Coal-fired unit collaborative battery energy storage deep peak regulation control method and device

Abstract

The invention provides a coal-fired unit collaborative battery energy storage deep peak regulation control method and device. The control method comprises the following steps: constructing a digital twinborn model of a coal-fired unit and a battery energy storage system; determining a quantile sequence of a load and frequency instruction of a future time domain by adopting a multi-quantile prediction method; a coal-fired unit equivalent life budget and a battery energy storage equivalent life budget are determined, and double life constraints are set; under the constraint condition including double-life constraint, taking the maximum comprehensive benefit as a target function, based on the quantile sequence of the load and frequency instruction of the future time domain, combining a digital twinborn model, and adopting a probabilistic model predictive control algorithm to determine the optimal output trajectory of the cooperative control of the coal-fired unit and the battery energy storage system of the future time domain; and executing the optimal output trajectory to realize collaborative deep peak regulation of the coal-fired unit and the battery energy storage system. And safe and efficient operation of the system under deep peak regulation is realized through fusion of multi-quantile prediction and probabilistic MPC in combination with double-life constraint.

Description

Technical Field

[0001] This disclosure relates to the field of generator set peak shaving control technology, specifically to a control method and device for deep peak shaving of coal-fired power units in conjunction with battery energy storage. Background Technology

[0002] With the large-scale grid connection of new energy sources such as wind and solar power, the load fluctuations and peak-shaving demands of the power system are becoming increasingly prominent. The loads from new energy sources like wind and solar are characterized by volatility, randomness, and intermittency, requiring generating units to handle deeper and wider peak-shaving loads. However, frequent ramp-up and parameter adjustments can impact unit lifespan. Coal-fired units suffer from slow ramp-up rates and high minimum technical output, making it difficult to independently meet the flexible peak-shaving demands under high-proportion new energy integration. Against this backdrop, introducing rapidly responsive and highly precise battery energy storage systems to form a collaborative peak-shaving system with coal-fired units has become a key technological direction for improving grid regulation capabilities and ensuring safe and stable operation.

[0003] Currently, most existing collaborative peak shaving methods employ deterministic control strategies. However, in deep peak shaving scenarios, the uncertainty of load and frequency commands is significant, leading to an imbalance between peak shaving demand and equipment safety, thus reducing system security and stability. Summary of the Invention

[0004] This disclosure addresses the problems existing in the prior art by providing a control method and device for deep peak shaving of coal-fired power units in conjunction with battery energy storage. It can solve the technical problem of imbalance between peak shaving demand and equipment safety caused by the large uncertainty of load and frequency commands, and improve the safety and stability of the system.

[0005] To achieve the above objectives, the technical solution adopted in this disclosure is as follows: The first aspect of this disclosure provides a control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage, comprising: constructing a digital twin model of the coal-fired power unit and the battery energy storage system; using a multi-quantile prediction method to determine the quantile sequence of load and frequency commands in the future time domain; determining the equivalent lifetime budget of the coal-fired power unit and the equivalent lifetime budget of the battery energy storage system, and setting dual lifetime constraints; under the constraints including dual lifetime constraints, taking maximizing comprehensive benefits as the objective function, and based on the quantile sequence of load and frequency commands in the future time domain, combined with the digital twin model, using a probabilistic model predictive control algorithm to determine the optimal output trajectory for coordinated control of the coal-fired power unit and the battery energy storage system in the future time domain; and executing the optimal output trajectory to achieve coordinated deep peak shaving of the coal-fired power unit and the battery energy storage system.

[0006] In one possible implementation, a multiquantile prediction method is used to determine the quantile sequence of load and frequency commands in the future time domain, including: setting multiple quantiles based on the current rolling time domain coal-fired unit load data and battery energy storage load data, and determining the quantile sequence of load and frequency commands in the future time domain through a pre-trained quantile predictor.

[0007] In one possible implementation, the multiple quantiles include a first quantile, a second quantile, and a third quantile; the quantile sequence of future time-domain load and frequency commands includes the future time-domain load and frequency command sequence corresponding to the first quantile, the future time-domain load and frequency command sequence corresponding to the second quantile, and the future time-domain load and frequency command sequence corresponding to the third quantile; wherein, the first quantile is a low-probability quantile, the second quantile is a medium-probability quantile, and the third quantile is a high-probability quantile.

[0008] In one possible implementation, the dual lifetime constraints include: the equivalent thermal fatigue damage of the coal-fired power unit is no greater than the equivalent lifetime budget of the coal-fired power unit, and the equivalent degradation damage of the battery energy storage is no greater than the equivalent lifetime budget of the battery energy storage; the equivalent thermal fatigue damage of the coal-fired power unit is obtained by rainflow counting on the wall temperature sequence and temperature difference sequence of the key heating surface of the coal-fired power unit; the equivalent degradation damage of the battery energy storage is obtained by mapping degradation damage on the historical sequence of battery rate, battery temperature, and battery state of charge.

[0009] In one possible implementation, the equivalent life budget of the coal-fired power unit and the equivalent life budget of the battery storage are updated on a rolling basis according to a preset period.

[0010] In one possible implementation, the constraints also include opportunity constraints; the opportunity constraints include at least the minimum technical output constraint of the unit, the unit ramp rate constraint, the minimum inlet temperature constraint for selective catalytic reduction, and the exhaust dryness constraint; wherein, the different constraints of the opportunity constraints adopt differentiated confidence levels, and the confidence levels of the opportunity constraints are in the confidence interval of 0.85 to 0.99.

[0011] In one possible implementation, the comprehensive benefits include: electricity revenue, capacity revenue, and accuracy rewards and penalties, minus the costs of power generation coal consumption, auxiliary equipment electricity consumption, energy storage system lifespan depreciation costs, and SCR low-temperature denitrification costs; the objective function is: In the formula, For overall benefits, For electricity revenue, For capacity revenue, Rewards and penalties based on accuracy. For the cost of coal consumption for power generation, To reduce the power consumption cost of auxiliary equipment, The depreciation cost over the lifespan of the energy storage system. Cost of SCR low-temperature denitrification.

