Photovoltaic energy storage collaborative control system considering multiple time scales

Abstract

The present application relates to photovoltaic energy storage control technical field, specifically to a kind of photovoltaic energy storage collaborative control system considering multiple time scales, comprising: constructing day-ahead prediction, day-ahead rolling and real-time control three-layer time scale hierarchical framework, each level corresponds different control cycle and execution frequency.Day-ahead layer combines grid dispatching instruction and twenty-four hours photovoltaic power prediction, to reduce the target of power purchase cost, reduce energy storage cycle loss, formulate the next day hour-level energy storage charge-discharge power baseline;Day-ahead layer relies on four hours ultra-short-term photovoltaic prediction, suppresses the deviation of actual operation and day-ahead plan, generates fifteen minutes level rolling adjustment instruction;Real-time layer collects photovoltaic actual output and energy storage operating state, outputs second-level control signal.The present application can realize multi-scale control instruction collaborative connection, adapt photovoltaic power fluctuation, reduce operating deviation, stabilize dc bus voltage, improve the stability and economy of system operation.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic energy storage control technology, and in particular to a photovoltaic energy storage collaborative control system that considers multiple time scales. Background Technology

[0002] Existing photovoltaic (PV) energy storage coordinated control systems mostly employ a single time scale or a two-layer control mode. Conventional technologies often rely solely on long-term PV power forecasts to formulate energy storage charging and discharging plans. Some solutions only perform simple adjustments based on short-term power forecasts, while others rely solely on real-time operational data to execute basic control of the energy storage converter. A complete control system covering long-term planning, short-term adjustment, and real-time execution has not been formed. The control cycles of existing control schemes are mostly concentrated in a single dimension, such as hours or minutes. The control cycle and execution frequency are not hierarchically matched, and the control objectives are solely focused on cost control or power mitigation, without adapting grid dispatch instructions, long-term and short-term forecast data, and real-time operational status to the control process in a dimensional manner.

[0003] Existing control schemes cannot match the fluctuation characteristics of photovoltaic power at different time scales. There is a lack of connection between long-term planning, short-term adjustment, and real-time control. Hourly planning is difficult to adapt to minute-level power deviations, and minute-level adjustment cannot support the second-level DC bus voltage stability control requirements. There is no progressive reference transmission relationship between energy storage charging and discharging plans, rolling adjustment commands, and real-time control signals. This easily leads to situations where the actual operation deviates significantly from the day-ahead plan, energy storage cycle losses are high, and DC bus voltage fluctuates. It is impossible to simultaneously meet the multiple control requirements of long-term dispatch cost optimization, short-term deviation mitigation, and real-time voltage stability.

[0004] This invention requires the construction of a time-scale hierarchical control framework that matches different control cycles and execution frequencies. It matches corresponding prediction data, control targets and output instructions according to the level, and establishes a progressive connection method in which the output of the previous level serves as the control reference for the next level, thereby solving the problems of disconnection between control instructions of different time scales and the inability of control targets to be coordinated and adapted. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a photovoltaic energy storage collaborative control system that considers multiple time scales.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a photovoltaic energy storage collaborative control system considering multiple time scales, comprising: At the forecast level, the system receives the dispatch command curve issued by the power grid and the photovoltaic power forecast sequence for the next 24 hours of the photovoltaic power plant. With the goal of minimizing the cost of electricity purchase and the cycle loss of energy storage equipment, the system formulates the baseline of planned charge and discharge power of the energy storage system at the hourly scale of the next day. At the intraday rolling level, ultra-short-term photovoltaic power prediction data for the next four hours of the photovoltaic power station is obtained. Based on the planned charge and discharge power baseline of the energy storage system at each hourly scale of the next day, with the goal of smoothing out the deviation between the actual operation and the previous day's plan, rolling adjustment power instructions for the energy storage system at the next fifteen-minute scale are formulated. At the real-time control level, the current actual output power of the photovoltaic array and the current actual operating status of the energy storage system are collected. With the goal of tracking the rolling power adjustment command of the energy storage system on a 15-minute scale and ensuring the stability of the DC bus voltage, real-time control signals of the energy storage converter on a second or sub-second scale are generated.

[0007] As a further aspect of the present invention, at the day-ahead forecasting level, receiving the dispatch command curve issued by the power grid and the photovoltaic power forecast sequence for the photovoltaic power plant for the next 24 hours, and using minimizing the electricity purchase cost and the cycle loss of the energy storage equipment as the comprehensive objective function, the planned charge and discharge power baseline of the energy storage system at each hourly scale of the next day is formulated, including: Receive and parse the grid's 96-point dispatch instructions for the next day, wherein the dispatch instruction curve includes the planned output power at the grid connection point; Numerical weather forecast data is acquired, and a photovoltaic power prediction sequence updated hourly for the next 24 hours is generated using a power conversion model; Read the rated capacity, current state of charge, maximum charge / discharge power limit, and charge / discharge efficiency parameters of the energy storage system; The comprehensive objective function is constructed, which includes the electricity cost item for interaction with the grid calculated based on time-of-use pricing, and the energy storage lifetime depreciation cost item calculated based on the number of charge-discharge cycles and depth. Under the constraints of the rated capacity, current state of charge, maximum charge / discharge power limit, and charge / discharge efficiency parameters, the optimal solution of the comprehensive objective function for the next 24 hours is solved with a time resolution of one hour. The optimal solution is a series of power values, forming the planned charge / discharge power baseline of the energy storage system at the hourly scale of the next day.

[0008] As a further aspect of the present invention, the system also includes a process for updating the reference state of charge of the energy storage system: After obtaining the planned charge and discharge power baseline of the energy storage system at each hourly scale of the next day, the current state of charge at the start time of the energy storage system is used as the initial value, and energy accumulation calculation is performed according to the planned charge and discharge power baseline of the energy storage system at each hourly scale of the next day. During the energy accumulation calculation, the charging power and the discharging power are calculated separately according to the charging and discharging efficiency parameters; The predicted state of termination of charge of the energy storage system after the end of the next 24 hours is calculated. The current state of charge at the start of the next day is compared with the ending state of charge. If the ending state of charge deviates from the preset ideal state of charge center value, the planned charging and discharging power baseline for the last few hours of the next day is corrected in a smooth adjustment manner so that the ending state of charge returns to the vicinity of the preset ideal state of charge center value. The planned charge and discharge power sequence, after being optimized and corrected for the state of charge, is finally determined as the baseline of the planned charge and discharge power of the energy storage system at the hourly scale of the next day.

[0009] As a further aspect of the present invention, the step of acquiring ultra-short-term photovoltaic power forecast data for the next four hours of the photovoltaic power station at the intraday rolling level, and based on the planned charge and discharge power baseline of the energy storage system at each hourly scale of the next day, with the goal of smoothing out the deviation between actual operation and the previous day's plan, formulating rolling adjustment power commands for the energy storage system at the next fifteen-minute scale, includes: A rolling optimization is triggered every 15 minutes to obtain the ultra-short-term photovoltaic power prediction data for the next four hours with a time resolution of five minutes, driven by real-time meteorological data. Read the actual state of charge of the energy storage system at the current moment, and the baseline of the planned charge and discharge power of the energy storage system at each hourly scale of the next day from the current moment to the next hour; Calculate the difference between the ultra-short-term photovoltaic power prediction data and the photovoltaic power prediction sequence for the corresponding time period in the day-ahead prediction level for the next hour starting from the current moment; the difference constitutes the photovoltaic power prediction deviation sequence. With the goal of minimizing the fluctuation of the total output power of the power plant caused by the photovoltaic power prediction deviation sequence, and under the premise of satisfying the power and energy constraints of the energy storage system, the power adjustment of the energy storage system in the next four 15-minute time periods is optimized. The average value of the planned charge and discharge power baseline of the energy storage system at each hourly scale on the next day within the corresponding 15-minute time period is added to the power adjustment amount of the time period to form the rolling adjustment power command of the energy storage system at the next 15-minute scale.

