Power management system and power management method

Abstract

A power management system is provided, including an inverter and a computing device. The inverter is operatively coupled to a photovoltaic array, a power grid, and a battery. The computing device is configured to: control the inverter to operate in either a rule-based control mode or a scheduling mode based on a cost evaluation table; periodically collect measurement data from the inverter; calculate a first future cost corresponding to the rule-based control mode and a second future cost corresponding to the scheduling mode based on the measurement data; and update the cost evaluation table based on the first future cost and the second future cost. The cost evaluation table is configured to store cost values ​​corresponding to the respective operating modes for each of a plurality of time intervals.

Description

Technical Field

[0001] This invention relates to energy management, and more particularly to a power management system for hybrid inverters that dynamically selects control modes based on cost assessment. Background Technology

[0002] In the energy management field of hybrid inverter applications, control strategies that coordinate photovoltaic (PV) power generation, load demand, and battery energy storage systems (BESS) play a crucial role in optimizing energy use and cost-effectiveness. Two common control strategies are rule-based control and scheduling-based control, each with its own characteristics and trade-offs.

[0003] Rule-based control dynamically adjusts BESS power in real time based on measurements such as PV power generation, load power, and battery state of charge. By responding immediately to real-time input data, rule-based control helps minimize interaction with the grid and enables configurations such as self-consumption mode to prioritize the use of locally generated renewable energy. However, this mode often leads to frequent battery cycling, potentially accelerating battery aging. Furthermore, due to a lack of foresight regarding future costs, rule-based control may not achieve optimal economic efficiency when facing variable pricing mechanisms such as time-of-use tariffs (TOU).

[0004] In contrast, scheduling-based control relies on predicted PV power generation, load curves, and electricity price data to calculate optimal charge / discharge schedules for future periods. This optimization process can employ model-based algorithms or reinforcement learning techniques, aiming to shift charging to low-price periods and discharging to peak-price periods to achieve smoother battery operation and lower energy costs. However, its effectiveness is highly dependent on the accuracy of the predictions. If the actual situation deviates significantly from the predictions, performance may degrade, and the ability to respond to immediate fluctuations in electricity demand or generation may decrease.

[0005] Since none of the above modes can provide a universally optimal solution in different situations, the industry is increasingly interested in strategies that combine the advantages of both. Therefore, there is a pressing need for a power management system that can adaptively switch between control modes to minimize overall energy costs while maintaining operational reliability and immediate responsiveness. Summary of the Invention

[0006] One embodiment of this disclosure provides a power management system including an inverter and a computing device. The inverter is operatively coupled to a photovoltaic array, a power grid, and a battery. The inverter is configured to operate in one of a plurality of operating modes to determine a power setpoint of the battery, wherein the operating modes include a rule-based control mode and a scheduling mode. The computing device is communicatively connected to the inverter and is configured to: control the inverter to operate in one of the rule-based control mode and the scheduling mode based on a cost assessment table; periodically collect measurement data from the inverter; calculate a first future cost corresponding to the rule-based control mode and a second future cost corresponding to the scheduling mode based on the measurement data; and update the cost assessment table based on the first future cost and the second future cost. The cost assessment table is configured to store a cost value corresponding to each of the operating modes for each of a plurality of time intervals.

[0007] In one embodiment, the computing device is further configured to: retrieve a first current cost corresponding to the rule-based control mode and a second current cost corresponding to the scheduling mode from the cost evaluation table based on a time interval index derived from a current time value. The computing device is further configured to compare the first current cost and the second current cost to select one of the rule-based control mode and the scheduling mode as a lower cost mode. The computing device is further configured to control the inverter to operate in the selected lower cost mode. The computing device is further configured to adjust the first current cost and the second current cost based on the first future cost and the second future cost, respectively, to update the cost evaluation table.

[0008] In one embodiment, the computing device is further configured to: control the inverter to operate in the scheduling mode in response to determining that the second current cost is less than the first current cost; and control the inverter to operate in the rule-based control mode in response to determining that the second current cost is not less than the first current cost.

[0009] In one embodiment, the computing device is further configured to update the cost assessment table by replacing the first current cost with a first weighted sum of the first future cost and the first current cost, and replacing the second current cost with a second weighted sum of the second future cost and the second current cost.

[0010] In one embodiment, the computing device is further configured to: during a past period, apply a first estimation model corresponding to the rule-based control mode to calculate a first estimated cost based on the measurement data corresponding to a previous time interval. For each subsequent time interval in the past period, the computing device is further configured to: extract cost values ​​corresponding to the rule-based control mode and the scheduling mode respectively from the cost assessment table; select one of the rule-based control mode and the scheduling mode as a lower cost mode for the subsequent time interval by comparing the extracted cost values; and apply an estimation model corresponding to the selected lower cost mode to calculate a first subsequent estimated cost based on the measurement data corresponding to the subsequent time interval. The computing device is further configured to sum the first estimated cost with the first subsequent estimated costs in the past period to obtain the first future cost. The computing device is further configured to: apply a second estimation model corresponding to the scheduling mode to calculate a second estimated cost based on the measurement data corresponding to the previous time interval. For each of the subsequent time intervals within the past period, the computing device is further configured to: extract cost values ​​from the cost assessment table corresponding to the rule-based control mode and the scheduling mode, respectively; select one of the rule-based control mode and the scheduling mode as a lower cost mode for the subsequent time interval by comparing the extracted cost values; and apply the estimation model corresponding to the selected lower cost mode based on the measurement data corresponding to the subsequent time interval to calculate a second estimated cost for the subsequent period. The computing device is further configured to sum the second estimated cost with the second estimated costs for the subsequent periods within the past period to obtain the second future cost.

[0011] In one embodiment, the measurement data includes an electricity demand, a power generation capacity of the photovoltaic array, and a battery charge status of the battery.

[0012] In one embodiment, in response to applying the first estimation model, the computing device is further configured to: obtain an estimated battery power corresponding to the previous time interval via the first estimation model based on the measurement data corresponding to the previous time interval; and calculate the first estimated cost based on an electricity price of the power grid corresponding to the previous time interval, the estimated battery power, and the measurement data. For each of the subsequent time intervals within the past period, in response to applying the first estimation model, the computing device is further configured to: obtain the estimated battery power corresponding to the subsequent time interval via the first estimation model based on the measurement data corresponding to the subsequent time interval; and calculate a subsequent estimated cost based on the electricity price of the power grid corresponding to the subsequent time interval, the estimated battery power, and the measurement data.

[0013] In one embodiment, the computing device is further configured to: periodically generate predictive data related to the electricity demand and the power generation of the photovoltaic array via a predictive model, based on historical measurement data related to the electricity demand and the power generation of the photovoltaic array. The computing device is further configured to: obtain an optimized battery power corresponding to the previous time period via the second estimation model based on the predictive data and the battery state of charge corresponding to the previous time period; and calculate the second estimated cost based on the electricity price of the grid corresponding to the previous time period, the optimized battery power, and the measurement data. For each of the subsequent time periods within the past period, in response to the application of the second estimation model, the computing device is further configured to: obtain the optimized battery power corresponding to the subsequent time period via the second estimation model based on the predictive data and the battery state of charge corresponding to the subsequent time period; and calculate the subsequent estimated cost based on the electricity price of the grid corresponding to the subsequent time period, the optimized battery power, and the measurement data.

