Lithium iron phosphate energy storage scheduling method for industrial park

Abstract

The invention relates to the technical field of energy storage and energy management, in particular to a lithium iron phosphate energy storage scheduling method for an industrial park. The equivalent life loss cost is introduced into a scheduling target and is determined by the discharge depth, the charge-discharge rate, the temperature and the health state parameters of the energy storage monomers, so that the long-term total cost of the park is taken as an optimization target. A multi-time-scale control framework is adopted, an upper-layer rolling robust optimizer generates an energy storage power predicted value and adds risk constraints, and a lower-layer real-time scheduler corrects power according to deviations and emergency states and distributes the power in a layered mode according to a load priority list. The parallel energy storage monomers distribute power in a weighted mode according to the health state, the charge state, the temperature and the internal resistance, the digital twin module predicts the attenuation rate and a maintenance window, and the health state is fed back to be optimized and dispatched. In the switching process, output is smoothly adjusted according to an energy retention boundary and a stepped discharge strategy, so that the comprehensive overhead of electric energy cost and service life cost is reduced, and the service life of the lithium iron phosphate battery is prolonged.

Description

Technical Field

[0001] This invention relates to the field of energy storage and energy management technology, and to a lithium iron phosphate energy storage scheduling method for industrial parks. Background Technology

[0002] This invention relates to the field of energy management and dispatch, particularly to dispatch methods and optimization technologies related to energy storage systems in industrial parks. Energy storage technology, especially lithium iron phosphate batteries, has become a key technology for addressing the instability of renewable energy and improving grid flexibility in industrial parks. With the increasing demand for energy efficiency and sustainable development in industrial parks, the dispatch optimization of energy storage systems has become a core element in achieving efficient energy management.

[0003] Industrial parks typically face unstable power supply demands, including cyclical load fluctuations, increased peak-hour demand, and uncertainties in renewable energy generation. Against this backdrop, energy storage systems, as a flexible regulation tool, can effectively regulate grid load, alleviate energy supply pressure, and play a crucial role in the electricity market. Lithium iron phosphate batteries are widely used in energy storage due to their high safety, long lifespan, and high energy density. However, how to efficiently schedule these energy storage units to maximize their performance and extend their lifespan remains a challenge in current research and application. Existing energy storage scheduling methods, especially in industrial parks, often rely on simple load forecasting and photovoltaic power generation forecasting data to make scheduling decisions. However, these methods neglect the impact of factors such as the health status of individual energy storage cells, charge / discharge status, and temperature on battery life and system performance. Therefore, in the actual operation of energy storage systems, over-discharging or over-charging of batteries can easily occur, leading to accelerated battery degradation and reduced long-term system efficiency. Summary of the Invention

[0004] The purpose of this invention is to provide a lithium iron phosphate energy storage scheduling method for industrial parks in order to solve the above-mentioned technical problems. This method balances the economy and reliability of energy storage systems in industrial parks, reduces the overall cost of electricity and lifespan, extends the lifespan of lithium iron phosphate batteries, and improves the safety and stability of energy use in the power grid and industrial parks.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A lithium iron phosphate energy storage dispatch method for industrial parks includes the following steps:

[0007] S1. Explicitly add an equivalent lifetime loss cost item to the scheduling target. The equivalent lifetime loss cost item is represented in the form of an identifiable parameter function. The equivalent lifetime loss cost item is jointly determined by the discharge depth, charge-discharge rate, unit temperature and health status parameters of the energy storage unit. This enables the scheduler to take the long-term total cost of the park as the optimization target. The long-term total cost includes the cost of electricity and the equivalent lifetime loss cost.

[0008] S2. A multi-timescale control architecture is adopted. In the first timescale, a rolling robust optimizer with conditional risk value constraints is executed to output the predicted value of energy storage power. In the second timescale, a real-time scheduler is executed to provide fault tolerance and rapid response based on actual operating deviations and emergency states. Energy storage power is allocated based on a load priority list, which includes critical continuous loads, batch loads, and peak-shaving loads.

[0009] S3. For multiple energy storage units operating in parallel, power is allocated according to individual unit weights. The individual unit weights are determined by a weighting function that takes the individual unit health status parameters, the individual unit state of charge, the individual unit temperature, and the individual unit internal resistance as inputs. The weighting function adopts a weak nonlinear rule to reduce the power allocation ratio of units with lower health status parameters. The attenuation rate and the next maintenance window of the energy storage units are predicted by a digital twin module, and the predicted health status parameters are used as input parameters of a rolling robust optimizer.