[0010] As a further aspect of the present invention, the system further includes a feasibility verification step for the rolling adjustment power command: After calculating the power adjustment amount of the energy storage system for the next four 15-minute time periods, starting from the actual state of charge of the energy storage system at the current moment, the trajectory of the change of the state of charge of the energy storage system after executing the rolling power adjustment command of the energy storage system at the next 15-minute scale is simulated and calculated time by time. Check whether the trajectory of the change in the state of charge is always within the preset safe operating range; If the trajectory of the change in the state of charge exceeds the safe operating range at any time, the rolling adjustment power command of the energy storage system in the next fifteen-minute time scale for the time period that caused the exceedance and the subsequent time period will be corrected by power limiting, and the correction direction is to bring the state of charge back to the safe operating range. The sequence of instructions, after feasibility verification and correction, will be used as the final execution command for the energy storage system to issue rolling power adjustment commands on a 15-minute timescale.

[0011] As a further aspect of the present invention, at the real-time control level, the current actual output power of the photovoltaic array and the current actual operating status of the energy storage system are collected to track the rolling power adjustment commands of the energy storage system on a 15-minute timescale and ensure the stability of the DC bus voltage, thereby generating a real-time control signal for the energy storage converter on a second- or sub-second timescale, including: The DC-side current and voltage of the photovoltaic array are collected at millisecond intervals to calculate the current actual output power of the photovoltaic array. The current, voltage, and DC bus voltage values ​​of the energy storage system are collected at the same interval to obtain the current actual operating status of the energy storage system. Receive the latest rolling power adjustment command issued by the energy storage system at the intraday rolling level for the next fifteen minutes; The difference between the rolling adjustment power command of the energy storage system on a 15-minute timescale and the current actual output power of the energy storage system is calculated as the power tracking error. Using the power tracking error and the offset of the DC bus voltage from the reference voltage as the input to the controller, the real-time control signal of the energy storage converter at the second or sub-second scale is calculated by the proportional-integral-derivative control algorithm. The real-time control signal is used to adjust the duty cycle of the power switching devices in the energy storage converter.

[0012] As a further aspect of the present invention, the system further includes a dynamic amplitude limiting protection step for the real-time control signal: Real-time monitoring of the current actual state of charge, temperature, and DC bus voltage of the energy storage system; Based on the current actual state of charge, the current allowable absolute power limit is obtained from the preset maximum allowable charge and discharge power curves corresponding to different state of charge ranges. Based on the temperature of the energy storage system, the power derating factor at the current temperature is obtained from the preset temperature-power derating factor table; Multiply the current allowed absolute power limit value by the power derating factor to obtain the dynamically adjusted real-time power limit value; The required power calculated from the real-time control signal of the energy storage converter at the second or sub-second scale is limited to the range of the dynamically adjusted real-time power limit value.

[0013] As a further aspect of the present invention, the system further includes the step of establishing information feedback between the day-ahead forecast level and the intraday rolling level: After each rolling optimization calculation at the intraday rolling level is completed, the statistical characteristics of the photovoltaic power prediction deviation sequence are recorded. The statistical characteristics include the average deviation and the standard deviation of the deviation. The statistical characteristics of the photovoltaic power prediction deviation sequence are uploaded to the day-ahead prediction level; Before the next optimization calculation is performed at the current prediction level, the internal parameters of the power conversion model are corrected or the prediction error compensation is applied to the photovoltaic power prediction sequence by utilizing the statistical characteristics of multiple photovoltaic power prediction deviation sequences uploaded in the past. Based on the corrected model or the compensated prediction sequence, the comprehensive objective function is solved again to generate a new round of planned charge and discharge power baselines for the energy storage system at the hourly scale of the next day.

[0014] As a further aspect of the present invention, the step of using the statistical characteristics of multiple historically uploaded photovoltaic power prediction deviation sequences to correct the internal parameters of the power conversion model or to apply prediction error compensation to the photovoltaic power prediction sequences includes: Extract the statistical characteristics of the photovoltaic power prediction deviation sequence uploaded at the same time within a set number of days in the past, and calculate its mean and trend; If the mean of the average deviation is consistently non-zero, it is determined that there is a systematic deviation in the power conversion model, and the conversion coefficient from meteorological data to photovoltaic power in the power conversion model is reverse-calibrated. If the standard deviation of the deviation shows an increasing trend, it is determined that the uncertainty of numerical weather forecast has increased, and a prediction error band based on the mean of the standard deviation of the deviation is superimposed on the photovoltaic power prediction sequence. When optimizing the solution of the comprehensive objective function, a robust optimization method is used to process the photovoltaic power prediction sequence superimposed with the prediction error band, so as to enhance the adaptability of the energy storage system to the prediction error at the hourly scale of each day.

[0015] As a further aspect of the present invention, the system also includes coordination, synchronization, and instruction switching management for multi-timescale control processes: A unified absolute time reference is set in the time-scale hierarchical framework, and the calculation and instruction issuance times of each of the day-ahead forecast level, intraday rolling level and real-time control level are all aligned with the absolute time reference. At the moment the instructions are issued at the intraday rolling level, check the actual instruction execution completion rate of the previous 15-minute rolling cycle; If the actual execution completion rate of the instruction is lower than the preset threshold, the execution deviation will be added to the currently calculated rolling adjustment power instruction. The execution deviation is the cumulative difference between the previous round of planned instructions and the actual execution power. At the real-time control level, a smooth transition mechanism for commands is set up. When the energy storage system detects an update of the rolling adjustment power command from the intraday rolling level on a 15-minute timescale, the real-time control level will smoothly transition the currently tracked command value to the new command value within several control cycles to avoid power command jumps.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: A time-scale hierarchical framework of day-ahead forecasting, intraday rolling, and real-time control is constructed, with each level corresponding to a differentiated control cycle and execution frequency. The day-ahead forecasting level takes the grid dispatch command curve and the 24-hour photovoltaic power forecast sequence as input, aiming to minimize electricity purchase costs and energy storage equipment cycle losses, forming an hourly planned charge and discharge power baseline. The intraday rolling level takes four-hour ultra-short-term photovoltaic power forecast data as input, aiming to smooth the deviation between actual operation and day-ahead plan, forming a 15-minute rolling adjustment power command. The real-time control level takes the actual output power of photovoltaics and the actual operating status of energy storage as input, aiming to track the rolling adjustment power command and stabilize the DC bus voltage, generating second-level or sub-second-level real-time control signals. The input data, control objectives, and output commands of each level form a precise adaptation relationship, and the control cycle and execution frequency are in line with the regulation requirements of the corresponding time scale and matched with the fluctuation characteristics of photovoltaic power at different time scales.

