not be repeated.

[0096] The linear correlation between thermal resistance change and heat exchange efficiency decay is a mature existing technology in the field of boiler thermal calculation, and its core principle is clearly recorded in public data such as "Industrial Boiler Thermal Performance Test Procedures" (GB / T10184) and "Boiler Principles" (Mechanical Industry Press, 2015 edition).

[0097] The value of the correction coefficient k is based on the current boiler industry public standard, mature technical literature and general test method, and technical personnel can realize complete reproduction by searching public data and executing standard test procedures.

[0098] On the basis of any one of the above technical solutions, further optimization is that the calculation method of the current actual maximum output load of a single boiler in step 3 is: actual maximum output load = boiler rated load x (actual heat exchange efficiency / rated heat exchange efficiency), wherein the boiler rated load is the rated evaporation capacity (unit: t / h) marked on the boiler nameplate, and the actual heat exchange efficiency is the corrected value in step 2.

[0099] In step 3, when selecting standby boilers according to the actual maximum output load from large to small, the boiler with a soot thickness growth of less than or equal to 0.2 mm in the past 24 hours is preferred; if there are multiple standby boilers with the same soot thickness growth, the boiler with a load adjustment frequency of less than or equal to 3 times in the past 12 hours is preferred to reduce equipment wear and tear.

[0100] In the conventional boiler actual maximum output load calculation, the rated load marked on the nameplate is directly used without considering the heat exchange efficiency decay caused by soot deposition on the heating surface and changes in fuel quality during operation, resulting in a disconnection between the theoretical output and the actual capacity.

[0101] The advantage of the optimization calculation method is that the corrected actual heat exchange efficiency in step 2 is associated with the rated efficiency, which makes the load capacity quantitatively match the current operating condition. The actual heat exchange efficiency reflects the real-time heat exchange capacity of the boiler, and the ratio of the actual heat exchange efficiency to the rated heat exchange efficiency accurately quantifies the influence of efficiency decay on the output. Multiplying the rated load can obtain the real maximum output under the current operating condition, avoiding overloading distribution or underloading waste.

[0102] On the basis of any one of the technical solutions above, the further optimization is that the specific logic for distributing the initial load of a single boiler according to the proportion of the actual maximum output load of each boiler in step 3 is: the initial load of a single boiler = the actual maximum output load of the boiler × (steam demand quantity / sum of actual maximum output loads of the started boiler group), and the initial load of a single boiler cannot exceed 90% of its actual maximum output load, leaving a 10% load adjustment margin; wherein the actual maximum output load is obtained from the above steps.

[0103] When obtaining the current steam demand quantity, the current actual steam demand quantity (denoted as Q, unit: t / h) is obtained through the existing steam pipe network monitoring system, production process demand record or operation scheduling instruction, ensuring that the data matches the load adjustment range of the started boiler group. For example: the current steam demand quantity Q = 65 t / h.

[0104] When obtaining the sum of the actual maximum output loads of the started boiler group, step 1: determine the started boiler group and the actual maximum output load of a single boiler (specifically, determine the started boiler group by combing the set of boilers that need to be started and running, and extract the actual maximum output load (denoted as Pa1, Pa2,..., Pan, unit: t / h, n is the number of started boilers) of each boiler in the started boiler group from the results of the above single boiler actual maximum output load calculation step). n

[0105] ​For example: start 2 boilers, the actual maximum output load of boiler 1 Pa1=38.16t / h, the actual maximum output load of boiler 2 Pa2=35.82t / h.); Step 2: calculate the sum of the actual maximum output load of the started boiler group (specifically, add the actual maximum output load of a single boiler extracted in step 1 to obtain the sum of the actual maximum output load of the started boiler group (denoted as P 总 a).

[0106] On the basis of any of the above technical solutions, further optimization is: the calculation logic of the actual steam production of the started boiler group in step 5 is: actual steam production = Σ(single boiler main steam flow rate × collection time length), wherein the collection time length is 1 min, and the measurement accuracy of the main steam flow rate is ±0.1 t / h; when calculating the deviation, if the steam quantity fluctuation frequency corresponding to the real-time heat demand is >1 time / 3 min, the average value of the steam quantity corresponding to the real-time heat demand in the last 3 min is taken to participate in the deviation calculation, so as to avoid frequent fine adjustment.

[0107] The specific steps for calculating the actual steam production of the started boiler group are: step 1: set the main steam flow rate collection parameters.