[0012] In one possible implementation, the digital twin model includes a coal-fired power unit sub-model and a battery energy storage sub-model; wherein, the coal-fired power unit sub-model includes at least one constraint among the following: minimum technical output of the unit, unit ramp rate, minimum inlet temperature for selective catalytic reduction, turbine back pressure, exhaust steam dryness, upper limit of flue gas temperature, and boiler stable combustion boundary; the battery energy storage sub-model includes the battery rate, battery state of charge, and the impact of battery temperature on internal resistance and available capacity.

[0013] In one possible implementation, the method further includes: during the execution of the optimal output trajectory, online calibration of key parameters of the coal-fired power unit and key parameters of the battery energy storage is performed using a joint state parameter estimation method, and synchronous updates are made to the next rolling time domain; and when communication is abnormal, prediction deviation exceeds the limit, or constraint confidence is lowered, an invariant set backoff strategy is automatically triggered to project the output trajectory to the safety domain, limit the unit ramp rate and battery rate, lock the protection window, and ensure that the coal-fired power unit and battery energy storage system operate within a safe range.

[0014] A second aspect of this disclosure provides a control device for deep peak shaving of coal-fired power units in conjunction with battery energy storage, comprising: a model building unit for building a digital twin model of the coal-fired power unit and the battery energy storage system; a load forecasting unit for determining the quantile sequence of load and frequency commands in the future time domain using a multi-quantile forecasting method; a lifetime budgeting unit for determining the equivalent lifetime budget of the coal-fired power unit and the equivalent lifetime budget of the battery energy storage system, and setting dual lifetime constraints; a probabilistic optimization unit for determining the optimal output trajectory of the coordinated control of the coal-fired power unit and the battery energy storage system in the future time domain, based on the quantile sequence of load and frequency commands in the future time domain, combined with the digital twin model, and using a probabilistic model predictive control algorithm, under the constraints including dual lifetime constraints, with the objective function of maximizing comprehensive benefits; and an online execution unit for executing the optimal output trajectory to achieve deep peak shaving of the coal-fired power unit and the battery energy storage system in conjunction with the quantile sequence of load and frequency commands in the future time domain.

[0015] This disclosure also provides an electronic device, comprising: a memory for storing at least one instruction; and a processor for invoking the instruction stored in the memory to execute the control method for deep peak shaving of coal-fired power unit in conjunction with battery energy storage in any possible embodiment of the first aspect.

[0016] This disclosure also provides a computer-readable storage medium storing at least one executable instruction, which is loaded and executed by a processor to implement the control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage in any possible embodiment of the first aspect.

[0017] This disclosure also provides a computer program product, which includes: computer program code, which, when executed by a computer, causes the computer to perform the control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage in the first aspect and any possible implementation thereof.

[0018] Compared with the prior art, this disclosure has the following beneficial effects: The control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage, as disclosed in this embodiment, provides a simulation environment by constructing a digital twin model, employs a multiquantile prediction method to cover the uncertainties of load and frequency commands, sets dual-lifetime constraints to control the lifespan loss range, and optimizes by maximizing comprehensive benefits. It solves for and executes the optimal output trajectory through a probabilistic model predictive control algorithm, thereby achieving deep peak shaving of coal-fired power units and battery energy storage systems. In deep peak shaving scenarios, it can effectively balance the uncertainties of load and frequency commands with equipment safety, ensuring system economy while reducing equipment maintenance and operation risks, and improving the overall safety and stability of the power system. Attached Figure Description

[0019] Figure 1 This is a schematic flowchart of a control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage, provided in Embodiment 1 of this disclosure. Figure 2 This is a schematic diagram of a dual lifetime budget provided in Embodiment 1 of this disclosure; Figure 3 This is a diagram of the hierarchical rolling optimization architecture of probabilistic MPC provided in Embodiment 1 of this disclosure; Figure 4 This is a schematic flowchart of a control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage, provided in Embodiment 2 of this disclosure. Figure 5 This is a structural block diagram of a control device for deep peak shaving of coal-fired power units in conjunction with battery energy storage, provided in Embodiment 3 of this disclosure. Detailed Implementation

[0020] The present disclosure will now be further described with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present disclosure and should not be construed as limiting the scope of protection of the present disclosure. It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application.

[0021] The acquisition, transmission, storage, use, and processing of data in this disclosed technical solution comply with relevant national laws and regulations. In the embodiments of this disclosure, certain existing industry solutions such as software, components, and models may be mentioned. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this disclosure, and do not imply that the applicant has already used or necessarily used such solutions.

[0022] All terms used in this disclosure have the same meaning as understood by one of ordinary skill in the art to which this disclosure pertains, unless otherwise specifically defined. It should also be understood that terms defined in general dictionaries should be interpreted as having meanings consistent with their meanings in the context of the relevant art, and not as idealized or highly formalized, unless expressly defined herein.

[0023] Figure 1 This is a flowchart illustrating a control method for deep peak shaving in conjunction with battery energy storage for coal-fired power units, as provided in Embodiment 1 of this disclosure. Figure 1 As shown, the control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage may include the following steps S11 to S15.

[0024] Step S11: Construct a digital twin model of the coal-fired power unit and the battery energy storage system.

[0025] It should be noted that a digital twin model refers to a virtual model constructed using digital means that maps and evolves synchronously with a physical entity in real time. It can accurately simulate the operating state, characteristics, and evolutionary patterns of the physical entity. The purpose of constructing this digital twin model is to provide accurate virtual simulation support for subsequent collaborative peak-shaving optimization decisions and control execution. By predicting the operating state of physical equipment in advance through the virtual model, blind adjustments during actual operation can be avoided. Simultaneously, the safety and stability of equipment operation during peak-shaving are improved, providing a reliable model foundation for load forecasting, lifetime estimation, and optimization algorithms.