[0017] The planned charge and discharge power baseline of the day-ahead forecast level is used as the basis for formulating the intraday rolling level, and the rolling adjustment power command output of the intraday rolling level is used as the tracking target of the real-time control level. This enables the step-by-step acceptance and reference transmission of control commands at different levels, reduces the degree of command disconnect between control links at different time scales, reduces the interference of photovoltaic power fluctuations on the control process, reduces the additional cycle losses of energy storage equipment caused by command mismatch, maintains the stable operation of DC bus voltage, and simultaneously achieves the control objectives of optimizing power purchase cost, smoothing operation deviation, and stabilizing DC bus voltage. The control functions of each level are interconnected to form a continuous and coordinated photovoltaic energy storage control mechanism, improving the consistency and adaptability of the overall control process. Attached Figure Description

[0018] Figure 1 This is a timing diagram of a photovoltaic energy storage collaborative control system considering multiple time scales as described in this invention. Figure 2A flowchart for establishing a baseline for planned charge and discharge power of energy storage systems at the day-ahead forecast level; Figure 3 A flowchart for formulating rolling power adjustment commands for intraday rolling levels. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0020] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0021] See Figure 1 This invention provides a photovoltaic energy storage collaborative control system that considers multiple time scales, specifically including: A hierarchical time-scale framework is established, comprising day-ahead forecasting, intraday rolling control, and real-time control. Within this framework, each level corresponds to a different control cycle and execution frequency. At the day-ahead forecasting level, this level receives the dispatch command curve from the grid and the photovoltaic power forecast sequence for the next 24 hours of the photovoltaic power plant. Using minimizing electricity purchase costs and energy storage device cycle losses as the comprehensive objective function, it formulates the planned charge / discharge power baseline for the energy storage system at the hourly scale of the following day. At the intraday rolling control level, this level acquires ultra-short-term photovoltaic power forecast data for the next four hours of the photovoltaic power plant. Based on the planned charge / discharge power baseline of the energy storage system at the hourly scale of the following day, and with the goal of mitigating the deviation between actual operation and day-ahead planning, it formulates rolling adjustment power commands for the energy storage system at the 15-minute scale of the next day. At the real-time control level, this level collects the current actual output power of the photovoltaic array and the current actual operating status of the energy storage system. With the goal of tracking the rolling adjustment power commands of the energy storage system at the 15-minute scale of the next day and ensuring DC bus voltage stability, it generates real-time control signals for the energy storage converter at the second or sub-second scale.

[0022] In one embodiment of the present invention, see [reference] Figure 2In the implementation of the current forecasting level, the system receives and parses the 96-point dispatch instructions issued by the power grid for the next day. These dispatch instructions include the planned output power at the connection point with the power grid. The system acquires numerical weather forecast data and generates an hourly updated photovoltaic power forecast sequence for the photovoltaic power plant for the next 24 hours using a power conversion model. The system reads the rated capacity, current state of charge, maximum charge / discharge power limit, and charge / discharge efficiency parameters of the energy storage system. The system constructs a comprehensive objective function, which includes the cost of electricity interaction with the grid calculated based on time-of-use pricing, and the energy storage lifetime depreciation cost calculated based on the number of charge / discharge cycles and depth. Under the constraints of rated capacity, current state of charge, maximum charge / discharge power limit, and charge / discharge efficiency parameters, the system solves for the optimal solution of the comprehensive objective function for the next 24 hours with a one-hour time resolution. This optimal solution is a series of power values, forming the planned charge / discharge power baseline of the energy storage system at each hourly scale of the next day.

[0023] After obtaining the planned charge / discharge power baseline of the energy storage system at each hourly scale of the following day, the system also performs a process to update the reference state of charge (SOC) of the energy storage system. This process uses the current SOC of the energy storage system at the start time as the initial value and performs energy accumulation calculations according to the planned charge / discharge power baseline of the energy storage system at each hourly scale of the following day. During the energy accumulation calculation, the system converts the charging power and discharging power separately according to the charge / discharge efficiency parameters. The system calculates the predicted final SOC of the energy storage system after the end of the next 24 hours. The system compares the current SOC at the start time of the next day with the final SOC. If the final SOC deviates from the preset ideal SOC center value, the planned charge / discharge power baseline for the last few hours of the following day is corrected in a smoothing adjustment manner, so that the final SOC returns to near the preset ideal SOC center value. The system finally determines the planned charge / discharge power sequence after SOC optimization and correction as the planned charge / discharge power baseline of the energy storage system at each hourly scale of the following day.

[0024] In practical implementation, the day-ahead forecasting level of a photovoltaic energy storage collaborative control system considering multiple time scales operates as follows: The system receives and parses the grid dispatch instructions for the next day at a fixed time each day from the upper-level dispatch master station. The dispatch instruction curve, at 15-minute intervals, specifies the planned output power of the photovoltaic energy storage at each time period of the next day's grid connection point. For example, the planned output power is 500 kW from 10:00 to 11:00 the next day and 200 kW from 19:00 to 20:00. Simultaneously, the system obtains numerical weather forecast data from a meteorological service interface. This data includes irradiance and ambient temperature sequences for the next 24 hours. The power conversion model, established based on the historical operating data and equipment parameters of the photovoltaic power station, converts the irradiance and temperature data from the numerical weather forecast data into an hourly updated photovoltaic power forecast sequence for the next 24 hours. For example, the predicted photovoltaic output power at 12:00 the next day is 800 kW. The system reads the rated capacity, current state of charge, maximum charge and discharge power limit, and charge and discharge efficiency parameters of the energy storage system from the local monitoring unit. For example, the rated capacity of the energy storage system is 1000 kWh, the current state of charge is 50%, the maximum charge and discharge power limit is 250 kW, the charging efficiency is 95%, and the discharging efficiency is 95%.

[0025] In some embodiments, the integrated objective function constructed by the system aims to minimize the total operating cost. The integrated objective function comprises two main cost terms. The first term is the grid interaction cost calculated based on time-of-use pricing, for example, a purchase price of RMB 1.0 / kWh during peak hours and RMB 0.3 / kWh during off-peak hours. The second term is the energy storage lifetime depreciation cost calculated based on the number of charge-discharge cycles and depth of charge, which is calculated using an aging model related to charge-discharge power and the current state of charge of the energy storage system. The mathematical expression of the integrated objective function aims to minimize the total cost over the next 24 hours, and its form is to minimize a weighted sum of the electricity cost term and the lifetime depreciation cost term. Under the constraints of the energy storage system's rated capacity, current state of charge, maximum charge-discharge power limit, and charge-discharge efficiency parameters, the system uses linear programming or quadratic programming algorithms with a one-hour time resolution to solve for the optimal solution of the integrated objective function over the next 24 hours. The optimal solution is a series of power values, each power value corresponding to the planned charging and discharging power of the energy storage system for one hour. Positive values ​​indicate discharging, and negative values ​​indicate charging. This series of power values ​​forms the baseline of the planned charging and discharging power of the energy storage system at the hourly scale of the next day. For example, the solution results indicate that the planned discharging power of the energy storage at 14:00 on the next day is 100 kW, and the planned charging power at 04:00 on the next day is -150 kW.

[0026] After obtaining the planned charge / discharge power baseline of the energy storage system at each hourly scale of the following day, the system performs a process to update the reference state of charge (SOC) of the energy storage system. This process uses the current SOC of the energy storage system at the start time as the initial value and performs hourly energy accumulation calculations according to the planned charge / discharge power baseline of the energy storage system at each hourly scale of the following day. During the energy accumulation calculation, the system converts the charging power and discharging power separately according to the charge / discharge efficiency parameters. For example, if the planned charging power is -100 kW, considering a charging efficiency of 95%, the actual energy added to the energy storage system increases by 95 kWh. The predicted final SOC of the energy storage system at the end of the next 24 hours is calculated, for example, a final SOC of 70%. The system compares the current SOC at the start time of the next day with the final SOC and refers to a preset ideal SOC center value, which is usually set to 50%. If the termination state of charge (SOC) deviates from the preset ideal SOC center value, for example, if the termination SOC is 70% while the ideal SOC center value is 50%, the system will adjust the planned charge / discharge power baseline for the last few hours of the following day in a smooth adjustment manner. The goal of the adjustment is to bring the termination SOC back to near the preset ideal SOC center value. For example, by fine-tuning the planned power for the last three hours, the termination SOC can be gradually adjusted from 70% to 52%. The planned charge / discharge power sequence after SOC optimization and adjustment is finally determined as the planned charge / discharge power baseline for the energy storage system at the hourly scale of the following day and is issued to the intraday rolling level.