[0010] S4. During the switching process between grid-connected and off-grid operation, a smooth switching logic is executed based on energy retention boundaries, individual cell temperature and health status parameter limits, and stepped discharge distribution to reduce the impact of the switching process on the power grid and equipment.

[0011] Furthermore, the equivalent lifetime loss cost item Determined by the following formula: in, This indicates the number of individual energy storage units included in the energy storage system. The depth of discharge of a single energy storage cell. The charge / discharge rate of the energy storage cell. The temperature of the energy storage cell, These are the health status parameters of individual energy storage cells. , , , For calibration parameters, Temperature factor; the equivalent lifetime loss cost item This is used to reflect the lifespan degradation cost of lithium iron phosphate batteries under different operating conditions; the scheduler estimates the equivalent lifespan loss cost of the energy storage system at each moment in real time. and the equivalent life loss cost item The total long-run cost is formed by adding the cost of electricity to the total cost of electricity. :

[0012] in, This represents the electricity price at time t. This indicates the amount of electricity purchased.

[0013] Furthermore, the rolling robust optimizer, based on the input load forecast, photovoltaic power generation forecast, electricity price information, purchased power, and battery status information, constructs a scheduling optimization model and calculates the charging and discharging power forecast. The rolling robust optimizer solves for the predicted energy storage power within the first time scale, incorporating power balance constraints, individual state of charge constraints, individual power constraints, and conditional value-at-risk constraints. The power balance constraints ensure that the total power supply balances the predicted load at each first time scale; the individual state of charge constraints maintain safe operation; and the individual power constraints prevent over-power operation. The first time scale interval is 1 hour. The supply shortage of critical continuous loads at time t is defined as... The supply shortage loss function is defined as follows: ;in, This indicates the permissible supply shortage threshold;

[0014] The conditional Value at Risk constraint: ;in, For confidence level, As an auxiliary variable, The predicted output charging and discharging power is the acceptable upper limit for the risk of supply shortage. Conditional Value at Risk (VaR) .

[0015] Furthermore, real-time operational data of the park is collected, including real-time load. Real-time photovoltaic power generation And energy storage system state parameters, including SOC, SOH, and temperature; calculation of the deviation between predictions and actual values: ;If the deviation If the threshold is exceeded, the deviation correction mechanism will be activated:

[0016] The predicted charge and discharge power output of the rolling robust optimizer Corrected to execution power:

[0017]

[0018] in To adjust the proportional coefficient; when the battery condition reaches its limit or there is a disturbance in the external power grid, the real-time dispatcher enters emergency mode to prioritize meeting the power supply needs of critical continuous loads;

[0019] A pre-defined load priority list for the industrial park is established, including critical continuous loads, batch loads, and peak-shaving loads, which are prioritized for reduction during power shortages. When energy storage capacity is insufficient to cover all loads, allocation is performed according to the following logic: For critical continuous loads, it must be ensured that their power supply meets the forecast value and does not fall below it: where, The actual power allocated to critical continuous loads. Power forecasts for critical continuous loads;

[0020] For batch loads, if the energy storage capacity is insufficient, the power supply can be adjusted to be lower than the predicted value based on the actual situation, thereby reducing the power demand of the batch load: where, where, The actual power allocated to the batch load. This represents the predicted power output for the batch load.

[0021] For peak-shaving loads, their power demand can be adjusted by reducing or shifting peak hours during periods of power shortage:

[0022]

[0023] in, The actual power allocated to peak-shaving loads. This represents the predicted power value for peak loads.

[0024] Furthermore, health status parameters are collected for the i-th energy storage cell. State of charge ,temperature Internal resistance This is used to reflect the degree of electrochemical degradation; a weakly nonlinear weighting function is set. Where α, β, γ, δ are non-negative distinguishable parameters; and These are the recommended state of charge and operating temperature, respectively; the weight of each individual cell is calculated based on the weighting function results, and this is ensured. =1:

[0025]

[0026] The actual power allocated to each of the aforementioned energy storage cells is:

[0027] .

[0028] Furthermore, based on the real-time operating data of the energy storage unit, the decay rate of the energy storage unit is predicted using a support vector machine model. The decay rate reflects the rate of change in the health status of the energy storage unit. Based on the predicted decay rate, the trend of the energy storage unit's health status over a future period is estimated. Based on the health status prediction results, the next maintenance window is determined. The next maintenance window is the time range within which the energy storage unit is expected to require maintenance or replacement in the future. The predicted health status parameters and the next maintenance window are used as input parameters for subsequent energy storage scheduling decision optimization processes.

