Thermal power generating unit deep peak regulation method, device, equipment and medium

Abstract

The invention belongs to the field of thermal power generation dispatching, and relates to a thermal power generating unit deep peak regulation method, device and equipment and a medium, and the method comprises the steps: processing multi-source data of a thermal power generating unit through a multi-source data processing model, and obtaining a predicted load; performing dual-objective optimization based on the predicted load to obtain a thermal power generating unit instruction; based on the boiler instruction, the heat storage instruction and the wind power frequency modulation standby load, main steam parameters and the actual air distribution ratio are obtained through cooperative combustion control; processing the steam turbine instruction and the main steam pressure to obtain steam turbine parameters; wind power rapid frequency modulation is combined, the boiler climbing time is shortened, the service life loss of equipment is reduced, wind-light-fire storage deep coupling is achieved, new energy fluctuation is stabilized, and the heat storage utilization rate is increased.

Description

Technical Field

[0001] This invention relates to the field of thermal power generation dispatching, and specifically discloses a method, apparatus, equipment, and medium for deep peak shaving of thermal power units. Background Technology

[0002] With the vigorous development of new energy sources, in scenarios involving multi-energy complementarity of wind, solar, thermal, and energy storage, and high proportions of new energy, traditional peak-shaving methods for thermal power units still rely on single-boiler combustion adjustments. Furthermore, the coordination between the thermal storage system and the unit is poor, failing to integrate wind and solar frequency regulation with the rapid heat release capacity of thermal storage. Boilers alone cannot meet the grid's response requirements, and the instability of boiler combustion limits the flexibility of thermal power peak-shaving, resulting in limited peak-shaving capacity for existing thermal power units. In addition, the high penetration of new energy sources necessitates frequent deep peak-shaving for thermal power, leading to low-cycle fatigue of turbine rotors and excessive thermal stress in the steam drum, making components highly susceptible to fatigue damage and exacerbating losses. Moreover, existing technologies do not enhance stratified heat release and waste heat utilization from thermal storage, resulting in weak thermal storage coordination, a lack of lifespan compensation and wind / solar frequency regulation, and insufficient equipment safety and peak-shaving speed.

[0003] In view of this, the present invention provides a method, device, equipment and medium for deep peak shaving of thermal power units, which is applicable to deep peak shaving of coal-fired units in the scenarios of multi-energy complementarity of wind, solar, thermal and storage and new energy consumption. Combined with wind power fast frequency regulation, it shortens the boiler ramp-up time, reduces equipment life loss, and realizes deep coupling of wind, solar, thermal and storage, smooths the fluctuation of new energy and improves the utilization rate of thermal storage. Summary of the Invention

[0004] The purpose of this invention is to provide a method, device, equipment, and medium for deep peak shaving of thermal power units, addressing the following problems: combining wind power rapid frequency regulation with thermal storage and heat release to shorten boiler ramp-up time and system response time, and reduce furnace temperature fluctuations at low loads; reducing rotor lifespan losses, increasing unit lifespan, and reducing steam drum thermal stress fluctuations; achieving deep coupling of wind, solar, thermal, and energy storage to mitigate new energy fluctuations; the specific solution is as follows:

[0005] A deep peak-shaving method for thermal power units includes: processing multi-source data of the thermal power unit using a multi-source data processing model to obtain predicted load; the predicted load includes predicted peak-shaving load, wind power frequency regulation reserve load, and predicted thermal storage load; performing bi-objective optimization based on the predicted load to obtain thermal power unit commands; the thermal power unit commands include boiler commands, turbine commands, and thermal storage commands; based on the boiler commands, the thermal storage commands, and the wind power frequency regulation reserve load, obtaining main steam parameters and actual air-fuel ratio through coordinated combustion control; the main steam parameters include main steam pressure and main steam temperature; processing the turbine commands and main steam pressure to obtain turbine parameters; the turbine parameters include turbine control valve opening and actual turbine output.

[0006] Furthermore, the step of processing multi-source data from thermal power units using a multi-source data processing model to obtain predicted load includes: acquiring multi-source data from thermal power units; the multi-source data includes grid commands, historical wind and solar power output, load demand, historical unit load, ambient temperature, and equipment status; processing the multi-source data to obtain uncertainty data intervals and uncertainty data weights; calculating system uncertainty based on the uncertainty of the uncertainty data and the uncertainty data weights; determining the wind power frequency regulation reserve load and the predicted thermal storage load based on the system uncertainty; and obtaining the predicted peak-shaving load based on the wind power frequency regulation reserve load and the predicted thermal storage load through a long short-term memory network.

[0007] Furthermore, the multi-source data processing model refers to a dual-mechanism prediction model designed based on uncertainty decision theory and long short-term memory networks. The multi-source data processing model includes a multi-source data acquisition layer, an uncertainty analysis layer, a long-short-term feature extraction layer, an uncertainty weight generation layer, and a result output layer. The uncertainty analysis layer is used to receive multi-source data from thermal power units collected by the multi-source data acquisition layer, quantify the impact of multi-source data fluctuations on uncertain data, and generate uncertain data intervals. The uncertainty weight generation layer is used to generate weights for uncertain data. The uncertain data includes wind power, photovoltaic power, and load power. The long-short-term feature extraction layer is used to generate predicted peak-shaving load, wind power frequency regulation reserve load, and predicted thermal storage load based on the uncertain data and its weights, and outputs the results through the result output layer.