[0027] Optionally, a specific formula framework can be introduced when constructing the comprehensive objective function. The comprehensive objective function aims to minimize the total cost, and its expression is:

[0028] in: This represents a time index, corresponding to each hour in the next 24 hours. Represents the moment Time-of-use electricity pricing. Represents the moment The power exchanged between the photovoltaic energy storage system and the power grid is represented by a positive value indicating that electricity is purchased from the grid, and a negative value indicating that electricity is sold to the grid. Weighting coefficients representing the cost of life loss. Represents the energy storage system at any time The lifespan depreciation cost function, this function is the energy storage system at time [time value missing]. State of charge and energy storage systems at all times absolute value of charging and discharging power The function. Represents the energy storage system at any time The planned charge and discharge power is the optimization variable. Solving this formula requires satisfying a series of constraints, including energy storage system power balance, energy storage system state of charge (SOC) changes, upper and lower limits of SOC, and upper and lower limits of power. Updating the energy storage system's baseline SOC can be understood as a closed-loop correction step. This process ensures that the planned charge and discharge power baseline, obtained based on 24-hour prediction and optimization, is self-consistent in terms of energy accumulation and allows the energy storage system to return to a desired baseline point at the end of the cycle. This avoids systematic drift of the energy storage system's SOC during daily operation, such as continuous accumulation towards a high or low SOC, thus ensuring the long-term reliability and regulation capability of the energy storage system. Smooth adjustments typically involve applying a small constant power offset or a linearly varying power offset in the last few time periods to ensure that changes to the day-ahead plan are gradual rather than abrupt.

[0029] In one embodiment of the present invention, see [reference] Figure 3 In the implementation of the intraday rolling tier, the system triggers a rolling optimization every 15 minutes, acquiring ultra-short-term photovoltaic (PV) power forecast data for the next four hours with a time resolution of five minutes, driven by real-time meteorological data. The system reads the actual state of charge (SOC) of the energy storage system at the current moment, and the baseline of the planned charge / discharge power of the energy storage system at the hourly scale of the next day for the corresponding time period from the current moment to the next hour. The system calculates the difference between the ultra-short-term PV power forecast data for the next hour starting from the current moment and the PV power forecast sequence for the corresponding time period in the previous day's forecast tier; this difference constitutes the PV power forecast deviation sequence. With the objective of minimizing the fluctuation in the total output power of the power plant caused by the PV power forecast deviation sequence, the system optimizes the solution for the power adjustment of the energy storage system in the next four 15-minute time periods, while satisfying the power and energy constraints of the energy storage system. The system adds the average value of the planned charge / discharge power baseline of the energy storage system at the hourly scale of the next day to the power adjustment amount for that time period, forming the rolling adjustment power command for the energy storage system at the next 15-minute scale for that time period.

[0030] After calculating the power adjustment amounts of the energy storage system for the next four 15-minute time periods, the system also performs a feasibility verification step for the rolling power adjustment command. This step starts with the actual state of charge (SBC) of the energy storage system at the current moment and simulates the trajectory of the SBC change after executing the rolling power adjustment command on a 15-minute scale for each time period. The system checks whether the SBC trajectory remains within the preset safe operating range. If the SBC trajectory exceeds the safe operating range at any time period, the rolling power adjustment command on a 15-minute scale for the energy storage system in the period that caused the exceedance and subsequent periods is corrected by limiting the power output to bring the SBC back to the safe operating range. The system then issues the command sequence after feasibility verification and correction as the final rolling power adjustment command for the energy storage system on a 15-minute scale.

[0031] In practical implementation, a daily rolling hierarchy of a photovoltaic energy storage collaborative control system considering multiple time scales operates as follows: The system automatically triggers a rolling optimization calculation every 15 minutes, such as 10:00:00 or 10:15:00. Upon triggering, the system immediately acquires ultra-short-term photovoltaic power forecast data for the next four hours, with a time resolution of five minutes, driven by real-time meteorological data. For example, it predicts the photovoltaic power value every five minutes between 10:15 and 14:15. Simultaneously, the system reads the actual state of charge of the energy storage system at the current time, such as 10:00, and also reads the planned charge and discharge power baseline for the energy storage system at the hourly scale of the next day, corresponding to the period from the current time to the next hour (i.e., from 10:00 to 11:00), from the results issued by the previous day's forecast hierarchy. This baseline represents the planned power value at the hourly level.

[0032] In some embodiments, the system calculates the power deviation sequence for the next hour starting from the current moment. During calculation, the system aligns and compares the predicted values ​​of the ultra-short-term photovoltaic (UVP) power forecast data at the corresponding five-minute time points with the hourly predicted values ​​of the UVP power forecast sequence generated in the day-ahead forecast level for the corresponding time period, after linear interpolation to the same time resolution. The difference between the two constitutes the UVP power forecast deviation sequence. For example, at 10:20, the UVP forecast power is 650 kW, while the corresponding day-ahead forecast interpolation is 620 kW, resulting in a deviation of +30 kW. The system establishes a rolling optimization model with the objective of minimizing the fluctuation in the total power output of the power plant caused by the UVP power forecast deviation sequence. Under the premise of satisfying the maximum charge / discharge power limit of the energy storage system and the upper and lower limits of the energy storage system's state of charge, the rolling optimization model optimizes the power adjustment of the energy storage system for each of the four consecutive fifteen-minute time periods, from 10:15 to 11:15. The power adjustment is a power value used to compensate for the forecast deviation. The rolling power regulation command for the energy storage system in the next 15-minute time period is formed by superimposing the average of the planned charge and discharge power baseline of the energy storage system at the hourly scale of the following day onto the optimized power regulation amount for that time period. For example, in the time period from 10:15 to 10:30, if the average power of the day-ahead baseline in that hour is 50 kW of discharge and the optimized power regulation amount is 20 kW of charging, then the final rolling power regulation command is 30 kW of discharge.

[0033] After calculating the power regulation of the energy storage system for the next four 15-minute time periods and forming a preliminary rolling power regulation command, the system executes a feasibility verification step for the rolling power regulation command. This step starts with the actual state of charge (SOC) of the energy storage system at the current moment, for example, if the current SOC is 60%, and simulates the trajectory of the SOC change after executing the rolling power regulation command at the next 15-minute scale. The simulation considers the charging and discharging efficiency of the energy storage system, for example, a charging efficiency of 95% and a discharging efficiency of 95%. The system checks whether the trajectory of the SOC change remains within a preset safe operating range, for example, a SOC ranging from 20% to 90%. If the SOC change trajectory exceeds the safe operating range at any time period, the rolling power regulation command for the energy storage system at the next 15-minute scale for the time period that caused the exceedance and subsequent time periods is corrected by power limiting. The correction aims to return the state of charge (SOC) to a safe operating range. For example, if simulations show that the SOC drops to 18% at 11:00 after executing the command, below the lower limit of 20%, the system will proportionally reduce the discharge power command value for the period from 10:45 to 11:00 and subsequent periods, or increase the charging power command value. The command sequence, after feasibility verification and correction, will be used as the final rolling power adjustment command issued to the real-time control level for the energy storage system on a 15-minute timescale. Optionally, a specific optimization objective function can be introduced when solving for the power adjustment amount in rolling optimization. The optimization objective function aims to minimize the combined cost of plan tracking deviation and energy storage system operation costs, and its expression is:

[0034] in: The index represents the time period that is scrolled forward, with values ​​from 1 to 4, corresponding to the next four 15-minute time periods. Representative at The photovoltaic power prediction deviation for the time period is a known input. Representative at The power regulation of the energy storage system to be solved during the time period. A positive value indicates that the discharge is increased or the charging is reduced to offset the positive deviation, and a negative value indicates that the charging is increased or the discharge is reduced to offset the negative deviation. and These are weighting coefficients, used to balance the accuracy of tracking error and the magnitude of power regulation in the energy storage system, respectively. Solving this optimization problem requires that the power regulation of the energy storage system in each time period does not exceed its maximum charge / discharge power limit, and that the simulated state-of-charge trajectory of the energy storage system meets energy constraints.

