Apparatus and method for cooling ESS

Abstract

According to the present disclosure, an apparatus and method for cooling an ESS can be provided, comprising: a plurality of cooling plates for cooling each of a plurality of battery packs stored in a container of the ESS; a plurality of temperature sensors for measuring the temperature of each of the plurality of battery packs; a plurality of pipelines for inputting refrigerant to the plurality of cooling plates; a chiller for supplying refrigerant to the cooling plates through the pipelines; a valve connected to each of the plurality of pipelines for controlling the flow rate of refrigerant supplied from the chiller to the cooling plates; and a controller for controlling the opening rate of the valve by considering the path of the plurality of pipelines in order to maintain the temperature of the plurality of battery packs measured using the temperature sensors uniformly.

Description

Technology Field

[0001] The present disclosure relates to an apparatus and method for cooling an ESS. Background Technology

[0002] Rechargeable batteries are rechargeable and dischargeable batteries. They are used in electric vehicles, Energy Storage Systems (ESS), and portable electronic devices. An Energy Storage System is a system capable of storing large-scale power using multiple battery models. Electricity generated from renewable energy power plants, such as solar or wind power, can be stored in the Energy Storage System and then supplied to the commercial power network when needed. Heat may be generated when the Energy Storage System performs charging or discharging. Excessive heat can accelerate the degradation of battery cells or cause fires. Prior art literature

[0003] (Patent Document 0001) KR 10-2024-0121937 A The problem to be solved

[0004] According to one aspect of the present disclosure, an apparatus and method for cooling an ESS using a proportional control solenoid valve to dynamically control the flow rate of cooling water for cooling a battery pack according to the real-time temperature of a plurality of battery packs of the ESS can be provided.

[0005] According to one aspect of the present disclosure, an apparatus and method for cooling an ESS can be provided, which controls the flow of cooling water to maintain temperature uniformity of a plurality of battery packs of an ESS while minimizing the power consumed to cool a plurality of battery packs of an ESS using a reinforcement learning algorithm. means of solving the problem

[0006] According to one aspect of the present disclosure, a device for cooling an ESS may include a plurality of cooling plates for cooling each of a plurality of battery packs housed in a container of the ESS, a plurality of temperature sensors for measuring the temperature of each of the plurality of battery packs, a plurality of pipelines for inputting refrigerant to the plurality of cooling plates, a chiller for supplying refrigerant to the cooling plates through the pipelines, a valve connected to each of the plurality of pipelines for controlling the flow rate of refrigerant supplied from the chiller to the cooling plates, and a controller for controlling the opening rate of the valve by considering the path of the plurality of pipelines in order to maintain the temperature of the plurality of battery packs measured using the temperature sensors uniformly.

[0007] According to one embodiment, the path of the pipeline may include one or more of the location of the battery pack, the pipe size of the pipeline, the distance of the pipeline from the chiller to the cooling plate, the branch pipe structure of the pipeline, and the ambient temperature of the pipeline, and the controller may, considering the path of the pipeline, control the opening rate of the valve connected to the pipeline where the temperature of the refrigerant reaching the cooling plate is relatively high to be relatively large, and control the opening rate of the valve connected to the pipeline where the temperature of the refrigerant reaching the cooling plate is relatively low to be relatively small.

[0008] According to one embodiment, the controller can control the opening rate of a valve connected to a pipeline supplying refrigerant to a cooling plate connected to a battery pack far from the chiller to be greater than the opening rate of a valve connected to a pipeline supplying refrigerant to a cooling plate connected to a battery pack close to the chiller.

[0009] According to one embodiment, the valve may be a solenoid valve whose opening rate is controlled by a control signal output from the controller.

[0010] According to one embodiment, the controller inputs the temperature of the plurality of battery packs, which is updated in real time, the opening rate of the valve connected to each of the plurality of pipelines, and the operating rate of the chiller as a starting state to a cooling control model, and controls the plurality of valves and chillers using the opening rate of the valve connected to each of the plurality of pipelines and the operating rate of the chiller output by the cooling control model as an action, and the cooling control model may be a reinforcement-learned artificial intelligence model that, upon receiving the temperature of the plurality of battery packs, the opening rate of the valve connected to each of the plurality of pipelines, and the operating rate of the chiller, outputs the opening rate of the valve connected to each of the plurality of pipelines and the operating rate of the chiller to maintain the temperature of the plurality of battery packs within a target temperature range.

[0011] According to one embodiment, the cooling control model may be generated by using a Deep Q-Learning algorithm to learn state-action pairs in which the temperature of the plurality of battery packs, the opening rate of the valve connected to each of the plurality of pipelines, and the operating rate of the chiller are in a starting state, and the opening rate of the valve connected to each of the plurality of pipelines and the operating rate of the chiller are in a behavior, and by using a Proximal Policy Optimization algorithm to continuously control the opening rate of the valve and the operating rate of the chiller.

