MESC-based boiler system closed-loop multi-objective optimization method and related equipment

Abstract

The invention discloses a boiler system closed-loop multi-objective optimization method based on MESC and related equipment, relates to the field of heat energy engineering control, and is applied to a coal-fired power generation unit adopting a layered secondary air structure. The method comprises the following steps: firstly, collecting boiler operation data, defining a comprehensive optimization objective function taking coal consumption of unit generating capacity and nitrogen oxide emission concentration as a core, and taking the air-coal ratio of each layer as a variable to be optimized; in the operation process, small-amplitude periodic disturbance signals with different frequencies are overlaid to different air distribution layers to excite system response, the real-time gradient of an objective function to all variables is extracted based on a synchronous detection and filtering algorithm, and an iterative updating strategy is adopted to adjust the set value of the air-coal ratio. And meanwhile, the self-adaptive adjustment of disturbance amplitude and step length parameters is realized by combining the working condition characteristics such as air-coal ratio operation boundary and load, coal quality change and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of thermal energy engineering control, in particular to a boiler system closed-loop multi-objective optimization method based on MESC and related equipment. BACKGROUND

[0002] As an important part of the energy system, coal-fired power plants face multiple challenges such as flexible operation, clean and efficient, and low carbon. With the continuous increase of the proportion of renewable energy grid-connected, the power grid has increasingly higher requirements for the peak shaving capacity of coal-fired units. The units need to frequently change operating conditions within a wide load range, while also need to meet strict pollutant emission limits such as nitrogen oxides (NOx). Under this background, improving the adaptability of the boiler system under uncertain conditions such as dynamic operating conditions, coal quality disturbances, and environmental protection standard upgrades has become a key technical requirement to ensure stable and efficient operation of the energy system.

[0003] Traditional boiler control systems mostly use PID control or single-variable optimization strategies based on mechanism models, which have three significant limitations: first, they rely on fixed system parameters and operating condition assumptions, making it difficult to adapt to dynamic changes such as coal heat value fluctuations and burner wear, resulting in decreased control accuracy; second, the control targets are single (such as focusing only on efficiency or emissions), and multi-objective collaborative optimization is not achieved, lacking a dynamic balancing mechanism when efficiency and environmental protection constraints conflict; third, the response to complex disturbances (such as sudden changes in combustion temperature) is lagging, and the robustness is insufficient, easily leading to unstable system operation. Although existing advanced control algorithms (such as model predictive control, MPC) can handle multi-variable constraints, they rely on accurate mechanism models, and the cost of model parameter calibration is high, making it difficult to accurately follow the changing flexible operation instructions, thereby bringing huge penalty costs to the power plant. At the same time, the optimization of the wind-coal ratio still contains an open-loop optimization mode adjusted according to artificial experience, which lacks accuracy and the optimization speed is difficult to match the future energy system demand. SUMMARY

[0004] The technical problem to be solved by the present application is how to achieve multi-input multi-objective dynamic optimization control without relying on accurate mechanism models. The purpose is to provide a boiler system closed-loop multi-objective optimization method based on MESC and related equipment, which solves the problem of insufficient adaptability of existing boiler systems under wide operating conditions and strong disturbances.

[0005] The present application is implemented by the following technical solutions:

[0006] A boiler system closed-loop multi-objective optimization method based on MESC, comprising:

[0007] Obtain historical operating data of a coal-fired power generating unit, establish a unit simulation model including a boiler combustion subsystem, an air volume control subsystem, and a coal feeding control subsystem, and verify the simulation model based on the historical operating data.

[0008] define a comprehensive objective function on the simulation model and / or the actual boiler system, which is used to comprehensively represent the boiler unit coal consumption and the boiler outlet nitrogen oxide concentration, and take the comprehensive objective function as a multi-objective optimization index of the boiler system;

[0009] divide the boiler secondary air system into multiple air distribution layers according to the nozzle arrangement, set the air-coal ratio of each air distribution layer as a control variable to be optimized, superimpose a low-amplitude periodic disturbance signal on the set value of each air-coal ratio to be optimized, and apply the air-coal ratio set value after superimposing the disturbance to the boiler secondary air distribution control;

[0010] collect the output responses of the boiler unit coal consumption and the nitrogen oxide concentration under the action of the disturbance signal, estimate the gradient of the comprehensive objective function with respect to each air-coal ratio control variable in real time based on the MESC algorithm, and iteratively update each air-coal ratio control variable according to the gradient estimation result;

[0011] establish a real-time data interaction channel between the MESC algorithm module and the power plant distributed control system through a data interaction interface, periodically collect the boiler unit coal consumption and the nitrogen oxide concentration as feedback signals, and write the updated air-coal ratio set value into the boiler air distribution control loop;

[0012] adaptively adjust the frequency and / or amplitude of the disturbance signal according to the current operating state of the boiler.

[0013] Further, the comprehensive objective function is:

[0014]

[0015] wherein MCCR represents the unit coal consumption, represents the boiler outlet nitrogen oxide concentration, and is a weight factor and satisfies ;

[0016] The weight factor and the weight factor are determined according to the proportion of the coal cost and the denitration cost in the total operating cost, the coal cost is calculated based on the unit coal consumption and the standard coal unit price, and the denitration cost is calculated based on the nitrogen oxide concentration, the coal feed rate, the dry flue gas volume, the theoretical ammonia amount required for nitrogen oxide removal, the ammonia-nitrogen ratio, and the liquid ammonia unit price.