[0035] It is understandable that verifying the feasibility of rolling power regulation commands is a necessary safety precaution. This process adds a forward simulation before issuing the commands calculated in the rolling optimization. The forward simulation, based on an accurate charge / discharge model of the energy storage system and its current state of charge, predicts the evolution of the system's state after executing the command sequence. This step can identify command combinations that might lead to overcharging or over-discharging of the energy storage system in advance, allowing for preventative correction before command issuance. Power limiting correction typically employs proportional scaling or priority adjustment methods to ensure that the corrected commands are technically feasible and as close as possible to the original optimization objective. Through this step, the rolling power regulation commands output at the intraday rolling level are not only theoretically optimal but also safely executable in practical engineering.

[0036] In one embodiment of the present invention, in the implementation of the real-time control level, the system collects the DC-side current and voltage of the photovoltaic array at millisecond intervals to calculate the current actual output power of the photovoltaic array. The system collects the DC-side current, voltage, and DC bus voltage of the energy storage system at the same intervals to obtain the current actual operating status of the energy storage system. The system receives the latest rolling power adjustment command for the energy storage system at the next fifteen-minute timescale from the intraday rolling level. The system calculates the difference between the rolling power adjustment command for the energy storage system at the next fifteen-minute timescale and the current actual output power of the energy storage system as the power tracking error. The system uses the power tracking error and the offset of the DC bus voltage from the reference voltage as inputs to the controller, and calculates the real-time control signal of the energy storage converter at the second or sub-second timescale using a proportional-integral-derivative control algorithm. This real-time control signal is used to adjust the duty cycle of the power switching devices in the energy storage converter.

[0037] The system also performs a dynamic limiting protection step for real-time control signals. This step monitors the current actual state of charge (SOC), temperature, and DC bus voltage of the energy storage system in real time. Based on the current SOC, the system retrieves the current allowable absolute power limit from a preset maximum allowable charge / discharge power curve corresponding to different SOC ranges. Based on the energy storage system's temperature, the system retrieves the power derating factor at the current temperature from a preset temperature-power derating factor table. The system multiplies the current allowable absolute power limit by the power derating factor to obtain the dynamically adjusted real-time power limit. The system then limits the power demand calculated from the energy storage converter's real-time control signals on a second- or sub-second scale to within the dynamically adjusted real-time power limit.

[0038] In practical implementation, a real-time control layer of a photovoltaic energy storage collaborative control system considering multiple time scales operates as follows: The system synchronously acquires the current and voltage measurements of the DC side of the photovoltaic array at millisecond intervals, for example, 1000 times per second. The current actual output power of the photovoltaic array is calculated by instantaneous multiplication. For example, if the DC side current is 200 amperes and the voltage is 500 volts at a certain moment, the current actual output power is calculated to be 100 kilowatts. The system acquires the current, voltage, and DC bus voltage values ​​of the energy storage system at the same acquisition interval, thereby obtaining the current actual operating status of the energy storage system, including the current actual output power, the estimated current actual state of charge, and the actual DC bus voltage. The system receives the latest rolling power adjustment commands for the energy storage system at the next fifteen-minute time scale from the intraday rolling level. For example, if a command is received at 10:15:00 requiring the energy storage system to discharge at a power of 50 kilowatts between 10:15:00 and 10:30:00.

[0039] In some embodiments, the system generates control signals by comparing a set target with the current state. The system calculates the difference between the rolling power adjustment command and the current actual output power of the energy storage system on a 15-minute timescale as the power tracking error. For example, if the rolling power adjustment command is a discharge of 50 kW, and the current actual output power is 45 kW, the power tracking error is +5 kW. The system also calculates the offset of the actual DC bus voltage from the reference DC bus voltage value. For example, if the reference DC bus voltage is 750 volts and the actual value is 745 volts, the offset is -5 volts. The system uses the power tracking error and the offset of the DC bus voltage from the reference voltage as inputs to the controller, which employs a proportional-integral-derivative (PID) control algorithm. The proportional-integral-derivative (PID) control algorithm calculates the power tracking error and voltage offset to obtain the real-time control signal of the energy storage converter on a second or sub-second scale. The real-time control signal is usually expressed as a duty cycle command of a pulse width modulation signal, which is used to directly adjust the on and off time ratio of power switching devices such as insulated-gate bipolar transistors in the energy storage converter.

[0040] In its implementation, the system also performs a dynamic limiting protection step for real-time control signals. This step monitors the current actual state of charge (SOC) of the energy storage system, its internal temperature, and the DC bus voltage in real time. Based on the SOC, the system retrieves the current allowable absolute power limit from preset maximum allowable charge / discharge power curves corresponding to different SOC ranges. For example, when the SOC is 85%, the maximum allowable charging power is -80 kW (the negative sign indicates charging), and the maximum discharging power is 100 kW, as shown in Table 1. Table 1: Relationship between State of Charge and Maximum Allowable Charge / Discharge Power of Energy Storage Systems State of charge range (%) Maximum permissible charging power (kW) Maximum permissible discharge power (kW) 0~20 -30 150 20~80 -100 100 80~90 -80 80 90~100 -20 20 The system retrieves the power derating factor for the current temperature from a preset temperature-power derating factor table based on the internal temperature of the energy storage system. For example, when the energy storage system temperature is 40 degrees Celsius, the power derating factor is found to be 0.9. The system multiplies the current allowable absolute power limit by the power derating factor to obtain the dynamically adjusted real-time power limit. For example, if the current allowable discharge power is 100 kW, multiplying by the derating factor of 0.9 yields a dynamically adjusted real-time discharge power limit of 90 kW. The system strictly limits the demand power calculated by the proportional-integral-derivative (PID) control algorithm within the dynamically adjusted real-time power limit. If the calculated demand discharge power is 95 kW, exceeding the 90 kW limit, the final real-time control signal will clamp the corresponding demand power to 90 kW. Optionally, the PID control algorithm can be designed as a dual-closed-loop structure. The inner loop is a current loop, and the outer loop is a power loop or voltage loop. The compensator output formula for the outer loop power control can be expressed as:

[0041] in: This represents the calculated pulse width modulation duty cycle command value. This represents the power point tracking error, which is the difference between the rolling power adjustment command of the energy storage system on a 15-minute timescale and the current actual output power of the energy storage system. This represents the proportional control gain coefficient. This represents the integral control gain coefficient. This represents the differential control gain coefficient. This formula describes how the power error is transformed into a direct control quantity for the energy storage converter's switching transistors through a combination of proportional, integral, and derivative operations. The integral stage eliminates steady-state errors, while the derivative stage predicts error trends to improve dynamic response. It can be understood that the dynamic limiting protection step is a key mechanism to ensure the real-time safe operation of the energy storage system. The maximum allowable charge / discharge power curve is formulated based on the electrochemical characteristics of the energy storage battery, typically limiting power at very high or very low states of charge to protect battery health. The temperature-power derating factor table considers the impact of temperature on battery internal resistance and heat dissipation; power needs to be reduced in high-temperature environments to prevent overheating. This step, by dynamically limiting the original demand power output from the proportional-integral-derivative control algorithm in real time, ensures that the power command corresponding to the real-time control signal acting on the energy storage converter is always within the safe operating range of the energy storage system under the current state of charge and temperature, thus ensuring operational safety and lifespan of the equipment while pursuing control performance.