[0029] Furthermore, the training method for the support vector machine model includes: training the model using historical charge and discharge data, wherein the historical charge and discharge data includes the decay rate of the energy storage cell under different operating conditions and the corresponding input features; and optimizing the performance of the support vector machine model by adjusting the hyperparameters of the support vector machine through the radial basis kernel function.

[0030] Furthermore, an energy retention boundary is defined to maintain a certain amount of remaining battery energy during the switching process. This energy retention boundary is dynamically adjusted based on the charge / discharge state of the individual energy storage cells and the current operating mode. During the switching process, the temperature and health status parameters of the individual energy storage cells are monitored, and limiting control is implemented based on these parameters. If the temperature of an individual energy storage cell exceeds a preset safety threshold, the system will pause the switching process and activate a temperature control mechanism. If the health status of an individual energy storage cell is low, the system limits the discharge power of that cell, prioritizing the use of cells with better health status for power supply. A tiered discharge distribution strategy is used to gradually adjust the battery power output. When switching to grid-connected mode, the power output of the energy storage system is gradually increased. When switching to off-grid mode, the output power of the energy storage system is gradually reduced. During the switching process, the operating status of the energy storage system is monitored and adjusted in real time, and SOC deviation is corrected.

[0031] A lithium iron phosphate energy storage and dispatch system for industrial parks, the system employing the above-mentioned methods, including:

[0032] The energy storage unit is composed of multiple parallel energy storage cells, and each energy storage cell has a state of charge, health status, temperature and internal resistance monitoring module.

[0033] A rolling robust optimizer is used to generate an energy storage power dispatch plan based on park load forecasts, photovoltaic output forecasts, and electricity price signals at the first time scale. Its optimization objective is to minimize the long-term total cost, including the cost of electricity and the equivalent lifetime loss cost, and it is calculated using an identifiable parameter function.

[0034] The real-time scheduler is used to correct the energy storage power output of the upper-level optimizer based on the actual load, photovoltaic output and energy storage unit status on the second time scale, so as to achieve power deviation fault tolerance and rapid response in emergency situations, and to perform hierarchical power allocation for key continuous loads, batch loads and peak-shaving loads according to the load priority list.

[0035] The power distribution module calculates weights based on the SOC, SOH, temperature, and internal resistance of each energy storage cell and distributes the energy storage power to each energy storage cell according to the weights.

[0036] The digital twin module is used to estimate the health status parameters of each energy storage unit online based on real-time acquired data and historical operating data, and to feed back the scheduling parameters to the rolling robust optimizer.

[0037] The grid-connected and off-grid switching control module is used to perform phased power allocation when the grid is abnormal or recovering, and to perform smooth switching by combining energy retention boundaries and stepped discharge strategies.

[0038] Compared with the prior art, the present invention has the following technical effects:

[0039] This invention explicitly introduces an equivalent lifetime loss cost term into the scheduling objective and combines it with a multi-timescale rolling robust optimization method. This enables the scheduler to optimize both energy cost and lifetime cost while ensuring the optimal energy cost for the industrial park and taking into account the long-term lifetime degradation characteristics of lithium iron phosphate batteries. Compared with existing scheduling methods based solely on electricity price or SOC constraints, this invention can more accurately quantify battery lifetime consumption, avoid additional degradation caused by deep discharge and high-rate operation, thereby effectively extending the lifespan of the energy storage system and reducing long-term operation and maintenance costs.

[0040] This invention achieves highly reliable adaptation to complex energy consumption scenarios in industrial parks through a real-time scheduler driven by a load priority list, individual power allocation rules driven by State of Health (SOH), and smooth control logic for grid-connected / off-grid switching. Under power shortages or sudden disturbances, this invention can prioritize power supply to critical continuous loads while also accommodating dynamic adjustments to batch loads and peak-shaving loads. When multiple energy storage units operate in parallel, weighted allocation is performed based on health status parameters, SOC, temperature, and internal resistance, effectively preventing excessive losses in units with low SOH. During grid-connected / off-grid switching, energy retention boundaries and a stepped discharge strategy reduce grid and equipment impacts, thereby improving the overall operational stability and safety of the system. Attached Figure Description

[0041] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0042] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0043] like Figure 1 The method for dispatching lithium iron phosphate energy storage in industrial parks, as shown, includes the following steps:

[0044] S1. An equivalent lifetime loss cost term is explicitly added to the scheduling objective. This equivalent lifetime loss cost term is represented in the form of an identifiable parameter function. The equivalent lifetime loss cost term is jointly determined by the discharge depth, charge / discharge rate, cell temperature, and health status parameters of the energy storage cells. This allows the scheduler to use the long-term total cost of the park as the optimization objective. The long-term total cost includes both the electricity cost and the equivalent lifetime loss cost. This quantifies the lifespan consumption of the energy storage system into an economic indicator that can be directly used for optimization decisions. It avoids considering only electricity price optimization while ignoring the long-term costs brought by battery life. It provides a comprehensive objective for multi-timescale rolling optimization.

[0045] S2. A multi-timescale control architecture is adopted. In the first timescale, a rolling robust optimizer with conditional risk value constraints is executed to output the predicted value of energy storage power. In the second timescale, a real-time scheduler is executed to provide fault tolerance and rapid response based on actual operating deviations and emergency states. Energy storage power is allocated based on a load priority list, which includes critical continuous loads, batch loads, and peak-shaving loads.

[0046] S3. For multiple energy storage units operating in parallel, power is allocated according to individual unit weights. The individual unit weights are determined by a weighting function that takes the individual unit health status parameters, the individual unit state of charge, the individual unit temperature, and the individual unit internal resistance as inputs. The weighting function adopts a weak nonlinear rule to reduce the power allocation ratio of units with lower health status parameters. The attenuation rate and the next maintenance window of the energy storage units are predicted by a digital twin module, and the predicted health status parameters are used as input parameters of a rolling robust optimizer.

[0047] S4. During the switching process between grid-connected and off-grid operation, a smooth switching logic is executed based on energy retention boundaries, limiting of individual cell temperature and health status parameters, and tiered discharge allocation to reduce the impact of the switching process on the power grid and equipment. This invention explicitly introduces an equivalent lifetime loss cost term into the scheduling objective and combines it with a multi-timescale rolling robust optimization method, enabling the scheduler to ensure optimal energy costs for the park while also considering the long-term lifetime degradation characteristics of lithium iron phosphate batteries, achieving comprehensive optimization of energy costs and lifetime costs. Compared with existing scheduling methods based solely on electricity price or SOC constraints, this invention can more accurately quantify battery lifetime consumption, avoid additional degradation caused by deep discharge and high-rate operation, thereby effectively extending the service life of the energy storage system and reducing long-term operation and maintenance costs.

[0048] This invention achieves highly reliable adaptation to complex energy consumption scenarios in industrial parks through a real-time scheduler driven by a load priority list, individual power allocation rules driven by State of Health (SOH), and smooth control logic for grid-connected / off-grid switching. Under power shortages or sudden disturbances, this invention can prioritize power supply to critical continuous loads while also accommodating dynamic adjustments to batch loads and peak-shaving loads. When multiple energy storage units operate in parallel, weighted allocation is performed based on health status parameters, SOC, temperature, and internal resistance, effectively preventing excessive losses in units with low SOH. During grid-connected / off-grid switching, energy retention boundaries and a stepped discharge strategy reduce grid and equipment impacts, thereby improving the overall operational stability and safety of the system.

[0049] Specifically, the equivalent lifetime loss cost item Determined by the following formula: in, This indicates the number of individual energy storage units included in the energy storage system. The depth of discharge of a single energy storage cell. The charge / discharge rate of the energy storage cell. The temperature of the energy storage cell, These are the health status parameters of individual energy storage cells. , , , For calibration parameters, Temperature factor; the equivalent lifetime loss cost item This is used to reflect the lifespan degradation cost of lithium iron phosphate batteries under different operating conditions; the scheduler estimates the equivalent lifespan loss cost of the energy storage system at each moment in real time. and the equivalent life loss cost item The total long-run cost is formed by adding the cost of electricity to the total cost of electricity. : ;

[0050] in, This represents the electricity price at time t. This represents the purchased power. The complex characteristics of battery life degradation are represented by a calculable function that can be directly used in dispatch optimization. By incorporating depth of discharge, charge / discharge rate, temperature, and state of equilibrium (SOH) into the lifespan cost calculation, the dispatcher can quantify the impact of battery degradation on economic costs. This avoids excessive battery life loss due to deep discharge and high-rate operation, extending the lifespan of the energy storage system. A comprehensive long-term total cost is formed as the dispatch optimization objective, ensuring that system dispatch considers both short-term electricity prices and long-term economic viability and sustainability.