[0017] Further, the air-coal ratio is a ratio of air mass flow rate to coal mass flow rate; the control signals corresponding to the auxiliary air air-coal ratio, the compact overfire air air-coal ratio and the separated overfire air air-coal ratio The low-amplitude periodic disturbance signals with different frequencies and amplitudes are injected respectively, and the control signals are expressed as:

[0018]

[0019] wherein, is an initial value of the air-coal ratio, is a disturbance amplitude, is an angular frequency of the disturbance signal;

[0020] The frequencies of the disturbance signals are different from each other, and the ratio of the frequencies of any two disturbance signals is a rational number and satisfies that the sum of any two frequencies is not equal to a third frequency.

[0021] Further, the gradient of the comprehensive objective function with respect to each air-coal ratio control variable is estimated in real time, and each air-coal ratio control variable is updated iteratively according to the gradient estimation result, including:

[0022] The system output signal obtained under the action of the disturbance is high-pass filtered to extract the high-frequency component containing the disturbance response, the filtered signal is multiplied by the corresponding disturbance signal and processed by a low-pass filter to obtain the gradient estimation value of the comprehensive objective function with respect to each air-coal ratio control variable;

[0023] The gradient descent method is used to update each air-coal ratio control variable according to the gradient estimation value, and the air-coal ratio control variable is updated to:

[0024]

[0025] wherein, is a step size coefficient, is the gradient estimation value obtained in the kth iteration, and the gradient estimation and control variable updating are repeatedly executed until the change of the comprehensive objective function is less than a preset convergence threshold.

[0026] Further, before applying the updated air-coal ratio set value to the boiler air distribution control, further comprising:

[0027] Minimum safety values and maximum safety values are set for the auxiliary air air-coal ratio, the compact overfire air air-coal ratio and the separated overfire air air-coal ratio respectively to form corresponding air-coal ratio safety constraint intervals;

[0028] ​​After each update of the wind-coal ratio control variable, it is determined whether the updated value is within the corresponding safety constraint interval, and if it is outside the safety constraint interval, the wind-coal ratio control variable is truncated to the upper limit value or the lower limit value of the safety constraint interval.

[0029] Further, the frequency and / or amplitude of the disturbance signal are adaptively adjusted according to the current operating state of the boiler, including:

[0030] In a flexible peak regulation scenario, the frequency of the disturbance signal is adjusted according to the rate of change of the AGC load instruction from the power grid, so that the period of the disturbance signal matches the time scale of the load change;

[0031] In a combustion temperature abnormality scenario, three types of initial operating conditions, high electricity coal consumption high pollution, high electricity coal consumption low pollution and low electricity coal consumption high pollution, are selected as the combustion abnormality types, and different disturbance amplitudes are selected according to the current electricity coal consumption and the level of nitrogen oxide concentration;

[0032] In a coal calorific value disturbance scenario, when fluctuations in the calorific value of the coal are monitored, the period of the disturbance signal is shortened to increase the response speed of the extremum search control algorithm.

[0033] Further, the variation trajectory of the comprehensive objective function and the nitrogen oxide concentration is recorded, and the operating performance of the unit using the extremum search control algorithm and the unit using the traditional control strategy is compared and evaluated based on the recorded results, wherein the evaluation includes:

[0034] In a flexible peak regulation scenario, the method is run during the process of adjusting the unit load from a first load set value to a second load set value, and the variation amplitude and convergence time of the comprehensive objective function and the nitrogen oxide concentration before and after optimization are compared;

[0035] In a complex combustion condition, three types of initial operating conditions, high electricity coal consumption high pollution, high electricity coal consumption low pollution and low electricity coal consumption high pollution, and a disturbance condition simulating fluctuations in the calorific value of the coal are constructed, the method is run in each condition, the variation trajectory of the electricity coal consumption and the nitrogen oxide concentration is recorded, and the reduction amplitude of the comprehensive objective function is quantified, which is used to evaluate the improvement ability of the method in the future adaptability of the boiler system under different initial conditions.

[0036] The application also provides a boiler system closed-loop multi-objective optimization device based on MESC, which is applied to a coal-fired generating unit boiler system equipped with a distributed control system DCS, to realize the boiler system closed-loop multi-objective optimization method based on MESC as described above, and includes:

[0037] A historical operating data acquisition and storage module is used to acquire and store the operating data of the boiler combustion subsystem, the air volume control subsystem and the coal feeding control subsystem from the historical operating records of the unit;

[0038] A unit simulation modeling module is configured to establish a unit simulation model including a boiler combustion subsystem, an air volume control subsystem and a coal feeding control subsystem based on the historical operation data, and to verify the simulation model in terms of steady state and dynamic characteristics by comparing simulation output with actual operation data;

[0039] A comprehensive objective function calculation module is configured to obtain unit coal consumption per kilowatt-hour and boiler outlet nitrogen oxide concentration on the simulation model and / or actual boiler system, and to calculate a comprehensive objective function according to a preset weight factor, which is used to represent a multi-objective optimization index of the boiler system;