[0026] In one possible implementation, the digital twin model includes a coal-fired power unit sub-model and a battery energy storage sub-model; wherein, the coal-fired power unit sub-model includes at least one constraint among the following: minimum technical output of the unit, unit ramp rate, minimum inlet temperature of SCR (Selective Catalytic Reduction), turbine back pressure, exhaust steam dryness, upper limit of flue gas temperature, and boiler stable combustion boundary; the battery energy storage sub-model includes battery rate, battery state of charge, and the impact of battery temperature on internal resistance and available capacity.

[0027] Additionally, it should be noted that in this embodiment, the coal-fired power unit sub-model is used to simulate the load response and boundary constraint dynamic characteristics of the unit during deep peak shaving; the energy storage sub-model is used to simulate the output characteristics, state evolution, and performance degradation of the battery during deep peak shaving charge and discharge. When constructing the two sub-models, it is necessary to base them on the actual operating data, equipment parameters, and characteristic curves of the coal-fired power unit and battery energy storage, using a combination of mechanistic modeling and data-driven modeling to ensure that the sub-models accurately reflect the actual operating state of the physical equipment, while also supporting collaborative linkage with the overall digital twin model to achieve real-time synchronization between the virtual and physical systems.

[0028] For example, in one specific implementation, the system constructs a sub-model of the coal-fired power unit based on fundamental data such as the boiler thermodynamic characteristics, turbine operating curves, and selective catalytic reduction system parameters. This sub-model includes boundary conditions such as minimum technical output constraints (e.g., 30% of rated capacity), ramp-up rate constraints (e.g., 2% of rated capacity / minute), and minimum SCR inlet temperature constraints (e.g., 300℃). It can simulate the changes in thermodynamic parameters of the unit under different loads and the triggering of constraint boundaries. Simultaneously, based on the individual battery characteristics, energy storage converter parameters, and battery management system data of the battery energy storage system, a battery energy storage sub-model is constructed to simulate the changes in the state of charge of the battery under different battery rates (e.g., 0.5C, 1C) and battery temperatures (e.g., 25℃, 45℃), as well as the dynamic evolution of internal resistance and available capacity. The two sub-models work together to form a complete digital twin model, providing support for subsequent steps.

[0029] Step S12: Use the multiquantile prediction method to determine the quantile sequence of future time-domain load and frequency commands.

[0030] It should be noted that, in this embodiment of the disclosure, the multiquantile prediction method refers to a probability and statistics-based prediction method. By setting multiple quantile levels, it predicts the range of values ​​of the target variable under different probabilities. Compared with traditional deterministic prediction methods, it can better characterize the uncertainty of the predicted object. The future time domain refers to the time period in which peak-shaving control and optimization decisions are required, and its duration can be set according to the actual peak-shaving needs and the system operating cycle.

[0031] In one possible implementation, a multi-quantile prediction method is used to determine the quantile sequence of load and frequency commands in the future time domain. Specifically, this may include: setting multiple quantiles based on current rolling time domain coal-fired power unit load data and battery storage load data; and using a pre-trained quantile predictor to determine the quantile sequence of load and frequency commands in the future time domain. It should be noted that the quantile predictor refers to a prediction model built and trained based on algorithms such as machine learning and statistical learning, capable of outputting prediction sequences corresponding to different quantiles based on input historical and real-time data. Pre-training refers to training, validating, and optimizing the quantile predictor based on sample data such as historical load data, frequency command data, and meteorological data before the model is put into actual use, ensuring that the prediction accuracy meets actual requirements.

[0032] In one possible implementation, the multiple quantiles include a first quantile, a second quantile, and a third quantile; the quantile sequence of future time-domain load and frequency commands includes the future time-domain load and frequency command sequence corresponding to the first quantile, the future time-domain load and frequency command sequence corresponding to the second quantile, and the future time-domain load and frequency command sequence corresponding to the third quantile. The first quantile, second quantile, and third quantile can be quantiles with different probability levels set according to the forecast demand.

[0033] For example, in one specific implementation, the system collects real-time output data of coal-fired power units and charging / discharging load data of battery energy storage systems within the current rolling time domain (e.g., the past 15 minutes), and combines this with historical load data from the same period and meteorological forecast data, inputting the data into a pre-trained quantile predictor. The first quantile is set as a low-probability quantile, such as 0.1 (corresponding to a predicted value with a 10% probability, representing the lower limit scenario for load and frequency commands), the second quantile as a medium-probability quantile, such as 0.5 (corresponding to a predicted value with a 50% probability, representing the baseline scenario for load and frequency commands), and the third quantile as a high-probability quantile, such as 0.9 (corresponding to a predicted value with a 90% probability, representing the upper limit scenario for load and frequency commands). Based on the input data, the quantile predictor outputs the load sequence and frequency command sequence for the next hour (future time domain) corresponding to the 0.1, 0.5, and 0.9 quantiles, respectively, forming a complete quantile sequence.

[0034] By using multiquantile prediction to characterize the uncertainty of load and frequency commands, multi-scenario prediction data is provided for subsequent probabilistic optimization algorithms, avoiding optimization decision bias caused by single deterministic prediction, and improving the adaptability of peak shaving control to load fluctuations.

[0035] Step S13: Determine the equivalent lifetime budget of the coal-fired power unit and the equivalent lifetime budget of the battery energy storage, and set dual lifetime constraints.

[0036] It should be noted that, in the embodiments disclosed herein, the dual lifespan constraint refers to the lifespan consumption constraint conditions set simultaneously for both coal-fired power units and battery energy storage. This constraint limits the lifespan loss of both during peak shaving, avoids a significant reduction in equipment lifespan due to excessive peak shaving, and achieves a balance between peak shaving demand and equipment lifespan safety.

