System and method for determining chargeable or dischargeable energy of battery energy storage system

Abstract

Systems and methods for determining the total chargeable / dischargeable energy of subsystems of a battery energy storage system (BESS) are disclosed. An iterative process is performed within a dynamic period of time divided into multiple iterations using a neural network model including an energy prediction sub-model and a state prediction sub-model. The energy prediction sub-model outputs chargeable / dischargeable energy for the subsystem of the current iteration. The state prediction sub-model outputs a voltage of the subsystem for the next iteration, a charge rate of the subsystem for the next iteration, a maximum temperature of the subsystem for the next iteration, and a charge rate difference for the next iteration. The total chargeable / dischargeable energy of the BESS subsystem is determined by summing the dischargeable energy determined for each iteration.

Description

[0001] Cross Reference to Related Applications

[0002] This application claims priority to and the benefit of U.S. Provisional Application No. 63 / 601,641, filed November 21, 2023, the disclosure of which is incorporated herein by reference. TECHNICAL FIELD

[0003] The present disclosure relates to managing battery energy storage systems (BESSs), and more particularly to accurately determining chargeable / dischargeable energy of a BESS. BACKGROUND

[0004] BESSs have become a critical component in modern energy management systems. As renewable power sources such as wind and solar, which are inherently intermittent, are increasingly integrated, energy storage solutions are necessary to ensure grid stability and efficient power distribution. BESS technology allows for excess power to be stored during periods of low demand and discharged during periods of high demand, thereby optimizing energy usage (by reducing curtailment of solar and wind power) and reducing dependence on fossil fuel-based power generation such as gas turbines. This capability is particularly valuable as the global shift to clean energy accelerates and as the share of intermittent power sources in the power supply mix grows.

[0005] As battery cells degrade (also referred to as “fade”) over time, it is critical to accurately predict the available charge capacity at a point in time in the life of a BESS. In some cases, a battery cell can fail a capacity test even when it is projected to pass the test. A conventional calculation to estimate the available charge capacity of a BESS is as follows: (site average state of charge (SOC) %) x (nameplate energy) x (capacity fade %) x (number of BESS racks). This calculation has inherent inaccuracies and can impact the ability to fully discharge a BESS and, thus, reduce market revenue. SUMMARY

[0006] Accordingly, the present disclosure describes a system and method for determining the chargeable / dischargeable energy of a subsystem of a BESS at a rated (specified) power. A neural network model takes into account the voltage, charge rate, and maximum temperature of the BESS subsystem to predict the total chargeable / dischargeable energy of the BESS over a dynamic time period divided into multiple iterations (i.e., time intervals). The neural network model includes an energy predictor model and a state predictor model. The energy predictor model predicts the chargeable / dischargeable energy for a current iteration, and the state predictor model predicts the input features (e.g., operating state including voltage, charge rate, and maximum temperature) for a next iteration. The neural network model iterates over the iterations of the time period and sums the chargeable / dischargeable energy for each iteration to arrive at the total chargeable / dischargeable energy of the subsystem of the BESS for the time period. The present system and method has several advantages over conventional BESS energy calculations.

[0007] First, users of BESSs using conventional energy calculations can bid in the electricity market with incorrect chargeable / dischargeable energy values (e.g., overestimated or underestimated), and thus can incur penalties and thus loss of revenue. In contrast, the present system and method can provide consistent and relatively accurate dischargeable (i.e., sellable) energy values, and thus the likelihood of penalties and loss of revenue is significantly reduced.

[0008] Second, since conventional energy calculations can incorrectly guess the chargeable / dischargeable energy of a BESS for a full SOC range (e.g., 95%), users can bid conservatively in the market, which can result in loss of revenue. In contrast, the present system and method can automatically calculate the chargeable / dischargeable energy at time intervals (e.g., every 2 minutes) based on the current system operating state. This real-time calculation means that there is no need to guess how much energy needs to be discharged (thereby increasing the revenue-generating opportunities of the electricity market).

[0009] Third, users can not be aware of the (significant) energy loss due to cell voltage imbalance (which can be a major factor), temperature variation, and degradation (i.e., battery deterioration) in their daily work. In contrast, the chargeable / dischargeable energy values generated by the present system and method automatically reveal the gap (relative to the installed or specified energy) and thus highlight any negative impact of cell voltage imbalance and other factors on the chargeable / dischargeable energy.

[0010] According to one aspect, the present disclosure relates to a system for determining a total chargeable / dischargeable energy of a subsystem of a battery energy storage system (BESS), the subsystem comprising battery cells, the system comprising: a controller comprising one or more processing modules and one or more non-transitory memory storage modules storing computing instructions that, when executed by the one or more processing modules, are configured to: perform an iterative process over a dynamic time period using a neural network model comprising an energy prediction submodel and a state prediction submodel, wherein the dynamic time period is divided into a plurality of iterations, wherein for each iteration of the plurality of iterations, the controller is configured to: (1) input into the energy prediction submodel: a voltage of the subsystem for a current iteration of the plurality of iterations, a charge rate of the subsystem for the current iteration, and a maximum temperature of the subsystem for the current iteration; wherein the energy prediction submodel is configured to output a chargeable / dischargeable energy of the subsystem for the current iteration; and (2) input into the state prediction submodel: the voltage of the subsystem for the current iteration, the charge rate of the subsystem for the current iteration, the maximum temperature of the subsystem for the current iteration, and a charge rate difference for the current iteration; wherein the charge rate difference for the current iteration is equal to the charge rate of the subsystem for the current iteration minus the charge rate of the subsystem for a previous iteration of the plurality of iterations; wherein the state prediction submodel is configured to output a voltage of the subsystem for a next iteration of the plurality of iterations, a charge rate of the subsystem for the next iteration, a maximum temperature of the subsystem for the next iteration, and a charge rate difference for the next iteration.

[0011] In some cases, the voltage of the subsystem for a first iteration of the plurality of iterations can be determined using a voltage imbalance for the first iteration of the subsystem according to the following equation: V t = CVmin t / (1-α ε) where CVmin t is a minimum cell voltage of the battery cells of the subsystem for the first iteration, ε is the voltage imbalance for the subsystem for the first iteration, and α is a scaling factor. The voltage imbalance is a function of the minimum cell voltage of the subsystem for the first iteration and a maximum cell voltage of the subsystem for the first iteration.

[0012] In some cases, the controller can be configured to repeat steps (1) and (2) until a last iteration of the iterative process is performed.

[0013] In some cases, the controller can be configured to determine the total chargeable / dischargeable energy of the subsystem of the BESS over the dynamic time period by summing the chargeable / dischargeable energy output by the energy predictor model for each iteration.

[0014] In some cases, the controller can be configured to repeat steps (1) and (2) until the voltage of the subsystem output by the state predictor model for a next iteration of the iteration process reaches the voltage limit.

[0015] In some cases, the dynamic time period can be determined by a time required to charge or discharge the subsystem of the BESS from a first iteration of the iteration process to a last iteration of the iteration process, wherein the last iteration of the iteration process is an iteration in which the voltage of the subsystem output by the state predictor model reaches the voltage limit.

[0016] In some cases, the subsystem is selected from at least one of: a plurality of battery packs comprising battery cells, or a plurality of battery racks comprising battery packs.