[0012] According to one embodiment, the cooling control model is an artificial intelligence model that has been reinforced using a policy that provides a reward when the temperature of the plurality of battery packs is maintained within a target range, increases the reward when the power required for the operation of the chiller is reduced, and imposes a penalty when the temperature of the plurality of battery packs is unbalanced or when overcooling or overheating occurs, and the controller can control the valve and the chiller according to the opening rate of the valve connected to each of the plurality of pipelines and the operating rate of the chiller output by the cooling control model in order to minimize the power consumption required for cooling while maintaining the temperature of the plurality of battery packs within the target temperature range.

[0013] According to one aspect of the present disclosure, a plurality of temperature sensors installed in each of the plurality of battery packs of an ESS may measure the temperature of the battery packs; a controller may determine the opening rate of a valve connected to a pipeline by considering the path of a pipeline that supplies a refrigerant to a cooling plate that cools the battery packs in order to maintain the temperature of the plurality of battery packs uniformly, based on the temperature of the battery packs received from the temperature sensors; and the controller may output a control signal to the valve according to the opening rate of the valve.

[0014] According to one embodiment, the path of the pipeline may include one or more of the location of the battery pack, the pipe size of the pipeline, the distance the pipeline passes from the chiller to the cooling plate, and the branch pipe structure of the pipeline, and the step of determining the opening rate of the valve may involve the controller considering the path of the pipeline and determining the opening rate of the valve connected to the pipeline where the temperature of the refrigerant reaching the cooling plate is relatively high to be relatively large, and determining the opening rate of the valve connected to the pipeline where the temperature of the refrigerant reaching the cooling plate is relatively low to be relatively small.

[0015] According to one embodiment, the step of determining the opening rate of the valve may determine the opening rate of the valve connected to the pipeline supplying refrigerant to a cooling plate connected to a battery pack far from the chiller to be greater than the opening rate of the valve connected to the pipeline supplying refrigerant to a cooling plate connected to a battery pack close to the chiller.

[0016] According to one embodiment, a method for cooling an ESS may further include the step of generating a cooling control model by reinforcing a state-action pair in which the temperature of the plurality of battery packs, the opening rate of the valve connected to each of the plurality of pipelines, and the operating rate of the chiller are in a starting state, and the opening rate of the valve connected to each of the plurality of pipelines and the operating rate of the chiller are in a action, and the step of determining the opening rate of the valve may be to input the temperature of the plurality of battery packs, the opening rate of the valve connected to each of the plurality of pipelines, and the operating rate of the chiller, which are updated in real time by the controller, into the cooling control model, and the opening rate of the valve connected to each of the plurality of pipelines and the operating rate of the chiller, which are actions output by the cooling control model, may be determined as the opening rate of the valve and the operating rate of the chiller for controlling the valve and the chiller.

[0017] According to one embodiment, a Deep Q-Learning algorithm may be used to learn state-action pairs in which the temperature of the plurality of battery packs, the opening rate of the valve connected to each of the plurality of pipelines, and the operating rate of the chiller are in a starting state, and the opening rate of the valve connected to each of the plurality of pipelines and the operating rate of the chiller are in a behavior, and a Proximal Policy Optimization algorithm may be used to continuously control the opening rate of the valve and the operating rate of the chiller.

[0018] According to one embodiment, the step of generating the cooling control model may reinforce the cooling control model with a policy of providing a reward when the temperature of the plurality of battery packs is maintained within a target range, increasing the reward when the power required for the operation of the chiller is reduced, and imposing a penalty when the temperature of the plurality of battery packs is unbalanced or when overcooling or overheating occurs; the step of determining the valve opening rate may determine the valve opening rate and chiller operating rate connected to each of the plurality of pipelines output by the cooling control model as the valve opening rate and chiller operating rate for controlling the valve and chiller in order to minimize the power consumed for cooling while maintaining the temperature of the plurality of battery packs within the target temperature range; and the step of outputting a control signal to the valve may perform the operation of the controller outputting a control signal to the valve and chiller according to the determined valve opening rate and chiller operating rate.

[0019] The features and advantages of the present disclosure will become more apparent from the following detailed description based on the accompanying drawings.

[0020] Prior to this, terms and words used in this specification and claims should not be interpreted in their ordinary and dictionary meanings, but should be interpreted in a meaning and concept consistent with the technical spirit of this disclosure, based on the principle that the inventor can appropriately define the concept of the terms to best describe his invention. Effects of the invention

[0021] According to one embodiment of the present disclosure, the temperature of a plurality of battery packs of an ESS can be maintained uniformly, and accordingly, the rate of degradation of battery cells can be reduced, thereby reducing the maintenance costs of the ESS.