[0040] A disturbance signal injection module is configured to select air-coal ratios corresponding to each air distribution layer of the secondary air as control variables to be optimized, to superimpose a low-amplitude periodic disturbance signal on each air-coal ratio set value, and to output the air-coal ratio set value after the disturbance to a boiler secondary air distribution control loop;

[0041] A MESC control module is configured to receive output responses of the coal consumption per kilowatt-hour and the nitrogen oxide concentration under the action of the disturbance signal, to estimate gradients of the comprehensive objective function with respect to each air-coal ratio control variable in real time, and to iteratively update each air-coal ratio control variable based on the gradient estimation result;

[0042] An operation state analysis and disturbance adaptive adjustment module is configured to obtain current operation state information of the boiler, and to adaptively adjust a frequency and / or amplitude of the disturbance signal according to operation scenarios such as flexible peak regulation, abnormal combustion and coal quality heat value disturbance;

[0043] A data interaction interface module is configured to periodically collect coal consumption per kilowatt-hour and boiler outlet nitrogen oxide concentration as feedback signals through a communication bus, and to write air-coal ratio set values of each air distribution layer updated by the multi-input extremum search control module into a boiler air distribution control loop in the distributed control system (DCS), so as to realize closed-loop multi-objective optimization control of the boiler system.

[0044] The application further provides an electronic device, including a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the MESC-based closed-loop multi-objective optimization method of the boiler system when executing the program.

[0045] The application further provides a computer readable storage medium, which stores a computer program, wherein the program is executable on a processor to implement the MESC-based closed-loop multi-objective optimization method of the boiler system.

[0046] Compared with the prior art, the application has the following advantages and beneficial effects:

[0047] The present application is based on a multi-input extremum seeking control (MESC) algorithm, which actively stimulates the boiler response through small amplitude periodic disturbance, estimates the gradient online and automatically adjusts the air distribution strategy without relying on the combustion mechanism model, so that the boiler can maintain the optimal operating state under different load and fuel change conditions.

[0048] The present application takes the layered air-coal ratio as the optimization variable and realizes multi-channel disturbance identification through frequency decoupling, which can independently identify the influence of different air distribution layers on combustion characteristics. Compared with the existing overall air volume regulation strategy, the optimization granularity is finer and the control precision is higher, which can significantly improve the flame shape, burnout rate and furnace temperature distribution.

[0049] The present application can automatically adjust the disturbance parameters and control step under flexible peak regulation, abnormal combustion and coal quality fluctuation, effectively avoid the performance degradation or even failure of traditional combustion optimization method under working condition change, and improve the robustness and sustainable optimization ability of the algorithm under high noise and large disturbance conditions.

[0050] In summary, the present application realizes sustainable optimization control of the boiler combustion system, has the characteristics of automatic optimization, noise adaptation, model independence, convenient deployment, etc., and can significantly improve the energy efficiency of the unit, reduce pollutant emissions and improve the system operation stability under complex operating scenarios. BRIEF DESCRIPTION OF DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the example embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor. In the drawings:

[0052] Figure 1 The figure is a schematic diagram of the boiler system closed-loop multi-objective optimization structure based on MESC of embodiment 1 of the present application;

[0053] Figure 2 The figure is a schematic diagram of the boiler system control deployment of embodiment 1 of the present application;

[0054] Figure 3 The figure is an air-coal ratio regulation curve and target function response curve of the present application under typical unit peak regulation operating scenarios;

[0055] Figure 4 The figure is an air-coal ratio dynamic response and performance change curve of the present application under multi-scenario working conditions. DETAILED DESCRIPTION

[0056] In order to make the objects, technical solutions and advantages of the present application clearer, further detailed description will be made to the present application in combination with embodiments and drawings, the illustrative embodiments and the description thereof are only used to explain the present application, and do not limit the present application.

[0057] Embodiment 1

[0058] A MESC-based closed-loop multi-objective optimization method for a boiler system, as shown in Figure 1 , comprises:

[0059] Obtaining historical operation data of a coal-fired power generating unit, establishing a unit simulation model including a boiler combustion subsystem, an air volume control subsystem and a coal feeding control subsystem, and verifying the simulation model based on the historical operation data;

[0060] Defining a comprehensive objective function on the simulation model and / or an actual boiler system, for comprehensively representing boiler unit coal consumption and boiler outlet nitrogen oxide concentration, and taking the comprehensive objective function as a multi-objective optimization index of the boiler system;

[0061] Dividing a boiler secondary air system into multiple air distribution layers according to nozzle arrangement, and setting air-coal ratios of the air distribution layers as control variables to be optimized, superimposing a low-amplitude periodic disturbance signal on the set values of the air-coal ratios to be optimized, and applying the air-coal ratio set values after superimposing the disturbance to boiler secondary air distribution control;

[0062] Collecting output responses of boiler unit coal consumption and nitrogen oxide concentration under the action of the disturbance signal, estimating gradients of the comprehensive objective function with respect to the air-coal ratio control variables in real time based on a MESC algorithm, and iteratively updating the air-coal ratio control variables according to the gradient estimation results;

[0063] Establishing a real-time data interaction channel between the MESC algorithm module and a power plant distributed control system through a data interaction interface, periodically collecting boiler unit coal consumption and nitrogen oxide concentration as feedback signals, and writing the updated air-coal ratio set values into a boiler air distribution control loop;

[0064] Adaptively adjusting the frequency and / or amplitude of the disturbance signal according to the current operating state of the boiler.