[0037] In one possible implementation, the dual lifetime constraints include: the equivalent thermal fatigue damage of the coal-fired power unit is no greater than the equivalent lifetime budget of the coal-fired power unit, and the equivalent degradation damage of the battery energy storage is no greater than the equivalent lifetime budget of the battery energy storage. The equivalent thermal fatigue damage of the coal-fired power unit is obtained by rainflow counting of the wall temperature sequence and temperature difference sequence of the key heating surfaces of the coal-fired power unit; the equivalent degradation damage of the battery energy storage is obtained by mapping degradation damage to the historical sequence of battery rate, battery temperature, and battery state of charge. Specifically, the equivalent thermal fatigue damage of the coal-fired power unit refers to the total amount of fatigue damage caused by alternating temperature changes on the key heating surfaces of the coal-fired power unit after equivalent conversion. Key heating surfaces include components that directly contact high-temperature flue gas and working fluid, such as boiler water-cooled walls, superheaters, and reheaters. The periodic changes in their wall temperature and temperature difference lead to thermal fatigue damage. The equivalent degradation damage of the battery energy storage refers to the total amount of performance degradation caused by changes in battery rate, temperature, and state of charge during the charging and discharging process of the battery energy storage system after equivalent conversion, directly reflecting the degree of lifespan consumption of the energy storage system.

[0038] For example, in one specific embodiment, rainflow counting is performed on the wall temperature and temperature difference of the key heating surfaces of the coal-fired power unit to obtain the equivalent thermal fatigue damage of the coal-fired power unit. And set an equivalent life budget for coal-fired power units. .

[0039] ,in, For the temperature difference of the heated surface, This represents the number of iterations.

[0040] Set life budget constraints for coal-fired power units: .

[0041] Degradation damage is mapped onto the battery rate, temperature, and state of charge (SOC) time-series data to obtain the equivalent degradation damage of the battery energy storage system. And set the battery storage equivalent lifetime budget. .

[0042] ,in, Battery rate, The battery temperature is denoted by , and the state of charge (SOC) is the ratio of the battery's stored energy to its rated capacity.

[0043] Set battery energy storage lifespan budget constraints: .

[0044] Specifically, such as Figure 2 As shown, based on the predicted load and parameter commands, the key heating surface wall temperature data of the coal-fired unit, as well as the battery energy storage rate, SOC, and temperature data, were simulated. Rainflow counting was performed on the key heating surface wall temperature data to obtain the equivalent thermal fatigue damage of the coal-fired unit. Degradation damage mapping was performed on battery energy storage rate, SOC, and temperature data to obtain the equivalent battery degradation damage of the battery energy storage system. They are used together in the probabilistic MPC optimization module.

[0045] By setting life safety boundaries, we can ensure the long-term stable operation of coal-fired power units and battery energy storage systems while pursuing peak-shaving benefits, thereby reducing equipment maintenance costs and replacement risks.

[0046] Additionally, it should be noted that in one possible implementation, the equivalent life budget of the coal-fired power unit and the equivalent life budget of the battery storage are updated on a rolling basis according to a preset period. For example, in one specific implementation, and Settlement is based on daily or weekly rolling settlements; when the budget is tight (e.g., actual damage reaches 80% of the budget), the deep peak shaving amplitude is automatically reduced and the price threshold is raised to reduce lifespan loss; when high-frequency switching of the steam-water separator is detected, the load reduction slope is automatically tightened to avoid drastic changes in the temperature difference of the heated surface that could lead to increased damage.

[0047] Step S14: Under the constraints including dual lifetime constraints, with the objective function of maximizing comprehensive benefits, based on the quantile sequence of load and frequency commands in the future time domain, combined with the digital twin model, a probabilistic model predictive control algorithm is used to determine the optimal output trajectory of the coordinated control of the coal-fired unit and the battery energy storage system in the future time domain.

[0048] It should be noted that in this embodiment, constraints refer to the boundary conditions and limiting rules that must be followed during the optimization decision-making process. These are used to ensure equipment operation safety, environmental compliance, and performance stability. In addition to dual lifetime constraints, they may also include opportunity constraints and operating parameter constraints. The objective function refers to the mathematical expression used to measure the degree of achievement of the optimization objective; here, maximizing the overall benefit is the core objective. The quantile sequence of future load and frequency commands in the future time domain is used to provide data on uncertain scenarios of future load and frequency. The digital twin model is used to simulate the equipment's operating status, constraint satisfaction, and lifetime loss under different output schemes, providing simulation support for optimization decisions. Probabilistic MPC (Model Predictive Control) refers to an optimization algorithm based on probability statistics and model predictive control. It can combine uncertain prediction data and determine the optimal control strategy through rolling optimization under the premise of satisfying constraints. Compared with traditional deterministic model predictive control algorithms, it is more adaptable to the uncertainties in the peak shaving process. The optimal output trajectory refers to the optimal curves of the output of the coal-fired power unit and the battery energy storage over time in the future time domain. It clarifies the output values ​​of the two at different times, providing precise control commands for subsequent execution steps. The purpose of this step is to maximize the overall peak-shaving benefits while taking into account equipment lifespan safety and operational constraints through probabilistic optimization algorithms. At the same time, it generates an optimal output scheme that can be directly used for execution, connecting the prediction, budgeting, and execution stages to form a complete optimization closed loop.

[0049] In one possible implementation, the constraints also include opportunity constraints; opportunity constraints refer to probabilistic constraints that ensure the probability of the constraint being met is not less than a set threshold at a certain confidence level, and are applicable to optimization problems with uncertainty; opportunity constraints include at least the minimum technical output constraint of the unit, the unit ramp rate constraint, the minimum inlet temperature constraint of the SCR, and the exhaust dryness constraint; wherein, different constraints of opportunity constraints adopt differentiated confidence levels, and the confidence levels of opportunity constraints are located in the confidence interval of 0.85 to 0.99.

[0050] For example, in one specific implementation, different confidence levels are applied to different constraints based on factors such as unit ramp rate, equipment life equivalent damage, and unit economy: the confidence level for the minimum technical output constraint is 0.9, the confidence level for the unit ramp rate constraint is 0.9, the confidence level for the minimum inlet temperature constraint of SCR is 0.95, and the confidence level for the exhaust dryness constraint is 0.9.