[0022] According to one embodiment of the present disclosure, energy consumed to cool a plurality of battery packs of an ESS can be minimized, performance degradation of the ESS can be prevented, operating costs can be reduced, and the lifespan can be extended. Brief explanation of the drawing

[0023] FIG. 1 is a drawing showing an ESS equipped with a cooling device for the ESS according to one embodiment. FIG. 2 is a drawing showing a device for cooling an ESS according to one embodiment. FIG. 3 is a diagram showing the path of a pipeline according to one embodiment. FIG. 4 is a diagram showing a cooling control model according to one embodiment. FIG. 5 is a flowchart showing each step of a method for cooling an ESS according to one embodiment. FIG. 6 is a diagram illustrating the steps of generating a cooling control model according to one embodiment. Specific details for implementing the invention

[0024] The purpose, specific advantages, and novel features of an embodiment of the present disclosure will become more apparent from the following detailed description and specific embodiments in conjunction with the accompanying drawings. It should be noted that in assigning reference numerals to the components of each drawing in this specification, the same components are assigned the same number whenever possible, even if they are shown in different drawings. In the following description of an embodiment of the present invention, detailed descriptions of related prior art that could unnecessarily obscure the essence of the embodiment of the present invention are omitted.

[0025] Hereinafter, an embodiment of the present disclosure will be described in detail with reference to the attached drawings.

[0026] FIG. 1 is a drawing showing an ESS (1) with a cooling device (10) installed according to one embodiment.

[0027] An Energy Storage System (ESS (1)) is a device that stores and releases electrical energy using a large number of battery packs (110). In this document, the Energy Storage System may be briefly referred to as ESS (1). ESS (1) may include a plurality of containers (100). A container (100) may include a plurality of battery packs (110). A battery pack (110) may include a plurality of battery cells (111). ESS (1) can perform charging or discharging using a plurality of battery packs (110) stored in a container (100). ESS (1) can perform charging or discharging by being connected to a commercial power network. ESS (1) can be connected to various power plants, such as solar power plants, wind power plants, and fuel cell power plants, to store electrical energy produced by the power plants and then discharge power to the commercial power network.

[0028] Heat may be generated when charging and discharging are performed on the battery cell (111). When heat is generated in the battery pack (110) containing multiple battery cells (111), cooling is required. A device (10) for cooling an ESS according to one embodiment can cool the battery pack (110) contained in the container (100) using a refrigerant cooled to a low temperature in a chiller (220). The refrigerant cooled to a low temperature in the chiller (220) can be supplied to the container (100) using a supply pipeline (210S), and the refrigerant, which has become high temperature after receiving heat from the battery pack (110), can be returned to the chiller (220) using a recovery pipeline (210R). The chiller (220) can supply the refrigerant through the pipeline (210) and perform the operation of cooling the refrigerant again after it has passed through the container (100) and its temperature has risen. The chiller (220) can supply refrigerant to a plurality of containers (100). The chiller (220) may include a pump. The pump of the chiller (220) can pump the refrigerant at a predetermined pressure so that the refrigerant is supplied to the cooling plate (120) along the pipeline (210). Thus, the flow rate of the refrigerant supplied to the cooling plate (120) can be controlled simply by controlling the opening rate of the valve (230).

[0029] A device (10) for cooling an ESS according to one embodiment may include a plurality of chillers (220). A single chiller (220) may supply refrigerant to a plurality of containers (100), a plurality of chillers (220) may supply refrigerant to a single container (100), or a plurality of chillers (220) may supply refrigerant to a plurality of containers (100).

[0030] The controller (300) can control a valve (230) that controls the amount of refrigerant supplied from the chiller (220) to the battery pack (110) through the pipeline (210). To individually control the amount of refrigerant supplied to the battery pack (110), one valve (230) may be connected to each pipeline (210) that is directly connected to the battery pack (110). When the controller (300) adjusts the opening rate of a specific valve (230), the flow rate of the refrigerant supplied to the cooling plate (120) that cools the battery pack (110) through the pipeline (210) to which the valve (230) is connected can be controlled. What is important for cooling the battery pack (110) is the supply pipeline (210S) that supplies refrigerant to the cooling plate (120). In the following, the path of the pipeline (210) that the controller (300) considers to control the opening rate of the valve (230) refers to the path of the supply pipeline (210S). Therefore, unless specifically referred to as the recovery pipeline (210R), the pipeline (210) can be understood as the supply pipeline (210S).