[0108] It is clear that the collection time length is 1 min, and it is confirmed that the main steam flow rate measurement device (such as a flow meter) of each boiler in the started boiler group meets the requirement of ±0.1 t / h, and it is ensured that the device is normally operated and the data can be collected in real time.

[0109] Step 2: collect the main steam flow rate of a single boiler for 1 min: in the same 1 min collection period, the main steam flow rates of each boiler in the started boiler group are collected synchronously (denoted as q1, q2, …, q n , unit: t / h, n is the number of started boilers), and the real-time flow rate data of each boiler is recorded.

[0110] Step 3: calculate the actual steam production of the started boiler group.

[0111] According to the actual steam production = Σ(single boiler main steam flow rate × collection time length), the collection time length (1 min = 1 / 60 h) is multiplied by the single main steam flow rate and then summed up, and the formula is simplified as: formula: actual steam production (t) = (q1+q2+…+q n )×(1 / 60)q n .

[0112] Example: the main steam flow rates of 3 boilers for 1 min are 30 t / h, 28 t / h and 25 t / h respectively, then the actual steam production = (30+28+25)×(1 / 60) = 83×(1 / 60)≈1.38 t.

[0113] Step 4: monitor the steam quantity fluctuation frequency corresponding to the real-time heat demand.

[0114] Continuously track the change of the real-time heat demand corresponding steam quantity (denoted as q1, unit: t / h), and count the fluctuation frequency (i.e. the number of value changes per unit time) to determine whether the fluctuation frequency satisfies the condition of > 1 times / 3 min.

[0115] Step 5: Determine the heat demand steam quantity for deviation calculation.

[0116] If the q1 fluctuation frequency is ≤ 1 times / 3 min: directly take the current q1 to participate in the deviation calculation;

[0117] If the q1 fluctuation frequency is > 1 times / 3 min: take the average value of all q1 data in the last 3 min (denoted as Qflat) to participate in the deviation calculation, to avoid frequent fine-tuning.

[0118] The front end of the present application builds a data foundation through step 1 multi-dimensional core parameter acquisition (heated surface ash fouling, conventional operation, flue gas composition, load demand), and relies on distributed optical fiber sensors (±0.05 mm precision) and laser gas analyzers (±0.1% oxygen precision) to ensure data reliability.

[0119] The middle end realizes the conversion from data to decision through step 2 dynamic heat resistance calculation and heat exchange efficiency correction (linear formula + fuel adaptation correction coefficient k), step 3 actual maximum output calculation and standby boiler selection (efficiency related output + state priority selection), and step 4 flue gas composition optimized load (double parameter judgment + quantitative adjustment).

[0120] The back end ensures the regulation and control landing and continuous optimization through step 5 real-time heat demand dynamic fine-tuning (deviation classification + high-efficiency boiler priority) and step 6 fuel air supply control and feedback iteration (load related fuel / air supply + daily optimization logic).

[0121] The above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or make equivalent substitutions for part or all of the technical features; and these modifications or substitutions do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application; any alternative improvement or change made by those skilled in the art to the embodiments of the present application falls within the protection scope of the present application.

[0122] The parts not described in detail in the present application are well-known to those skilled in the art.

Claims

1. A method for rapid response and stable control of a boiler load, characterized by, The method comprises the following steps: Step 1, real-time acquisition of core operating parameters and construction of a boiler parameter acquisition library, wherein the core operating parameters comprise: The ash fouling thickness of the heating surface of the boiler; The pressure of the steam drum, the water level of the steam drum, the main steam flow, the main steam temperature, and the flue gas component parameters; The oxygen volume fraction and the carbon monoxide volume fraction in the tail flue of the boiler; The load demand parameter: real-time heat demand is acquired through a user-side metering device, and a prediction value of the heat demand in the next 10-15 minutes is acquired through a prediction model based on historical heat demand data and environmental parameters, and the prediction update frequency is 1 time / 10 minutes; Step 2, calculation of dynamic thermal resistance based on the ash fouling thickness of the heating surface, and correction of the actual heat exchange efficiency of the boiler, specifically: According to the acquired ash fouling thickness and the preset ash fouling thermal conductivity coefficient, the dynamic thermal resistance of the heating surface is calculated; and then based on the ratio of the dynamic thermal resistance to the reference thermal resistance, the actual heat exchange efficiency of the boiler is linearly corrected; Step 3, determination of an initial load distribution scheme of the boiler in combination with the corrected actual heat exchange efficiency and the prediction value of the heat demand; Step 4, optimization of the initial load distribution scheme based on the flue gas component parameters to determine a target load; Step 5, dynamic fine adjustment of the target load based on the real-time heat demand, and selection of boilers with the top two actual heat exchange efficiencies, if the load deviation after the fine adjustment is greater than 10% of the steam quantity corresponding to the real-time heat demand, steps 3-4 are re-executed to update the opened boiler group and the target load; Step 6, control of the fuel input and the air supply of the boiler according to the load after the fine adjustment.