[0017] In some cases, the battery type of the battery cells of the subsystem is selected from a group consisting of: lithium nickel manganese cobalt (NMC), lithium iron phosphate (LFP), lithium sulfur (Li-S), solid state, nickel cadmium, nickel metal hydride (NiMH), zinc air, iron air, vanadium redox flow, sodium ion, potassium ion, aluminum ion, lead acid, silicon anode, or a combination thereof.

[0018] In some cases, the architecture of the neural network model is selected from a group consisting of: a multi-layer perceptron (MLP), a recurrent neural network (RNN), a long short-term memory network, a gated recurrent unit (GRU), or a combination thereof.

[0019] In another aspect, the present disclosure relates to a method for determining a chargeable / dischargeable energy of a subsystem of a battery energy storage system (BESS), the subsystem comprising a battery cell, the method comprising: performing an iterative process over a dynamic time period by using a neural network model comprising an energy predictor submodel and a state predictor submodel, wherein the dynamic time period is divided into a plurality of iterations, wherein each iteration of the plurality of iterations comprises the following steps: (1) inputting into the energy predictor submodel: a voltage of the subsystem for a current iteration of the plurality of iterations, a charge rate of the subsystem for the current iteration, and a maximum temperature of the subsystem for the current iteration; wherein the energy predictor submodel outputs a chargeable / dischargeable energy of the subsystem for the current iteration; and (2) inputting into the state predictor submodel: the voltage of the subsystem for the current iteration, the charge rate of the subsystem for the current iteration, the maximum temperature of the subsystem for the current iteration, and a charge rate difference for the current iteration; wherein the charge rate difference for the current iteration is equal to the charge rate of the subsystem for the current iteration minus the charge rate of the subsystem for a previous iteration of the plurality of iterations; wherein the state predictor submodel outputs a voltage of the subsystem for a next iteration of the plurality of iterations, a charge rate of the subsystem for the next iteration, a maximum temperature of the subsystem for the next iteration, and a charge rate difference for the next iteration.

[0020] It should be noted that the technical effects obtainable by the present disclosure are not limited to the above-mentioned effects, and other effects not mentioned herein will be clearly understood by those skilled in the art from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0021] The accompanying drawings illustrate exemplary aspects of the present disclosure and together with the following detailed description, provide further understanding for the technical spirit of the present disclosure. However, the present disclosure should not be understood to be limited only to the drawings.

[0022] Figure 1 is a perspective view schematically showing a configuration of a battery container according to an aspect of the present disclosure.

[0023] Figure 2 is a perspective view schematically showing a form in which some components of a battery container according to an aspect of the present disclosure are separated or moved.

[0024] Figure 3 is a view showing an internal configuration of a battery container according to an aspect of the present disclosure, viewed from above.

[0025] Figure 4 is a chart showing a difference between an estimated chargeable / dischargeable energy of a BESS subsystem and an actual discharged energy.

[0026] Figure 5is a conceptual diagram illustrating the effect of unbalanced voltage levels among battery monoblocks on the charging and discharging capabilities of a BESS subsystem.

[0027] Figure 6A is a schematic diagram illustrating a BESS subsystem including power blocks according to an aspect of the present disclosure.

[0028] Figure 6B is a schematic diagram illustrating a BESS site including multiple power blocks according to an aspect of the present disclosure.

[0029] Figure 7 is a chart illustrating the effect of unbalanced voltage levels among racks on the operational power of a power block.

[0030] Figure 8A is a conceptual diagram illustrating a model for determining chargeable / dischargeable energy of a BESS subsystem according to an aspect of the present disclosure.

[0031] Figure 8B is a flowchart illustrating an implementation of a model for determining chargeable / dischargeable energy of a BESS subsystem according to an aspect of the present disclosure.

[0032] Figure 9 is a schematic diagram illustrating a BESS subsystem including power blocks according to an aspect of the present disclosure, the BESS subsystem including a controller implementing a model of Figure 6A and Figure 6B . DETAILED DESCRIPTION

[0033] The present disclosure can be modified in various ways and has various aspects, and the specific aspects disclosed in detail herein are intended to facilitate the understanding of the present disclosure for those skilled in the art.

[0034] Therefore, it should be understood that the present disclosure is not intended to be limited to the particular aspects disclosed, but rather, the present disclosure covers all modifications, equivalents, and alternatives falling within the spirit and scope of the present disclosure.

[0035] In the present application, it should be understood that terms such as "include" or "have" are intended to indicate the existence of described features, numbers, steps, operations, components, parts, or combinations thereof, and are not intended to exclude the possibility of the existence or addition of one or more other features, numbers, steps, operations, components, parts, or combinations thereof.

[0036] Figure 1 is a perspective view schematically illustrating a configuration of a battery container 1000 according to an aspect of the present disclosure. Furthermore, Figure 2 is a perspective view schematically illustrating a form in which some components of a battery container 1000 according to an aspect of the present disclosure are separated or moved.Figure 3 FIG. 1 is a diagram illustrating an internal configuration of a battery container 1000 according to an aspect of the disclosure, viewed from above.

[0037] Referring to Figures 1 to 3 The battery container 1000 according to the disclosure includes a battery rack 100, a container housing 200, a main connector 300, and a main busbar 400.

[0038] The battery rack 100 can include a plurality of battery modules 110. Among them, each battery module 110 can be configured in a form in which a plurality of battery cells (secondary batteries) are accommodated in a module housing. In addition, the battery modules 110 can be stacked in one direction, such as upward and downward directions, to form the battery rack 100. Specifically, the battery rack 100 can include a rack housing to facilitate stacking of the battery modules 110. In this case, the plurality of battery modules 110 can be accommodated in respective storage spaces provided in the rack housing to form a module stack. In some aspects, the battery modules 110 can be arranged in other configurations, such as side by side or in a matrix style. The rack housing can include features such as cooling channels or structural reinforcements to support the weight of the stacked modules. In some cases, the battery rack 100 can incorporate sensors to monitor temperature, voltage, or other parameters of the battery modules 110.

[0039] The battery modules 110 in the battery rack 100 can also include control units, such as a battery management system (BMS) for each group or a specific group. For example, a separate group BMS can be provided for each battery module 110. In this case, each battery module 110 can be referred to as a battery pack. That is, the battery rack 100 can be considered to include a plurality of battery packs. In various descriptions below, the battery modules 110 can be replaced with battery packs. In some cases, the battery rack 100 can incorporate sensors to monitor parameters such as temperature, voltage, or current of the battery modules 110. The BMS of each battery module or group can communicate with a superior rack BMS to coordinate battery rack performance and safety.

[0040] One or more battery racks 100 can be included in the battery container 1000. Specifically, a plurality of battery racks 100 can be included in the battery container 1000. Also, the plurality of battery racks 100 can be disposed in at least one direction, for example, in a horizontal direction. For example, eight battery racks 100 can be included in the battery container 1000, and the plurality of battery racks 100 can be arranged in a left-right direction (X-axis direction) inside the battery container 1000. When a plurality of battery racks 100 are included, a separate control unit, such as a rack BMS, can be provided for each battery rack 100. In this case, the rack BMS can be connected to the plurality of group BMSs to exchange data and control the plurality of group BMSs. Meanwhile, when the battery container 1000 includes at least one rack BMS, the rack BMS can be connected to a separate control device, such as a control container, disposed outside the battery container 1000. Also, the control container can be connected to the rack BMS or the group BMS of the battery container 1000 to control the same or exchange data therewith.