[0051] Specifically, the input includes load forecast, photovoltaic power generation forecast, electricity price information, purchased power, and battery status information. The rolling robust optimizer constructs a scheduling optimization model based on these data and calculates the predicted charging and discharging power. The rolling robust optimizer is designed to consider the actual operating conditions of the park's load, renewable power generation, and energy storage status, supporting subsequent power prediction calculations and the establishment of constraints. It solves for the predicted power values ​​of energy storage and photovoltaic power generation within the first time scale. The predicted energy storage power value is based on historical electricity consumption data and real-time operating data, predicted through time series analysis, and the output is the predicted power value for various load types within the first time scale. The photovoltaic power generation forecast is based on historical power generation data, real-time meteorological data, and weather forecast information. It is predicted using a numerical meteorological model, and the output is the predicted photovoltaic power generation at each time step. It incorporates power balance constraints, individual state-of-charge constraints, individual power constraints, and conditional value-at-risk constraints; the power balance constraints are used to maintain a balance between the total power supply and the predicted load at each first time scale; the individual state-of-charge constraints are used to maintain safe operation; and the individual power constraints prevent over-power operation; the first time scale interval is 1 hour; the supply shortage of critical continuous loads at time t is defined as... The supply shortage loss function is defined as follows: ;in, The threshold for allowable power shortages is indicated; power balance constraints ensure supply and demand matching in each time period to avoid power shortages or over-dispatch; SOC constraints ensure the safe operation of individual energy storage units to avoid overcharging / over-discharging; power constraints prevent individual energy storage units from operating at overpower and improve equipment reliability; CVaR constraints reduce the extreme economic risk of critical load power shortages and improve system robustness.

[0052] The conditional Value at Risk constraint:

[0053]

[0054] in, For confidence level, As an auxiliary variable, The predicted output charging and discharging power is the acceptable upper limit for the risk of supply shortage. Conditional Value at Risk (VaR) Under the conditions of satisfying power balance constraints, energy storage operation constraints and the aforementioned CVaR risk constraints, the charging and discharging power trajectory of the park's energy storage is output.

[0055] Specifically, real-time operational data of the park is collected, including real-time load. Real-time photovoltaic power generation And the energy storage system's state parameters, including individual cell health status parameters (SOH), individual cell state of charge (SOC), and temperature; the calculated deviation between predictions and actual values: ;If the deviation If the threshold is exceeded, the deviation correction mechanism is activated to incorporate the deviation between the forecast and the actual situation into the control, thereby reducing the situation of supply shortage or oversupply caused by forecast errors.

[0056] The predicted charge and discharge power output of the rolling robust optimizer Corrected to execution power:

[0057]

[0058] in To adjust the proportional coefficient; when the battery condition reaches its limit or there is a disturbance in the external power grid, the real-time dispatcher enters emergency mode to prioritize meeting the power supply needs of critical continuous loads; power correction ensures that the energy storage system operates within safety constraints; emergency mode can guarantee continuous power supply to critical loads and improve power supply reliability.

[0059] A pre-defined load priority list for the industrial park is established, including critical continuous loads, batch loads, and peak-shaving loads, which are prioritized for reduction during power shortages. When energy storage capacity is insufficient to cover all loads, allocation is performed according to the following logic: For critical continuous loads, it must be ensured that their power supply meets the forecast value and does not fall below it: where, The actual power allocated to critical continuous loads. Power forecasts for critical continuous loads;

[0060] For batch loads, if the energy storage capacity is insufficient, the power supply can be adjusted to be lower than the predicted value based on the actual situation, thereby reducing the power demand of the batch load: where, where, The actual power allocated to the batch load. This represents the predicted power output for the batch load.

[0061] For peak-shaving loads, their power demand can be adjusted by reducing or shifting peak hours during periods of power shortage:

[0062]

[0063] in, The actual power allocated to peak-shaving loads. This represents the predicted power value for peak-shaving loads. Prioritize critical loads to ensure continuous power supply to production lines or core equipment; flexibly adjust batch loads and peak-shaving loads to achieve peak shaving and valley filling; improve the utilization efficiency of energy storage power to reduce the overall electricity cost and supply shortage risk of the park.