[0065] As shown in Figure 1As shown, the closed-loop optimization system of the present application is deployed on the boiler combustion control level, wherein "Boiler" represents the main structure of the boiler; the mill (Mill) is used to grind the coal powder and send it into the furnace through the primary air. Multiple secondary air distribution layers are arranged in the furnace, including the auxiliary air layer AA (Auxiliary Air), the close-coupled over-fire air CCOFA (Close-Coupled Over-Fire Air), and the separated over-fire air SOFA (Separated Over-Fire Air), which together constitute the combustion air supply gradient structure.

[0066] The overall control flow of the disturbance signal injection, boiler DCS execution, cost function calculation, and gradient information extraction is shown. Among them, the disturbance signal module outputs sinusoidal disturbance signals with different frequencies and amplitudes to different air distribution layers, and the boiler operation response signals including nitrogen oxides (NOx) and coal consumption per kilowatt-hour (CCR) are fed back to the gradient estimation module, after synchronous detection and filtering processing, output to the MESC controller, and finally update the air-coal ratio set value, realize the combustion optimization and closed-loop control.

[0067] θ(t+1) in the figure represents the next time update value of each air-coal ratio set value calculated by the multi-input extremum seeking control module (MESC, Multi-Input Extremum Seeking Control). The MESC control module executes optimization decision based on the real-time feedback of the coal consumption per kilowatt-hour CCR (Coal Consumption Rate) and the nitrogen oxides emission concentration CNOx (NOx Emission Concentration), and writes the adjusted air-coal ratio set value into the boiler air distribution control channel. The exhaust gases generated by combustion are fed back by the denitration system and online monitoring system for closed-loop optimization control.

[0068] In this embodiment, first, the boiler operation data is collected through the unit historical database and the DCS real-time database, including: coal consumption per kilowatt-hour (CCR, Coal Consumption Rate), nitrogen oxides emission concentration (NOx), dry flue gas flow (Gy), load regulation instruction (AGC, Automatic Generation Control), and air volume and coal quantity measurement point signals;

[0069] Based on the above data, a boiler simulation model is established and dynamic and steady-state verification is performed to ensure that the model is consistent with the actual operation response, which is used to determine the air-coal ratio safe operation boundary and algorithm initialization parameters.

[0070] As shown in Figure 1 The optimization objective of the present application is represented by a comprehensive function related to the operating cost as follows:

[0071]

[0072] wherein MCCR represents the unit degree electricity coal consumption, represents the boiler outlet nitrogen oxide concentration, and is a weight factor and satisfies ;

[0073] The weight factor and the weight factor are determined according to the coal combustion cost and the denitration cost in the total operation cost, the coal combustion cost is calculated based on the unit degree electricity standard coal consumption and the standard coal unit price, and the denitration cost is calculated based on the nitrogen oxide concentration, the coal feed rate, the dry flue gas volume, the theoretical ammonia amount required for removing nitrogen oxide, the ammonia nitrogen ratio and the liquid ammonia unit price. It is expressed as:

[0074]

[0075]

[0076]

[0077] wherein b is the unit degree electricity standard coal consumption (g / kWh); PRICE coal is the standard coal price (CNY / t); C NOx is the boiler outlet NOx concentration (mg / m 3 ); B is the coal feed rate (t / h); V gy is the dry flue gas volume (m 3 / kg); Q m,NH3 is the theoretical ammonia amount required for removing NOx; β is the ammonia nitrogen ratio; PRICE NH3 is the liquid ammonia cost (CNY / t).

[0078] Subsequently, the present application selects the air-coal ratio (Air-Coal Ratio, abbreviated as ACR) corresponding to each air distribution layer as the optimization control variable, and superimposes periodic disturbance signals with different frequencies and amplitudes on these variables: the air-coal ratio is the ratio of air mass flow rate to coal mass flow rate; the control signals corresponding to the auxiliary air air-coal ratio, the compact overfire air air-coal ratio and the separated overfire air air-coal ratio are respectively injected with low-amplitude periodic disturbance signals with different frequencies and amplitudes, and the control signals are expressed as:

[0079] ;

[0080] wherein, is an initial value of the wind-coal ratio, is a disturbance amplitude, is a disturbance signal angular frequency;

[0081] The wind-coal ratio is a core factor affecting the combustion state, and is expressed as:

[0082]

[0083] wherein Q air and Q coal represent the mass flow rates of air and coal powder, respectively;

[0084] The disturbance signal is input to the boiler actuator through a control link to stimulate system response, thereby realizing model-free gradient estimation.

[0085] The frequencies of the disturbance signals are different from each other, and the ratio of the frequencies of any two disturbance signals is a rational number and satisfies the condition that the sum of any two frequencies is not equal to a third frequency.