[0051] In one possible implementation, the comprehensive benefits include: electricity revenue, capacity revenue, and accuracy rewards and penalties, minus the costs of power generation coal consumption, auxiliary equipment electricity consumption, energy storage system lifespan depreciation costs, and SCR low-temperature denitrification costs; the objective function is: In the formula, For overall benefits, For electricity revenue, For capacity revenue, Rewards and penalties based on accuracy. For the cost of coal consumption for power generation, To reduce the power consumption cost of auxiliary equipment, The depreciation cost over the lifespan of the energy storage system. The objective function represents the cost of SCR low-temperature denitrification. It comprehensively covers the benefits and costs of the peak-shaving process, accurately measuring the overall benefit level and providing precise target guidance for optimization decisions.

[0052] For example, in one specific implementation, under the model predictive control framework, the goal is to maximize the comprehensive benefits of "economy, safety, and flexibility": the goal includes energy, capacity and mileage benefits and accuracy rewards and penalties, unit heat rate and auxiliary power consumption, ammonia or urea consumption and ammonia escape penalties, SCR low temperature risk costs, energy storage charging and discharging price difference and degradation costs, and standby opportunity costs; and opportunity constraints (such as minimum technical output, ramp rate, SCR inlet temperature, final stage humidity, upper limit of flue gas temperature and acid dew point, maximum back pressure, boiler stable combustion boundary) and dual lifetime budget constraints are applied to solve for the unit output and energy storage charging and discharging trajectory in the future time domain. Specifically, the system inputs the load and frequency command sequences corresponding to the 0.1, 0.5, and 0.9 quantiles, calls the constructed digital twin model, sets four types of opportunity constraints and dual lifetime constraints, and sets differentiated confidence levels based on the degree of constraint impact. Simultaneously, based on data such as electricity market prices, coal prices, and reducing agent prices, it calculates the various benefits and costs in the objective function. A probabilistic model predictive control algorithm is used for rolling optimization, updating the optimization window every 15 minutes to obtain the optimal output trajectory of the coal-fired unit and battery energy storage system for the next hour. For example, during peak load periods, the coal-fired unit maintains a high output while the battery energy storage system discharges to replenish energy; during off-peak load periods, the coal-fired unit operates near its minimum technical output while the battery energy storage system charges and stores energy, ensuring that all constraints are met and maximizing overall benefits.

[0053] Additionally, it should be noted that the probabilistic MPC optimization method in this embodiment adopts a hierarchical rolling optimization architecture, such as... Figure 3 As shown, this hierarchical rolling optimization architecture includes a perception layer, a cloud optimizer, and an execution layer. The perception layer corresponds to the data input and state estimation stage of probabilistic MPC; the cloud optimizer corresponds to the rolling optimization solution stage of probabilistic MPC; and the execution layer corresponds to the control command issuance and feedback correction stage of probabilistic MPC.

[0054] Step S15: Execute the optimal output trajectory to achieve coordinated deep peak shaving between the coal-fired power unit and the battery energy storage system.

[0055] Specifically, control commands for the coal-fired power units and battery energy storage systems are generated based on the optimal output trajectory and then issued for execution. This step translates optimization decisions into actual control actions, achieving a balance between peak-shaving demand, equipment safety, and overall benefits through precise execution of the optimal output trajectory, thereby ensuring the stable operation of the power system.

[0056] In one possible implementation, the online calibration step in step S16 and the safety rollback step in step S17 may also be included to address uncertainties and abnormal situations during the execution process.

[0057] Step S16: During the execution of the optimal output trajectory, the key parameters of the coal-fired power unit and the key parameters of the battery energy storage are calibrated online through the joint state parameter estimation method and updated synchronously to the next rolling time domain.

[0058] It should be noted that the joint state parameter estimation method refers to an algorithm that simultaneously estimates and calibrates the system state and parameters. It can combine real-time operating data to correct deviations between model parameters and state variables. Key parameters for coal-fired power units may include the boiler's effective heat transfer coefficient, reheater hysteresis coefficient, and valve characteristic parameters, while key parameters for battery energy storage may include equivalent internal resistance, available capacity, and SOC estimation parameters. The next rolling time domain refers to the subsequent optimization and execution cycle. Updating the calibrated parameters to the next cycle improves the accuracy of subsequent optimization decisions and ensures real-time matching between the digital twin model and the physical equipment state.

[0059] For example, the optimal output trajectory is executed, and the boiler's effective heat transfer coefficient, reheater hysteresis, valve characteristics, and battery equivalent parameters are calibrated online using joint state parameter estimation methods such as UKF (Unscented Kalman Filter) or EnKF (Ensemble Kalman Filter). The calibrated parameters are then synchronized to the next 15-minute rolling time domain for subsequent optimization decisions.

[0060] Step S17: When communication is abnormal, prediction deviation exceeds the limit, or constraint confidence falls below the limit, the invariant set backoff strategy is automatically triggered to project the output trajectory to the safety domain, limit the unit ramp rate and battery rate, lock the protection window, and ensure that the coal-fired unit and battery energy storage system operate within the safe range.

[0061] It should be noted that the invariant set backoff strategy refers to a pre-set safety control strategy. When an abnormal situation occurs, the system state and output trajectory are adjusted to within the preset safety set to prevent the risk from escalating. The safety domain refers to the parameter range and output range within which the coal-fired power unit and battery energy storage system can operate safely and stably. The protection window refers to the operating boundaries and control ranges that are set to prevent exceeding the limits to ensure equipment safety. The invariant set backoff strategy is automatically triggered by the edge controller.

[0062] In one possible implementation, a terminal convergence band is set: the main and reheat steam temperatures and pressures, reheater metal temperature, and condenser back pressure at the end of the window fall within a preset range; when the load crosses the critical range, a supercritical or subcritical mode switching constraint is activated. Here, the terminal convergence band refers to the parameter convergence range set to ensure the stability of the operating parameters of the equipment at the end of the optimization window, avoiding excessive fluctuations in terminal parameters that could affect the operation of the next cycle.

[0063] Figure 4 This is a flowchart illustrating a control method for deep peak shaving in conjunction with battery energy storage in coal-fired power units, as provided in Embodiment 2 of this disclosure. Figure 4 As shown, the specific process includes the following: Peak load forecasting: Based on historical operating data and real-time power grid dispatching demand, peak load forecasting for the future time domain is completed, providing basic data support for subsequent collaborative optimization.