[0064] Specifically, health status parameters are collected for the i-th energy storage cell. It reflects the remaining lifetime of the monomer, the degree of electrochemical decay, and the state of charge. Overcharge / over-discharge reaction, temperature Internal resistance This is used to reflect the degree of electrochemical degradation; a weakly nonlinear weighting function is set. Where α, β, γ, δ are non-negative distinguishable parameters; and These are the recommended state of charge and operating temperature; real-time monitoring of individual cell operating status to provide data for power allocation; and ensuring that each energy storage cell participates in scheduling under safe operating conditions.

[0065] The weight of each unit is calculated based on the weighting function results, dynamically reflecting the health status and current operating conditions of each unit, reducing the power allocation of deteriorated or abnormal units, supporting the safety and balance of multi-unit parallel operation, and ensuring... =1:

[0066]

[0067] The actual power allocated to each of the aforementioned energy storage cells is:

[0068] .

[0069] Specifically, based on the real-time operating data of the energy storage units, the decay rate of the energy storage units is predicted using a support vector machine model. The decay rate reflects the rate of change in the health status of the energy storage units. Based on the predicted decay rate, the trend of the health status of the energy storage units over a future period is estimated. Based on the health status prediction results, the next maintenance window is determined, which is the expected time range within which the energy storage units will need maintenance or replacement in the future. The predicted health status parameters and the next maintenance window are used as input parameters for subsequent energy storage scheduling decision optimization processes. This enables dynamic allocation of unit power according to health status, balancing short-term power supply needs with long-term lifespan protection. It improves the overall power supply capacity and reliability of the energy storage system, avoids unit overload, and enhances the system's economy and safety.

[0070] Specifically, the training method for the support vector machine model includes: training the model using historical charge and discharge data, which includes the decay rate of energy storage cells under different operating conditions and corresponding input features; adjusting the hyperparameters of the support vector machine to optimize the model's performance through radial basis kernel functions; learning the health decay patterns of energy storage cells under different operating conditions; achieving high prediction accuracy, which can be used as a forward prediction input for the SOH and the next maintenance window in a digital twin module; and supporting a rolling robust optimizer to make long-term total cost optimization energy storage scheduling decisions.

[0071] Specifically, an energy retention boundary is defined, and a certain amount of remaining battery energy is maintained during the switching process to ensure continuous power supply to critical loads during off-grid or grid-connected switching. This energy retention boundary is dynamically adjusted based on the charge / discharge status of the energy storage cells and the current operating mode. During the switching process, the temperature and health status parameters of the energy storage cells are monitored, and limiting control is implemented based on these parameters. Overheating or overloading of degraded batteries is avoided, reducing the risk of thermal runaway and lifespan degradation. If the temperature of an energy storage cell exceeds a preset safety threshold, the system will pause the switching process and activate the temperature control mechanism. If the health status of an energy storage cell is low, the system limits the discharge power of the cell, prioritizing the use of cells with better health status for power supply. A tiered discharge distribution strategy is used to gradually adjust the battery power output, smoothly switching the energy storage system output and reducing the impact on the grid and equipment. This improves the continuity of the energy storage system's output power and load stability. Through gradual adjustment, the risk of cell lifespan degradation is reduced, improving long-term reliability. When switching to grid-connected mode, the power output of the energy storage system is gradually increased; when switching to off-grid mode, the output power of the energy storage system is gradually reduced; during the switching process, the operating status of the energy storage system is monitored and adjusted in real time, and the SOC deviation is corrected to improve the system's fault tolerance to emergencies.

[0072] A lithium iron phosphate energy storage and dispatch system for industrial parks, the system employing the above-mentioned methods, including:

[0073] The energy storage unit is composed of multiple parallel energy storage cells, and each energy storage cell has a state of charge, health status, temperature and internal resistance monitoring module.

[0074] A rolling robust optimizer is used to generate an energy storage power dispatch plan based on park load forecasts, photovoltaic output forecasts, and electricity price signals at the first time scale. Its optimization objective is to minimize the long-term total cost, including the cost of electricity and the equivalent lifetime loss cost, and it is calculated using an identifiable parameter function.

[0075] The real-time scheduler is used to correct the energy storage power output of the upper-level optimizer based on the actual load, photovoltaic output and energy storage unit status on the second time scale, so as to achieve power deviation fault tolerance and rapid response in emergency situations, and to perform hierarchical power allocation for key continuous loads, batch loads and peak-shaving loads according to the load priority list.

[0076] The power distribution module calculates weights based on the SOC, SOH, temperature, and internal resistance of each energy storage cell and distributes the energy storage power to each energy storage cell according to the weights.