[0008] Furthermore, the step of performing dual-objective optimization based on the predicted load to obtain thermal power unit instructions includes: splitting the base load and peak load based on the predicted load to obtain the base load and thermal storage peak load; constructing an optimization objective function and constraints based on the base load and the thermal storage peak load; and optimizing the optimization objective function based on the constraints through multi-objective particle swarm optimization to obtain the thermal power unit instructions.

[0009] Furthermore, the optimization objective function is:

[0010] minf=α·C total (t)+β·d xh (t)+γ·△P dev (t)+δ·(1-η heat (t));

[0011] P boiler =P base +△P boiler_peak ;

[0012] P heat =P peak -△Pboiler_peak -△P turbine_peak ;

[0013] Where, min represents taking the minimum value; f represents the optimization objective function; α, β, γ, and δ represent the weights of coal consumption cost, equipment lifespan, load deviation, and thermal storage utilization rate, respectively; C total Indicates the total peak-shaving cost; d xh Indicates the rotor life loss rate; △P dev t) represents the load deviation; η heat (t) represents the thermal energy storage utilization rate; P boiler Indicates boiler instruction; P base This indicates the minimum stable combustion load that a thermal power unit needs to maintain; △P boiler_peak Indicates the peak load shaving undertaken by the boiler; P heat Indicates thermal storage command; P peak Indicates the fluctuating load that needs to be borne; △P turbine_peak This indicates the peak-shaving load borne by the steam turbine;

[0014] The constraints include minimum stable combustion constraints for the boiler, overpressure prevention constraints for the steam turbine, safety constraints for the thermal storage system, equipment life protection constraints, and start-up and shutdown constraints.

[0015] P boiler ≥0.18P rated ;

[0016] P turbine ≤1.05P boiler ;

[0017] 0≤P beat ≤P beat，max ;

[0018] T heat，min ≤T heat (t)≤T heat，max ;

[0019] d xh ≤0.03;

[0020] When P TH When (t) = 0, P TH (t+1)~P TH (t+10)=0;

[0021] P TH =P base +P peak +P wind_reg ;

[0022] Among them, P turbine Indicates a steam turbine instruction; T heat，min Indicates the lowest lava flow temperature; Theat (t) represents the real-time molten salt temperature; T heat，max P represents the highest temperature of lava. TH (t) represents the total load of the unit; P TH (t+1)~P TH (t+10)=0 means that if the unit shuts down at a certain time t, the unit must remain shut down for the next 10 hours; P TH (t+1) represents the total unit load at time t+1; P TH (t+10) represents the total load of the unit at time t+10.

[0023] Furthermore, the step of obtaining the main steam parameters and actual air distribution ratio based on the boiler command, the thermal storage command, and the wind power frequency regulation standby load through coordinated combustion control includes: calculating the air distribution ratio of the boiler command, the thermal storage command, and the wind power frequency regulation standby load to obtain the actual air distribution ratio; correcting the initial burner sway angle based on the furnace temperature and system uncertainty to obtain the corrected burner sway angle; calculating the waste heat recovery of the boiler tail flue gas temperature and the thermal storage command to obtain the preheated primary air temperature; calculating the supplementary energy of the boiler's actual ramp rate and maximum ramp rate to obtain the actual wind power frequency regulation power; and applying the actual air distribution ratio, the corrected burner sway angle, and the actual wind power frequency regulation power to boiler combustion to obtain the main steam temperature and main steam pressure, thereby obtaining the main steam parameters.

[0024] Furthermore, the actual air distribution ratio is:

[0025]

[0026] Where γ(t) represents the actual air distribution ratio; γ0 represents the basic air distribution ratio; Δγ(t) represents the air distribution ratio increment; k w Represents the wind power synergy coefficient; k h This represents the thermal storage compensation coefficient.

[0027] A deep peak-shaving device for a thermal power unit utilizing the aforementioned deep peak-shaving method includes a boiler, a steam turbine, a molten salt thermal energy storage system, a wind turbine generator, a photovoltaic power generation array, a furnace temperature monitor, a boiler tail flue gas temperature monitor, and a rotor life monitoring unit. The boiler provides steam through combustion adjustment, undertaking both base load and part of the peak-shaving load. The steam turbine converts the boiler steam thermal energy into mechanical energy to drive the generator. The molten salt thermal energy storage system stores and / or releases thermal energy to supplement and mitigate fluctuations in thermal power generation. The wind turbine generator generates wind power and supplements the boiler's peak-shaving capacity when it is insufficient. The photovoltaic power generation array generates photovoltaic power and participates in the uncertainty quantification of load forecasting. The furnace temperature monitor collects the boiler furnace combustion temperature. The boiler tail flue gas temperature monitor collects the flue gas temperature. The rotor life monitoring unit monitors and calculates the turbine rotor life loss rate.