[0014] In one embodiment, the inverter is further configured to: in the scheduling mode, determine the power setpoint of the battery via an optimization model based on the predicted data, the battery's state of charge, and the electricity price; and in the rule-based control mode, determine the power setpoint of the battery based on the measurement data.

[0015] In one embodiment, the computing device controls the inverter to operate in the rule-based control mode during an initial operation period before selecting the operating mode based on the cost assessment table.

[0016] In one embodiment, the power management system further includes an edge device, which is connected between the inverter and the computing device, for real-time transmission of the measurement data and multiple control signals, wherein the computing device is a cloud server.

[0017] In one embodiment, the inverter is configured to adjust a power flow more frequently in the rule-based control mode compared to the scheduling mode.

[0018] One embodiment of this disclosure provides a power management method. The method is performed by a computing device communicatively connected to an inverter. The inverter is operatively coupled to a photovoltaic array, a power grid, and a battery, and is configured to operate in one of a plurality of operating modes to determine a power setpoint of the battery, wherein the operating modes include a rule-based control mode and a scheduling mode. The method includes: controlling the inverter to operate in one of the rule-based control mode and the scheduling mode based on a cost evaluation table, wherein the cost evaluation table is configured to store a cost value corresponding to each of the operating modes for each of a plurality of time intervals. The method further includes: periodically collecting measurement data from the inverter; calculating a first future cost corresponding to the rule-based control mode and a second future cost corresponding to the scheduling mode based on the measurement data; and updating the cost evaluation table based on the first future cost and the second future cost.

[0019] In one embodiment, the method further includes: retrieving a first current cost corresponding to the rule-based control mode and a second current cost corresponding to the scheduling mode from the cost evaluation table based on a time interval index derived from a current time value; comparing the first current cost and the second current cost to select one of the rule-based control mode and the scheduling mode as a lower cost mode; controlling the inverter to operate in the selected lower cost mode; and adjusting the first current cost and the second current cost based on the first future cost and the second future cost to update the cost evaluation table.

[0020] In one embodiment, the method further includes: controlling the inverter to operate in the scheduling mode in response to determining that the second current cost is less than the first current cost; and controlling the inverter to operate in the rule-based control mode in response to determining that the second current cost is not less than the first current cost.

[0021] In one embodiment, the method further includes updating the cost assessment table by replacing the first current cost with a first weighted sum of the first future cost and the first current cost, and replacing the second current cost with a second weighted sum of the second future cost and the second current cost.

[0022] In one embodiment, the method further includes: during a past period, applying a first estimation model corresponding to the rule-based control mode based on measurement data corresponding to a previous time interval to calculate a first estimated cost corresponding to the previous time interval; for each subsequent time interval in the past period: extracting cost values ​​corresponding to the rule-based control mode and the scheduling mode respectively from the cost assessment table; selecting one of the rule-based control mode and the scheduling mode as a lower cost mode for the subsequent time interval by comparing the extracted cost values; and applying an estimation model corresponding to the selected lower cost mode based on the measurement data corresponding to the subsequent time interval to calculate a first subsequent estimated cost; and comparing the first estimated cost with the first subsequent estimated costs in the past period. Costs are summed to obtain the first future cost; based on the measurement data corresponding to the previous time interval, a second estimation model corresponding to the scheduling pattern is applied to calculate a second estimated cost; for each of the subsequent time intervals in the past period: cost values ​​corresponding to the rule-based control pattern and the scheduling pattern are extracted from the cost assessment table; by comparing the extracted cost values, one of the rule-based control pattern and the scheduling pattern is selected as a lower cost pattern for the subsequent time interval; and based on the measurement data corresponding to the subsequent time interval, the estimation model corresponding to the selected lower cost pattern is applied to calculate a second subsequent estimated cost; and the second estimated cost is summed with the second subsequent estimated costs in the past period to obtain the second future cost.

[0023] In one embodiment, in response to applying the first estimation model, the method further includes: obtaining an estimated battery power corresponding to the previous time interval via the first estimation model based on the measurement data corresponding to the previous time interval; and calculating the first estimated cost based on an electricity price of the power grid corresponding to the previous time interval, the estimated battery power, and the measurement data. For each of the subsequent time intervals within the past period, in response to applying the first estimation model, the method further includes: obtaining the estimated battery power corresponding to the subsequent time interval via the first estimation model based on the measurement data corresponding to the subsequent time interval; and calculating a subsequent estimated cost based on the electricity price of the power grid corresponding to the subsequent time interval, the estimated battery power, and the measurement data.

[0024] In one embodiment, the method further includes: periodically generating predictive data related to the electricity demand and the power generation of the photovoltaic array via a predictive model based on historical measurement data related to the electricity demand and the power generation of the photovoltaic array; obtaining an estimated battery power corresponding to the previous time period via a second estimation model based on the predictive data and the battery charging state corresponding to the previous time period; calculating a second estimated cost based on the electricity price of the grid corresponding to the previous time period, the estimated battery power, and the measurement data; and for each of the subsequent time periods in the past period: in response to the application of the second estimation model, obtaining the estimated battery power corresponding to the subsequent time period via the second estimation model based on the predictive data and the battery charging state corresponding to the subsequent time period; and calculating the subsequent estimated cost based on the electricity price of the grid corresponding to the subsequent time period, the estimated battery power, and the measurement data.

[0025] As described above, the power management system and method provide an adaptive control architecture that intelligently switches between rule-based control and scheduling modes based on a learned cost assessment table. By utilizing real-time measurement data, predicted demand and power generation, and mode-specific estimation models, the system disclosed herein can dynamically minimize energy costs while maintaining operational flexibility. This hybrid control strategy not only improves the efficiency and responsiveness of integrating photovoltaic and energy storage systems but also offers scalability, enabling deployment in various energy management scenarios. Attached Figure Description

[0026] This disclosure can be more fully understood by referring to the accompanying drawings and reading the following detailed description and embodiments.

[0027] Figure 1 A schematic diagram of an electric power management system according to this disclosure is shown.

[0028] Figure 2 A system block diagram of a computing device according to an embodiment of the present disclosure is shown.

[0029] Figure 3 This is a flowchart according to an embodiment of the present disclosure, showing further steps in controlling the inverter to operate in either a rule-based control mode or a scheduling mode.

[0030] Figure 4 A flowchart according to an embodiment of the present disclosure is shown, illustrating further steps in calculating a first future cost and a second future cost.

[0031] Figure 5AA schematic flowchart illustrating the application of a first estimation model to calculate a first estimated cost according to an embodiment of the present disclosure is shown.

[0032] Figure 5B This illustration shows a process for calculating subsequent estimated costs when, according to an embodiment of the present disclosure, the rule-based control mode is selected as the lower-cost mode in a subsequent time interval of a past period.