[0077] The digital twin module is used to estimate the health status parameters of each energy storage unit online based on real-time acquired data and historical operating data, and to feed back the scheduling parameters to the rolling robust optimizer.

[0078] The grid-connected and off-grid switching control module is used to perform phased power allocation when the grid is abnormal or recovering, and to perform smooth switching by combining energy retention boundaries and stepped discharge strategies.

[0079] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0080] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A lithium iron phosphate energy storage scheduling method for industrial parks, characterized by, The method comprises the following steps: S1, adding an equivalent life loss cost item in the scheduling target, the equivalent life loss cost item is expressed in the form of a recognizable parameter function, and the equivalent life loss cost item is determined by the depth of discharge, the charge-discharge rate, the temperature and the health state parameter of the energy storage unit, so that the scheduler takes the long-term total cost of the park as the optimization target, and the long-term total cost comprises an electric energy cost and an equivalent life loss cost; S2, using a multi-time scale control architecture, executing a rolling robust optimizer with a conditional value-at-risk constraint in a first time scale to output an energy storage power prediction value, executing a real-time scheduler in a second time scale to perform fault tolerance and rapid response according to actual operation deviations and emergency states, and performing energy storage power distribution based on a load priority list, the load priority list comprising critical continuous loads, batch loads and peak-cut loads; S3, distributing power to multiple energy storage units operating in parallel according to a unit weight, the unit weight being determined by a weight function taking the unit health state parameter, the unit state of charge, the unit temperature and the unit internal resistance as inputs, and the weight function adopting a weak non-linear rule to reduce the power distribution proportion of a unit with a low health state parameter; predicting the decay rate and the next maintenance window of the energy storage unit through a digital twin module, and taking the predicted health state parameter as an input parameter of the rolling robust optimizer; S4, during the switching process between grid-connected operation and off-grid operation, performing smooth switching logic based on an energy retention boundary, unit temperature and health state parameter limiting and step discharge distribution to reduce the impact of the switching process on the power grid and equipment.

2. The method of claim 1, wherein, the equivalent life loss cost term is determined by the formula: ； wherein, represents the number of energy storage cells contained in the energy storage system, represents the depth of discharge of the energy storage cell, represents the charge-discharge rate of the energy storage cell, represents the temperature of the energy storage cell, represents the state of health parameter of the energy storage cell, , , , represents a calibration parameter, represents a temperature factor; the equivalent life consumption cost term is used to reflect the life decay cost of lithium iron phosphate battery under different operating conditions; the scheduler estimates the equivalent life consumption cost term of the energy storage system at each moment in real time , and adds the equivalent life consumption cost term and the electricity cost to form the long-term total cost : ； wherein, denotes the electricity price at time t, denotes the purchased power.

3. The method of claim 2, wherein, The input load prediction value, photovoltaic power generation prediction value, electricity price information, power purchase power, and battery state information, and the rolling robust optimizer construct a scheduling optimization model according to the load prediction value, photovoltaic power generation prediction value, electricity price information, power purchase power, and battery state information, and calculate the charge and discharge power prediction value ; The rolling robust optimizer solves the energy storage power prediction value in a first time scale, and adds power balance constraints, single cell state of charge constraints, single cell power constraints and conditional risk value constraints; the power balance constraints are used to balance the total power supply and the load prediction value at each first time scale, the single cell state of charge constraints are used to maintain safe operation, and the single cell power constraints prevent over-power operation; the interval of the first time scale is 1 hour; the shortage amount of key continuous load at time t is defined as , and the shortage loss function is defined as: ; wherein, represents the allowed shortage threshold. the conditional value at risk constraint: ; where, is a confidence level, is an auxiliary variable, is an acceptable upper limit of the supply and demand risk, and the output is a charge and discharge power prediction value and the conditional value at risk constraint .