2. The boiler load quick response and stabilization control method according to claim 1, characterized by, The ash thickness measurement accuracy of each sensing node of the distributed optical fiber sensor in step 1 is ±0.05 mm; the sampling point of the laser gas analyzer is set at 1-2 m before the outlet of the boiler tail flue, and the oxygen volume fraction measurement accuracy is ±0.1%, and the carbon monoxide volume fraction measurement accuracy is ±5x10 -6 .

3. The boiler load quick response and stabilization control method according to claim 2, characterized by, The prediction model in step 1 is an existing LSTM neural network model, the input layer of the LSTM neural network model comprises historical heat demand data, outdoor temperature and wind speed, the hidden layer is set to 2-3 layers, the output layer is a prediction value of the heat demand in the next 10-15 minutes, and the prediction error is controlled within ±5%.

4. The boiler load quick response and stabilization control method according to claim 3, characterized by, The calculation steps of the dynamic thermal resistance of the heating surface are as follows: Step a: acquisition of ash fouling thickness original data; Step b: screening of effective ash fouling thickness data; The ash fouling thickness values of each node in the original data list are checked one by one to determine whether they are out of the normal range: If the ash fouling thickness of a node is less than or equal to 3 mm, the data of the node is retained and included in the effective data set; If the ash fouling thickness of a node is greater than 3 mm, the data of the node is excluded and not included in the effective data set; Step c: calculation of the average value of the ash fouling thickness; The number of data in the effective data set is counted, the ash fouling thickness values of all effective nodes are added to obtain the total ash fouling thickness, and then the average value is calculated through the formula δ=(S×0.001) / m; Step d: calculation of the dynamic thermal resistance; The ash fouling thermal conductivity coefficient needs to be determined according to the actual ash fouling type of the boiler, and the average value of the ash fouling thickness calculated in step 3 is combined to calculate the dynamic thermal resistance through the formula Rd=δ / λ.

5. The boiler load quick response and stabilization control method according to claim 4, characterized by, The specific formula for linearly correcting the actual heat exchange efficiency of the boiler in step 2 is: actual heat exchange efficiency=rated heat exchange efficiency×[1-k×(dynamic thermal resistance / reference thermal resistance)]; wherein the correction coefficient k is determined according to the fuel type.

6. The boiler load quick response and stabilization control method according to claim 5, characterized by, The calculation method of the current actual maximum output load of the single boiler in step 3 is: actual maximum output load = boiler rated load × (actual heat exchange efficiency / rated heat exchange efficiency), wherein the boiler rated load is the rated evaporation capacity marked on the boiler nameplate, and the actual heat exchange efficiency is the value corrected in step 2.

7. The boiler load quick response and stabilization control method according to claim 6, characterized by, The specific logic of distributing the initial load of the single boiler according to the proportion of the actual maximum output load of each boiler in step 3 is: single boiler initial load = actual maximum output load of the boiler × (steam demand / sum of actual maximum output loads of the started boiler group), and the single boiler initial load cannot exceed 90% of the actual maximum output load, and 10% of the load adjustment margin is reserved; wherein the actual maximum output load is obtained in the above steps.

8. The boiler load quick response and stabilization control method according to claim 7, characterized by, The calculation logic of the current actual steam production of the started boiler group in step 5 is: actual steam production = Σ(single boiler main steam flow × collection time length), wherein the collection time length is 1 min, and the measurement accuracy of the main steam flow is ±0.1 t / h.

9. The boiler load quick response and stabilization control method according to claim 8, characterized by, In the deviation calculation in step 5, if the steam quantity fluctuation frequency corresponding to the real-time heat demand is >1 time / 3 min, the average value of the steam quantity corresponding to the real-time heat demand in the last 3 min is taken to participate in the deviation calculation.