[0033] Figure 6A A schematic flowchart illustrating the application of a second estimation model to calculate a second estimated cost according to an embodiment of the present disclosure is shown.

[0034] Figure 6B This illustration shows a process for calculating subsequent estimated costs when the scheduling mode is selected as the lower cost mode for a subsequent time interval in the past period, according to an embodiment of the present disclosure.

[0035] Figure 7 A flowchart illustrating dynamic mode selection according to an embodiment of the present disclosure is shown.

[0036] List of reference numerals

[0037] 10: Power Management System

[0038] 101: Computing device

[0039] 102: Inverter

[0040] 103: PV array

[0041] 104: Power Grid

[0042] 105: Battery

[0043] 106: Load

[0044] 201: Measurement Data

[0045] 202: Cost Assessment Table

[0046] 203: First Future Cost

[0047] 204: Second Future Costs

[0048] 205: RBC Mode

[0049] 206: Scheduling Mode

[0050] O21, O22, O23, O24: Operation

[0051] S301, S302, S303: Steps

[0052] S401, S402, S403, S404, S405: Steps

[0053] 501: Measurement Data

[0054] 502: Power Requirements

[0055] 503: Power generation

[0056] 504: Battery charging status

[0057] 505: First Estimation Model

[0058] 506: Estimating battery power

[0059] 507: Electricity Price

[0060] 508: First estimated cost

[0061] 511: Measurement Data

[0062] 512: Power Requirements

[0063] 513: Power generation

[0064] 514: Battery charging status

[0065] 516: Estimating battery power

[0066] 517: Electricity Price

[0067] 518: Subsequent estimated costs

[0068] O51, O52, O53, O54: Operation

[0069] 601: Historical Measurement Data

[0070] 602: Predictive Model

[0071] 603: Forecast Data

[0072] 604: Second Estimation Model

[0073] 605: Optimized battery power

[0074] 606: Measurement data

[0075] 607: Electricity Price

[0076] 608: Second estimated cost

[0077] 613: Forecast Data

[0078] 615: Optimized battery power

[0079] 616: Measurement Data

[0080] 617: Electricity Price

[0081] 618: Subsequent estimated costs

[0082] O61, O62, O63, O64, O65, O66: Operation

[0083] 71: Initialization Phase

[0084] 72: Mode Selection Phase

[0085] 73: Table Update Phase Detailed Implementation

[0086] The following description is intended to illustrate the general principles of the invention and should not be construed as limiting. The scope of the invention is defined by the appended claims.

[0087] Figure 1 A schematic diagram of a power management system 10 according to this disclosure is shown. Figure 1 As shown, the power management system 10 includes a computing device 101 and an inverter 102, wherein the computing device 101 and the inverter 102 are communicatively connected to each other. The inverter 102 is operatively coupled to a photovoltaic (PV) array 103, a power grid 104, and a battery 105. Furthermore, a load 106 receives power under the control of the inverter 102.

[0088] The computing device 101 may include a processor, a memory, and one or more input / output interfaces, configured to execute software instructions for managing the operation of the inverter 102. Specifically, the memory stores computer-readable instructions, and the processor is configured to execute these instructions to perform functions related to the power management system 10. The computing device 101 may be implemented as an embedded controller, an industrial computer, a microcontroller-based unit, or a system-on-a-chip. In some embodiments, the computing device 101 may be deployed at the system edge, for example, integrated with the inverter 102, or configured in a local power management infrastructure. Alternatively, the computing device 101 may also be implemented as a remote server connected to the inverter 102 via a communication network.

[0089] Inverter 102 is configured to regulate the power flow between battery 105, PV array 103, power grid 104, and load 106. In particular, inverter 102 is capable of performing DC / AC or AC / DC power conversion and controlling power distribution paths to ensure energy efficiency and overall system stability.

[0090] PV array 103 is configured to convert solar energy into electrical energy and supply it to inverter 102. Grid 104 serves as an external power source and charging station, allowing inverter 102 to draw power from or output power to the grid. Battery 105 is configured to store electrical energy and can supply or absorb power according to control strategies. Load 106 represents an electrical device or system that consumes electrical energy. Inverter 102 regulates the power flow between PV array 103, battery 105, grid 104, and load 106.

[0091] In particular, inverter 102 manages bidirectional power exchange with battery 105. (Symbol) Indicates that battery 105 is in time The discharge power flowing to inverter 102, and Indicates that inverter 102 is in time Charging power flowing to battery 105. Time parameters. This represents a discrete-time index within a power control cycle or time interval. Inverter 102 can operate within each time interval. Based on the system state and control objectives, a decision is made as to whether to charge or discharge the battery.

[0092] Inverter 102 also manages the power flow between the PV array 103 and the power grid 104, symbol This indicates that the PV array 103 is in time The generated electricity. The power exchange between inverter 102 and grid 104 involves two directions, denoted by [symbol missing]. Indicates that power grid 104 is in time The power input to inverter 102, and the symbol Indicates that inverter 102 is in time The electricity supplied to power grid 104. (Symbol) This indicates that the load is 106 at time. The inverter 102, based on system measurement and control logic, performs real-time and dynamic coordination and balance of power consumption among power generation, energy storage, grid interaction, and power consumption.

[0093] Inverter 102 is configured to operate in one of several operating modes to determine the battery's power setpoint. This power setpoint refers to the power point at a specific time. The target charging or discharging power level to be delivered to or absorbed by battery 105. Based on this power setpoint, inverter 102 adjusts the charging power. or discharge power This is to coordinate battery operation with other energy sources and loads in the system. Determining the power setpoint directly impacts energy costs, battery life, and interaction with the power grid.

[0094] A commonly used operating mode is rule-based control (RBC). In this mode, the inverter 102 dynamically determines the battery power setpoint based on real-time measurements, including PV power. Load requirements And the battery charging status. Specifically, net demand will be calculated. If the load exceeds the PV power generation (i.e. If the PV power generation is excessive (i.e., when the battery can discharge to supply the difference), then the battery can discharge to supply the difference. Conversely, if the PV power generation is excessive (i.e., when the battery can discharge to supply the difference), then the battery can discharge to supply the difference. This allows the battery to be charged to absorb excess energy. The desired battery energy storage system (BESS) power... Initial settings And is limited to the feasible range defined by battery power limits and battery state-of-charge limits. The inverter applies this feasible power to battery 105, while any remaining net demand or surplus power is input or output from grid 104, which respectively... and This mode offers instant responsiveness and is easy to implement, but may result in frequent battery cycles, thus reducing cost-effectiveness.

[0095] Another commonly used operating mode is the scheduling mode, which utilizes forecast data and optimization techniques to generate planned charge / discharge curves over future time steps. In this mode, the inverter 102 collaborates with the computing device 101 to process the predicted PV power generation. Forecasted load demand And electricity price data, such as buying and selling electricity prices. Then, optimization or reinforcement learning algorithms are applied to determine a set of power setpoints. and The goal is to minimize the expected energy cost within the planning time domain while satisfying power balance and battery constraints. The resulting setpoint can be applied using a receding horizon approach, which can lead to smoother battery operation and potentially reduce energy costs, but will reduce responsiveness to immediate changes.