[0031] The controller (300) can also control the valve (230) that controls the supply of refrigerant on a container (100) basis. When the controller (300) controls the opening rate of the valve (230) of the pipeline (210) connected to a specific container (100), the flow rate of the refrigerant supplied to the container (100) can be controlled. The temperature of the refrigerant supplied to the container (100) may vary depending on the location of the container (100), the distance between the container (100) and the chiller (220), and the ambient temperature at the location where the pipeline (210) passes from the chiller (220) to the container (100). The controller (300) can control the opening rate of the valve (230) to increase the flow rate of the refrigerant supplied to the container (100) at a relatively high temperature. The controller (300) can adjust the opening rate of the valve (230) to reduce the flow rate of the refrigerant in the container (100) supplied with the refrigerant at a relatively low temperature. The statement that the refrigerant temperature is relatively high or low may mean that the refrigerant temperature is different for each pipeline (210), and that the temperature of the refrigerant transported by a specific pipeline (210) is higher or lower than the average of the temperatures of the refrigerants transported by other pipelines (210). Alternatively, the statement that the refrigerant temperature is relatively high or low may mean that the temperature of the refrigerant transported by a specific pipeline (210) exceeds or falls below the temperature range in which the temperatures of the refrigerants transported by multiple other pipelines (210) are most distributed.

[0032] FIG. 2 is a drawing showing a device (10) for cooling an ESS according to one embodiment. Refer to FIG. 1 together.

[0033] A device (10) for cooling an ESS may include a plurality of cooling plates (120) for cooling each of a plurality of battery packs (110) stored in a container (100) of an ESS (1), a plurality of temperature sensors (112) for measuring the temperature of each of the plurality of battery packs (110), a plurality of pipelines (210) for inputting refrigerant to the plurality of cooling plates (120), a chiller (220) for supplying refrigerant to the cooling plates (120) through the pipelines (210), a valve (230) connected to each of the plurality of pipelines (210) for controlling the flow rate of refrigerant supplied from the chiller (220) to the cooling plates (120), and a controller (300) for controlling the opening rate of the valve (230) by considering the path of the plurality of pipelines (210) to maintain the temperature of the plurality of battery packs (110) measured using the temperature sensors (112) uniformly.

[0034] The container (100) may include a cooling plate (120) to cool each of the plurality of battery packs (110). The cooling plate (120) can cool the heat generated by the battery cell (111) and the heat generated by the pack BMS (113). The cooling plate (120) may be in contact with one side of the battery pack (110). The cooling plate (120) can cool the battery pack (110) using a refrigerant received through a supply pipeline (210S). The cooling plate (120) can receive heat from the battery pack (110) and discharge the refrigerant, which has become relatively high temperature, through a recovery pipeline (210R).

[0035] The cooling plate (120) may be located inside the battery pack (110). In this case, the battery pack (110) may include a battery cell, a pack BMS (113), and a cooling plate (120). To allow the battery pack (110) to be connected to a container (100) so that a refrigerant can be introduced into the cooling plate (120), the battery pack (110) may include a connector to which the cooling plate (120) can be connected to a supply pipeline (210S) and a recovery pipeline (210R). Although FIG. 2 depicts the cooling plate (120) as being located outside the battery pack (110), it can be understood that the technical content presented in this document can be applied in the same way even if the cooling plate (120) is located inside the battery pack (110).

[0036] The chiller (220) can supply a relatively low-temperature refrigerant to the cooling plate (120) through the pipeline (210). The chiller (220) can cool the relatively high-temperature refrigerant received through the recovery pipeline (210R).

[0037] A supply pipeline (210S) can connect between each cooling plate (120) in a chiller (220). A valve (230) can be connected to the supply pipeline (210S). The valve (230) can have its opening rate adjusted according to a control signal provided by a controller (300). The valve (230) may include a proportional control solenoid valve (230) in which the opening rate is controlled by a control signal output from the controller (300). A valve (230) is connected to each of the multiple pipelines (210) so that the flow rate of the refrigerant supplied to each of the multiple cooling plates (120) can be controlled.

[0038] The battery pack (110) may include a temperature sensor (112). The temperature sensor (112) may measure the temperature of the battery pack (110). The temperature of the battery pack (110) may vary depending on the heat generated by the battery cell (111) and the heat generated by the pack BMS (113). The temperature sensor (112) may measure the temperature of the battery pack (110) in real time and transmit the measured temperature to the controller (300). Alternatively, the temperature sensor (112) may transmit the measured temperature to the pack BMS (113), and the pack BMS (113) may transmit the temperature to the controller (300). Alternatively, the temperature sensor (112) may transmit the measured temperature to the pack BMS (113), the pack BMS (113) may transmit the temperature to a container BMS (not shown) that manages the container (100), and the container (100) BMS may transmit the temperature to the controller (300). In the process of providing the temperature measured by a plurality of temperature sensors (112) to the controller (300), the method of collecting multiple temperature data and transmitting them to the controller (300) may be changed.

[0039] The controller (300) may include a PC, a server computer, a PLC, a PANEL, or other information processing device. The controller (300) may include a processor (310) and a memory (320) that is connected to the processor (310) to transmit and receive data and stores program code and data. The processor (310) may include a CPU, a GPU, an ASIC, or other information processing integrated circuit or semiconductor device. The memory (320) may include RAM, ROM, temporary or non-temporary memory (320), a hard disk, magnetic tape, etc. The memory (320) may store program code written to perform each step of a method for cooling the ESS (1). The processor (310) may read and execute the program code stored in the memory (320) to perform a method for cooling the ESS (1).