[0064] Collaborative optimization of load allocation: Based on the peak load forecast results, and combined with the operating characteristics and constraints of coal-fired power units and battery energy storage systems, the peak load in the future time domain is collaboratively optimized and allocated to preliminarily determine the load allocation ratio between the two.

[0065] Layered control strategy: Based on the load allocation results, a layered control strategy is generated, which outputs load and parameter instructions for coal-fired units and load instructions for battery energy storage systems. At the same time, it triggers real-time calculation of the life budget of coal-fired units and the life budget of battery energy storage systems, and clarifies the upper limit of life loss of the two in the current peak shaving cycle.

[0066] Probabilistic MPC optimization solution: The load command and lifetime budget are input into the probabilistic model predictive control (MPC) optimization module. This module takes the maximization of the comprehensive benefits of "economy, safety and flexibility" as the objective function, and solves the optimal output trajectory of the coal-fired unit and battery energy storage coordinated control in the future time domain under the constraints of dual lifetime constraints and opportunity constraints.

[0067] Online calibration and multi-dimensional judgment: The optimal output trajectory is calibrated online, and model parameters are corrected based on real-time operating data. The trajectory is then judged from three dimensions: economy, safety, and flexibility. If the judgment does not meet the requirements, the process returns to the hierarchical control strategy stage to regenerate control instructions; if the judgment meets the requirements, the process proceeds to the execution stage.

[0068] Execute the optimal output trajectory: Send control commands to the coal-fired power unit and the battery energy storage system to execute the load increase / decrease of the unit and the charging / discharging operation of the energy storage system, so that the actual output of the two is consistent with the optimal output trajectory.

[0069] Safety rollback mechanism trigger: During execution, if abnormal situations such as communication failure, prediction deviation exceeding limits, or constraint confidence falling below limits occur, the edge controller will automatically trigger the invariant set rollback strategy, project the power trajectory to the safety domain, limit the unit ramp rate and battery rate, lock the protection window, and ensure that the coal-fired unit and battery energy storage system operate within the safe range.

[0070] For ease of understanding, the following detailed description is provided in conjunction with specific embodiments.

[0071] This embodiment is based on a 660 MW supercritical coal-fired unit and a 60 MW / 120 MWh lithium battery energy storage system. It adopts probabilistic model predictive control (MPC) and dual lifetime budget constraints to achieve rapid load increase and deep peak shaving operation, taking into account economy, equipment life and system safety.

[0072] The coal-fired power unit has a rated capacity of 660 MW, a minimum load of 20%, and a maximum load increase rate of 5%P / min. The battery energy storage system has a power of 60 MW, a capacity of 120 MWh, a rate limit of 1.5C, a state of charge (SOC) range of 15-90%, and a temperature window of 10-40℃. The degradation model uses a combination of cyclic (rainfall counting) and calendar (Arrhenius) superposition. The system employs rolling optimization control to achieve rapid peak shaving.

[0073] Unit load forecast: Multiquantile prediction, such as quantiles Only four types of constraints directly related to deep peak shaving can be retained: minimum technical output of the unit (to avoid boiler shutdown risk), unit ramp rate (considering equipment thermal inertia and safety limits), minimum SCR inlet temperature (to ensure denitrification efficiency), low-pressure cylinder flow rate and exhaust dryness (for turbine flow safety).

[0074] Set the step size to 5 min and the time domain H=12; the chance confidence level is: minimum SCR inlet temperature is 0.95, exhaust dry constraint is 0.9, unit ramp rate constraint and minimum technical output constraint are both 0.9.

[0075] We performed governance (missing / drift detection) on historical AGC / load and unit process quantity data for the past two years and trained a quantile predictor; we determined the minimum inlet temperature of SCR, terminal convergence tube and invariant set boundary through offline simulation; we accessed 1-second data on-site and deployed fast MPC on the edge side; and we conducted a 30-day A / B test run (compared with the "no lifetime budget" baseline).

[0076] Data Acquisition and Interfaces: Sampling and execution: Edge control 1~2 s; Cloud optimization 5 min.

[0077] Key measurement parameters: main / reheat steam temperature, pressure, and flow rate; feedwater flow rate; extraction steam rate; furnace outlet temperature; primary / secondary air volume and ratio; oxygen content; SCR inlet and outlet temperatures; flue gas temperature; condenser back pressure; flame detection / combustion stabilization signals; valve position and valve speed; unit load; AGC (Automatic Generation Control) / DEH (Digital Electro-Hydraulic Control System) commands; energy storage side SOC (State of Charge), SOH (State of Health), temperature, current, voltage, and PCS (Power Conversion System) status.

[0078] Execution interface: AGC / DEH load setting; fuel and air volume ratio; injection reduction and bypass valves; feedwater and steam extraction; energy storage active power commands; reheater / air preheater temperature boosting bypass when necessary.

[0079] Models and constraints: 1) Lightweight digital twin model: Boiler-reheater-turbine coupled model (online calibration: effective heat transfer coefficient, reheater hysteresis, valve characteristics); Battery equivalent circuit and thermal model (the influence of temperature, rate, and SOC on internal resistance and available capacity).

[0080] Parameter Adjustment: 1) Data Governance and Baseline Inventory: Extract second-level and minute-level data from the past two years to construct two baselines: "No Lifetime Budget" and "Conventional AGC + Battery Allocation"; 2) Opportunity Constraint Threshold Tuning: Set the confidence levels and buffers for the minimum SCR inlet temperature, lower limit of exhaust steam dryness, and upper limit of back pressure based on the statistical quantile of 30-day historical residuals; 3) Lifetime Budget Calibration: Calibrate the equivalent lifetime budget of coal-fired units using offline simulations of typical weeks (flat, peak, and valley). Battery energy storage equivalent lifespan budget .

[0081] Probabilistic MPC optimization: Under the constraints (minimum unit output, ramp-up / pull-down rate, energy storage SOC boundary, system power balance), solve for the optimal power allocation: ,in, For total load, For the power generation of steam turbines, This provides the output power for the battery energy storage system.