[0041] A vacant space can be formed inside the container housing 200. Also, the container housing 200 can accommodate the battery rack 100 in the internal space. More specifically, the container housing 200 can be formed in a substantially cuboid shape, as shown in FIG. 1A, for example. In this case, the container housing 200 can include an upper housing 201, a lower housing, a front housing 203, a rear housing, a left housing 205, and a right housing, which surround the internal space. Also, the container housing 200 can accommodate the battery rack 100 in the internal space defined by the six unit housings. Figure 1

[0042] The container housing 200 can be made of a material capable of ensuring a certain level of rigidity and stably protecting internal components from external physical and chemical factors. For example, the container housing 200 can be made of, or can have, a metallic material such as steel, aluminum, or titanium. In some aspects, the container housing 200 can be made of a composite material such as carbon fiber reinforced polymer or glass fiber, which provides a high strength-to-weight ratio. In areas exposed to harsh environmental conditions, the housing can also incorporate corrosion-resistant alloys, such as stainless steel or galvanized steel. In some cases, the container housing 200 can utilize a combination of materials, such as a steel frame with aluminum panels, to balance strength, weight, and cost considerations. Also, the housing can include a special coating or treatment, such as a powder coating or anodization, to enhance durability and weather resistance.

[0043] ​The container housing can have the same or similar size as that of a shipping container. Also, the container housing can follow a standard of a shipping container predetermined according to ISO standards or the like. For example, the container housing can be designed to have the same or similar size as that of a 20-foot container or a 40-foot container. However, the size of the container housing can be appropriately designed according to the situation. Specifically, the size or shape of the container housing can be set in various ways according to the construction scale, shape, form, or the like of a system to which the battery container is applied, such as an energy storage system. The present disclosure is not limited by the size or shape of the container housing. For example, in some aspects, the container housing can have other shapes, such as a cylindrical shape, a spherical shape, or a custom polygonal shape. The housing can also be modular, allowing for expansion or contraction based on capacity needs. In some cases, the container housing can incorporate features such as a sloped roof for drainage or reinforced walls for improved durability in harsh environments.

[0044] The main connector 300 can be configured to be electrically connected to the outside. That is, with respect to the battery container 1000, the main connector 300 can be configured to be connected to another component outside the battery container 1000, for example, another battery container 1000 or a control container equipped with a control unit such as a battery system controller (BSC).

[0045] The main connector 300 can be located on at least one side of the container housing 200. For example, the main connector 300 can be located on the left or right side of the container housing 200. Also, a plurality of main connectors 300 can be included in the battery container 1000. For example, as shown in FIG. 1, the battery container 1000 can include a first connector 301 and a second connector 302. Figure 2 and Figure 3 As shown in FIG. 1, the main connector 300 can include two main connectors 300, i.e., a first connector 301 and a second connector 302.

[0046] The plurality of main connectors 300 can be located on different sides of the container housing 200. Also, the plurality of main connectors 300 can be located on opposite sides of the container housing 200. For example, as shown in FIG. 1, the first connector 301 and the second connector 302 can be disposed on the left and right sides of the container housing 200, respectively. In some aspects, the main connector 300 can be located on the top or bottom of the container housing 200. In some cases, the main connector 300 can be located at a corner or edge of the container housing 200. The main connector 300 can also be arranged in various configurations, such as in a staggered pattern or vertically along the sides of the container housing 200. In some embodiments, additional main connectors can be included on the front or back of the container housing 200 to provide additional connection options. Figures 1 to 3

[0047] ​The main bus bar 400 can be configured to transmit power. Specifically, the main bus bar 400 can serve as a path through which charging power and discharging power of the battery holder 100 included in the corresponding battery container 1000 are transmitted. To this end, the main bus bar 400 can be electrically connected to each terminal of the battery module 110 provided in the battery holder 100. In addition, the main bus bar 400 can be connected to the main connector 300. Accordingly, the main bus bar 400 can serve as a path through which the charging power is transmitted from the main connector 300 to the battery module 110. In addition, the main bus bar 400 can serve as a path through which the discharging power is transmitted from the battery module 110 to the main connector 300.

[0048] In addition, the main bus bar 400 can serve as a power transmission line between a plurality of main connectors 300. To this end, different ends of the main bus bar 400 can be connected to different main connectors 300. For example, the main bus bar 400 can be a power line elongated in one direction (e.g., the left-right direction). In this case, both ends of the main bus bar 400 can be connected to different main connectors 300, e.g., the first connector 301 and the second connector 302. In addition, the main bus bar 400 can serve as a path for transmitting power between different main connectors 300 (e.g., between the first connector 301 and the second connector 302).

[0049] The main bus bar 400 can include two unit bus bars, i.e., a positive bus bar 410 and a negative bus bar 420, in order to serve as a power transmission path. The positive bus bar 410 can be connected to the positive terminal of the battery holder 100 or the positive terminal of the battery module 110 included therein. In addition, the negative bus bar 420 can be connected to the negative terminal of the battery holder 100 or the negative terminal of the battery module 110 included therein.

[0050] In addition, the main connector 300 can be provided at each end of the positive bus bar 410 and the negative bus bar 420, respectively. For example, the first connector 301 and the second connector 302 can be provided at the left end and the right end of the positive bus bar 410, respectively. The first connector 301 and the second connector 302 provided at both ends of the positive bus bar 410 can be positive connectors 310. In addition, the first connector 301 and the second connector 302 can be provided at the left end and the right end of the negative bus bar 420, respectively. The two connectors, i.e., the first connector 301 and the second connector 302, provided at both ends of the negative bus bar 420 can all be negative connectors 320.

[0051] Further, the battery container 1000 according to the disclosure can include a cable cover CC. The cable cover CC can be configured to surround a cable connected to the battery container 1000. For example, a plurality of power cables can be connected to the terminal bus bar TB to transfer power. In this case, the cable cover CC can be located at one end (e.g., the lower end) of the terminal cover TC to protect the plurality of power cables connected to the terminal bus bar TB. Alternatively, the battery container 1000 can be connected to a data cable to exchange various data with other external components such as a control container 2000. In this case, the cable cover CC can be configured to protect the data cable connected to the battery container 1000 or the like from external influences.

[0052] Specifically, the cable cover CC can include a cable tray CC1 and a tray cover CC2. The cable tray CC1 can include a body portion attached to the outer wall of the container housing 200 and a side wall portion protruding outward from the edge of the body portion. For example, the side wall portion can be formed to protrude leftward from the front and rear edges of the body portion. The tray cover CC2 can be coupled to the end of the side wall portion protruding from the body portion of the cable tray CC1 to form an empty space therein together with the body portion and the side wall portion. Specifically, the empty space can be formed in a hollow shape. Accordingly, the cable can extend outward from the battery container 1000 through the empty space of the cable cover CC. Further, the cable extended to the outside can be connected to other external components such as the control container 2000 or another battery container 1000.

[0053] According to this aspect, by minimizing the exposure of the cable extended from the battery container 1000 to the outside, the cable can be protected and prevented from damage or breakage. Further, the cable cover CC is designed to form a hollow portion downward at the side surface of the container housing, so that the cable accommodated inside can be exposed downward to the outside. In this case, this can be advantageous for the installation, management, and underground burying of the cable.