[0040] The controller (300) may further include an input / output interface (330). The input / output interface (330) may include an input device and an output device. The input device may include a mouse, keyboard, USB port, touchscreen, button, etc., used by a user to input commands or data into the controller (300). The output device may include a display, speaker, lamp, printer, etc., used by the controller (300) to provide information to the user.

[0041] The controller (300) may further include a communication interface (340). The communication interface (340) may be connected to a wired or wireless network to communicate with another remote control system, or to communicate with a container BMS or a pack BMS (113). The communication interface (340) may include a circuit, an antenna, and a semiconductor chip capable of performing LAN, WAN, Ethernet, IPv4, IPv6, Wi-Fi, Zigbee, 5G, 6G, or other wired or wireless communication methods.

[0042] The controller (300) can control each of the plurality of valves (230) that control the flow rate of the refrigerant input to the cooling plate (120). The controller (300) can recognize the temperature of the battery pack (110) received from the temperature sensor (112). Based on the real-time temperature of the plurality of battery packs (110), the controller (300) can adjust the amount of refrigerant supplied to the cooling plate (120) so that the temperature of the plurality of battery packs (110) becomes uniform. The controller (300) can adjust the opening rate of the plurality of valves (230) to control the amount of refrigerant supplied to each of the plurality of cooling plates (120) that cool the plurality of battery packs (110). The opening rate of the valve (230) refers to the degree to which the valve (230) is open. If the opening rate is 100%, it means that the valve (230) is fully open, and if the opening rate is 0%, it means that the valve (230) is closed.

[0043] The controller (300) can adjust the opening rate of the valve (230) so that the temperature of a plurality of battery packs (110) is within a target range. The controller (300) can determine a large opening rate of the valve (230) to supply a large flow rate of refrigerant to the cooling plate (120) that cools the battery pack (110) with a high temperature, and a small opening rate of the valve (230) to supply a relatively small flow rate of refrigerant to the cooling plate (120) that cools the battery pack (110) with a relatively low temperature.

[0044] The controller (300) may consider the path of multiple pipelines (210) to determine the opening rate of the valve (230). This is because the temperature of the refrigerant supplied to each of the multiple cooling plates (120) varies depending on the path of the pipeline (210). If the same flow rate of refrigerant is supplied to all cooling plates (120) without considering the path of the pipeline (210), it is difficult to maintain the temperature of each battery pack (110) uniformly.

[0045] FIG. 3 is a diagram showing the path of a pipeline (210) according to one embodiment. FIG. 1, FIG. 2, and FIG. 3 are referenced together.

[0046] The path of the pipeline (210) is a concept that includes factors that can affect the temperature of the refrigerant during the process of transporting the refrigerant from the chiller (220) to the cooling plate (120). The path of the pipeline (210) may include one or more of the following: the location of the battery pack (110), the pipe size of the pipeline (210), the distance of the pipeline (210) from the chiller (220) to the cooling plate (120), the branch pipe structure of the pipeline (210), and the ambient temperature of the pipeline (210). The path of the pipeline (210) may include additional factors in addition to the factors described.

[0047] The position of the battery pack (110) refers to the location where the battery pack (110) is stored in the container (100). If the battery pack (110) is located at the edge (e.g., battery pack A (A in FIG. 3)), the temperature of the battery pack (110) may be affected by external wind, sunlight, and ambient temperature. If the battery pack (110) is located in the center of the container (100) (e.g., battery pack C (C in FIG. 3)), other battery packs (110) are located above, below, to the left, and to the right of the battery pack (110), making heat dissipation difficult and causing the temperature of the battery pack (110) to rise quickly. If the position of the battery pack (110) in the container (100) is different, the distance from the chiller (220) to the cooling plate (120) is different. Therefore, if the position of the battery pack (110) is different, the temperature of the refrigerant delivered to the cooling plate (120) may be different.

[0048] The temperature, flow rate, and pressure of the refrigerant delivered to the cooling plate (120) may differ between a pipeline (210) with a large pipe size and a pipeline (210) with a small pipe size. Therefore, it is necessary to consider the pipe size (diameter).

[0049] The longer the distance of the pipeline (210) from the chiller (220) to the cooling plate (120), the higher the temperature of the refrigerant supplied to the cooling plate (120). This is because the refrigerant cooled in the chiller (220) can absorb ambient temperature as it travels to the cooling plate (120). The distance of the pipeline (210) from the chiller (220) to the cooling plate (120) can be a major cause of the difference in the temperature of the refrigerant supplied to the cooling plate (120).

[0050] The branch pipe structure of the pipeline (210) can be located in the portion supplying refrigerant from the chiller (220) to a plurality of containers (100), and can be located in the portion supplying refrigerant to a plurality of cooling plates (120) within the container (100). Since the branch pipe structure of the pipeline (210) affects the pressure and flow rate of the refrigerant supply, it becomes a factor that the controller (300) considers when determining the opening rate of the valve (230).