[0042] In one embodiment of the present invention, the system further includes a step of establishing information feedback between the day-ahead forecast level and the intraday rolling level. After each rolling optimization calculation at the intraday rolling level is completed, the system records the statistical characteristics of the photovoltaic power prediction deviation sequence, which includes the average deviation and the standard deviation of the deviation. The system uploads the statistical characteristics of the photovoltaic power prediction deviation sequence to the day-ahead forecast level. Before the next optimization calculation at the day-ahead forecast level, the system uses the statistical characteristics of multiple historically uploaded photovoltaic power prediction deviation sequences to correct the internal parameters of the power conversion model or apply prediction error compensation to the photovoltaic power prediction sequence. Based on the corrected model or the compensated prediction sequence, the system re-solves the comprehensive objective function to generate a new round of planned charge and discharge power baselines for the energy storage system at the hourly scale of the next day. The specific process of using the statistical characteristics of multiple historically uploaded photovoltaic power prediction deviation sequences to correct the internal parameters of the power conversion model or apply prediction error compensation to the photovoltaic power prediction sequence includes: the system extracts the statistical characteristics of photovoltaic power prediction deviation sequences uploaded at the same time within a set number of days in the past, and calculates their mean and trend. If the mean of the average deviation remains non-zero, a systematic bias is identified in the power conversion model, and the conversion coefficients from meteorological data to photovoltaic power in the power conversion model are reverse-calibrated. If the standard deviation of the deviation shows an increasing trend, the uncertainty of numerical weather prediction is identified as increasing, and a prediction error band based on the mean of the standard deviation is superimposed on the photovoltaic power prediction sequence. When optimizing the comprehensive objective function, a robust optimization method is used to process the photovoltaic power prediction sequence with the superimposed prediction error band to enhance the adaptability of the energy storage system to the prediction error at the hourly scale of the next day.

[0043] In practice, after each rolling optimization calculation at the daily rolling level, the system records the statistical characteristics of the photovoltaic power prediction deviation sequence. These characteristics include the average deviation and standard deviation of the photovoltaic power prediction deviation sequence within a one-hour window. The system uploads these statistical characteristics, along with the corresponding timestamp information, to the data area accessible at the day-ahead prediction level, for example, uploading statistical characteristic data for 96 time points daily (once every 15 minutes). Before the next optimization calculation at the day-ahead prediction level, the system calls upon the statistical characteristics of multiple historically uploaded photovoltaic power prediction deviation sequences, using these historical statistical characteristics to correct the internal parameters of the power conversion model or apply prediction error compensation to the photovoltaic power prediction sequence. Based on the corrected model or compensated prediction sequence, the system re-solves the comprehensive objective function to generate a new round of planned charge and discharge power baselines for the energy storage system at the hourly scale of the following day.

[0044] In some embodiments, the process of correcting the internal parameters of the power conversion model or applying prediction error compensation to the photovoltaic power prediction sequence by utilizing the statistical characteristics of multiple historically uploaded photovoltaic power prediction deviation sequences is executed according to the following logic: The system extracts the statistical characteristics of photovoltaic power prediction deviation sequences uploaded at the same time within a set number of days, such as the past 7 days. The system calculates the mean and trend of these historical statistical characteristics over a period of time. For example, for the time of 10:00 AM each day, the average deviation of the photovoltaic power prediction deviation sequence calculated between 10:00 AM and 11:00 AM over the past 7 days is extracted, forming a set containing 7 values. The average value of this set is then calculated to be +5.2 kW, and its trend is analyzed to be gradual. The system executes a judgment and corresponding correction strategy. If the mean of the average deviation of the photovoltaic power prediction deviation sequence is consistently non-zero, for example, if the mean of the average deviation at the same time for several consecutive days is significantly positive, then it is determined that there is a systematic bias in the power conversion model. The system performs reverse calibration on the conversion coefficients from meteorological data to photovoltaic power in the power conversion model. For example, the conversion efficiency coefficient used to calculate power in the model is adjusted from the original value of 0.18 to 0.182 based on the average deviation. If the standard deviation of the photovoltaic power prediction deviation sequence shows an increasing trend, for example, if the standard deviation sequence for the same time in the past week is [8,9,10,12,11,13,15] kW and shows an upward trend, then the uncertainty of numerical weather prediction is considered to have increased. The system superimposes a prediction error band based on the mean of the standard deviation on the photovoltaic power prediction sequence. When optimizing the comprehensive objective function, a robust optimization method is used to process the photovoltaic power prediction sequence with the superimposed prediction error band to enhance the adaptability of the energy storage system to the prediction error at the hourly scale of the next day.

[0045] In practical implementation, the quantification and compensation of prediction errors can be achieved through a specific mechanism. The system maintains a historical deviation characteristic table to store and analyze data from intraday rolling hierarchical feedback. This table can record key information, such as the statistical characteristics of the photovoltaic power prediction deviation sequence over the past N days for each 15-minute period, as shown in Table 2: Table 2: Historical Characteristics Analysis of Photovoltaic Power Prediction Deviation Historical Dates Average deviation of photovoltaic power forecast (kW) Photovoltaic power prediction deviation standard deviation (kW) Numerical weather forecast irradiance (W / m²) 2026-04-02 +3.5 8.2 650 2026-04-03 -1.2 6.5 720 2026-04-04 +6.8 12.1 580 2026-04-05 +4.1 9.7 610 2026-04-06 +5.5 11.3 590 2026-04-07 +7.0 14.5 560 2026-04-08 +4.8 10.8 630 Optionally, an adaptive boundary model based on the historical deviation standard deviation can be used when constructing the prediction error band. The prediction error band defines the possible upper and lower fluctuation range of the photovoltaic power prediction sequence at each time step. Its upper boundary... and lower boundary It can be represented as:

[0046]

[0047] in: Represents a specific hour in the coming day. Represents the moment The original photovoltaic power prediction sequence values. Represents the value calculated based on historical data at a given time. The mean of the standard deviation of the photovoltaic power prediction bias for the nearby time period. and This is a positive coefficient used to adjust the conservatism of the error band, typically set based on the tolerance for uncertainty. This error band provides the uncertainty set for robust optimization, where the optimization process seeks the optimal conditions for photovoltaic power output. A low-cost, feasible baseline for the planned charge and discharge power of an energy storage system, applicable to any value within the specified range. This can be understood as a closed-loop learning and improvement process utilizing historical feedback information before the next optimization calculation at the day-ahead forecast level. Internal parameter correction of the power conversion model directly addresses the model's systematic errors, a calibration process aimed at improving the average accuracy of long-term forecasts. Applying forecast error compensation to the photovoltaic power forecast sequence, particularly constructing a forecast error band and combining it with robust optimization, is a strategy to address the random uncertainty of forecasts. Robust optimization methods consider worst-case photovoltaic power fluctuations in the optimization model, ensuring that the calculated baseline for the planned charge and discharge power of the energy storage system at the hourly scale of the following day still meets constraints (such as power balance and ensuring the energy storage system's state of charge does not exceed limits) even when actual photovoltaic power fluctuates within the error band, thus enhancing the robustness of the day-ahead plan. These two approaches complement each other, improving the accuracy and reliability of the day-ahead plan by reducing systematic bias and addressing random fluctuations, respectively, using operational feedback data.

[0048] In one embodiment of the invention, the system further includes coordination, synchronization, and command switching management for multi-timescale control processes. A unified absolute time reference is set within the timescale hierarchical framework, and the calculation and command issuance times of each step at the day-ahead forecast level, intraday rolling level, and real-time control level are aligned with this absolute time reference. At the command issuance time of the intraday rolling level, the system checks the actual command execution completion rate of the previous 15-minute rolling cycle. If the actual command execution completion rate is lower than a preset threshold, the execution deviation is accumulated into the currently calculated rolling adjustment power command; this execution deviation is the cumulative difference between the planned command and the actual executed power of the previous cycle. At the real-time control level, the system sets up a command smoothing transition mechanism. When an update to the rolling adjustment power command from the energy storage system at the intraday rolling level for the next 15-minute scale is detected, the real-time control level smoothly transitions the currently tracked command value to the new command value over several control cycles, avoiding power command jumps.