[0086] The gradient of the comprehensive objective function with respect to each wind-coal ratio control variable is estimated in real time, and each wind-coal ratio control variable is iteratively updated according to the gradient estimation result, including:

[0087] The system output signal obtained under the action of the disturbance is high-pass filtered to extract the high-frequency component containing the disturbance response, the filtered signal is multiplied by the corresponding disturbance signal and processed by a low-pass filter to obtain the gradient estimation value of the comprehensive objective function with respect to each wind-coal ratio control variable; expressed as:

[0088]

[0089] The gradient descent method is used to update each wind-coal ratio control variable according to the gradient estimation value, and the wind-coal ratio control variable is updated to:

[0090] ;

[0091] wherein, is a step size coefficient, is the gradient estimation value obtained in the kth iteration, and the gradient estimation and control variable updating are repeatedly performed until the change in the comprehensive objective function is less than a preset convergence threshold.

[0092] Before applying the updated wind-coal ratio set value to the boiler air distribution control, it further includes:

[0093] Minimum and maximum safety values are set for the auxiliary air wind-coal ratio, the compact overfire air wind-coal ratio, and the separated overfire air wind-coal ratio, respectively, to form corresponding wind-coal ratio safety constraint intervals;

[0094] After each update of the wind-coal ratio control variable, it is judged whether the updated value is within the corresponding safety constraint interval. If it is outside the safety constraint interval, the wind-coal ratio control variable is truncated to the upper limit value or the lower limit value of the safety constraint interval.

[0095] The frequency and / or amplitude of the disturbance signal are adaptively adjusted according to the current operating state of the boiler, including:

[0096] In the flexible peak regulation scenario, the frequency of the disturbance signal is adjusted according to the AGC load instruction change rate from the power grid, so that the disturbance signal period matches the time scale of the load change;

[0097] In the abnormal combustion temperature scenario, three types of initial working conditions, high electricity coal consumption high pollution, high electricity coal consumption low pollution and low electricity coal consumption high pollution, are selected as the combustion abnormal type, and different disturbance amplitudes are selected according to the current electricity coal consumption and the nitrogen oxide concentration level;

[0098] In the coal quality heat value disturbance scenario, when the fluctuation of the coal heat value is monitored, the disturbance signal period is shortened to improve the response speed of the extreme value search control algorithm.

[0099] is expressed as:

[0100]

[0101] That is, the wind-coal ratios corresponding to the three layers of air distribution (AA, CCOFA, SOFA) must all work within their respective allowed safety intervals, and cannot be too low or exceed the upper limit.

[0102] It also includes recording the change trajectory of the comprehensive target function and the nitrogen oxide concentration, and comparing and evaluating the operating performance of the unit using the extreme value search control algorithm and the unit using the traditional control strategy based on the recorded results, wherein the evaluation includes:

[0103] In the flexible peak regulation scenario, the method is run in the process of adjusting the unit load from the first load set value to the second load set value, and the change amplitudes and convergence times of the comprehensive target function and the nitrogen oxide concentration before and after optimization are compared;

[0104] In the complex combustion working condition, three types of initial working conditions, high electricity coal consumption high pollution, high electricity coal consumption low pollution and low electricity coal consumption high pollution, and disturbance working conditions simulating coal quality heat value fluctuations are constructed, and the method is run in each working condition, the change trajectory of the electricity coal consumption and the nitrogen oxide concentration is recorded, and the reduction amplitude of the comprehensive target function is quantified according to the change trajectory, which is used to evaluate the improvement ability of the method in the future adaptability of the boiler system under different initial conditions.

[0105] In this embodiment, a system verification environment is constructed based on the operating data of a certain 1000MW ultra-supercritical coal-fired unit, such asFigure 2 The unit is composed of a boiler combustion subsystem, an air volume control subsystem, a coal feeding control subsystem, and a water feeding control subsystem, and its regenerative system includes three high-pressure heaters, three low-pressure heaters, and an oxygen remover. The unit operation mode follows an AGC (Automatic Generation Control) strategy, and the electric load control loop is kept in a closed-loop operation state.

[0106] In terms of combustion air supply, the unit air supply system includes primary air (PA) and secondary air (SA). The primary air is heated by an air preheater and then sent to a coal mill, mixed with coal powder, and then delivered to a furnace for combustion and outlet temperature adjustment; the secondary air is sent to different height areas according to the layered structure of the furnace, used to provide main combustion air and participate in the combustion process, and the secondary air accounts for a large proportion in the total air volume.

[0107] During operation, key control parameters are collected, including but not limited to boiler load, unit power coal consumption (CCR), nitrogen oxide emission concentration (NOx), and air-coal ratio (ACR) of each layer, which are used for algorithm optimization, model training, and closed-loop control execution.

[0108] The simulation parameters obtained by simulation under the closed-loop control state are compared with the actual thermal parameters of the operation process of a power plant, and the error is within a controllable range, as shown in Table 1.

[0109]

[0110] The multi-input extremum search control (MESC) algorithm proposed in the application mainly consists of four functions: signal generation, gradient estimation, parameter optimization, and safety constraint. In each iteration period, the algorithm first collects the current unit power coal consumption and NOx concentration as feedback signals, calculates the comprehensive objective function combined with the set weight factor; then, different frequency disturbance signals are injected into the three types of secondary air; then, the Fourier transform is used to extract the fundamental component of the target function response, and the gradient direction of each input is calculated; finally, the control parameters are updated based on the gradient descent method, and the safety constraint module is used to check whether the parameters are within the safety range, and finally the optimized value is output.