[0051] The ambient temperature of the pipeline (210) is a factor to consider because the refrigerant is supplied along the pipeline (210) and absorbs heat from the surroundings. The ambient temperature of the pipeline (210) may include the ambient temperature or the temperature inside the container (100).

[0052] The controller (300) can control the opening rate of the valve (230) connected to the pipeline (210) where the temperature of the refrigerant reaching the cooling plate (120) is relatively high by considering the path of the pipeline (210), and can control the opening rate of the valve (230) connected to the pipeline (210) where the temperature of the refrigerant reaching the cooling plate (120) is relatively low by controlling it relatively low.

[0053] The controller (300) can determine the opening rate of the valve (230) according to the temperature of the refrigerant supplied to the cooling plate (120) by considering the path of the pipeline (210). The temperature of the refrigerant may be higher as the distance between the chiller (220) and the battery pack (110) increases. The temperature of the refrigerant may be higher as the temperature around the path through which the pipeline (210) passes increases. Also, the temperature of the refrigerant may be higher as the time it travels inside the pipeline (210) increases depending on the pipe size of the pipeline (210). By considering these factors, the controller (300) can determine the opening rate of the valve (230).

[0054] When the temperature of the refrigerant supplied to the cooling plate (120) is relatively high, a large flow rate of the refrigerant is required to cool the battery pack (110) to a desired temperature. Therefore, the controller (300) can determine a relatively large opening rate of the valve (230) connected to the pipeline (210). When the temperature of the refrigerant supplied to the cooling plate (120) is relatively low, a relatively small flow rate of the refrigerant may be required to cool the battery pack (110) to a desired temperature. Therefore, the controller (300) can determine a relatively small opening rate of the valve (230) connected to the pipeline (210).

[0055] Specifically, the controller (300) can control the opening rate of the valve (230) connected to the pipeline (210) that supplies refrigerant to the cooling plate (120) connected to the battery pack (110) that is far from the chiller (220) to be greater than the opening rate of the valve (230) connected to the pipeline (210) that supplies refrigerant to the cooling plate (120) connected to the battery pack (110) that is close to the chiller (220).

[0056] The refrigerant supplied to the cooling plate (120) cooling battery pack A (A) may have a relatively low temperature. The refrigerant supplied to the cooling plate (120) cooling battery pack B (B) may have a relatively high temperature. The temperature of the refrigerant supplied to the cooling plate (120) cooling battery pack C (C) may be approximately the middle temperature between the temperature of the refrigerant supplied to the cooling plate (120) cooling battery pack A (A) and the temperature of the refrigerant supplied to the cooling plate (120) cooling battery pack B (B). This difference in refrigerant temperature is because the temperature rise of the refrigerant along the pipeline (210) path may be low because the distance between battery pack A (A) and the chiller (220) is short, and the temperature rise of the refrigerant along the pipeline (210) path may be high because the distance between battery pack B (B) and the chiller (220) is long. Therefore, in order to control the flow rate of the refrigerant supplied to the cooling plate (120) that cools battery pack B (B) to be greater than the flow rate of the refrigerant supplied to the cooling plate (120) that cools battery pack A (A), the opening rate of the valve (230) of the pipeline (210) connected to the cooling plate (120) that cools battery pack B (B) can be determined to be relatively large.

[0057] The location of the battery pack (110), the distance of the pipeline (210), the branch pipe structure, etc., may be data stored in the controller (300). The controller (300) can determine the opening rate of the valve (230) such that the flow rate supplied to the cooling plate (120) is greater the greater the distance between the chiller (220) and the cooling plate (120).

[0058] When the opening rate of the valve (230) is determined, the controller (300) can control the opening rate of the valve (230) by outputting a control signal to the valve (230). The valve (230) can be partially opened or partially closed according to the control signal, thereby controlling the flow rate of the refrigerant supplied to the cooling plate (120).

[0059] FIG. 4 is a drawing showing a cooling control model (350) according to one embodiment. Refer to FIG. 1, FIG. 2, FIG. 3, and FIG. 4 together.

[0060] Because the number of containers (100) included in the ESS (1) is large, and the number of battery packs (110) in each container (100) is large, and the path of the pipeline (210) is different for each container (100) and battery pack (110), it may be difficult for the controller (300) to determine the opening rate of all valves (230) based on rules. In one embodiment, the controller (300) can determine the opening rate of multiple valves (230) using an artificial intelligence model generated using reinforcement learning.

[0061] The controller (300) inputs the temperature of a plurality of battery packs (110) that are updated in real time, the opening rate of a valve (230) connected to each of a plurality of pipelines (210), and the operating rate of a chiller (220) as a starting state to the cooling control model (350), and can control the valve (230) and the chiller (220) using the opening rate of the valve (230) connected to each of a plurality of pipelines (210) and the operating rate of the chiller (220) that the cooling control model (350) outputs as an action.