[0049] In practical implementation, a photovoltaic energy storage collaborative control system considering multiple time scales includes the coordination, synchronization, and command switching management of control processes across multiple time scales. A unified absolute time reference is set within the time scale hierarchical framework, typically using Coordinated Universal Time (UTC) or a synchronized local clock. The calculation and command issuance times for each stage of the day-ahead forecasting level, the intraday rolling level, and the real-time control level are all aligned with the absolute time reference. For example, the absolute time reference indicates that 0:00 daily, triggering the optimization calculation of the day-ahead forecasting level; 0:00, 6:00, 12:00, and 18:00 daily, triggering the day-ahead plan update based on the latest weather forecast; every 15 minutes indicated by the absolute time reference (e.g., 10:00:00, 10:15:00), triggering the rolling optimization and command issuance of the intraday rolling level; and the real-time control level uses the millisecond-level pulse of the absolute time reference as the starting point for the control cycle timing.

[0050] In some embodiments, at the time of instruction issuance at the intraday rolling level, the system checks the actual instruction execution completion rate of the previous 15-minute rolling cycle. The system obtains the actual power sequence executed by the energy storage system in the previous cycle from the real-time control level or the data acquisition and monitoring control system, and compares it with the rolling adjustment power instruction sequence for the energy storage system at the next 15-minute scale issued in the previous cycle at the intraday rolling level. The system calculates the cumulative deviation between the actual executed power and the planned instruction power to assess the actual instruction execution completion rate. If the actual instruction execution completion rate is lower than a preset threshold, for example, a preset threshold of 90% and a calculated actual instruction execution completion rate of 80%, the system adds the execution deviation to the currently calculated rolling adjustment power instruction. The execution deviation is the cumulative difference between the planned instruction and the actual executed power in the previous cycle. For example, if the planned total discharge energy in the previous cycle was 15 kWh and the actual total discharge energy was 12 kWh, resulting in a 3 kWh undischarged energy deviation, the system can convert this 3 kWh energy into a success rate difference, distribute it evenly across the next one or several 15-minute time periods, and add it to the newly calculated rolling adjustment power instruction.

[0051] In practical implementation, a smooth transition mechanism for commands is set at the real-time control level. When a rolling power adjustment command update for the energy storage system on a 15-minute timescale is detected from the intraday rolling level, for example, if a new rolling power adjustment command is received at 10:15:00 requiring a constant power discharge of 60 kW for the next 15 minutes, while the previously tracked command value was 50 kW, the real-time control level will not immediately jump the tracked target from 50 kW to 60 kW. Instead, it will smoothly transition the currently tracked command value to the new command value over several control cycles. The smooth transition can be achieved through a first-order inertial filter or a ramp function. For example, within 1 second, the command value can be linearly increased from 50 kW to 60 kW at a rate of 2 kW per second, thereby avoiding the impact of power command jumps on the energy storage converter and the power grid. Optionally, the command smooth transition can be calculated using a discrete form with a first-order low-pass filter, and the smoothed command value... We obtain it from the following formula:

[0052] in: Representative at the The power command value, after a smooth transition, output during each control cycle is used to replace the original jump command value as the input of the proportional-integral-derivative controller. This represents the original value of the newly issued rolling adjustment power command from the intraday rolling level, such as 60 kilowatts. Representative at the The power command value that has been smoothed over a control cycle will be the old command value the instant a new command is received, for example, 50 kilowatts. This is a smoothing coefficient, with a value between 0 and 1, determining the speed of the transition process. For example, a value of 0.1 indicates a slower transition, while a value of 0.5 indicates a faster transition. It can be understood that setting a unified absolute time base is fundamental to ensuring multi-level collaborative operation. Absolute time base alignment ensures a definite time relationship between the daily forecast level's plan refresh, the intraday rolling level's cycle initiation, and the real-time control level's control cycle, avoiding instruction timing errors or logical conflicts caused by clock asynchrony at different levels. Checking the actual instruction execution completion rate of the previous cycle and accumulating the deviation is a feedforward-feedback composite mechanism. This mechanism incorporates the deviation between planning and execution caused by hardware limitations, response delays, or sudden constraints as a correction factor into the new round of rolling optimization, thereby closing the "planning-execution-deviation compensation" loop on the time scale and preventing deviation accumulation. The instruction smoothing transition mechanism is a protective measure for the control level. A sudden change in power command requires the energy storage converter to output a huge rate of current change within a very short time, which can damage power electronic devices and cause drastic fluctuations in DC bus voltage. Smooth transition transforms a sudden command into a continuously changing command, allowing the energy storage converter to adjust its output smoothly, thus improving the transient stability of the system and the safety of the equipment.

[0053] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A photovoltaic energy storage collaborative control system considering multiple time scales, characterized in that, The system establishes a time-scale hierarchical framework comprising day-ahead forecasting, intraday rolling, and real-time control. Each level in the time-scale hierarchical framework corresponds to a different control cycle and execution frequency, specifically including: At the forecast level, the system receives the dispatch command curve issued by the power grid and the photovoltaic power forecast sequence for the next 24 hours of the photovoltaic power plant. With the goal of minimizing the cost of electricity purchase and the cycle loss of energy storage equipment, the system formulates the baseline of planned charge and discharge power of the energy storage system at the hourly scale of the next day. At the intraday rolling level, ultra-short-term photovoltaic power prediction data for the next four hours of the photovoltaic power station is obtained. Based on the planned charge and discharge power baseline of the energy storage system at each hourly scale of the next day, with the goal of smoothing out the deviation between the actual operation and the previous day's plan, rolling adjustment power instructions for the energy storage system at the next fifteen-minute scale are formulated. At the real-time control level, the current actual output power of the photovoltaic array and the current actual operating status of the energy storage system are collected. With the goal of tracking the rolling power adjustment command of the energy storage system on a 15-minute scale and ensuring the stability of the DC bus voltage, real-time control signals of the energy storage converter on a second or sub-second scale are generated.

2. The photovoltaic energy storage collaborative control system considering multiple time scales according to claim 1, characterized in that, At the day-ahead forecasting level, the system receives the dispatch command curve from the power grid and the photovoltaic power forecast sequence for the next 24 hours of the photovoltaic power plant. Using minimizing electricity purchase costs and energy storage equipment cycle losses as the comprehensive objective function, it establishes the planned charge and discharge power baseline for the energy storage system at each hourly scale of the following day, including: Receive and parse the grid's 96-point dispatch instructions for the next day, wherein the dispatch instruction curve includes the planned output power at the grid connection point; Numerical weather forecast data is acquired, and a photovoltaic power prediction sequence updated hourly for the next 24 hours is generated using a power conversion model; Read the rated capacity, current state of charge, maximum charge / discharge power limit, and charge / discharge efficiency parameters of the energy storage system; The comprehensive objective function is constructed, which includes the electricity cost item for interaction with the grid calculated based on time-of-use pricing, and the energy storage lifetime depreciation cost item calculated based on the number of charge-discharge cycles and depth. Under the constraints of the rated capacity, current state of charge, maximum charge / discharge power limit, and charge / discharge efficiency parameters, the optimal solution of the comprehensive objective function for the next 24 hours is solved with a time resolution of one hour. The optimal solution is a series of power values, forming the planned charge / discharge power baseline of the energy storage system at the hourly scale of the next day.

3. A photovoltaic energy storage collaborative control system considering multiple time scales according to claim 2, characterized in that, The system also includes a process for updating the energy storage system's reference state of charge: After obtaining the planned charge and discharge power baseline of the energy storage system at each hourly scale of the next day, the current state of charge at the start time of the energy storage system is used as the initial value, and energy accumulation calculation is performed according to the planned charge and discharge power baseline of the energy storage system at each hourly scale of the next day. During the energy accumulation calculation, the charging power and the discharging power are calculated separately according to the charging and discharging efficiency parameters; The predicted state of termination of charge of the energy storage system after the end of the next 24 hours is calculated. The current state of charge at the start of the next day is compared with the ending state of charge. If the ending state of charge deviates from the preset ideal state of charge center value, the planned charging and discharging power baseline for the last few hours of the next day is corrected in a smooth adjustment manner so that the ending state of charge returns to the vicinity of the preset ideal state of charge center value. The planned charge and discharge power sequence, after being optimized and corrected for the state of charge, is finally determined as the baseline of the planned charge and discharge power of the energy storage system at the hourly scale of the next day.