[0054] Further, the battery container 1000 according to the disclosure can also include a cable cover CC as Figure 1 and Figure 2The air conditioning module 600 can be configured to adjust the air inside the container housing 200. Specifically, the air conditioning module 600 can control the temperature state of the internal air. In addition, the air conditioning module 600 can be configured to circulate the air inside the container housing 200 to control the temperature of various electronic devices included in the battery container 1000, such as the battery rack 100 or the BMS, within a certain range. Specifically, the air conditioning module 600 can cool the air inside the container housing 200. For example, the air conditioning module 600 can be configured to absorb heat from the air inside the container housing 200 and discharge it to the outside. In addition, the air conditioning module 600 can be configured to remove dust or foreign matter from the air inside the container housing 200.

[0055] Representatively, the air conditioning module 600 can include at least one HVAC (Heating, Ventilation, and Air Conditioning). For example, the battery container 1000 according to the present disclosure can include four HVAC systems. The HVAC system can allow air to circulate inside the container housing 200. In this case, the temperature of the battery rack 100 can be lowered, and the temperature difference between the battery racks 100 or between the battery modules 110 included in the container housing 200 can be reduced.

[0056] Specifically, the container housing 200 can include at least one door, as indicated by E in FIGS. 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, and 100 to facilitate the installation, maintenance, or repair of the battery rack 100. For example, the container housing 200 can have eight doors E on the front surface. In addition, two doors E can be opened and closed in the form of a casement window in pairs. In addition, such a door E can be additionally provided at another portion of the container housing 200, for example, at the rear surface. Figure 1 and Figure 2

[0057] In this way, when the door E is provided for the container housing 200, the HVAC can be installed in the door E of the container housing 200. For example, when two doors E are configured as a pair, the HVAC can be provided on one of the two doors E. In addition, the HVAC, that is, the air conditioning module 600 can be configured to penetrate the container housing 200, particularly the door E. In this case, one surface of the air conditioning module 600 can be exposed to the outside of the container housing 200, and the other surface of the air conditioning module 600 can be exposed to the inside of the container housing 200. Accordingly, the inner surface of the air conditioning module 600 can contact the internal air of the container housing 200 to absorb heat, and the outer surface of the air conditioning module 600 can contact the external air of the container housing 200 to discharge heat.

[0058] ​The air conditioning module 600 can be configured to prevent direct contact between the internal air and the external air. That is, the air conditioning module 600 can be configured to prevent the internal air from being discharged to the outside and to prevent the external air from being introduced into the inside. Accordingly, even if the temperature inside the container housing 200 increases, the air conditioning module 600 only absorbs heat from the internal air and discharges the heat to the outside, without directly discharging the internal air to the outside. According to this aspect, even if a fire or a toxic gas is generated inside the battery container 1000, it is possible to prevent the fire or the toxic gas from being discharged to the outside and causing damage to other devices (such as other nearby battery containers 1000 or workers) outside.

[0059] In addition, the battery container 1000 according to the present disclosure can further include a ventilation module 700 as shown in FIGS. 11A and 11B. Figure 1 and Figure 2 The ventilation module 700 can be configured to discharge the gas inside the container housing 200 to the outside. In addition, the ventilation module 700 can also introduce the external air inside the container housing 200 to the inside. Accordingly, the ventilation module 700 can function as a ventilation device. That is, the ventilation module 700 can exchange or circulate the gas between the inside and the outside of the container housing 200.

[0060] Specifically, the ventilation module 700 can be configured to operate in an abnormal situation, such as when a ventilation gas or a fire is generated in a specific battery module 110. In addition, when a gas or the like is generated inside the container housing 200 due to a thermal runaway phenomenon of the battery rack 100 or the like, the ventilation module 700 can be configured to discharge the gas to the outside. In addition, the ventilation module 700 can be configured to be in a closed state in a normal state and to be switched to an open state in an abnormal state such as a thermal runaway situation. In this case, since the ventilation module 700 performs active ventilation, the ventilation module 700 can be referred to as an AVS (Active Ventilation System) or include such a system.

[0061] In this case, it is possible to prevent a greater problem such as an explosion from occurring due to an increase in the internal pressure of the battery container 1000. In addition, by rapidly discharging the flammable gas inside the container housing 200 to the outside, it is possible to reduce the possibility of a fire occurring in the battery container 1000 or to delay the occurrence of a fire, and it is possible to reduce the scale of a fire.

[0062] Meanwhile, in an aspect including both the ventilation module 700 and the air conditioning module 600, in normal cases, the ventilation module 700 can not operate, but the air conditioning module 600 can operate. In this case, in the process of cooling, foreign matter or moisture can be prevented from flowing into the container housing 200 through the ventilation module 700. According to this aspect, since the air conditioning module 600, the ventilation module 700, etc. are included in the battery container 1000, it is only necessary to transport and install the battery container 1000, i.e., the air conditioning module 600 or the ventilation module 700 can be transported and installed together. Therefore, it is possible to minimize on-site installation work for installing an energy storage system, and it is possible to simplify a connection structure.

[0063] In this aspect, the air conditioning module 600 and / or the ventilation module 700 can operate under the control of the control container 2000. Alternatively, the air conditioning module 600 and / or the ventilation module 700 can be controlled by a control unit included in the battery container 1000, such as a rack BMS that controls the charging / discharging operation of each battery rack 100 or another separate control unit.

[0064] Further, the battery container 1000 according to the present disclosure can include at least one sensor and provide sensing information to the rack BMS included in the battery container 1000, another separate control unit, or the control container 2000. For example, a temperature sensor, a smoke sensor, an H2 sensor, and / or a CO sensor can be included in the battery container 1000. In this case, the operation of the air conditioning module 600 and / or the ventilation module 700 can be controlled based on information sensed by these sensors. The battery container 1000 can further include a fire fighting connector 810 to a fire fighting module (not shown in the drawings).

[0065] As an example, Figure 4 A graph 1100 showing the discharge of a BESS site specified at 10 MWhr (energy) and 2.5 MW (power) with a discharge time of 4 hours at the beginning of life (BOL) is shown. Based on a conventional calculation (represented by curve 1102), the BESS was estimated to discharge about 10 MWhr. However, when the BESS site was fully discharged at the specified power, the actual energy discharged before reaching the power limit was about 9 MWhr in 4 hours (represented by curve 1104). Therefore, even though the BESS should have passed the capacity test, the BESS site did not pass the capacity test.

[0066] The difference between the estimated discharge energy and the actual discharge energy can be explained by the cell voltage imbalance, which is the difference between the maximum cell voltage and the minimum cell voltage across a given BESS subsystem (e.g., a pack, a module, a rack, a string, or a group of racks). The BESS takes these minimum and maximum cell voltages into account to ensure that the batteries are not discharged below the minimum cell voltage or charged above the maximum cell voltage. When discharging, the cell voltage imbalance captures energy that could have otherwise been discharged (in cells with a voltage above the minimum cell voltage), and when charging, the cell voltage imbalance captures energy that could have otherwise been charged (in cells with a voltage below the maximum cell voltage). The cell voltage imbalance can lead to balancing grid supply and demand issues and can require the use of polluting and inefficient backup power sources, such as gas turbines. To address the cell voltage imbalance, a battery management system (BMS) can passively balance the battery cells, which can occur when the BESS subsystem SOC is above a threshold and the current consumed is below a threshold. Thus, the continuous use of the BESS without a rest period exacerbates the cell voltage imbalance.