[0111] The MESC algorithm takes the ACR of each layer of secondary air as the input variable (u ACR_AA , u ACR_CCOFA , u ACR_SOFA ), and the comprehensive objective function J MESCTo optimize the index, the gradient estimation of system output response is realized by injecting periodic disturbance signal, and then the wind-coal ratio parameter is dynamically adjusted based on gradient descent method to minimize the objective function. The weight coefficients p and q are determined by the ratio of coal cost to denitration cost. When the standard coal price is 1000 yuan / ton, p=0.95 and q=0.05 are calculated.

[0112] The core of the proposed MESC algorithm architecture is the multivariate gradient optimization mechanism, including obtaining boundary conditions and initial conditions from historical data, running the physical model according to the initial parameters, reading the data of power consumption and pollutant concentration, modulation and demodulation process, returning the optimized wind-coal ratio, and the iteration process of the physical model receiving data for the next simulation. The design of the disturbance signal meets three conditions: each frequency is different, the ratio of any two frequencies is a rational number, and the sum of any two frequencies is not equal to the third frequency, which ensures the time scale separation of the optimization loop and the controlled process.

[0113] To verify the performance of the algorithm, the traditional control method is selected as the benchmark to compare and analyze the optimization effect under different scenarios.

[0114] Case 1 (effect see Figure 3 ): In the 1000-900 MW flexible operation test, the proposed algorithm reduces the comprehensive objective function by 1.53%, NO x concentration by 27%, and converges to the optimal value within 1800 seconds.

[0115] Case 2 (effect see Figure 4 abc):

[0116] The test results of complex conditions show that:

[0117] In the high power consumption and high NOx scenario, the optimized power consumption reduces by 1.46%, NOx reduces by 7.42%, and the comprehensive objective function reduces by 1.81%;

[0118] In the high power consumption and low NOx scenario, the power consumption reduces by 1.02%, and the comprehensive objective function reduces by 0.30%;

[0119] In the low power consumption and high NOx scenario, NOx reduces by 25.47%, and the comprehensive objective function reduces by 1.48%.

[0120] Case 3 (effect see Figure 4 d): In the coal quality disturbance test, the algorithm realizes stable optimization within 3500 seconds, and the comprehensive objective function reduces by 1.21% compared with the initial state of disturbance.

[0121] The dynamic performance index shows that the rise time of each layer ACR adjustment is less than 700 seconds, the overshoot is controlled in a reasonable range, and the stable time is not more than 1500 seconds, which is significantly superior to the adjustment speed of the traditional control method in optimization, and the optimization from several tens of minutes to several hours is changed into a closed-loop system which can be roughly adjusted to the optimal value near the optimal value in a few minutes and complete the whole optimization in tens of minutes.

[0122] The MESC algorithm based on the Simulink-APROS closed-loop control architecture provided by the application, through the synergistic effect of the disturbance signal generation, gradient estimation and parameter adjustment modules, constructs a multivariable optimization framework suitable for a boiler system, and successfully realizes the dynamic optimization of the secondary air-fuel ratio in the 1000-900 MW flexible operation and complex combustion scenarios. The algorithm effectively solves the open-loop optimization limitations of the traditional boiler control system under load fluctuations and fuel blending conditions and the AGC signal response delay problem, and exhibits fast convergence ability and stable optimization effect under complex combustion conditions and coal quality disturbances, fully verifying its effectiveness and superiority.

[0123] Embodiment 2

[0124] The MESC-based boiler system closed-loop multi-objective optimization device is applied to a coal-fired generating unit boiler system equipped with a distributed control system DCS to realize the MESC-based boiler system closed-loop multi-objective optimization method as described in Embodiment 1, and includes:

[0125] A historical operation data acquisition and storage module is configured to acquire and store operation data of a boiler combustion subsystem, an air volume control subsystem and a coal feeding control subsystem from historical operation records of the unit;

[0126] A unit simulation modeling module is configured to establish a unit simulation model including the boiler combustion subsystem, the air volume control subsystem and the coal feeding control subsystem based on the historical operation data, and to perform steady-state and dynamic characteristic verification on the simulation model by comparing simulation output and actual operation data;

[0127] A comprehensive objective function calculation module is configured to acquire unit degree-electricity coal consumption and boiler outlet nitrogen oxide concentration on the simulation model and / or the actual boiler system, and to calculate a comprehensive objective function according to a preset weight factor, the comprehensive objective function being used to represent multi-objective optimization indexes of the boiler system;

[0128] A disturbance signal injection module is configured to select air-fuel ratios corresponding to each air distribution layer of the secondary air as to-be-optimized control variables, to superimpose low-amplitude periodic disturbance signals on each air-fuel ratio set value, and to output the air-fuel ratio set value with the disturbance to a boiler secondary air distribution control loop;

[0129] a MESC control module configured to receive output responses of the degree electric coal consumption and the nitrogen oxide concentration under the action of the disturbance signal, to perform real-time estimation of gradients of the comprehensive target function with respect to the wind-to-coal ratio control variables, and to iteratively update the wind-to-coal ratio control variables based on the gradient estimation results;

[0130] a running state analysis and disturbance adaptive adjustment module configured to acquire current running state information of the boiler, and to adaptively adjust the frequency and / or amplitude of the disturbance signal according to running scenarios such as flexible peak regulation, abnormal combustion, and coal quality heat value disturbance;

[0131] a data interaction interface module configured to periodically collect the degree electric coal consumption and the boiler outlet nitrogen oxide concentration of the unit as feedback signals through a communication bus, and to write the wind-to-coal ratio set values of the wind distribution layers updated by the multi-input extreme value search control module into a boiler wind distribution control loop in the distributed control system (DCS), so as to realize closed-loop multi-target optimization control of the boiler system.