[0082] Tiered regulation strategy: Upper layer: Intraday scheduling layer (1-hour level) determines the unit output curve and energy storage power plan; Middle layer: Real-time optimization layer (minute level) corrects the allocation based on forecast deviation and actual load; Lower layer: Execution control layer (second level) controls power output by coordinating boiler fuel valves, turbine valve positions and energy storage converters.

[0083] The control objective function is: In the formula, For overall benefits, For electricity revenue, For capacity revenue, Rewards and penalties based on accuracy. For the cost of coal consumption for power generation, To reduce the power consumption cost of auxiliary equipment, The depreciation cost over the lifespan of the energy storage system. Cost of SCR low-temperature denitrification.

[0084] Rapid load increase and economic synergy strategy: Based on electricity price forecast signals and load forecast models, the system achieves economic synergy between electromechanical systems and energy storage in peak shaving and frequency regulation: 1) During periods of high electricity prices or rapid load increases: coal-fired units rapidly increase load, and energy storage discharges to support system response; 2) During periods of low electricity prices or low load: energy storage charges at low power to reduce the inefficient combustion time of the units; 3) When SOC > 80% or the units are under high load: the system maintains balance, and energy storage is in standby or low-power cycling; 4) When SOC < 30%, priority is given to charging to avoid degradation caused by deep discharge.

[0085] When the grid load increases, the energy storage system prioritizes discharging 60 MW to compensate for the unit's response lag. The unit's load increase rate is a maximum of 5% of the rated load / min to prevent excessive thermal stress on the heating surfaces. The energy storage provides power support within 6 minutes, and then gradually switches to recharging to maintain stable boiler steam temperature and SCR temperature.

[0086] During the deep peak shaving phase, the unit operates at 20% load, and the energy storage system charges according to the electricity price forecast to ensure that the coal consumption rate does not exceed 1.7 times the design value (increase ≤ 70%). Stable combustion is maintained through joint control algorithms, reducing inefficient operating periods.

[0087] When the load crosses the critical range, mode switching constraints are activated: the steam-water separator opening and closing hysteresis is 2~3% of the rated load; the reheater cold end corrosion protection temperature window is 280~300℃; stricter constraints are applied to the load slope and valve speed during the load reduction phase. Confidence level configuration: SCR inlet temperature ≥ lower limit, confidence level 0.97, others 0.9 (e.g., maximum back pressure ≤ upper limit, exhaust steam dryness ≥ lower limit, low-pressure cylinder flow rate ≥ lower limit, boiler stable combustion ≥ lower limit); degradation weight: NMC is temperature sensitive, increase the penalty for high temperature and high rate (above 35℃ or above 1.2℃).

[0088] Beneficial effects: This control scheme achieves ≥2 times the ramp-up capability and a deeper load floor under probabilistic guarantees; hard constraints on lifespan budgets balance unit thermal fatigue and battery degradation, avoiding "trading performance for excessive energy storage lifespan"; and constant cascade rollback ensures rapid and safe landing in case of anomalies. This control scheme enables the unit to achieve a 5% P / min load increase and a 20% P deep adjustment while maintaining an SCR inlet temperature ≥310℃. Thermal fatigue equivalent damage is reduced by 20-35% compared to the baseline. System coal consumption is reduced by approximately 3-8% compared to conventional AGC, energy storage degradation rate is reduced by 30%, and annual economic benefits are increased by approximately 5-10%.

[0089] Figure 5 This is a structural block diagram of a control device for deep peak shaving of coal-fired power units in conjunction with battery energy storage, provided in Embodiment 3 of this disclosure. Figure 5 As shown, the control device 100 for deep peak shaving of coal-fired power unit in conjunction with battery energy storage may include a model building unit 110, a load prediction unit 120, a lifetime budget unit 130, a probabilistic optimization unit 140, and an online execution unit 150.

[0090] The system includes: a model building unit 110 for constructing a digital twin model of the coal-fired power unit and the battery energy storage system; a load forecasting unit 120 for determining the quantile sequence of load and frequency commands in the future time domain using a multiquantile forecasting method; a lifetime budgeting unit 130 for determining the equivalent lifetime budget of the coal-fired power unit and the equivalent lifetime budget of the battery energy storage system, and setting dual lifetime constraints; a probabilistic optimization unit 140 for determining the optimal output trajectory of the coordinated control of the coal-fired power unit and the battery energy storage system in the future time domain, based on the quantile sequence of load and frequency commands in the future time domain, combined with the digital twin model, using a probabilistic model predictive control algorithm, under constraints including dual lifetime constraints, with the objective function of maximizing comprehensive benefits; and an online execution unit 150 for executing the optimal output trajectory to achieve coordinated deep peak shaving of the coal-fired power unit and the battery energy storage system.

[0091] For specific details and benefits of the control method and apparatus for deep peak shaving of coal-fired power units in conjunction with battery energy storage provided in the embodiments of this disclosure, please refer to the above description of the control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage, which will not be repeated here.

[0092] This disclosure also provides an electronic device, comprising: a memory for storing at least one instruction; and a processor for calling the instruction stored in the memory to execute the control method for deep peak shaving of coal-fired power unit in conjunction with battery energy storage in any of the above embodiments.

[0093] This disclosure also provides a computer-readable storage medium storing at least one executable instruction, which is loaded and executed by a processor to implement the control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage in any of the above embodiments.

[0094] This disclosure also provides a computer program product, which includes computer program code. When the computer program code is run by a computer, it causes the computer to execute the control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage in any of the above embodiments.

[0095] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, systems, or computer program products. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0096] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0097] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0098] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0099] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0100] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0101] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0102] It should be noted that the terms "first," "second," and similar terms used in this disclosure do not indicate any order, quantity, or importance, but are merely used to distinguish different parts. Terms such as "including" or "contains" mean that the element preceding the word covers the element listed after the word, and do not exclude the possibility of covering other elements as well.

[0103] Although operations are described in a specific order in the accompanying drawings in this disclosure, it should not be construed as requiring these operations to be performed in the specific order or serial order shown, or requiring all of the shown operations to obtain the desired result. In certain environments, multitasking and parallel processing may be advantageous.