[0028] A deep peak-shaving device for thermal power units utilizing the aforementioned deep peak-shaving method includes a predicted load determination module, a thermal power unit instruction determination module, a boiler parameter determination module, and a turbine parameter determination module. The predicted load determination module processes multi-source data from the thermal power unit using a multi-source data processing model to obtain the predicted load. The predicted load includes predicted peak-shaving load, wind power frequency regulation reserve load, and predicted thermal storage load. The thermal power unit instruction determination module performs bi-objective optimization based on the predicted load to obtain thermal power unit instructions. The thermal power unit instructions include boiler instructions, turbine instructions, and thermal storage instructions. The boiler parameter determination module, based on the boiler instructions, the thermal storage instructions, and the wind power frequency regulation reserve load, obtains main steam parameters and the actual air distribution ratio through coordinated combustion control. The turbine parameter determination module processes the turbine instructions and main steam pressure to obtain turbine parameters, including the turbine control valve opening and the actual turbine output.

[0029] A storage medium storing a computer program, wherein the computer program is configured to execute the aforementioned deep peak shaving method for thermal power units during runtime.

[0030] The present invention has the following advantages and beneficial effects:

[0031] This invention is applicable to deep peak shaving of coal-fired power units in scenarios of multi-energy complementarity of wind, solar, thermal and energy storage and consumption of new energy sources. Combined with wind power rapid frequency regulation, it shortens boiler ramp-up time, reduces equipment lifespan loss, and achieves deep coupling of wind, solar, thermal and energy storage, smoothing out new energy fluctuations and improving thermal storage utilization.

[0032] This invention employs a multi-source data processing model based on uncertain decision theory and long short-term memory networks, which can effectively integrate multi-dimensional data from the power grid, quantify data uncertainty and generate weights, significantly improve the accuracy of load forecasting, and better adapt to the fluctuations in power output from new energy sources such as wind and solar.

[0033] This invention, through dual-objective optimization, can achieve a multi-dimensional balance between cost, equipment lifespan, and peak-shaving effect while ensuring the safe operation of core equipment in thermal power units (including boilers, steam turbines, and thermal storage systems), thus improving the sustainability of deep peak shaving.

[0034] This invention integrates boiler commands, thermal storage commands, and wind power frequency regulation standby loads. Through methods such as optimized air distribution ratio, burner angle correction, and flue gas waste heat recovery, it not only improves boiler combustion efficiency but also achieves coordinated response of multiple power generation methods, enhancing the flexibility and convenience of thermal power peak shaving. Furthermore, based on the coordinated processing of turbine commands and main steam pressure, it can precisely control the valve opening and actual output while taking into account rotor life loss, ensuring the safe and efficient operation of the turbine under deep peak shaving. Attached Figure Description

[0035] Figure 1 This is an exemplary flowchart of a deep peak shaving method for thermal power units according to the present invention;

[0036] Figure 2 This is an exemplary data flow diagram of a deep peak-shaving device for thermal power units according to the present invention;

[0037] Figure 3 This is an exemplary block diagram of a deep peak-shaving device for thermal power units according to the present invention;

[0038] Reference numerals: 1-boiler, 2-steam turbine, 3-molten salt thermal storage system, 4-wind turbine generator set, 5-photovoltaic power generation array, 6-furnace temperature monitor, 7-boiler tail flue gas temperature monitor, and 8-rotor life monitoring unit. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0040] Figure 1 This is an exemplary flowchart of a deep peak-shaving method for thermal power units according to the present invention. Figure 1 As shown, the deep peak shaving method for thermal power units includes the following:

[0041] A multi-source data processing model is used to process multi-source data from thermal power units to obtain predicted loads. These predicted loads include predicted peak-shaving loads, wind power frequency regulation reserve loads, and predicted thermal storage loads. The multi-source data processing model is a dual-mechanism prediction model designed based on uncertainty decision theory and long short-term memory (LSTM) networks. By processing the input multi-source data, the model predicts the load of thermal power units, achieving high-precision load forecasting and deep peak-shaving decision support. The multi-source data processing model includes a multi-source data acquisition layer, an uncertainty analysis layer, a long-short-term feature extraction layer, an uncertainty weight generation layer, and a result output layer. The uncertainty analysis layer receives multi-source data from thermal power units acquired by the multi-source data acquisition layer and quantifies the impact of multi-source data fluctuations on uncertain data, generating uncertainty data intervals. The uncertainty weight generation layer generates weights for the uncertain data. The long-short-term feature extraction layer generates predicted peak-shaving loads, wind power frequency regulation reserve loads, and predicted thermal storage loads based on the uncertain data and their weights, and outputs the results through the result output layer. Uncertain data includes wind power, photovoltaic power, and load power. Multi-source data refers to various types of data related to thermal power units, which can be acquired through data collection. This includes grid commands, historical wind and solar power output, load demand, historical unit load, ambient temperature, and equipment status. Forecasted load refers to the load over a future period obtained by predicting various loads of thermal power units. This can include predicted peak-shaving load, wind power frequency regulation reserve load, and predicted thermal storage load. Predicted peak-shaving load refers to the predicted peak-shaving power undertaken by the thermal power units. Wind power frequency regulation reserve load refers to the predicted reserve power reserved from wind power resources, used to supplement energy when boiler peak-shaving capacity is insufficient, and to smooth out fluctuations in thermal power output. Predicted thermal storage load refers to the predicted power command value that the thermal storage system needs to release in advance, used to reduce reliance on single-combustion adjustments of the boiler, reduce equipment lifespan loss, and improve thermal storage utilization.