[0096] Figure 2 A system block diagram of a computing device 101 according to one embodiment of the present disclosure is shown. Figure 2 As shown, the computing device 101 is configured to implement a power management method involving operations O21-O24. Each operation will be described in detail below.

[0097] Operation O21 involves controlling inverter 102 based on cost assessment table 202 to operate in either RBC mode or scheduling mode. Cost assessment table 202 is configured to store cost values ​​corresponding to each operating mode for each of a plurality of time intervals.

[0098] In one embodiment, Table 1 below shows how cost values ​​are stored and indexed according to time and operating mode.

[0099]

[0100] As shown in Table 1, each column corresponds to a time index (e.g., from 0 to N-1) within the scheduling time domain, and each column represents a candidate operation mode; for example, mode 0 represents the RBC mode, and mode 1 is the scheduling mode. The item Q(a,n) represents the cost value corresponding to selecting operation mode a∈{0,1} at time index n. The computing device 101 is configured to select the optimal operation mode in each time interval based on a comparison of the corresponding cost values, thereby forming a mode sequence σ(t)=a within the operation time domain. In one embodiment, if measurement data 201 is collected every 30 minutes, there will be 48 time intervals in a day, therefore N equals 48. The structure of the cost assessment table 202 allows for efficient finding and updating of mode-related cost curves, thereby supporting online assessment and learning-based adaptation.

[0101] Operation O22 involves periodically collecting measurement data 201 from inverter 102. Measurement data 201 may include, for example, the instantaneous output power of PV array 103. Power consumption of load 106 Battery charging status data of battery 105, charging power of battery 105 With discharge power and the input / output power of the power grid , This information can be stored in the computing device 101 and used as input for subsequent cost estimation procedures.

[0102] Operation O23 involves calculating a first future cost 203 corresponding to RBC mode 205 and a second future cost 204 corresponding to scheduling mode 206 based on measurement data 201. The term "future cost" here refers to the cumulative cost value over the future optimization time domain, calculated under the assumption that the system initially selects one of two operating modes at a specific time step. Once the initial action is fixed at the selected mode, subsequent actions in the time domain are assumed to be based on the optimal strategy determined in the cost assessment table. Therefore, the future cost reflects the assumed total cost incurred if the selected mode is adopted at the current time step, and then the optimal mode is dynamically selected at each subsequent time step. The purpose of this calculation is not to predict actual energy costs, but to support cost-based mode comparisons for decision-making.

[0103] Operation O24 involves updating the cost assessment table 202 based on the first future cost 203 and the second future cost 204. Specifically, items Q(0,n) and Q(1,n) can be updated separately to reflect the latest calculated first and second future cost values ​​for each future time index n. The purpose of updating the cost assessment table 202 is to maintain an up-to-date reference value reflecting the current or predicted system status, thereby enabling adaptive and information-based mode selection at each time step. This update procedure allows the computing device 101 to progressively improve its mode switching strategy according to the constantly changing operating environment.

[0104] Although Figure 1 and Figure 2 Not shown, in one embodiment, when the computing device 101 is implemented as a cloud server, the power management system 10 may further include an edge device as an intermediary interface between the inverter 102 and the cloud server. The edge device may be configured to locally collect real-time measurement data, such as voltage, current, power, and battery state of charge, from various system components including the PV array 103, battery 105, and load 106. The edge device may also be configured to preprocess, buffer, or compress the data before transmitting it over the network to the cloud server, thereby reducing communication latency and bandwidth consumption.

[0105] In addition to data acquisition and transmission, edge devices can also receive control commands or operating parameters generated by the cloud server based on higher-level optimization algorithms, and promptly forward these commands to the inverter 102. In some embodiments, the edge device may further possess limited computing power to execute fallback control logic or temporarily assume local control functions when communication with the cloud server is interrupted. Therefore, the addition of edge devices can enhance the robustness, scalability, and responsiveness of the system, especially in distributed energy systems employing remote cloud computing.

[0106] In one embodiment, inverter 102 adjusts the frequency of power flow more frequently in RBC mode 205 than in scheduling mode 206. Specifically, in RBC mode 205, inverter 102 can update its power setpoint over relatively short time intervals, such as every few seconds or minutes, based on real-time measurement data including immediate power demand, PV power generation, and battery state of charge. This allows the system to react instantly to rapid changes in load or power generation, thereby maintaining operational stability and preventing battery overcharging or over-discharging.

[0107] In contrast, in scheduling mode 206, inverter 102 can be configured to follow a predetermined or optimized power dispatch plan over longer control intervals (e.g., 15 or 30 minutes), thereby reducing switching frequency and associated control burden. The lower adjustment frequency in scheduling mode reflects its reliance on forecast data and planned energy dispatch, making it less sensitive to short-term fluctuations. This difference in control frequency between the two modes allows the power management system to strike a balance between responsiveness and computational efficiency, while leveraging the advantages of real-time RBC and forward-looking scheduling strategies.

[0108] Figure 3 This is a flowchart according to an embodiment of the present disclosure, showing further steps in controlling the inverter 102 to operate in either RBC mode or scheduling mode during operation O21. Figure 3 As shown, operation O21 may further include steps S301 to S303. The contents of each step will be described in detail below.

[0109] In step S301, the computing device 101 retrieves the first current cost and the second current cost corresponding to the RBC mode and the scheduling mode, respectively, from the cost evaluation table 202. This retrieval is based on a time interval index derived from the current time value, for example, by applying modulo to the current time step t to calculate mod(t, N), where N represents the number of time intervals in a day. This indexing method allows the cost evaluation table 202 to maintain a cyclical representation of costs within repeating time intervals each day, thereby enabling adaptive mode selection based on different times of the day.

[0110] In step S302, the computing device 101 compares a first current cost with a second current cost retrieved from the cost assessment table 202. Based on this comparison, the computing device 101 selects the operating mode corresponding to the lower of the two cost values. That is, if the cost of the RBC mode is lower than that of the scheduling mode, the system selects the RBC mode for that time interval, and vice versa. This decision-making procedure allows the system to dynamically adjust its operating mode according to cost trends that change over time.

[0111] In step S303, the computing device 101 controls the inverter 102 to operate in the selected lower-cost mode. In one embodiment, when the inverter 102's operating modes only include RBC mode and scheduling mode, if it is determined in step S302 that the second current cost is less than the first current cost, then in step S303, the computing device 101 controls the inverter 102 to operate in scheduling mode. Conversely, if it is determined in step S302 that the second current cost is not less than the first current cost, then in step S303, the computing device 101 controls the inverter 102 to operate in RBC mode.

[0112] In one embodiment, Table 2 below illustrates example items in cost assessment table 202. Cost values ​​are indexed by time of day, with each column representing a specific time interval and each row corresponding to a different operating mode.