[0104] Finally, it should be noted that the above content is only used to illustrate the technical solution of this disclosure, and is not intended to limit the scope of protection of this disclosure. Simple modifications or equivalent substitutions made by those skilled in the art to the technical solution of this disclosure do not depart from the substance and scope of the technical solution of this disclosure.

Claims

1. A control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage, characterized in that, include: Construct a digital twin model of a coal-fired power unit and a battery energy storage system; The quantile sequence of future time-domain load and frequency commands is determined using the multiquantile forecasting method. Determine the equivalent lifetime budget for coal-fired power units and the equivalent lifetime budget for battery energy storage, and set dual lifetime constraints; Under the constraints including the dual lifetime constraints, with the objective function of maximizing comprehensive benefits, based on the quantile sequence of the load and frequency commands in the future time domain, combined with the digital twin model, a probabilistic model predictive control algorithm is used to determine the optimal output trajectory of the coordinated control of the coal-fired power unit and the battery energy storage system in the future time domain. By executing the optimal output trajectory, the coal-fired power unit and the battery energy storage system can achieve coordinated deep peak shaving.

2. The control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage according to claim 1, characterized in that, The method of determining the quantile sequence of future time-domain load and frequency commands using the multiquantile forecasting method includes: Based on the current rolling time-domain load data of coal-fired power units and battery energy storage, multiple quantiles are set, and a pre-trained quantile predictor is used to determine the quantile sequence of load and frequency commands in the future time domain.

3. The control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage according to claim 2, characterized in that, The plurality of quantiles includes a first quantile, a second quantile, and a third quantile; the quantile sequence of future time-domain load and frequency commands includes the future time-domain load and frequency command sequence corresponding to the first quantile, the future time-domain load and frequency command sequence corresponding to the second quantile, and the future time-domain load and frequency command sequence corresponding to the third quantile; wherein, the first quantile is a low-probability quantile, the second quantile is a medium-probability quantile, and the third quantile is a high-probability quantile.

4. The control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage according to claim 1, characterized in that, The dual lifetime constraints include: the equivalent thermal fatigue damage of the coal-fired power unit is no greater than the equivalent lifetime budget of the coal-fired power unit, and the equivalent degradation damage of the battery energy storage is no greater than the equivalent lifetime budget of the battery energy storage; the equivalent thermal fatigue damage of the coal-fired power unit is obtained by rainflow counting on the wall temperature sequence and temperature difference sequence of the key heating surface of the coal-fired power unit; the equivalent degradation damage of the battery energy storage is obtained by mapping degradation damage on the historical sequence of battery rate, battery temperature, and battery state of charge.

5. The control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage according to any one of claims 1, characterized in that, The equivalent life budget of the coal-fired power unit and the equivalent life budget of the battery energy storage are updated on a rolling basis according to a preset cycle.

6. The control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage according to claim 1, characterized in that, The constraints also include opportunity constraints; the opportunity constraints include at least the minimum technical output constraint of the unit, the unit ramp rate constraint, the minimum inlet temperature constraint for selective catalytic reduction, and the exhaust steam dryness constraint. The different constraints of the opportunity constraint adopt differentiated confidence levels, and the confidence level of the opportunity constraint is located in the confidence interval of 0.85 to 0.

99.

7. The control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage according to claim 1, characterized in that, The comprehensive benefits include: electricity revenue, capacity revenue, and accuracy rewards and penalties, minus the costs of power generation coal consumption, auxiliary equipment electricity consumption, energy storage system life depreciation costs, and SCR low-temperature denitrification costs; the objective function is: ， In the formula, For overall benefits, For electricity revenue, For capacity revenue, Rewards and penalties based on accuracy. For the cost of coal consumption for power generation, To reduce the power consumption cost of auxiliary equipment, The depreciation cost over the lifespan of the energy storage system. Cost of SCR low-temperature denitrification.

8. The control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage according to claim 1, characterized in that, The digital twin model includes a coal-fired power unit sub-model and a battery energy storage sub-model; The coal-fired unit sub-model includes at least one constraint among the following: minimum technical output of the unit, unit ramp rate, minimum inlet temperature of selective catalytic reduction, turbine back pressure, exhaust steam dryness, upper limit of flue gas temperature, and boiler stable combustion boundary; the battery energy storage sub-model includes the effects of battery rate, battery state of charge, and battery temperature on internal resistance and available capacity.

9. The control method for deep peak shaving of coal-fired power units in conjunction with battery energy storage according to claim 1, characterized in that, Also includes: During the execution of the optimal output trajectory, the key parameters of the coal-fired power unit and the key parameters of the battery energy storage are calibrated online through the joint state parameter estimation method and updated synchronously to the next rolling time domain. as well as When communication is abnormal, prediction deviation exceeds the limit, or constraint confidence falls below the limit, the invariant set backoff strategy is automatically triggered to project the output trajectory to the safety domain, limit the unit ramp rate and battery rate, lock the protection window, and ensure that the coal-fired unit and battery energy storage system operate within the safe range.

10. A control device for deep peak shaving of coal-fired power units in conjunction with battery energy storage, characterized in that, include: The model building unit is used to build digital twin models of coal-fired power units and battery energy storage systems. The load forecasting unit is used to determine the quantile sequence of future time-domain load and frequency commands using the multiquantile forecasting method. The lifetime budget unit is used to determine the equivalent lifetime budget of the coal-fired power unit and the equivalent lifetime budget of the battery storage, and to set dual lifetime constraints. A probabilistic optimization unit is used to determine the optimal output trajectory of the coordinated control of the coal-fired power unit and the battery energy storage system in the future time domain, with the objective function of maximizing comprehensive benefits, based on the quantile sequence of the load and frequency commands in the future time domain, combined with the digital twin model, and using a probabilistic model predictive control algorithm, under the constraints including the dual lifetime constraints. An online execution unit is used to execute the optimal output trajectory to achieve coordinated deep peak shaving between the coal-fired power unit and the battery energy storage system.

Images