4. A photovoltaic energy storage collaborative control system considering multiple time scales according to claim 2, characterized in that, The method involves acquiring ultra-short-term photovoltaic power forecast data for the next four hours of the photovoltaic power plant at the intraday rolling level. Based on the planned charge and discharge power baseline of the energy storage system at each hourly scale of the following day, and with the goal of mitigating the deviation between actual operation and the previous day's plan, rolling adjustment power commands for the energy storage system at the next fifteen-minute scale are formulated, including: A rolling optimization is triggered every 15 minutes to obtain the ultra-short-term photovoltaic power prediction data for the next four hours with a time resolution of five minutes, driven by real-time meteorological data; Read the actual state of charge of the energy storage system at the current moment, and the baseline of the planned charge and discharge power of the energy storage system at each hourly scale of the next day from the current moment to the next hour; Calculate the difference between the ultra-short-term photovoltaic power prediction data and the photovoltaic power prediction sequence for the corresponding time period in the day-ahead prediction level for the next hour starting from the current moment; the difference constitutes the photovoltaic power prediction deviation sequence. With the goal of minimizing the fluctuation of the total output power of the power plant caused by the photovoltaic power prediction deviation sequence, and under the premise of satisfying the power and energy constraints of the energy storage system, the power adjustment of the energy storage system in the next four 15-minute time periods is optimized. The average value of the planned charge and discharge power baseline of the energy storage system at each hourly scale on the next day within the corresponding 15-minute time period is added to the power adjustment amount of the time period to form the rolling adjustment power command of the energy storage system at the next 15-minute scale.

5. A photovoltaic energy storage collaborative control system considering multiple time scales according to claim 4, characterized in that, The system also includes a feasibility verification step for the rolling power adjustment command: After calculating the power adjustment amount of the energy storage system for the next four 15-minute time periods, starting from the actual state of charge of the energy storage system at the current moment, the trajectory of the change of the state of charge of the energy storage system after executing the rolling power adjustment command of the energy storage system at the next 15-minute scale is simulated and calculated time by time. Check whether the trajectory of the change in the state of charge is always within the preset safe operating range; If the trajectory of the change in the state of charge exceeds the safe operating range at any time, the rolling adjustment power command of the energy storage system in the next fifteen-minute time scale for the time period that caused the exceedance and the subsequent time period will be corrected by power limiting, and the correction direction is to bring the state of charge back to the safe operating range. The sequence of instructions, after feasibility verification and correction, will be used as the final execution command for the energy storage system to issue rolling power adjustment commands on a 15-minute timescale.

6. A photovoltaic energy storage collaborative control system considering multiple time scales according to claim 4, characterized in that, At the real-time control level, the current actual output power of the photovoltaic array and the current actual operating status of the energy storage system are collected. The goal is to track the rolling power adjustment commands of the energy storage system over the next fifteen minutes and ensure DC bus voltage stability. This generates real-time control signals for the energy storage converter on a second- or sub-second scale, including: The DC-side current and voltage of the photovoltaic array are collected at millisecond intervals, and the current actual output power of the photovoltaic array is calculated. The current, voltage, and DC bus voltage values ​​of the energy storage system are collected at the same interval to obtain the current actual operating status of the energy storage system. Receive the latest rolling power adjustment command issued by the energy storage system at the intraday rolling level for the next fifteen minutes; The difference between the rolling adjustment power command of the energy storage system on a 15-minute timescale and the current actual output power of the energy storage system is calculated as the power tracking error. Using the power tracking error and the offset of the DC bus voltage from the reference voltage as the input to the controller, the real-time control signal of the energy storage converter at the second or sub-second scale is calculated by the proportional-integral-derivative control algorithm. The real-time control signal is used to adjust the duty cycle of the power switching devices in the energy storage converter.

7. A photovoltaic energy storage collaborative control system considering multiple time scales according to claim 6, characterized in that, The system also includes a dynamic amplitude limiting protection step for real-time control signals: Real-time monitoring of the current actual state of charge, temperature, and DC bus voltage of the energy storage system; Based on the current actual state of charge, the current allowable absolute power limit is obtained from the preset maximum allowable charge and discharge power curves corresponding to different state of charge intervals. Based on the temperature of the energy storage system, the power derating factor at the current temperature is obtained from the preset temperature-power derating factor table; Multiply the current allowed absolute power limit value by the power derating factor to obtain the dynamically adjusted real-time power limit value; The required power calculated from the real-time control signal of the energy storage converter at the second or sub-second scale is limited to the range of the dynamically adjusted real-time power limit value.

8. A photovoltaic energy storage collaborative control system considering multiple time scales according to claim 1, characterized in that, The system also includes a step of establishing information feedback between the day-ahead forecast level and the intraday rolling level: After each rolling optimization calculation at the intraday rolling level is completed, the statistical characteristics of the photovoltaic power prediction deviation sequence are recorded. The statistical characteristics include the average deviation and the standard deviation of the deviation. The statistical characteristics of the photovoltaic power prediction deviation sequence are uploaded to the day-ahead prediction level; Before the next optimization calculation is performed at the current prediction level, the internal parameters of the power conversion model are corrected or the prediction error compensation is applied to the photovoltaic power prediction sequence by utilizing the statistical characteristics of multiple photovoltaic power prediction deviation sequences uploaded in the past. Based on the corrected model or the compensated prediction sequence, the comprehensive objective function is solved again to generate a new round of planned charge and discharge power baselines for the energy storage system at the hourly scale of the next day.

9. A photovoltaic energy storage collaborative control system considering multiple time scales according to claim 8, characterized in that, The step of using statistical characteristics of multiple historically uploaded photovoltaic power prediction deviation sequences to correct the internal parameters of the power conversion model or to apply prediction error compensation to the photovoltaic power prediction sequences includes: Extract the statistical characteristics of the photovoltaic power prediction deviation sequence uploaded at the same time within a set number of days in the past, and calculate its mean and trend; If the mean of the average deviation is consistently non-zero, it is determined that there is a systematic deviation in the power conversion model, and the conversion coefficient from meteorological data to photovoltaic power in the power conversion model is reverse-calibrated. If the standard deviation of the deviation shows an increasing trend, it is determined that the uncertainty of numerical weather forecast has increased, and a prediction error band based on the mean of the standard deviation of the deviation is superimposed on the photovoltaic power prediction sequence. When optimizing the solution of the comprehensive objective function, a robust optimization method is used to process the photovoltaic power prediction sequence superimposed with the prediction error band, so as to enhance the adaptability of the energy storage system to the prediction error at the hourly scale of each day.

10. A photovoltaic energy storage collaborative control system considering multiple time scales according to claim 1, characterized in that, The system also includes coordination, synchronization, and instruction switching management for multi-timescale control processes: A unified absolute time reference is set in the time-scale hierarchical framework, and the calculation and instruction issuance times of each of the day-ahead forecast level, intraday rolling level and real-time control level are all aligned with the absolute time reference. At the moment the instructions are issued at the intraday rolling level, check the actual instruction execution completion rate of the previous 15-minute rolling cycle; If the actual execution completion rate of the instruction is lower than the preset threshold, the execution deviation will be added to the currently calculated rolling adjustment power instruction. The execution deviation is the cumulative difference between the previous round of planned instructions and the actual execution power. At the real-time control level, a smooth transition mechanism for commands is set up. When the energy storage system detects an update of the rolling adjustment power command from the intraday rolling level on a 15-minute timescale, the real-time control level will smoothly transition the currently tracked command value to the new command value within several control cycles to avoid power command jumps.

Images