[0067] Figure 5 is a graph illustrating the impact of cell voltage imbalance. A group of imbalanced battery cells 1202 can include a cell 1204 with the highest voltage level 1208, a cell 1206 with the lowest voltage level 1210, and cells 1203 and 1205 with intermediate voltage levels 1207 and 1209. As shown on the left side, when discharging, energy cannot be fully discharged from cells 1203, 1204, and 1205 (represented by the shaded area) due to being limited by cell 1206 with the lowest voltage level 1210. As shown on the right side, energy cannot be fully charged to cells 1203, 1205, and 1206 (represented by the shaded area) due to being limited by cell 1204 with the highest voltage level 1208.

[0068] Figure 6Ais a diagram illustrating a BESS subsystem 1300a according to one aspect of the present disclosure. The BESS subsystem 1300a can include a power block 1301 that includes a plurality of battery racks 1302a, 1302b, and 1302c, which in turn include a plurality of battery packs 1303aa-ar, 1303ba-br, and 1303ca-cr and battery protection units (BPUs) 1310a, 1310b, and 1310c, respectively. The racks 1302a-c can include a physical structure (e.g., a steel or aluminum frame) with a standardized form, allowing for simple installation, management, and expansion. A suitable example battery rack can be, for example, a TR1300 (model ERT5422CN201) manufactured by LG New Energy. In some aspects, the racks can be constructed of other materials, such as carbon fiber composite, fiberglass, or reinforced plastic, to reduce weight while maintaining strength. Each of the battery packs 1303aa-ar, 1303ba-br, and 1303ca-cr can include one or more battery modules and can be connected to monitoring and management electronics, such as a battery management system (BMS). Each of the battery modules can include a plurality of battery cells connected together, which can be packaged and managed as a single unit. A battery cell is the smallest unit of an electrical BESS in which electrochemical reactions occur to store and release energy. Cells can have different form factors, such as cylindrical, pouch, or prismatic. In each of the battery racks 1302a-c, the battery packs 1303aa-ar, 1303ba-br, and 1303ca-cr can be electrically connected in series with respect to one another, although the present disclosure is not so limited. The battery racks 1302a-c can be electrically connected in parallel with respect to one another, although the present disclosure is not so limited. The battery racks 1302a-c and the battery packs 1303aa-ar, 1303ba-br, and 1303ca-cr can be connected in any series or parallel arrangement to achieve a target power output. While battery packs and / or modules are described in this particular aspect, other racks that do not include packs and / or modules are contemplated within the scope of the present disclosure. For example, a rack can include a plurality of battery cells without any modules.

[0069] The BPUs 1310a-c, which can be referred to as rack BMSs, include electrical and communication interfaces connected to the packs 1303aa-ar, 1303ba-br, and 1303ca-cr within each respective rack 1302a, 1302b, and 1302c. The BPUs 1310a-c can be electrically and communicatively connected to the voltage line 1304 and one or more power block controllers 1308.

[0070] The battery racks 1302a-c can be electrically connected to the grid 1307 via voltage lines 1304. DC switches 1305 can be used to disconnect or isolate the batteries from other components for maintenance, safety, or in the event of a fault, particularly from the power conversion system (PCS) 1306 or grid tie inverter. In the event of an overvoltage, overcurrent, or other fault in the system, the DC switches 1305 can quickly interrupt current to prevent damage to the power blocks 1301 or other components. The PCS 1306 manages the conversion between DC power from the power blocks 1301 and AC power for use by the grid 1307 (i.e., the load). The PCS 1306 can include both inverters (DC to AC) and rectifiers (AC to DC), enabling bidirectional energy flow between the power blocks 1301 and the grid 1307. The PCS 1306 synchronizes the output from the power blocks 1301 with the voltage, frequency, and phase of the grid 1307, allowing the power blocks 1301 to smoothly inject or absorb power from the grid 1307.

[0071] An energy management system (EMS) 1309 can coordinate and optimize overall energy flow in the BESS subsystem 1300a. The EMS 1309 can handle strategic decisions of when and how energy should be stored or discharged, and can integrate multiple sources of energy (e.g., co-located solar and wind power connected in a microgrid and / or the grid 1307). The EMS 1309 can decide when the BESS subsystem 1300a should store or discharge power based on load demand, market signals (e.g., prices such as locational marginal price (LMP)), and availability of renewable power, and can manage the interaction between the power blocks 1301 and the grid 1307, providing services such as frequency regulation, voltage support, demand response.

[0072] The power block controller (PBC) 1308 can control and operate the various components within the BESS subsystem 1300a, such as the battery packs 1303aa-ar, 1303ba-br, and 1303ca-cr, and the battery modules and cells therein, and ensure safe and efficient operation of the BESS subsystem 1300a at the hardware level. The PBC 1308 can work in conjunction with one or more BMS to ensure safe operation of the battery cells, preventing overcharging, deep discharging, or temperature issues. In coordination with the EMS 1309, the PBC 1308 can manage the conversion of DC power from the power block 1301 to AC power to the grid 307, and vice versa (in coordination with the EMS to follow power set points), and can make adjustments in real-time, ensuring that the power output from the PB 1301 meets the voltage and frequency requirements of the grid 1307. The PBC 1308 can monitor the power block 1301 for faults and execute protection mechanisms, e.g., in conjunction with the DC switch 1305, in response to issues such as overvoltage, overcurrent, or overheating. The PBC 1308 can be responsible for executing commands from the EMS 1309 at the hardware level.

[0073] Figure 6B A BESS site 1300b is shown that includes multiple power blocks 1301a-h (e.g., similar to the power block 1301), each connected to a respective PCS 1307a-h. A single cell in a given power block 1301a-h can reduce the capacity of the entire BESS site 1300b. For example, if a single cell in the power block 1301a reaches the minimum voltage faster than the other cells in the power block 1301, the capacity of the entire BESS site 1300b can potentially reduce its designated capacity by about 1 / 8, thus illustrating the importance of managing cell voltage imbalances.

[0074] Figure 7 Graph 1400a and graph 1400b are charts showing the impact of imbalanced voltage levels between racks on the operating power of a power block. As shown in graph 1400a, racks Al, A2, and A3 have about the same minimum voltage level between them. Over time, the operating power of the power block including racks Al-A3 is about the same at the same point in time. In contrast, as shown in graph 1400b, racks Bl, B2, and B3 have different minimum voltage levels between them. Rack B2 forces the power block to reach a certain operating power faster over time, limiting the ability of racks Bl and B3 to discharge at a higher operating power at the same point in time.

[0075] Figure 8A Graph 1400c is a chart showing a method for determining a BESS subsystem (e.g., BESS subsystem 1300a) according to the present disclosure. The method includes receiving 1401 a voltage level of each cell in each rack of the BESS subsystem. The method also includes determining 1402 a minimum voltage level of each rack based on the voltage level of each cell in each rack. The method also includes determining 1403 an operating power of each rack based on the minimum voltage level of each rack. The method also includes determining 1404 an imbalance between the operating power of each rack. The method also includes determining 1405 a power imbalance between the operating power of each rack based on the imbalance between the operating power of each rack. Figure 6Aa conceptual diagram of a model of chargeable / dischargeable energy 1507 of a BESS subsystem 1300a shown in FIG. 13. The model determines the chargeable / dischargeable energy of the BESS subsystem at a rated power and is a function of the operating state (voltage, maximum cell temperature, and charge rate) of the BESS subsystem.