4. The method of claim 3, wherein, Collecting real-time running data of the park, the real-time running data including real-time load , real-time photovoltaic power generation , and energy storage system state parameters, including SOC, SOH, temperature; calculating the deviation between prediction and actual value: ; if the deviation exceeds the threshold value, starting the deviation correction mechanism; Charging and discharging power prediction values output by a rolling robust optimizer Corrected to perform power: ； wherein is a regulation factor; when the battery state reaches the limit or the external power grid is disturbed, the real-time scheduler enters an emergency mode, and the power supply demand of the key continuous load is preferentially met; Pre-set the park load priority list, including key continuous load, batch load, peak shaving load, and the priority is reduced in power shortage; when the energy storage power is insufficient to cover all loads, the following logic is used for distribution: for key continuous load, it must be ensured that the power supply meets the predicted value and cannot be lower than the value: wherein, is the actual power allocated to the key continuous load, is the power prediction value of the key continuous load; For batch load, if the energy storage power is insufficient, the power supply can be adjusted according to the actual situation to be lower than the predicted value, reducing the power demand of the batch load: wherein, wherein, is the actual power allocated to the batch load, is the power prediction value of the batch load; For the peak-cut load, the power supply demand can be adjusted by cutting or peak-shifting during power shortage: ； wherein, Pact is the actual power assigned to the peakable load, Ppred is the power prediction value of the peakable load.

5. The method of claim 4, wherein, Collecting health state parameters of the i-th energy storage unit , state of charge , temperature , internal resistance , for reflecting the degree of electrochemical attenuation; setting a weak non-linear weight function ; wherein: α, β, γ, δ are non-negative identifiable parameters; and are the recommended state of charge and operating temperature respectively; according to the weight function result, the weight of each unit is calculated, and =1: ； The actual allocated power of each energy storage unit is: 。 6. The method of claim 5, wherein, Based on the real-time operation data of the energy storage unit, the decay rate of the energy storage unit is predicted through a support vector machine model, the decay rate reflecting the rate of change of the health state of the energy storage unit; according to the predicted decay rate, the trend of the health state of the energy storage unit in a future period of time is estimated; based on the prediction result of the health state, the next maintenance window is determined, the next maintenance window being a time range in which the energy storage unit is predicted to need maintenance or replacement in the future; the predicted health state parameter and the next maintenance window are taken as input parameters for subsequent energy storage scheduling decision optimization process.

7. The method of claim 6, wherein, The training method of the support vector machine model comprises: using charge-discharge historical data to train the model, the charge-discharge historical data containing the decay rate of the energy storage unit under different working conditions and the corresponding input features; and adjusting the hyperparameter optimization model of the support vector machine through a radial basis kernel function to improve the performance of the model.

8. The method as claimed in claim 4, wherein, An energy retention boundary is defined, and a certain amount of battery remaining energy is maintained during the switching process; the energy retention boundary is dynamically adjusted according to the charge-discharge state of the energy storage unit and the current operation mode; during the switching process, the temperature and the health state parameter of the energy storage unit are monitored, and limiting control is performed according to these parameters; if the temperature of the energy storage unit exceeds a preset safety threshold, the system will pause the switching process and start a temperature control mechanism; If the state of health of the energy storage cells is low, the system limits the discharge power of the energy storage cells, and preferentially uses cells with better state of health for power supply; in the stepwise discharge distribution strategy, the power output of the battery is gradually adjusted; when switching to grid-connected mode, the power output of the energy storage system is gradually increased; when switching to off-grid mode, the output power of the energy storage system is gradually reduced; during the switching process, the operating state of the energy storage system is monitored and adjusted in real time, and the SOC deviation is corrected.

9. A lithium iron phosphate energy storage scheduling system for industrial parks, characterized in that, The system uses the method of any one of claims 1-8, and the system comprises: an energy storage unit composed of a plurality of parallel energy storage cells, each energy storage unit having a state of charge, a state of health, a temperature, and an internal resistance monitoring module; a rolling robust optimizer for generating an energy storage power scheduling plan according to park load prediction, photovoltaic output prediction, and electricity price signals at a first time scale, with the optimization objective being to minimize long-term total cost, including electricity cost and equivalent life loss cost, and using a discernible parameter function form for calculation; a real-time scheduler for correcting the energy storage power output of the upper optimizer according to actual load, photovoltaic output, and energy storage cell state at a second time scale, achieving power deviation fault tolerance and emergency state rapid response, and performing hierarchical power distribution on key continuous load, batch load, and peak-shaving load according to a load priority list; a power distribution module for calculating weights according to the SOC, SOH, temperature, and internal resistance of each energy storage cell and distributing energy storage power to each energy storage cell according to the weights; a digital twin module for online estimation of the state of health parameters of each energy storage cell based on real-time acquisition data and historical operation data, and feedback of scheduling parameters to the rolling robust optimizer; a grid-connected and off-grid switching control module for performing phased power distribution when the power grid is abnormal or restored, and combining energy retention boundaries and stepwise discharge strategies for smooth switching.

Images