[0042] In some embodiments, obtaining predicted data for thermal power units includes: acquiring multi-source data of the thermal power units; the multi-source data includes grid commands, historical wind and solar power output, load demand, historical unit load, ambient temperature, and equipment status. Grid commands are used to indicate the target power output that the thermal power units need to achieve. Historical wind and solar power output refers to historical wind power output data. Load demand reflects the total electricity demand of the power system. Historical unit load represents the actual power generation load of the unit at various past times. Ambient temperature reflects the temperature of the operating environment of the unit. Equipment status represents the status of the thermal power equipment, which may include steam drum pressure and rotor temperature. Steam drum pressure reflects the pressure status of the steam-water system of the thermal power unit. Rotor temperature is used to monitor the rotor temperature status.

[0043] The multi-source data is processed to obtain uncertainty data intervals and uncertainty data weights. An uncertainty data interval refers to the range of values ​​for the uncertain data. An uncertainty data weight refers to the weight of the uncertain data. In some embodiments, the uncertainty of the uncertain data can be determined based on the historical deviation distribution, meteorological patterns, and equipment fluctuations in historical data. Then, an uncertainty data interval is constructed based on the uncertainty. This uncertainty data interval includes wind power intervals, photovoltaic power intervals, and load power intervals.

[0044]

[0045] Among them, U(P WT α WT ) represents the wind power range model; P WT Indicates wind power output; α WT Indicates the uncertainty of wind power; This represents the predicted wind power output. U(P) represents the actual value of wind power output. PV ,α PV ) represents the photovoltaic power range model; P PV Indicates photovoltaic power; α PV Indicates photovoltaic uncertainty; This represents the predicted photovoltaic power output. U(P) represents the actual value of photovoltaic power. LD ,α LD ) represents the load power range model; P LD Indicates load power; α LD Indicates the uncertainty of the load; This represents the predicted load power. This indicates the actual value of the load power.

[0046] In some embodiments, the weights of each uncertain data point can be calculated using information entropy and mutual information:

[0047]

[0048] Among them, w i The weight represents the i-th type of uncertain data; i represents the first type of uncertain data variable, including wind power data, photovoltaic data, load data, and temperature data; j represents the second type of uncertain data variable; x i x represents the non-normalized value of the i-th class of data; j E represents the non-normalized value of the j-th class of data; i The information entropy of the i-th type of uncertain data reflects the inherent uncertainty of the data; I ij It represents the mutual information between the i-th type of uncertainty data and the j-th type of uncertainty data, reflecting the degree of correlation between the two types of data.

[0049] The system uncertainty is calculated based on the uncertainty of the uncertain data and the weights of the uncertain data. The system uncertainty reflects the overall uncertainty of wind, solar, and environmental loads. In some embodiments, the sum of the products of the uncertainties of multiple uncertain data and their corresponding uncertain data weights can be used as the system uncertainty.

[0050] Based on the system uncertainty, the wind power frequency regulation reserve load and the predicted thermal storage load are determined; the prediction range of the LSTM is constrained by the wind-solar load interval model. Based on the wind power frequency regulation reserve load and the predicted thermal storage load, the predicted peak-shaving load is obtained through a long short-time memory network.

[0051] P pred (t)=σ(W o PSO-Attention(h) t )+b o )-P wind-reg (t)-P heat-pre (t);

[0052]

[0053] Among them, P pred (t) represents the predicted peak load at time t; σ represents the Sigmoid activation function; W o This represents the model output weight matrix of the LSTM model; PSO-Attention(h t ) represents the attention mechanism function based on particle swarm optimization; h t b represents the hidden state vector of the LSTM at time t; o P represents the model output bias term; wind-reg (t) represents the wind power frequency regulation reserve load at time t; P heat-pre (t); k reg α represents the wind power frequency regulation reserve factor; sys Indicates the system uncertainty; denoted as the predicted wind power at time t; min indicates taking the minimum value; λ1 represents the proportionality coefficient of the maximum heat release power of thermal storage, used to limit the maximum output limit of thermal storage pre-command, and can be taken as 0.3; P heat，max λ represents the maximum heat release power of the thermal storage system, which is determined by the physical parameters of the thermal storage equipment and is an inherent property of the molten salt thermal storage system; λ2 represents the predicted peak load ratio coefficient, which is used to limit the upper limit of the load association of the thermal storage pre-command, and can be 0.2.