[0062] The cooling control model (350) may be a reinforcement-learned artificial intelligence model that, upon receiving inputs of the temperature of a plurality of battery packs (110), the opening rate of a valve (230) connected to each of a plurality of pipelines (210), and the operating rate of a chiller (220), outputs the opening rate of the valve (230) connected to each of a plurality of pipelines (210) and the operating rate of the chiller (220) in order to maintain the temperature of the plurality of battery packs (110) within a target temperature range.

[0063] The temperature of the multiple battery packs (110) can be measured in real time from the temperature sensor (112) of each battery pack (110). The opening rate of the valve (230) connected to each of the multiple pipelines (210) is substantially the same as the flow rate of the refrigerant flowing into the cooling plate (120) through the multiple pipelines (210). The operating rate of the chiller (220) may refer to the performance of the chiller (220) in cooling the refrigerant, the performance of the chiller (220) in pumping the refrigerant into the pipeline (210), and the amount of heat that the chiller (220) can cool overall. The cooling control model (350) may also learn the ambient temperature of the container (100) as a starting state. The temperature of the multiple battery packs (110), the opening rate of the valve (230), the operating rate of the chiller (220), the ambient temperature of the container (100), and other environmental factors can collectively be referred to as the real-time state of the ESS.

[0064] The cooling control model (350) is an artificial intelligence model trained using a reinforcement learning algorithm. When the cooling control model (350) receives the temperature of a plurality of battery packs (110) that is updated in real time, the opening rate of a valve (230) connected to each of a plurality of pipelines (210), and the operating rate of a chiller (220) as inputs as a starting state, it can output the opening rate of the valve (230) connected to each of a plurality of pipelines (210) and the operating rate of the chiller (220) as actions based on the learned information.

[0065] The cooling control model (350) can be generated by learning state-action pairs, evaluating selected actions by a reward policy, and modifying the action selection method by feedback. Specific details for generating the cooling control model (350) will be described later. The cooling control model (350) can be generated in the controller (300). Alternatively, the cooling control model (350) can be generated in a computer device other than the controller (300) and then stored in the controller (300) for use.

[0066] The cooling control model (350) may be a reinforcement-learned artificial intelligence model that uses a policy to provide a reward when the temperature of a plurality of battery packs (110) is maintained within a target range, increase the reward when the power required for the operation of the chiller (220) is reduced, and impose a penalty when the temperature of the plurality of battery packs (110) is unbalanced or when overcooling or overheating occurs. The controller (300) can control the valve (230) and the chiller (220) according to the opening rate of the valve (230) connected to each of the plurality of pipelines (210) and the operating rate of the chiller (220) output by the cooling control model (350) in order to minimize the power consumption required for cooling while maintaining the temperature of the plurality of battery packs (110) within a target temperature range.

[0067] The cooling control model (350) can be trained to select an action by a policy. The policy may include not only maintaining the temperature of multiple battery packs (110) within a target temperature range, but also increasing the reward when the power consumed for the operation of the chiller (220) is reduced. Thus, the cooling control model (350) can be trained to select an action that minimizes the power consumed to cool the ESS.

[0068] The controller (300) can control the valve (230) and the chiller (220) together based on the operating rate of the chiller (220) along with the opening rate of the valve (230). By controlling the valve (230) and the chiller (220) together, the temperature of multiple battery packs (110) can be uniformly controlled at a low cost.

[0069] A cooling control model (350) can be generated by using a Deep Q-Learning algorithm to learn state-action pairs in which the temperature of a plurality of battery packs (110), the opening rate of a valve (230) connected to each of a plurality of pipelines (210), and the operating rate of a chiller (220) are in a starting state, and the opening rate of a valve (230) connected to each of a plurality of pipelines (210) and the operating rate of a chiller (220) are in an action, and by using a Proximal Policy Optimization algorithm to continuously control the opening rate of the valve (230) and the operating rate of the chiller (220).

[0070] The Deep Q-Learning algorithm may include a neural network comprising one input layer having multiple input nodes, one output layer having multiple output nodes, and multiple hidden layers having multiple nodes located between the input layer and the output layer. The Deep Q-Learning algorithm may be used in a network for selecting actions.

[0071] The Proximal Policy Optimization algorithm can be used to update a policy network that evaluates selected actions. The Proximal Policy Optimization algorithm can evaluate the opening rate of the valve (230) and the operating rate of the chiller (220) in a continuous manner.

[0072] The cooling control model (350) is a model learned by a policy that receives a reward for maintaining the temperature of multiple battery packs (110) within a target range, receives a reward for reducing the power required for the operation of the chiller (220), and receives a penalty for temperature imbalance, overcooling, or overheating of the battery packs (110). Therefore, by using the cooling control model (350), the controller (300) can determine the opening rate of the valve (230) and the operating rate of the chiller (220) to consume minimal power while maintaining the temperature of the battery packs (110) within a set target range.