[0113]

[0114] refer to Figure 3 Referring to Table 2, the figure illustrates an example scenario where the current time step t corresponds to time interval 2, i.e., mod(t, N) = 2. Under this index, the first current cost (corresponding to RBC mode) is 2.4, while the second current cost (corresponding to scheduling mode) is 1.3. Based on this comparison, since the second current cost is lower, computing device 101 selects scheduling mode as the lower-cost operating mode for this time step. Therefore, inverter 102 is controlled to operate in scheduling mode.

[0115] In this example, the mode selection for each time step t can be expressed mathematically as follows:

[0116] Where, σ (t) represents the mode selected at time step t, and Q(a, mod(t, N)) represents the cost value retrieved from the cost evaluation table within the time interval indexed by mod(t, N) and corresponding to action a∈{0,1} (i.e., RBC mode or scheduling mode). The computing device 101 determines the optimal mode by selecting actions with lower costs, enabling the system to perform time-adaptive and cost-efficient control.

[0117] After performing step S303, the cost assessment table 202 is updated by adjusting the first current cost and the second current cost based on the first future cost and the second future cost, respectively. Through this update mechanism, the system can incorporate newer information into the cost assessment table, thereby improving the accuracy of future mode selection.

[0118] In one embodiment, the cost evaluation table 202 is updated by replacing the first current cost with a first weighted sum of a first future cost and a first current cost, and replacing the second current cost with a second weighted sum of a second future cost and a second current cost. For a given action a∈{0,1} and time interval index... The updated cost value can be expressed mathematically as follows:

[0119] Where α is the learning rate coefficient, used to determine the initial cost value. With the future cost of new calculations The weighted ratio between them. This update rule allows cost assessment table 202 to be iteratively optimized based on historical performance, striking a balance between stability and responsiveness to environmental changes.

[0120] It should be noted that the update of cost assessment table 202 is not based on the current time t, but on the previous time interval. Its definition is ,in This is called the update delay time domain. The purpose of introducing this update delay time domain is to ensure that the system has sufficient time to collect the necessary historical data (such as PV power generation, load consumption, and battery charging status) during the look-ahead period, allowing for accurate cost assessment before updating the cost assessment table. Therefore, at time t, the system will update the cost assessment table 202 with the data related to time t. The associated Q value.

[0121] Figure 4 A flowchart according to an embodiment of the present disclosure is shown, illustrating further steps in calculating the first future cost 203 and the second future cost 204 in operation O23. Figure 4 As shown, operation O23 may further include steps S401 to S405. Each step will be described in detail below.

[0122] In step S401, the computing device 101 applies either a first estimation model corresponding to the RBC mode or a second estimation model corresponding to the scheduling mode to calculate the previous time interval in the past period. The estimated cost. Specifically, when calculating the first future cost 203, the first estimation model is applied to the corresponding... Historical measurement data was used to calculate the first estimated cost. Similarly, in calculating the second future cost 204, the second estimation model is applied to calculate the second estimated cost. .

[0123] Steps S402 to S404 involve iterating through multiple subsequent time intervals over a past period to calculate the corresponding estimated costs for each. In step S402, the computing device 101 retrieves the cost values ​​corresponding to the RBC mode and the scheduling mode from the cost assessment table 202, respectively. and These cost values ​​correspond to subsequent time intervals. The time index, and T represents the optimization rolling time domain, which is the period during which energy costs are optimized. For example, if T=48 and measurement data 201 is collected every 30 minutes, then the cost can be optimized for the next 24 hours.

[0124] In step S403, the computing device 101 compares the extracted cost values. and To determine the lower-cost pattern corresponding to each subsequent time interval. The lower cost mode is dynamically selected based on the stored cost values ​​and is not necessarily the same as the original mode a used to initiate the cost assessment in step S401.

[0125] In step S404, the computing device 101 applies an estimation model corresponding to the selected lower-cost mode to calculate a subsequent estimated cost. , where a∈{0,1} corresponds to the initial pattern. The estimation model can be a first estimation model associated with the RBC pattern, or a second estimation model associated with the scheduling pattern. These models are based on time intervals. The actual historical measurement data is used as input to calculate the estimated energy cost. Specifically, when calculating the first future cost 203 corresponding to RBC mode 205, the computing device 101 calculates the cost in each subsequent time interval. Use the corresponding lower cost mode. The estimation model calculates the first subsequent estimated cost. Similarly, when calculating the second future cost 204 corresponding to scheduling pattern 206, the computing device 101 also calculates the cost in each time interval. Using the estimation model corresponding to the selected lower-cost model, a second subsequent estimated cost is calculated. These subsequent cost estimates all reflect the dynamic impact of the chosen operating mode under historical conditions.

[0126] In step S405, the computing device 101 summarizes the estimated costs over the past period. That is, the computing device 101 sums the estimated costs of the first time interval with subsequent estimated costs to calculate the complete future cost. Regarding the first future cost 203, the calculation result is as follows:

[0127] Similarly, the calculation for the second future cost 204 is as follows:

[0128] These future cost values ​​reflect the total energy costs incurred when choosing the RBC mode or scheduling mode in the initial time interval, while allowing for a subsequent mode switch based on the current cost assessment table.

[0129] In one embodiment, measurement data 201 includes power demand, the power output of the PV array, and the battery's state of charge. These data items reflect the load's energy consumption behavior, the available renewable energy provided by the PV array, and the battery's current energy capacity state, respectively. This measurement data serves as a key input to the cost estimation process, enabling the computing device 101 to assess the feasibility of battery scheduling and evaluate cost performance under different operating modes.

[0130] Figure 5A A schematic flowchart illustrating the application of a first estimation model 505 to calculate a first estimated cost according to an embodiment of the present disclosure is shown. Figure 5A As shown, operation O51 involves applying the first estimation model 505 to obtain the corresponding previous time interval. The estimated battery power 506. Specifically, the first estimation model 505 receives measurement data 501 as input, which may include power demand 502 (in the form of...). (indicated), power generation 503 (in) (indicated), and the battery charging status 504 (indicated by) (This indicates that each of these corresponds to a previous time interval). Based on these inputs, the first estimation model 505 generates the corresponding estimated battery power. The first estimation model 505 can be implemented through rule-based decision logic, a simulation-based inverter behavior simulator, or a lightweight machine learning model trained on historical control actions, but this disclosure is not limited thereto.

[0131] Operation O52 involves the estimated battery power 506 and grid electricity price 507 obtained based on self-operation O51. (represented), and corresponding to the previous time interval. The measurement data 501 is used to calculate the first estimated cost 508 (in the form of measurement data 501). (Indicated). The grid electricity price of 507 indicates the corresponding time interval. The unit electricity price can reflect the dynamic pricing structure adopted by the grid operator, such as time-of-use pricing or real-time market prices.

[0132] Figure 5B This illustration shows a process for calculating subsequent estimated costs when the RBC mode is selected as the lower cost mode in a subsequent time interval of a past period, according to an embodiment of this disclosure. Figure 5B As shown, corresponding to the time interval The measurement data 511 includes power requirements 512 (in... (indicated), power generation 513 (in) (indicated), and battery charging status 514 (indicated) (Represented). This data is provided as input to the first estimation model 505 in operation O53. Based on the input data, the first estimation model 505 outputs an estimated battery power 516 (in words). express).