[0054] Based on the predicted load, a bi-objective optimization function and constraints are constructed. Solving the bi-objective optimization function yields the thermal power unit commands. These commands include boiler commands, turbine commands, and thermal storage commands. The bi-objective optimization function is an optimization model constructed based on the particle swarm optimization algorithm, with the objective of minimizing coal consumption costs and equipment lifespan losses. The constraints are the boundary conditions ensuring the safe and stable operation of the thermal power unit and thermal storage system. The thermal power unit commands are the core control commands guiding the operation of the thermal power unit and its associated thermal storage system, including boiler commands, turbine commands, and thermal storage commands. Boiler commands guide boiler combustion adjustments. Turbine commands guide turbine sliding pressure curve optimization and valve opening adjustments. Thermal storage commands guide molten salt stratified heat release.

[0055] In some embodiments, receiving instructions from a thermal power unit includes:

[0056] Based on the predicted load, the base load and peak load are separated to obtain the base load and thermal storage peak load. The base load refers to the basic load of the thermal power unit, that is, the minimum stable combustion load that the thermal power unit needs to maintain. The thermal storage peak load is also called the fluctuating load, which represents the fluctuating load that needs to be borne by the thermal power unit.

[0057] P base (t)=max(0.18P rated P pred (t)-P wind-reg (t)-P heat，max );

[0058] P peak (t)=P pred (t)-P base (t)-P wind-reg (t);

[0059] Among them, P base (t) represents the minimum stable combustion load that the thermal power unit needs to maintain at time t; max represents taking the maximum value; P rated P represents the rated power of a thermal power unit. pred (t) represents the predicted peak load at time t; P wind-reg (t) represents the wind power frequency regulation reserve load at time t; P heat，max P represents the maximum heat release power of the thermal storage system. peak (t) represents the fluctuating load that needs to be borne at time t.

[0060] Based on the base load and the thermal storage peak-shaving load, an optimization objective function and constraints are constructed. Then, based on the constraints, the optimization objective function is optimized using multi-objective particle swarm optimization to obtain the thermal power unit command. The optimization objective function is:

[0061] minf=α·Ctotal (t)+β·d xh (t)+γ·△P dev (t)+δ·(1-η heat (t));

[0062] P boiler =P base +△P boiler_peak ;

[0063] P beat =P peak -△P boiler_peak -△P turbine_peak ;

[0064] Where, min represents taking the minimum value; f represents the optimization objective function; α, β, γ, and δ represent the weights of coal consumption cost, equipment lifespan, load deviation, and thermal storage utilization rate, respectively. These weights can be obtained by considering the importance of coal consumption cost, equipment lifespan, load deviation, and thermal storage utilization rate, and can be specifically set according to actual needs; C total This represents the total peak-shaving cost, including coal consumption cost and equipment lifespan depreciation cost; d xh Indicates the rotor life loss rate. N xh (t) represents the number of cycles that cause cracking in the turbine rotor, referring to the total operating time or number of cycles from the initial commissioning of the rotor to the appearance of the first macroscopic crack; △P dev (t) represents the load deviation, which is the difference between the predicted peak load and the total load. The total load is the sum of the turbine command, the thermal storage command, and the wind power frequency regulation reserve load. η heat (t) represents the thermal energy storage utilization rate, which is the ratio of the thermal energy storage command to the maximum thermal energy storage power; P boiler Indicates boiler instruction; P base This indicates the minimum stable combustion load that a thermal power unit needs to maintain; △P boiler_peak This represents the peak load borne by the boiler, which is usually very small and can be ignored when calculating thermal storage commands; P heat Indicates thermal storage command; P peak Indicates the fluctuating load that needs to be borne; △P turbine_peak This indicates the peak load shaving undertaken by the steam turbine.

[0065] The constraints include minimum stable combustion constraints for boilers, overpressure protection constraints for steam turbines, safety constraints for thermal storage systems, equipment life protection constraints, and start-up and shutdown constraints.

[0066] Boiler minimum combustion stability constraint:

[0067] P boiler ≥0.18P rated ;

[0068] Steam turbine overpressure protection constraints:

[0069] P turbine ≤1.05P boiler ;

[0070] Safety constraints for thermal storage systems:

[0071] 0≤P heat ≤P heat,max ;

[0072] T heat,min ≤T heat (t)≤T heat,max ;

[0073] Equipment life protection constraints:

[0074] d xh ≤0.03;

[0075] Start-stop constraints:

[0076] When P TH When (t) = 0, P TH (t+1)~P TH (t+10)=0;

[0077] P TH =P base +P peak +P wind_reg ;

[0078] Among them, P turbine Indicates a steam turbine instruction; T heat,min Indicates the lowest lava flow temperature; T heat (t) represents the real-time molten salt temperature; T heat,max This represents the highest temperature of the lava, i.e., the upper limit of the tank's tolerance; P TH (t) represents the total load of the unit; P TH (t+1)~P TH (t+10) = 0 means that if the total load of the unit drops to 0 at a certain time t (i.e., it stops), then for the next 10 hours from time t+1 to time t+10, the unit needs to remain in a stopped state (total load remains 0) to avoid frequent start-ups and shutdowns and reduce its lifespan; P TH (t+1) represents the total unit load at time t+1; P TH (t+10) represents the total load of the unit at time t+10.