[0076] The chargeable / dischargeable energy can be determined for each subsystem in a BESS site, respectively. The chargeable / dischargeable energy computed at each subsystem can be summed together to determine the total chargeable / dischargeable energy of the BESS site (or the total chargeable / dischargeable energy of a group of subsystems). In some cases, the chargeable / dischargeable energy is computed for each battery pack in a group of battery packs (battery racks), and the chargeable / dischargeable energy computed for each battery pack is summed together to determine the chargeable / dischargeable energy of the battery racks. In some cases, the chargeable / dischargeable energy is computed for each battery rack in a group of battery racks (power blocks), and the chargeable / dischargeable energy computed for each battery rack is summed together to determine the chargeable / dischargeable energy of the power blocks. In some cases, the chargeable / dischargeable energy is computed for each power block in a group of power blocks (BESS site), and the chargeable / dischargeable energy computed for each power block is summed together to determine the chargeable / dischargeable energy of the BESS site.

[0077] The model can include an energy predictor sub-model 1501 and a state predictor sub-model 1502. One or both of the sub-models 1501 or 1502 can be implemented as a neural network model that implements machine learning. The energy predictor sub-model 1501 predicts the chargeable / dischargeable energy for a current iteration, and the state predictor sub-model 1502 predicts the operating state (e.g., voltage, charge rate, and maximum cell temperature) for the next iteration. The model can be executed as an iterative process over a dynamic time period (i.e., cycle) that is divided into multiple iterations (i.e., time intervals). The time period can be one minute, one hour, one day, one month, etc. Each iteration can be defined or configured in units of seconds, minutes, hours, etc. (smaller iterations can yield more accurate results). In some cases, each iteration can be defined or configured to be any value in a range from 1 minute to 15 minutes. In one example, the dynamic time period is determined to be 24 hours, and each iteration is defined or configured to be 2 minutes. In this way, the chargeable / dischargeable energy of a BESS subsystem can be computed every 2 minutes, and the total chargeable / dischargeable energy of the BESS subsystem can be computed by summing the chargeable / dischargeable energy computed for each 2-minute iteration together.

[0078] As Figure 8BThe model can be implemented by inputting features 1503 (i.e., the operating state of the BESS subsystem) into an energy prediction submodel 1501 and a state prediction submodel 1502, as shown in flowchart 1500. The features 1503 can include the voltage of the subsystem for the current iteration (i.e., the operating voltage), the charge rate of the subsystem for the current iteration, and the maximum temperature of the subsystem for the current iteration (e.g., the maximum temperature among the battery monoblocks). For the state prediction submodel 1502, the charge rate difference (the charge rate of the subsystem for the current iteration minus the charge rate of the subsystem for the last iteration) is additionally inputted.

[0079] The energy prediction submodel 1501 outputs the chargeable / dischargeable energy 1504 of the subsystem for the current iteration. The state prediction submodel 1502 outputs the operating state 1506 for the next iteration (to become the input features 1503 for the next iteration), which can include the voltage of the subsystem for the next iteration, the charge rate of the subsystem for the next iteration, the maximum temperature of the subsystem for the next iteration, and the charge rate difference for the next iteration. After inputting the initial voltage, the initial charge rate, and the initial maximum temperature for the first iteration of the time period, the model can predict the chargeable / dischargeable energy 1504 and the next operating state 1506 for the remaining iterations within the time period.

[0080] The initial voltage of the subsystem for the first iteration of the time period can be determined using the voltage imbalance of the subsystem for the first iteration according to the following equation: V t = CVmin t / (1 - a e) where CVmin t is the minimum voltage of the battery monoblocks of the subsystem for the first iteration, e is the voltage imbalance of the subsystem for the first iteration, and a is a scaling factor. The voltage imbalance is a function (i.e., the difference) of the minimum monoblock voltage of the battery monoblocks of the subsystem for the first iteration and the maximum monoblock voltage of the subsystem for the first iteration. The minimum monoblock voltage and the maximum monoblock voltage can be determined (i.e., measured) using one or more voltage sensors (which can be integrated with the BMS). The initial charge rate of the subsystem for the first iteration of the time period can be determined according to the following equation: Charge rate = (rated power / rated energy capacity).

[0081] The initial maximum temperature of the subsystem for the first iteration of the time period can be determined (i.e., measured) using one or more temperature sensors (which can be integrated with the BMS).

[0082] The energy predictor model 1501 and the state predictor model 1502 can be executed until a last iteration of the iterative process is performed (in other words, a last iteration of the dynamic time period), where a decision can be determined at decision point 1505. In some cases, the energy predictor model 1501 and the state predictor model 1502 can be executed until a maximum operating voltage (i.e., a voltage limit) is output by the state predictor model 1502, which can be determined at decision point 1505. The maximum operating voltage can depend on the cell type and can be, for example, between about 1 V to about 10 V. In some cases, the maximum operating voltage can be between about 2.5 V to 4 V (e.g., about 3.2 V). In some cases, the dynamic time period can be determined by the time required to charge or discharge the subsystem of the BESS from the first iteration of the iterative process to the last iteration of the iterative process, where the last iteration of the iterative process is the iteration in which the subsystem voltage output by the state predictor model 1502 reaches the maximum operating voltage.

[0083] The total chargeable / dischargeable energy 1507 of the subsystem of the BESS over the dynamic time period can be determined by summing the chargeable / dischargeable energy 1504 output by the energy predictor model 1501 for each iteration. For a BESS site, the chargeable / dischargeable energy of the entire BESS site over the dynamic time period can be determined by summing the total dischargeable energy 1507 determined for each power block (group of battery racks).

[0084] In some cases, the battery type (i.e., chemical composition) of the battery cells of the BESS subsystem can be one or more of lithium ion, lithium nickel manganese cobalt (NMC), lithium iron phosphate (LFP), lithium sulfur (Li-S), solid state, nickel cadmium, nickel metal hydride (NiMH), zinc air, iron air, vanadium redox flow, sodium ion (e.g., sodium sulfur (NaS), nickel sodium chloride), potassium ion, aluminum ion, lead acid, silicon anode, or combinations thereof, although the present disclosure is not limited thereto. Other battery types can include lithium titanate (LTO), lithium air, magnesium ion, calcium ion, organic radical battery, flow battery (such as zinc bromine or iron chromium), sodium air, lithium organic, and solid polymer electrolyte battery. In some aspects, emerging technologies such as graphene-based batteries, glass batteries, or sand batteries can also be utilized in the BESS subsystem.

[0085] The architecture of the neural network model can be one or more of the following: a multilayer perceptron (MLP), a recurrent neural network (RNN), a long short-term memory network (LSTM), a gated recurrent unit (GRU), a convolutional neural network (CNN), a transformer network, a deep belief network (DBN), a radial basis function network (RBFN), an echo state network (ESN), a Hopfield network, a Boltzmann machine, a restricted Boltzmann machine (RBM), a deep autoencoder, a variational autoencoder (VAE), a generative adversarial network (GAN), or a combination thereof, although the present disclosure is not limited thereto. In some cases, the architecture of the model is not a neural network, but a tree-based model such as a decision tree, a random forest, a gradient boosting machine (GBM), or a combination thereof, although the present disclosure is not limited thereto.