[0132] Embodiment 3

[0133] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the MESC-based closed-loop multi-target optimization method of the boiler system as described in Embodiment 1 when executing the program.

[0134] Embodiment 4

[0135] A computer readable storage medium having a computer program stored thereon, and the program is executable on a processor to implement the MESC-based closed-loop multi-target optimization method of the boiler system as described in Embodiment 1.

[0136] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system or a computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) containing computer-usable program code.

[0137] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks.

[0138] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks.

[0139] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks.

[0140] Those skilled in the art can understand that all or part of the steps of the above-mentioned facts and methods can be instructed by programs to relevant hardware, and the programs involved or the programs can be stored in a computer-readable storage medium. When the program is executed, the steps include the corresponding method steps, and the storage medium can be ROM / RAM, a magnetic disc, an optical disc, etc.

[0141] The above specific embodiments further illustrate the purpose, technical solutions and advantages of the present application. It should be understood that the above description is only a specific embodiment of the present application and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for closed loop multi-objective optimization of a boiler system based on MESC, characterized in that, The method comprises the following steps: acquiring historical operation data of a coal-fired generator unit, establishing a unit simulation model including a boiler combustion subsystem, an air volume control subsystem and a coal feeding control subsystem, and verifying the simulation model based on the historical operation data; defining a comprehensive objective function on the simulation model and / or an actual boiler system, which is used to comprehensively represent boiler unit coal consumption and boiler outlet nitrogen oxide concentration, and taking the comprehensive objective function as a multi-objective optimization index of the boiler system; dividing the boiler secondary air system into multiple air distribution layers according to nozzle arrangement, setting air-coal ratios of each air distribution layer as control variables to be optimized, superimposing a low-amplitude periodic disturbance signal on the set value of each air-coal ratio to be optimized, and applying the air-coal ratio set value after superimposing the disturbance to boiler secondary air distribution control; under the action of the disturbance signal, collecting the output responses of boiler unit coal consumption and nitrogen oxide concentration, estimating the gradient of the comprehensive objective function with respect to each air-coal ratio control variable in real time based on the MESC algorithm, and iteratively updating each air-coal ratio control variable according to the gradient estimation result; establishing a real-time data interaction channel between the MESC algorithm module and the power plant distributed control system through a data interaction interface, periodically collecting boiler unit coal consumption and nitrogen oxide concentration as feedback signals, and writing the updated air-coal ratio set value into the boiler air distribution control loop; adapting the frequency and / or amplitude of the disturbance signal according to the current operating state of the boiler.

2. The MESC-based boiler system closed loop multi-objective optimization method according to claim 1, wherein, The integrated objective function is: ； wherein MCCR represents the unit degree electricity coal consumption, represents the boiler outlet nitrogen oxide concentration, and is a weight factor and satisfies ; The weight factor And the weight factor According to the coal-fired cost With the denitration cost In the total operating cost, the proportion of the coal-fired cost Based on the unit power standard coal consumption and the standard coal unit price, the denitration cost is calculated Based on the nitrogen oxide concentration, the coal supply, the dry flue gas volume, the theoretical ammonia required for nitrogen oxide removal, the ammonia nitrogen ratio, and the liquid ammonia unit price.

3. The MESC-based boiler system closed loop multi-objective optimization method of claim 1, wherein, The wind-coal ratio is the ratio of air mass flow rate to coal mass flow rate; control signals corresponding to the auxiliary wind-coal ratio, the compact overfire air wind-coal ratio, and the separated overfire air wind-coal ratio Periodic disturbance signals with different frequencies and amplitudes are injected respectively, and the control signals It is represented as: ； wherein is the wind to coal ratio initial value, is the perturbation amplitude, is the perturbation signal angular frequency; The frequencies of the disturbance signals are different from each other, and the ratio of the frequencies of any two disturbance signals is a rational number and satisfies that the sum of any two frequencies is not equal to a third frequency.

4. The MESC-based boiler system closed loop multi-objective optimization method of claim 1, wherein, The gradient of the comprehensive objective function with respect to each air-coal ratio control variable is estimated in real time, and each air-coal ratio control variable is iteratively updated according to the gradient estimation result, which comprises: high-pass filtering the system output signal obtained under the action of the disturbance to extract the high-frequency component containing the disturbance response, multiplying the filtered signal by the corresponding disturbance signal and processing it through a low-pass filter to obtain the gradient estimation value of the comprehensive objective function with respect to each air-coal ratio control variable; The wind-to-coal ratio control variable is updated according to the gradient estimate value by using the gradient descent method, and the wind-to-coal ratio control variable in the kth iteration is updated to: is updated to: ； wherein, is a step coefficient, is a gradient estimate obtained in the kth iteration, and the gradient estimation and the control variable update are repeatedly performed until the variation of the integrated objective function is less than a preset convergence threshold.