[0079] Based on the boiler command, the thermal storage command, and the wind power frequency regulation standby load, the main steam parameters and the actual air distribution ratio are obtained through coordinated combustion control. The main steam parameters refer to the thermodynamic parameters of the steam supplied from the boiler to the turbine in a thermal power unit. The actual air distribution ratio refers to the proportion of primary air to the total air volume during boiler combustion.

[0080] In some embodiments, obtaining the main steam parameters and the actual air distribution ratio includes:

[0081] The actual air distribution ratio is obtained by calculating the air distribution ratio of the boiler command, the thermal storage command, and the wind power frequency regulation standby load.

[0082]

[0083] Where γ(t) represents the actual air distribution ratio; γ0 represents the basic air distribution ratio; Δγ(t) represents the air distribution ratio increment, which is related to the rate of change of the boiler command at the current moment, for example, a certain proportion of the boiler load change rate; k w This represents the wind power coordination coefficient, determined based on actual conditions. A larger wind power reserve allows for higher fine-tuning of the wind distribution ratio; k h This represents the thermal storage compensation coefficient, which is determined based on actual operating conditions. The higher the thermal storage load, the lower the air distribution ratio adjustment.

[0084] The initial burner swing angle is corrected based on furnace temperature and system uncertainty to obtain the corrected burner swing angle. Furnace temperature refers to the temperature of the combustion zone inside the furnace of a thermal power unit boiler. The initial burner swing angle is the reference angle for burner swing angle control. The corrected burner swing angle is the actual operating angle of the burner after dynamic correction based on the initial burner swing angle and considering real-time furnace temperature deviation and system uncertainty. In some embodiments, the corrected swing angle can be determined based on the furnace temperature difference between the target furnace temperature and the current furnace temperature, and then the sum of the corrected swing angle and the initial burner swing angle is taken as the corrected burner swing angle.

[0085] The waste heat recovery calculation is performed on the boiler tail flue gas temperature and the aforementioned heat storage command to obtain the preheated primary air temperature. The preheated primary air temperature refers to the final temperature of the primary air after recovering waste heat from the boiler tail flue gas and correcting it with the heat storage command.

[0086] The actual wind power frequency regulation power is obtained by calculating the supplementary energy from the boiler's actual ramp rate and maximum ramp rate. The actual ramp rate refers to the actual load change rate achieved by the boiler per unit time during load adjustment. The maximum ramp rate is the upper limit of the boiler's designed load adjustment capacity, i.e., the maximum load change rate the boiler can achieve per unit time. The actual wind power frequency regulation power refers to the actual frequency regulation supplementary energy power invested by the wind power.

[0087]

[0088] Among them, P wind_reg，actual (t) represents the actual wind power frequency regulation power; P wind_reg (t) represents the wind power frequency regulation standby load; ΔP boiler，max Indicates the boiler's maximum ramp rate; ΔP boiler(t) represents the current actual ramp-up rate of the boiler.

[0089] By applying the actual air distribution ratio, burner sway angle, and wind power supplementation to boiler combustion, the main steam temperature and main steam pressure are obtained, and thus the main steam parameters are derived.

[0090] The turbine commands and main steam pressure are processed to obtain turbine parameters, including turbine valve opening and actual turbine output. Main steam pressure refers to the steam pressure at which the boiler of the thermal power unit supplies steam to the turbine. Turbine parameters refer to the turbine's operating parameters, which may include turbine valve opening and actual turbine output. The turbine efficiency coefficient is determined by the valve opening and main steam pressure, and the actual turbine output is determined by the turbine efficiency coefficient and turbine commands.

[0091] Figure 2 This is an exemplary data flow diagram of a deep peak-shaving device for thermal power units according to the present invention. Figure 2 As shown, the deep peak-shaving device for thermal power units provided by this invention includes a boiler 1, a steam turbine 2, a molten salt thermal storage system 3, a wind turbine generator set 4, a photovoltaic power generation array 5, a furnace temperature monitor 6, a boiler tail flue gas temperature monitor 7, and a rotor life monitoring unit 8. The boiler provides steam through combustion adjustment, undertaking base load and part of the peak-shaving load, reducing reliance on single combustion; the steam turbine converts the boiler steam thermal energy into mechanical energy to drive the generator, undertaking the core regulation task of peak load; the molten salt thermal storage system stores and / or releases thermal energy to supplement and smooth thermal power fluctuations, reducing life loss caused by frequent boiler adjustments; the wind turbine generator set generates wind power and supplements energy when the boiler's peak-shaving capacity is insufficient; the photovoltaic power generation array generates photovoltaic power, participating in the uncertainty quantification of load forecasting; the furnace temperature monitor collects the boiler furnace combustion temperature for correcting the burner angle; the boiler tail flue gas temperature monitor collects the flue gas temperature for waste heat recovery calculation and optimizing the primary air preheating temperature; and the rotor life monitoring unit monitors and calculates the turbine rotor life loss rate.