[0086] An MLP model can be composed of multiple layers, including an input layer, one or more hidden layers, and an output layer. Each layer can be fully connected, meaning that each neuron in one layer is connected to every neuron in the subsequent layer through a weighted connection. The network can operate by processing input data through these layers, applying an activation function at each neuron to introduce nonlinearity, and generating an output at the final layer. The weights of the connections can be adjusted during training through a process known as backpropagation, which can minimize the difference between the predicted output and the actual target. An MLP model is capable of performing complex tasks such as classification or regression by learning to map inputs to outputs based on provided training data. In some aspects, an MLP model can incorporate additional techniques like dropout or batch normalization to improve generalization and training stability. The architecture of an MLP can be customized with different numbers and sizes of hidden layers to suit different problem complexities. In some cases, an MLP model can be combined with other neural network types in an ensemble method to further enhance prediction performance.

[0087] RNN models can process sequences of data by maintaining connections between neurons across different time steps, enabling the network to retain information from previous inputs. This processing can be achieved through the use of recurrent connections, where the output from one time step is fed back into the network as input for the next time step, allowing the network to capture temporal dependencies in the data. The network can be trained using backpropagation through time (BPTT), which adjusts the weights to minimize the error between the predicted output and the actual output across the entire sequence. In some aspects, RNN models can incorporate gating mechanisms to better handle long-term dependencies. These gating mechanisms can allow the network to selectively update, forget, or output information at each time step. Additionally, RNN models can be stacked into multiple layers to create a deep RNN architecture, potentially increasing its capacity to model complex temporal patterns. In some cases, attention mechanisms can be integrated into RNN models to allow the network to focus on different parts of the input sequence when making predictions, which can be particularly useful for tasks involving long sequences or where certain parts of the input are more relevant than others.

[0088] LSTM network models are a type of RNN model that address the limitations of traditional RNN models in capturing long-range dependencies in sequential data. LSTM network models can incorporate a unique cell structure that includes a memory cell and gating mechanisms—i.e., input, output, and forget gates. These gates regulate the flow of information, allowing the network to selectively retain or forget data over time, enabling it to maintain important information across longer sequences and avoid issues like gradient vanishing or exploding. LSTM networks can be particularly effective for tasks involving time series data or natural language processing. In some cases, LSTM models can be stacked or combined with other neural network architectures to create more complex and powerful models for specific applications.

[0089] GRU models are similar to LSTM models, but can have a simpler architecture and can be computationally less demanding while still being effective at sequence data prediction. GRU models can handle sequence data by incorporating a gating mechanism that can regulate the flow of information through the network. The GRU structure is characterized by two main gates: an update gate that determines the amount of past information to retain, and a reset gate that controls the impact of new input data. These gates allow the GRU to dynamically maintain or modify its internal state, enabling the network to effectively capture temporal dependencies without the complexity of separate input and forget gates found in other architectures. In some aspects, GRU models can be particularly suitable for tasks involving shorter sequences or prioritizing computational efficiency. Furthermore, in certain applications, GRU models can be easier to train and can converge faster than LSTM models, making them a popular choice for various sequence modeling tasks in natural language processing and time series analysis.

[0090] Figure 9 is a schematic diagram illustrating a BESS subsystem 1300a including a power block 1301 according to one aspect of the present disclosure, which includes an implementation of the neural network model described with reference to Figure 8A and Figure 8B controller 1600.

[0091] The controller 1600 can include one or more processors 1602 (i.e., processing modules) configured to execute program instructions maintained in a memory 1604 (i.e., memory module). In this regard, the one or more processors 1602 of the controller 1600 can perform any of the various methods, processes, steps, or algorithms described throughout the present disclosure, e.g., implementation of the neural network model, including inputting the features 1503 into the energy prediction sub-model 1501 and the state prediction sub-model 1502, determining the chargeable / dischargeable energy 1504 for each iteration of a dynamic time period, outputting by the state prediction sub-model 1502 an operating state 1506 for the next iteration, determining whether to end execution of the sub-models 1501 and 1502 at a decision point 1505, and determining the total chargeable / dischargeable energy 1507.

[0092] Further, the controller 1600 can be configured to receive data, including but not limited to historical data of the operating states of the BESS subsystem (from the power block controller 1308 and / or the EMS 1309). The historical data of the BESS subsystem with a specific cell type (i.e., chemical composition such as Li NMC or LFP) can be used to train the sub-models 1501 and 1502. The historical data can include the voltage, the charge rate, and the maximum cell temperature of the BESS subsystem over a period of time (e.g., a few months). The historical data can be measured using sensors (e.g., voltage and temperature sensors). The sub-models 1501 and 1502 can be trained using the historical data 1605 derived from the same BESS site whose chargeable / dischargeable energy is to be determined. In some cases, the sub-models 1501 and 1502 can be trained using the historical data 1606 from different BESS sites (whose chargeable / dischargeable energy is to be determined). In some cases, the sub-models 1501 and 1502 can be trained using a combination of the historical data 1605 derived from the same BESS site and the historical data 1606 derived from different BESS sites. In some cases, the controller 1600 can receive initial operating states 1607 of the BESS subsystem 1300a, such as initial minimum cell voltage, initial maximum cell voltage, initial cell voltage imbalance, initial operating voltage, initial charge rate, initial maximum temperature, but the present disclosure is not limited thereto. In some cases, the controller 1600 can provide the total chargeable / dischargeable energy 1507 data to the EMS 1309 and / or the power block controller 1308, and then the EMS 1309 and / or the power block controller 1308 can decide when to discharge and how much energy to discharge.

[0093] The controller 1600 (i.e., computing device) can include a desktop computer, a mainframe computer system, a workstation, an image computer, a parallel processor, or any other computer system (e.g., networked computers). The one or more processors 1602 of the controller 1600 can include any processing element known in the art. In this regard, the one or more processors 1602 can include any microprocessor-type device configured to execute algorithms and / or instructions, such as an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a parallel processor, a graphics processing unit (GPU), a central processing unit (CPU), other chipsets, logic circuitry, and / or electronic processors. It is further recognized that the term “processor” can be broadly defined to encompass any device having one or more processing elements that execute program instructions from a non-transitory memory 1604. Moreover, the steps described throughout this disclosure can be performed by a single controller 1600 or, alternatively, by multiple controllers. For example, the power block controller 1308, the EMS 1309, the PCS 1306, and the controller 1600 can be the same controller or multiple controllers. Moreover, the controller 1600 can include one or more controllers housed in a common housing or within multiple housings. In this manner, any controller or combination of controllers can be individually packaged as a module suitable for integration into the BESS subsystem 1300a.

[0094] The memory 1604 can include any storage medium known in the art suitable for storing program instructions executable by the associated one or more processors 1602. For example, the memory 1604 can include a non-transitory storage medium. As another example, the storage medium 1604 can include, but is not limited to, read-only memory, random access memory, magnetic or optical storage devices (e.g., magnetic disks), magnetic tape, solid state drives, and the like. It is further noted that the memory 1604 can be installed in a common controller housing with the processors 1602. In some cases, the memory 1604 can be remotely located relative to the physical location of the processors 1602 and the controller 1600. For example, the one or more processors 1602 of the controller 1600 can access a remote memory (e.g., a server or cloud) accessible over a network (e.g., the Internet, an intranet, etc.).

[0095] In the foregoing, the present disclosure has been described in some detail by way of illustration with the accompanying drawings and aspects. However, the configurations described in the drawings and aspects herein are merely by way of example and not limiting of the aspects of the present disclosure. Therefore, it should be understood that various changes and modifications might be suggested by one skilled in the art, and many of which will be known or appreciated by those skilled in the art, that will fall within the spirit and scope of the appended claims.