[0133] In operation O54, the estimated battery power 516 is used, along with the corresponding measurement data and the grid electricity price 517 (in... (This is represented by a variable) to calculate the subsequent estimated cost 518. This subsequent estimated cost 518 reflects the cost of the inverter over time. Expected costs when operating in RBC mode.

[0134] It should be noted that, Figure 5B Only the selected lower-cost model was described. This is the case for the RBC (Reduced Cost by BC) scheduling pattern. For the time interval where the selected scheduling pattern is the lower cost pattern, a second estimation model will be applied, which will be discussed later. Figure 6A and Figure 6B Please provide an explanation.

[0135] Figure 6A A schematic flowchart illustrating the application of a second estimation model 604 to calculate a second estimated cost according to an embodiment of the present disclosure is shown. In this embodiment, historical measurement data 601, including previously observed power demand data and PV array power generation data, is provided to the prediction model 602 in operation O61. The prediction model 602 is configured to... to During this period, predictive data 603 related to power demand and PV array power generation are generated periodically. The prediction model 602 can be implemented using statistical prediction techniques (e.g., Autoregressive Integrated Moving Average (ARIMA) models), neural network models (e.g., Long Short-Term Memory (LSTM)), or other machine learning-based time series predictors, but this disclosure is not limited thereto.

[0136] In operation O62, the predicted data 603, along with its corresponding previous time interval, is used. The actual state of battery charging, i.e. Together, they are provided as input to the second estimation model 604. The second estimation model 604 can be implemented using optimization-based control algorithms, such as model predictive control, mixed-integer programming (MIP), or reinforcement heuristics, but this disclosure is not limited thereto. The output of the second estimation model 604 corresponds to the previous time interval. Optimized battery power 605 (with express).

[0137] In operation O63, it is based on the optimized battery power 605, actual measurement data 606, and grid electricity price 607 (in... (indicated) to calculate the second estimated cost of 608 (in terms of) (This indicates that) all of these data correspond to previous time intervals. The second estimated cost 608 represents the expected cost that would have occurred if the scheduling pattern had been applied at the start point of the past period.

[0138] Figure 6B This illustration shows a process for calculating subsequent estimated costs when the scheduling pattern is selected as the lower-cost pattern for a subsequent time interval in the past period, according to an embodiment of this disclosure. Figure 6B As shown, historical measurement data 601 is provided as input to prediction model 602 in operation O64. Prediction model 602 generates predictions corresponding to subsequent time intervals. The forecast data 613 includes expected power demand and PV array output.

[0139] In operation O65, the predicted data 613, along with its corresponding subsequent time interval, is used. The actual state of battery charging, i.e. Together, they are provided as input to the second estimation model 604 to generate corresponding values ​​for subsequent time intervals. Optimized battery power 615 (with express).

[0140] In operation O66, the subsequent estimated cost 618 is based on the grid electricity price 617 (in... The data is calculated from the optimized battery power 615 and the measurement data 616, all of which correspond to subsequent time intervals. The subsequent estimated cost of 618 represents the cost if, within this specific time interval... Expected costs when applying scheduling patterns.

[0141] In one embodiment, inverter 102 is configured to determine the battery power setpoint using different strategies based on the selected operating mode. In scheduling mode, the power setpoint is determined through an optimization model that takes predicted data (e.g., predicted power demand and PV generation), the battery's state of charge, and grid electricity prices as inputs. This optimization model can be implemented using techniques such as linear programming, quadratic programming, or model predictive control to minimize expected operating costs in the future time domain while satisfying system constraints, but this disclosure is not limited thereto. The optimization output provides the optimal battery power scheduling plan for each time interval within the scheduling window.

[0142] In contrast, when operating in RBC mode, the inverter 102 determines the battery's power setpoint solely based on real-time measurement data. This measurement data may include real-time power demand, PV power generation, and battery state of charge. Rule-based control logic can be implemented using threshold-based decision rules, such as charging the battery when PV power generation exceeds demand and the battery state of charge is below a certain threshold, or discharging the battery when demand exceeds PV power generation and the battery state of charge is above another threshold.

[0143] In one embodiment, before enabling operating mode selection based on the cost assessment table 202, the inverter 102 is controlled by the computing device 101 to operate in RBC mode 205 during the initial operation period, which may be referred to as the warm-up phase. During this warm-up phase, the system accumulates measurement data and generates initial cost estimates corresponding to different time intervals without actively switching control modes. Since the cost assessment table 202 is not yet fully populated with reliable cost data in the early stages of operation, forcing a dynamic mode switch could lead to unstable or suboptimal operating behavior. The warm-up phase ensures safe and consistent operation by relying solely on RBC mode 205, while allowing the system to gradually build up the historical and estimated cost values ​​required for subsequent decisions.

[0144] Figure 7 This diagram shows the overall flowchart of dynamic mode selection according to an embodiment of the present disclosure. Figure 7 As shown, the process begins with initialization phase 71, in which the system initializes internal variables, parameters, and data structures, including cost assessment table 202. Specifically, the Q value stored in cost assessment table 202 is initially set to zero, applicable to all control modes and time intervals. This initialization allows the system to accumulate meaningful cost assessments from scratch, unaffected by arbitrary prior values.

[0145] The system then checks whether the current time step t exceeds the threshold defined by x×N, where N represents the number of time intervals per day, and x is a predetermined multiplier representing the duration of the warm-up phase. If the condition t>x×N is not yet met, the system will remain in RBC mode 205 while collecting additional data and filling in the cost assessment table 202.

[0146] Once this condition is met, the system transitions to the mode selection phase 72. In this phase, the computing device compares the cost values ​​in cost assessment table 202 and selects either the RBC mode or the scheduling mode for each time interval based on which incurs the lower cost. Subsequently, in the table update phase 73, the cost assessment table 202 is updated using the latest calculated future costs to continuously refine and adjust the control strategy. After each iteration, the time step increases, and the process is repeated, allowing the system to continuously optimize and learn throughout the overall operation.

[0147] The power management system and method proposed in this paper provide an adaptive control architecture that intelligently switches between RBC (Resource-Based Cost) and scheduling modes based on a learned cost assessment table. By utilizing real-time measurement data, predicted demand and generation, and mode-specific estimation models, the system dynamically minimizes energy costs while maintaining operational flexibility. This hybrid control strategy not only improves the efficiency and responsiveness of PV and energy storage system integration but also offers scalability, enabling deployment in various energy management scenarios.

[0148] In some embodiments, the system operates on recurring time intervals within a daily cycle. Let N represent the number of time intervals per day, and T represent the optimized time domain used for cumulative cost assessment. Indicates the update delay time domain, where The historical index to be updated has been identified, and This represents the forecast time domain used by the forecasting model to generate forecasted demand and PV power generation. These parameters satisfy... ≥T and ≥T, such that the update depends on the implementation data within the evaluation window, and the prediction coverage is also consistent across windows. The learning rate α∈(0,1], as described in the weighted update of the previous cost evaluation table 202, determines the mixing ratio between the newly calculated cumulative cost and the previously stored value.