[0092] Figure 3 This is an exemplary block diagram of a deep peak-shaving device for thermal power units according to the present invention. Figure 3As shown, the deep peak-shaving equipment for thermal power units provided by this invention includes a predicted load determination module, a thermal power unit instruction determination module, a boiler parameter determination module, and a turbine parameter determination module. The predicted load determination module processes multi-source data from the thermal power unit using a multi-source data processing model to obtain the predicted load. The predicted load includes predicted peak-shaving load, wind power frequency regulation reserve load, and predicted thermal storage load. The thermal power unit instruction determination module performs dual-objective optimization based on the predicted load to obtain thermal power unit instructions. The thermal power unit instructions include boiler instructions, turbine instructions, and thermal storage instructions. The boiler parameter determination module, based on the boiler instructions, the thermal storage instructions, and the wind power frequency regulation reserve load, obtains the main steam parameters and the actual air distribution ratio through coordinated combustion control. The turbine parameter determination module processes the turbine instructions and the main steam pressure to obtain turbine parameters, including the turbine control valve opening and the actual turbine output.

[0093] The present invention also provides a storage medium storing a computer program, wherein the computer program is configured to execute at runtime. Figure 1 A deep peak shaving method for thermal power units as described in any one of the following.

[0094] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A method for deep peak shaving of thermal power units, characterized in that, include: The multi-source data of thermal power units is processed by a multi-source data processing model to obtain the predicted load; the predicted load includes the predicted peak load, the wind power frequency regulation reserve load and the predicted thermal storage load. Based on the predicted load, a dual-objective optimization is performed to obtain thermal power unit commands; the thermal power unit commands include boiler commands, turbine commands, and thermal storage commands. Based on the boiler command, the thermal storage command, and the wind power frequency regulation standby load, the main steam parameters and the actual air distribution ratio are obtained through coordinated combustion control; the main steam parameters include the main steam pressure and the main steam temperature. The turbine commands and the main steam pressure are processed to obtain turbine parameters; the turbine parameters include the turbine control valve opening and the actual turbine output.

2. The deep peak shaving method for thermal power units according to claim 1, characterized in that, The process of processing multi-source data from thermal power units using a multi-source data processing model to obtain predicted load includes: Acquire multi-source data from thermal power units; the multi-source data includes grid commands, historical wind and solar power output, load demand, historical unit load, ambient temperature, and equipment status. The multi-source data is processed to obtain the uncertain data range and the uncertain data weight; The system uncertainty is calculated based on the uncertainty of the uncertainty data and the weight of the uncertainty data. Based on the system uncertainty, the wind power frequency regulation reserve load and the predicted thermal storage load are determined; Based on the wind power frequency regulation reserve load and the predicted thermal storage load, the predicted peak-shaving load is obtained through a long short-time memory network.

3. The deep peak shaving method for thermal power units according to claim 2, characterized in that, The multi-source data processing model refers to a dual-mechanism prediction model designed based on uncertainty decision theory and long short-term memory network; the multi-source data processing model includes a multi-source data acquisition layer, an uncertainty analysis layer, a long short-term feature extraction layer, an uncertainty weight generation layer, and a result output layer; The uncertainty analysis layer is used to receive multi-source data from the thermal power unit collected by the multi-source data acquisition layer, quantify the impact of multi-source data fluctuations on uncertain data, and generate uncertain data ranges. The uncertainty weight generation layer is used to generate weights for uncertain data; the uncertain data includes wind power, photovoltaic power, and load power. The long-term and short-term feature extraction layer is used to generate predicted peak load, wind power frequency regulation reserve load and predicted thermal storage load based on uncertain data and the weights of uncertain data, and outputs the results through the result output layer.

4. The deep peak shaving method for thermal power units according to claim 1, characterized in that, The process of performing bi-objective optimization based on the predicted load to obtain thermal power unit commands includes: Based on the predicted load, the base load and peak load are separated to obtain the base load and thermal storage peak load; Based on the base load and the thermal storage peak-shaving load, an optimization objective function and constraints are constructed, and the optimization objective function is optimized by multi-objective particle swarm optimization based on the constraints to obtain the thermal power unit command.