Claims

1. A system for determining a total chargeable / dischargeable energy of a subsystem of a battery energy storage system, the subsystem comprising battery cells, the system comprising: a controller comprising one or more processing modules and one or more non-transitory memory storage modules storing computational instructions configured to, when executed by the one or more processing modules: perform an iterative process over a dynamic time period using a neural network model comprising an energy prediction submodel and a state prediction submodel, wherein the dynamic time period is divided into a plurality of iterations, wherein for each iteration of the plurality of iterations, the controller is configured to: (1) input into the energy prediction submodel: a voltage of the subsystem for a current iteration of the plurality of iterations, a charge rate of the subsystem for the current iteration, and a maximum temperature of the subsystem for the current iteration; wherein the energy prediction submodel is configured to output a chargeable / dischargeable energy of the subsystem for the current iteration; and (2) input into the state prediction submodel: the voltage of the subsystem for the current iteration, the charge rate of the subsystem for the current iteration, the maximum temperature of the subsystem for the current iteration, and a charge rate difference for the current iteration; wherein the charge rate difference for the current iteration is equal to the charge rate of the subsystem for the current iteration minus a charge rate of the subsystem for a previous iteration of the plurality of iterations; wherein the state prediction submodel is configured to output a voltage of the subsystem for a next iteration of the plurality of iterations, a charge rate of the subsystem for the next iteration, a maximum temperature of the subsystem for the next iteration, and a charge rate difference for the next iteration.

2. The system of claim 1, wherein, the voltage of the subsystem for a first iteration of the plurality of iterations is determined using a voltage imbalance of the subsystem for the first iteration according to the following equation: V t = CVmin t / (1-α ε) where CVmin t is the minimum cell voltage of the battery cells of the subsystem for the first iteration, ε is the voltage imbalance of the subsystem for the first iteration, and a is a scaling factor, wherein the voltage imbalance is a function of the minimum cell voltage of the subsystem for the first iteration and a maximum cell voltage of the subsystem for the first iteration.

3. The system of claim 1, wherein, the controller is configured to: repeat steps (1) and (2) until a last iteration of the iterative process is performed.

4. The system of claim 3, wherein, the controller is configured to: determine the total chargeable / dischargeable energy of the subsystem of the battery energy storage system over the dynamic time period by summing the chargeable / dischargeable energy output by the energy prediction submodel for each iteration.

5. The system of claim 1, wherein, the controller is configured to: repeat steps (1) and (2) until the voltage of the subsystem for the next iteration of the iterative process output by the state prediction submodel reaches a voltage limit.

6. The system of claim 5, wherein, the controller is configured to: determining the total chargeable / dischargable energy of the subsystem of the battery energy storage system over the dynamic time period by summing the chargeable / dischargable energy output by the energy predictor model for each iteration.

7. The system of claim 5, wherein, the dynamic time period is determined by the time required to charge or discharge the subsystem of the battery energy storage system from the first iteration of the iterative process to the last iteration of the iterative process, wherein the last iteration of the iterative process is the iteration in which the voltage of the subsystem output by the state predictor model reaches the voltage limit.

8. The system of claim 1, wherein, the subsystem is selected from at least one of: a plurality of battery packs comprising the battery monoblocs, or a plurality of battery racks comprising the battery packs.

9. The system of claim 1, wherein, the battery type of the battery monoblocs of the subsystem is selected from the group comprising: lithium nickel manganese cobalt (NMC), lithium iron phosphate (LFP), lithium sulfur (Li-S), solid state, nickel cadmium, nickel metal hydride (NiMH), zinc air, iron air, vanadium redox flow, sodium ion, potassium ion, aluminum ion, lead acid, silicon anode, or a combination thereof.

10. The system of claim 1, wherein, the architecture of the neural network model is selected from the group consisting of: a multi-layer perceptron (MLP), a recurrent neural network (RNN), a long short-term memory network, a gated recurrent unit (GRU), or a combination thereof.

11. A method for determining a total chargeable / dischargable energy of a subsystem of a battery energy storage system, the subsystem comprising battery monoblocs, the method comprising: performing an iterative process over a dynamic time period using a neural network model comprising an energy predictor model and a state predictor model, wherein the dynamic time period is divided into a plurality of iterations, wherein each iteration of the plurality of iterations comprises the steps of: (1) inputting into the energy predictor model: a voltage of the subsystem for a current iteration of the plurality of iterations, a charge rate of the subsystem for the current iteration, and a maximum temperature of the subsystem for the current iteration; wherein the energy predictor model outputs a chargeable / dischargable energy of the subsystem for the current iteration; and (2) inputting into the state predictor model: the voltage of the subsystem for the current iteration, the charge rate of the subsystem for the current iteration, the maximum temperature of the subsystem for the current iteration, and a charge rate difference for the current iteration; wherein the charge rate difference for the current iteration is equal to the charge rate of the subsystem for the current iteration minus a charge rate of the subsystem for a previous iteration of the plurality of iterations; wherein the state predictor model outputs a voltage of the subsystem for a next iteration of the plurality of iterations, a charge rate of the subsystem for the next iteration, a maximum temperature of the subsystem for the next iteration, and a charge rate difference for the next iteration.

12. The method of claim 11, wherein, The voltage of the subsystem for the first iteration of the plurality of iterations is determined using a voltage imbalance for the subsystem for the first iteration according to the following equation: V t = CVmin t / (1-α ε) where CVmin t is the minimum cell voltage of the battery cells of the subsystem for the first iteration, ε is the voltage imbalance of the subsystem for the first iteration, and a is a scaling factor, where the voltage imbalance is a function of the minimum cell voltage for the subsystem for the first iteration and a maximum cell voltage for the subsystem for the first iteration.

13. The method of claim 11, further comprising: repeating steps (1) and (2) until a last iteration of the iterative process is performed.

14. The method of claim 13, further comprising: determining the total chargeable / dischargeable energy of the subsystem of the battery energy storage system over the dynamic time period by summing the chargeable / dischargeable energy output by the energy predictor submodel for each iteration.

15. The method of claim 11, further comprising: repeating steps (1) and (2) until the voltage of the subsystem for a next iteration of the iterative process output by the state predictor submodel reaches a voltage limit.

16. The method of claim 15, further comprising: determining the total chargeable / dischargeable energy of the subsystem of the battery energy storage system over the dynamic time period by summing the chargeable / dischargeable energy output by the energy predictor submodel for each iteration.

17. The method of claim 15, wherein, the dynamic time period is determined by a time required to charge or discharge the subsystem of the battery energy storage system from a first iteration of the iterative process to a last iteration of the iterative process, wherein the last iteration of the iterative process is an iteration in which the voltage of the subsystem output by the state predictor submodel reaches the voltage limit.

18. The method of claim 11, wherein, the subsystem is selected from at least one of a plurality of battery packs comprising the battery cells, or a plurality of battery racks comprising the battery packs.

19. The method of claim 11, wherein, the battery type of the battery cells of the subsystem is selected from a group consisting of lithium nickel manganese cobalt (NMC), lithium iron phosphate (LFP), lithium sulfur (Li-S), solid state, nickel cadmium, nickel metal hydride (NiMH), zinc air, iron air, vanadium redox flow, sodium ion, potassium ion, aluminum ion, lead acid, silicon anode, or a combination thereof.

20. The method of claim 11, wherein, the architecture of the neural network model is selected from a group comprising a multilayer perceptron (MLP), a recurrent neural network (RNN), a long short-term memory network, a gated recurrent unit (GRU), or a combination thereof.

Images