[0149] As used in this article, "past period" refers to a continuous time interval series in which cost estimation is performed using the first estimation model 505 and the second estimation model 604 (see reference). Figures 5A to 6B "Future costs" refers to the cumulative cost calculated over the optimal time domain based on realized or predicted operating conditions (see reference). Figure 4The "Cost Assessment Table" is a data structure that stores cost values ​​associated with both the RBC (Real-Time Commitment) and scheduling models for each time interval. It is updated over time based on realized measurement data and forecast coverage. Figures 3 to 6B As revealed, to achieve deterministic daily time alignment, a time interval index is derived from the current time value, including calculating the modulus corresponding to the current time value to the total number of daily time intervals, such as... Figure 3 As stated above.

[0150] Will Linking it to T ensures that the cost assessment uses a complete forecast coverage, while... Setting the learning rate to at least T ensures that updates reflect the actual conditions throughout the evaluation window. As mentioned earlier, mixed learning rates strike a balance between reaction speed and stability, thereby reducing oscillating behavior and improving economic efficiency. Figure 3 The explicit modulo-based indexing avoids slot drift and ensures that price-sensitive comparisons and updates are performed within the expected daily time intervals, thereby improving predictability and simplifying table maintenance in time-of-use pricing environments. In some embodiments, the computing device 101 delays replacing the storage cost value of the time interval index corresponding to the previous time value until all measurement data used to calculate the corresponding future cost has been collected and covers every subsequent time interval in the past period. Once this condition is met, a weighted sum of the future cost and the storage cost value is used for replacement. This method of delaying updates until the actual data is complete reduces bias caused by prediction errors and telemetry delays, improves the accuracy of subsequent cost comparisons, and enhances robustness in scenarios with fluctuating loads and intermittent renewable energy generation. In summary, these features enable the system to achieve cost-aware adaptive control while maintaining stability and immediate responsiveness, improving economic efficiency, protecting battery health, and supporting reliable operation in environments with dynamic pricing and uncertain forecasting.

[0151] While this disclosure has been described by way of examples and preferred embodiments, it should be understood that the invention is not limited to the disclosed embodiments. Rather, the invention is intended to cover various modifications and similar arrangements (which will be apparent to those skilled in the art). Therefore, the scope of the appended claims should be given the broadest interpretation to cover all such modifications and similar arrangements.

Claims

1. A power management system, comprising: An inverter, operatively coupled to a photovoltaic array, a power grid, and a battery, is configured to operate in one of several operating modes to determine a power setpoint for the battery, wherein the several operating modes include a rule-based control mode and a scheduling mode; and A computing device, communicatively connected to the inverter, is configured to: The inverter is controlled to operate in either the rule-based control mode or the scheduling mode based on a cost evaluation table, wherein the cost evaluation table is configured to store a cost value corresponding to each of the multiple operating modes for each of the multiple time intervals. Periodically collect measurement data from the inverter; Based on the measurement data, a first future cost corresponding to the rule-based control mode and a second future cost corresponding to the scheduling mode are calculated; and The cost assessment table is updated based on the first future cost and the second future cost.

2. The power management system of claim 1, wherein the computing device is further configured to: Based on a time interval index derived from a current time value, a first current cost corresponding to the rule-based control mode and a second current cost corresponding to the scheduling mode are retrieved from the cost evaluation table; Compare the first current cost with the second current cost to select either the rule-based control mode or the scheduling mode as the lower cost mode; Control the inverter to operate in the selected lower-cost mode; and Based on the first future cost and the second future cost, the first current cost and the second current cost are adjusted respectively to update the cost assessment table.

3. The power management system of claim 2, wherein the computing device is further configured to: In response to the determination that the second current cost is less than the first current cost, the inverter is controlled to operate in the scheduling mode; and When it is determined that the second current cost is not less than the first current cost, the inverter is controlled to operate in the rule-based control mode.

4. The power management system of claim 2, wherein the computing device is further configured to: The cost assessment table is updated by replacing the first current cost with a first weighted sum of the first future cost and the first current cost, and by replacing the second current cost with a second weighted sum of the second future cost and the second current cost.

5. The power management system of claim 1, wherein the computing device is further configured to: In a past period, based on the measurement data corresponding to a previous time interval, a first estimation model corresponding to the rule-based control mode is applied to calculate a first estimated cost; For each subsequent time interval within this past period: Multiple cost values ​​corresponding to the rule-based control mode and the scheduling mode are extracted from the cost assessment table; By comparing the multiple cost values ​​obtained, the rule-based control mode and the scheduling mode are selected as the lower cost mode for the subsequent time interval; as well as Based on the measurement data corresponding to the subsequent time interval, an estimation model corresponding to the selected lower cost pattern is applied to calculate a first subsequent estimated cost. The first estimated cost is summed with the first subsequent estimated cost in the past period to obtain the first future cost; Based on the measurement data corresponding to the previous time interval, a second estimation model corresponding to the scheduling pattern is applied to calculate a second estimated cost; For each of the multiple subsequent time intervals within that past period: The cost values ​​corresponding to the rule-based control mode and the scheduling mode are extracted from the cost assessment table. By comparing the multiple cost values ​​obtained, the rule-based control mode and the scheduling mode are selected as the lower cost mode for the subsequent time interval; as well as Based on the measurement data corresponding to the subsequent time interval, the estimation model corresponding to the selected lower cost mode is applied to calculate a second subsequent estimated cost. as well as The second estimated cost is summed with a plurality of second subsequent estimated costs in the past period to obtain the second future cost.

6. The power management system of claim 5, wherein the measurement data includes an electricity demand, a power generation capacity of the photovoltaic array, and a battery charging status of the battery.

7. The power management system of claim 6, wherein the computing device is further configured to: Reflecting on the application of this first estimation model, Based on the measurement data corresponding to the previous time interval, an estimated battery power corresponding to the previous time interval is obtained through the first estimation model; The first estimated cost is calculated based on an electricity price of the power grid corresponding to the previous time interval, the estimated battery power, and the measurement data. For each of the multiple subsequent time intervals within that past period: When this first estimation model is applied, Based on the measurement data corresponding to the subsequent time interval, the estimated battery power corresponding to the subsequent time interval is obtained via the first estimation model; and Based on the electricity price of the power grid corresponding to the subsequent time interval, the estimated battery power, and the measurement data, a subsequent estimated cost is calculated.

8. The power management system of claim 7, wherein the computing device is further configured to: Based on historical measurement data related to the electricity demand and the power generation of the photovoltaic array, a predictive model periodically generates predictive data related to the electricity demand and the power generation of the photovoltaic array. Based on the predicted data and the battery charging state corresponding to the previous time interval, an optimized battery power corresponding to the previous time interval is obtained through the second estimation model; Based on the electricity price of the power grid corresponding to the previous time interval, the optimized battery power, and the measurement data, the second estimated cost is calculated; and For each of the multiple subsequent time intervals within that past period: When this second estimation model is applied, Based on the predicted data and the battery's state of charge corresponding to the subsequent time interval, the optimized battery power corresponding to the subsequent time interval is obtained through the second estimation model; and The estimated cost is calculated based on the electricity price of the power grid corresponding to the subsequent time interval, the optimized battery power, and the measurement data.