5. The deep peak shaving method for thermal power units according to claim 4, characterized in that, The optimization objective function is: minf=α·C total (t)+β·d xh (t)+γ·ΔP dev (t)+δ·(1-η heat (t)); P boiler =P base +ΔP boiler_peak ; P heat =P peak -ΔP boiler_peak -ΔP turbine_peak ; Where, min represents taking the minimum value; f represents the optimization objective function; α, β, γ, and δ represent the weights of coal consumption cost, equipment lifespan, load deviation, and thermal storage utilization rate, respectively; C total Indicates the total peak-shaving cost; d xh Indicates the rotor life loss rate; ΔP dev (t) represents the load deviation; η heat (t) represents the thermal energy storage utilization rate; P boiler Indicates boiler instruction; P base This represents the minimum stable combustion load that a thermal power unit needs to maintain; ΔP boiler_peak Indicates the peak load shaving undertaken by the boiler; P heat Indicates thermal storage command; P peak Indicates the fluctuating load that needs to be borne; ΔP turbine_peak This indicates the peak-shaving load borne by the steam turbine; The constraints include minimum stable combustion constraints for the boiler, overpressure prevention constraints for the steam turbine, safety constraints for the thermal storage system, equipment life protection constraints, and start-up and shutdown constraints. P boiler ≥0.18P rated ; P turbine ≤1.05P boiler ; 0≤P heat ≤P heat，max ; T heat,min ≤T heat (t)≤T heat,max ; d xh ≤0.03； When P TH When (t) = 0, P TH (t+1)~P TH (t+10)=0; P TH =P base +P peak +P wind_reg ; Among them, P turbine Indicates a steam turbine instruction; T heat,min Indicates the lowest lava flow temperature; T heat (t) represents the real-time molten salt temperature; T heat,max P represents the highest temperature of lava. TH (t) represents the total load of the unit; P TH (t+1)~P TH (t+10)=0 means that if the unit shuts down at a certain time t, the unit must remain shut down for the next 10 hours; P TH (t+1) represents the total unit load at time t+1; P TH (t+10) represents the total load of the unit at time t+10.

6. The deep peak shaving method for thermal power units according to claim 1, characterized in that, The process of obtaining main steam parameters and actual air distribution ratio based on the boiler command, the thermal storage command, and the wind power frequency regulation standby load through coordinated combustion control includes: The actual air distribution ratio is obtained by calculating the air distribution ratio of the boiler command, the thermal storage command, and the wind power frequency regulation standby load. The initial burner swing angle is corrected based on the furnace temperature and system uncertainty to obtain the corrected burner swing angle. The waste heat recovery calculation is performed on the flue gas temperature at the tail end of the boiler and the heat storage command to obtain the preheated primary air temperature. The actual wind power frequency regulation power is obtained by calculating the supplementary energy for the actual ramp rate and maximum ramp rate of the boiler. The actual air distribution ratio, the corrected burner angle, and the actual wind power frequency regulation power are applied to boiler combustion to obtain the main steam temperature and main steam pressure, and thus the main steam parameters.

7. The deep peak shaving method for thermal power units according to claim 1, characterized in that, The actual air distribution ratio is: Where γ(t) represents the actual air distribution ratio; γ0 represents the basic air distribution ratio; Δγ(t) represents the air distribution ratio increment; k w Represents the wind power coordination coefficient; k h This represents the thermal storage compensation coefficient.

8. A deep peak-shaving device for a thermal power unit utilizing the deep peak-shaving method for thermal power units as described in any one of claims 1-7, characterized in that, This includes boilers, steam turbines, molten salt thermal storage systems, wind turbine generators, photovoltaic power generation arrays, furnace temperature monitors, boiler tail flue gas temperature monitors, and rotor life monitoring units. The boiler provides steam through combustion adjustment, and undertakes the base load and part of the peak load; The steam turbine is used to convert the thermal energy of boiler steam into mechanical energy to drive the generator to generate electricity; The molten salt thermal energy storage system stores and / or releases thermal energy to supplement and mitigate fluctuations in thermal power generation. The wind turbine generator set is used to generate wind power and supplement the boiler's peak-shaving capacity when it is insufficient. The photovoltaic power generation array generates photovoltaic power, which is used to quantify the uncertainty of load forecasting; The furnace temperature monitor is used to collect the combustion temperature in the boiler furnace. The boiler tail flue gas temperature monitor is used to collect flue gas temperature. The rotor life monitoring unit is used to monitor and calculate the life loss rate of the steam turbine rotor.

9. A deep peak-shaving device for a thermal power unit utilizing the deep peak-shaving method for thermal power units as described in any one of claims 1-7, characterized in that, It includes a load prediction determination module, a thermal power unit instruction determination module, a boiler parameter determination module, and a steam turbine parameter determination module; The predicted load determination module is used to process multi-source data of thermal power units through a multi-source data processing model to obtain the predicted load; the predicted load includes predicted peak-shaving load, wind power frequency regulation reserve load and predicted thermal storage load; The thermal power unit instruction determination module is used to perform dual-objective optimization based on the predicted load to obtain thermal power unit instructions; the thermal power unit instructions include boiler instructions, turbine instructions and thermal storage instructions. The boiler parameter determination module is used to obtain the main steam parameters and the actual air distribution ratio based on the boiler command, the thermal storage command and the wind power frequency regulation standby load through coordinated combustion control. The turbine parameter determination module is used to process the turbine commands and main steam pressure to obtain turbine parameters; the turbine parameters include the turbine control valve opening and the actual turbine output.

10. A storage medium, characterized in that, The storage medium stores a computer program, wherein the computer program is configured to execute, when running, a deep peak shaving method for thermal power units as described in any one of claims 1-7.

Images