5. The MESC-based boiler system closed loop multi-objective optimization method of claim 1, wherein, before applying the updated air-coal ratio set value to the boiler air distribution control, further comprising: setting minimum and maximum safety values for the air-coal ratios of auxiliary air, compact overfire air and separate overfire air respectively to form corresponding air-coal ratio safety constraint intervals; after updating the air-coal ratio control variable each time, judging whether the updated value is within the corresponding safety constraint interval, if it exceeds the safety constraint interval, truncating the air-coal ratio control variable to the upper limit value or the lower limit value of the safety constraint interval.

6. The MESC-based boiler system closed loop multi-objective optimization method of claim 1, wherein, Adapting the frequency and / or amplitude of the disturbance signal according to the current operating state of the boiler, which comprises: in the flexible peak regulation scenario, adjusting the disturbance signal frequency according to the AGC load instruction change rate from the power grid to match the disturbance signal period with the load change time scale; in the abnormal combustion temperature scenario, taking three types of initial working conditions of high unit coal consumption and high pollution, high unit coal consumption and low pollution, and low unit coal consumption and high pollution as combustion abnormality types, and selecting different disturbance amplitudes according to the current unit coal consumption and nitrogen oxide concentration levels. In the coal quality heat value disturbance scenario, when the fluctuation of the coal heat value is monitored, the disturbance signal period is shortened to improve the response speed of the extreme value search control algorithm.

7. The MESC-based boiler system closed loop multi-objective optimization method of claim 1, wherein, Also included is recording the change trajectory of the comprehensive target function and the nitrogen oxide concentration, and based on the recording results, comparing and evaluating the operation performance of the unit adopting the extreme value search control algorithm and the unit adopting the traditional control strategy, wherein the evaluation includes: In the flexible peak regulation scenario, the method is run in the process of adjusting the unit load from the first load set value to the second load set value, and the change amplitudes of the comprehensive target function and the nitrogen oxide concentration before and after optimization and the convergence time are compared; In the complex combustion condition, three types of initial conditions of high electricity coal consumption and high pollution, high electricity coal consumption and low pollution, and low electricity coal consumption and high pollution are constructed respectively, and a disturbance condition of coal quality heat value fluctuation is simulated, the method is run in each condition, the change trajectory of the electricity coal consumption and the nitrogen oxide concentration is recorded, and the reduction amplitude of the comprehensive target function is quantified based on this, which is used to evaluate the improvement ability of the method in the future adaptability of the boiler system under different initial conditions.

8. A MESC-based closed-loop multi-objective optimization device for a boiler system, applied to a coal-fired power generating unit boiler system equipped with a distributed control system (DCS), to implement the MESC-based closed-loop multi-objective optimization method according to any one of claims 1 to 7, characterized in that, It includes: A historical operation data acquisition and storage module for acquiring and storing operation data of a boiler combustion subsystem, an air volume control subsystem and a coal feeding control subsystem from historical operation records of the unit; A unit simulation modeling module for establishing a unit simulation model including the boiler combustion subsystem, the air volume control subsystem and the coal feeding control subsystem based on the historical operation data, and performing steady-state and dynamic characteristic verification on the simulation model by comparing simulation output with actual operation data; A comprehensive target function calculation module for acquiring unit electricity coal consumption and boiler outlet nitrogen oxide concentration on the simulation model and / or actual boiler system, and calculating a comprehensive target function according to a preset weight factor, which is used to represent multi-objective optimization indicators of the boiler system; A disturbance signal injection module for selecting air-coal ratios of each air distribution layer of the secondary air as control variables to be optimized, superimposing low-amplitude periodic disturbance signals on each air-coal ratio set value, and outputting the air-coal ratio set value after adding the disturbance to the boiler secondary air distribution control loop; A MESC control module for receiving output responses of the electricity coal consumption and the nitrogen oxide concentration under the action of the disturbance signal, estimating gradients of the comprehensive target function with respect to each air-coal ratio control variable in real time, and iteratively updating each air-coal ratio control variable based on the gradient estimation results; An operation state analysis and disturbance adaptive adjustment module for acquiring current operation state information of the boiler, and adaptively adjusting the frequency and / or amplitude of the disturbance signal according to flexible peak regulation, abnormal combustion and coal heat value disturbance and other operation scenarios; A data interaction interface module for periodically collecting unit electricity coal consumption and boiler outlet nitrogen oxide concentration as feedback signals through a communication bus, and writing each air distribution layer air-coal ratio set value updated by the multi-input extreme value search control module into the boiler air distribution control loop in the distributed control system (DCS), so as to realize closed-loop multi-objective optimization control of the boiler system.

9. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the MESC-based boiler system closed-loop multi-objective optimization method according to any one of claims 1-7 when executing the program.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the MESC-based boiler system closed-loop multi-objective optimization method according to any one of claims 1-7.

Images