9. The power management system of claim 8, wherein the inverter is further configured to: In this scheduling mode, based on the predicted data, the battery's state of charge, and the electricity price, an optimization model determines the battery's power setpoint; and In this rule-based control mode, the power setpoint of the battery is determined based on the measurement data.

10. The power management system as described in claim 1, wherein: Before selecting the operating mode based on the cost assessment table, the computing device controls the inverter to operate in the rule-based control mode during an initial operation period.

11. The power management system as described in claim 1, further comprising: An edge device, connected between the inverter and the computing device, is used to transmit the measurement data and multiple control signals in real time, wherein the computing device is a cloud server.

12. The power management system of claim 1, wherein the inverter is configured to: In this rule-based control mode, power flow is adjusted more frequently compared to the scheduling mode.

13. The power management system of claim 2, wherein the operation of deriving the time interval index from the current time value includes: Calculate the modulus of the total number of time intervals corresponding to the current time value for each day.

14. The power management system of claim 5, wherein updating the cost assessment table includes: Only after the measurement data used to calculate the corresponding future cost has been collected for each of the plurality of subsequent time intervals in the past period, is the storage cost value corresponding to a time interval index of a previous time value replaced by a weighted sum of the corresponding future cost and a storage cost value.

15. A power management method, performed by a computing device communicatively connected to an inverter, the method comprising: Based on a cost assessment table, the inverter is controlled to operate in either a rule-based control mode or a scheduling mode, wherein the inverter is operatively coupled to a photovoltaic array, a power grid, and a battery, and the inverter is configured to operate in one of a plurality of operating modes to determine a power setpoint of the battery, wherein the plurality of operating modes include the rule-based control mode and the scheduling mode, and wherein the cost assessment table is configured to store a cost value corresponding to each of the plurality of operating modes for each of the plurality of time intervals; Periodically collect measurement data from the inverter; Based on the measurement data, a first future cost corresponding to the rule-based control mode and a second future cost corresponding to the scheduling mode are calculated; and The cost assessment table is updated based on the first future cost and the second future cost.

16. The power management method as described in claim 15, further comprising: Based on a time interval index derived from a current time value, a first current cost corresponding to the rule-based control mode and a second current cost corresponding to the scheduling mode are retrieved from the cost evaluation table; Compare the first current cost with the second current cost to select either the rule-based control mode or the scheduling mode as the lower cost mode; Control the inverter to operate in the selected lower-cost mode; and Based on the first future cost and the second future cost, the first current cost and the second current cost are adjusted respectively to update the cost assessment table.

17. The power management method as described in claim 16, further comprising: When it is determined that the second current cost is less than the first current cost, the inverter is controlled to operate in the scheduling mode. as well as When it is determined that the second current cost is not less than the first current cost, the inverter is controlled to operate in the rule-based control mode.

18. The power management method as described in claim 16, further comprising: The cost assessment table is updated by replacing the first current cost with a first weighted sum of the first future cost and the first current cost, and by replacing the second current cost with a second weighted sum of the second future cost and the second current cost.

19. The power management method as described in claim 15, further comprising: In a past period, based on the measurement data corresponding to a previous time interval, a first estimation model corresponding to the rule-based control mode is applied to calculate a first estimated cost; For each subsequent time interval within this past period: Multiple cost values ​​corresponding to the rule-based control mode and the scheduling mode are extracted from the cost assessment table; By comparing the multiple cost values ​​obtained, the rule-based control mode and the scheduling mode are selected as the lower cost mode for the subsequent time interval; as well as Based on the measurement data corresponding to the subsequent time interval, an estimation model corresponding to the selected lower cost pattern is applied to calculate a first subsequent estimated cost. The first estimated cost is summed with the plurality of first subsequent estimated costs in the past period to obtain the first future cost; Based on the measurement data corresponding to the previous time interval, a second estimation model corresponding to the scheduling pattern is applied to calculate a second estimated cost; For each of the multiple subsequent time intervals within that past period: The cost values ​​corresponding to the rule-based control mode and the scheduling mode are extracted from the cost assessment table. By comparing the multiple cost values ​​obtained, the rule-based control mode and the scheduling mode are selected as the lower cost mode for the subsequent time interval; as well as Based on the measurement data corresponding to the subsequent time interval, the estimation model corresponding to the selected lower cost mode is applied to calculate a second subsequent estimated cost. as well as The second estimated cost is summed with multiple second subsequent estimated costs over the past period to obtain the second future cost.

20. The power management method of claim 19, wherein the measurement data includes a power demand, a power generation capacity of the photovoltaic array, and a battery charging state of the battery.

21. The power management method as described in claim 20, further comprising: Reflecting on the application of this first estimation model, Based on the measurement data corresponding to the previous time interval, an estimated battery power corresponding to the previous time interval is obtained through the first estimation model; The first estimated cost is calculated based on an electricity price of the power grid corresponding to the previous time interval, the estimated battery power, and the measurement data. For each of the multiple subsequent time intervals within that past period: When this first estimation model is applied, Based on the measurement data corresponding to the subsequent time interval, the estimated battery power corresponding to the subsequent time interval is obtained via the first estimation model; and Based on the electricity price of the power grid corresponding to the subsequent time interval, the estimated battery power, and the measurement data, a subsequent estimated cost is calculated.

22. The power management method as described in claim 21, further comprising: Based on historical measurement data related to the electricity demand and the power generation of the photovoltaic array, a predictive model periodically generates predictive data related to the electricity demand and the power generation of the photovoltaic array. Based on the predicted data and the battery charging state corresponding to the previous time interval, an estimated battery power corresponding to the previous time interval is obtained through the second estimation model. Based on the electricity price of the power grid corresponding to the previous time interval, the estimated battery power, and the measurement data, the second estimated cost is calculated; and For each of the multiple subsequent time intervals within that past period: When this second estimation model is applied, Based on the predicted data and the battery's state of charge corresponding to the subsequent time interval, the estimated battery power corresponding to the subsequent time interval is obtained through the second estimation model; and The estimated cost is calculated based on the electricity price of the power grid corresponding to the subsequent time interval, the estimated battery power, and the measurement data.

23. The power management method as described in claim 16, further comprising: The index of a time interval is derived from the current time value by calculating a modulus of the total number of time intervals corresponding to the current time value for each day.

24. The power management method as described in claim 19, further comprising: Only after the measurement data used to calculate a corresponding future cost has been collected for each of the multiple subsequent time intervals in the past period, is the storage cost value corresponding to a time interval index of a previous time value replaced by a weighted sum of the corresponding future cost and a storage cost value, in order to update the cost assessment table.

Images