[0073] The controller (300) can control the valve (230) according to the opening rate of the valve (230) output by the cooling control model (350) at set intervals, and can control the chiller (220) according to the operating rate of the chiller (220).

[0074] FIG. 5 is a flowchart illustrating each step of a method for cooling an ESS (1) according to one embodiment. Refer to FIG. 1 to 5 together.

[0075] A method for cooling the ESS (1) may include a step (S10) in which a plurality of temperature sensors (112) installed in each of the plurality of battery packs (110) of the ESS (1) measure the temperature of the battery pack (110); a step (S20) in which a controller (300), based on the temperature of the battery pack (110) received from the temperature sensors (112), determines the opening rate of a valve (230) connected to a pipeline (210) by considering the path of a pipeline (210) that supplies a refrigerant to a cooling plate (120) that cools the battery pack (110) to maintain the temperature of the plurality of battery packs (110) uniformly; and a step (S30) in which the controller (300) outputs a control signal to the valve (230) according to the opening rate of the valve (230).

[0076] The step (S10) of measuring the temperature of the battery pack (110) involves a temperature sensor (112) measuring the temperature of the battery pack (110) and providing it to the controller (300). The temperature sensor (112) included in each battery pack (110) can measure the temperature caused by heat generated by the battery cell (111) or the pack BMS (113) and transmit it to the pack BMS (113), and the pack BMS (113) can transmit the temperature to the controller (300) using a wired or wireless network. The step (S10) of measuring the temperature of the battery pack (110) can be performed in real time.

[0077] The step (S20) of determining the opening rate of the valve (230) is for the controller (300) to determine the opening rate of the valve (230) connected to the pipeline (210) that supplies refrigerant to maintain the temperature of the plurality of battery packs (110) uniformly based on the temperature of the plurality of battery packs (110). The controller (300) may receive the temperature of each of the plurality of battery packs (110) from the plurality of temperature sensors (112). The controller (300) may select a battery pack (110) that exceeds the upper limit of the target temperature range among the plurality of battery packs (110). The controller (300) may determine the opening rate of the valve (230) of the pipeline (210) connected to the cooling plate (120) that cools the battery pack (110) with a high temperature among the plurality of battery packs (110) significantly. The controller (300) may consider the path of the pipeline (210) in the process of determining the opening rate of the pipeline (210).

[0078] The path of the pipeline (210) may include one or more of the location of the battery pack (110), the pipe size of the pipeline (210), the distance the pipeline (210) passes from the chiller (220) to the cooling plate (120), and the branch pipe structure of the pipeline (210).

[0079] The controller (300) can determine a large opening rate for the valve (230) of the pipeline (210) connected to the cooling plate (120) located in the center of the container (100). The controller (300) can determine a large opening rate for the valve (230) of the pipeline (210) connected to the cooling plate (120) located far from the chiller (220). The controller (300) can determine a large opening rate for the valve (230) of the pipeline (210) passing through an environment with a high ambient temperature. The controller (300) can determine different opening rates for multiple valves (230) by considering the branch pipe structure of the pipeline (210). The controller (300) can determine the opening rate of the valve (230) by considering the pipe size of the pipeline (210) and the time it takes for the refrigerant to be delivered from the chiller (220) to the cooling plate (120) according to the pressure of the refrigerant.

[0080] In the step (S20) of determining the opening rate of the valve (230), the controller (300) may determine the opening rate of the valve (230) connected to the pipeline (210) where the temperature of the refrigerant reaching the cooling plate (120) is relatively high by considering the path of the pipeline (210), and the opening rate of the valve (230) connected to the pipeline (210) where the temperature of the refrigerant reaching the cooling plate (120) is relatively low by determining the opening rate of the valve (230) of the corresponding pipeline (210) to be relatively low. When the temperature of the refrigerant supplied to the cooling plate (120) through the pipeline (210) of a specific path is relatively high, the controller (300) may determine the opening rate of the valve (230) of the corresponding pipeline (210) to be high, thereby increasing the flow rate of the refrigerant supplied to the cooling plate (120). When the temperature of the refrigerant supplied to the cooling plate (120) through the pipeline (210) of a specific path is relatively low, the controller (300) can reduce the flow rate of the refrigerant supplied to the cooling plate (120) by determining a small opening rate of the valve (230) of the pipeline (210).

[0081] The step (S20) of determining the opening rate of the valve (230) allows the controller (300) to determine the opening rate of the valve (230) connected to the pipeline (210) that supplies refrigerant to the cooling plate (120) connected to the battery pack (110) that is far from the chiller (220) to be greater than the opening rate of the valve (230) connected to the pipeline (210) that supplies refrigerant to the cooling plate (120) connected to the battery pack (110) that is close to the chiller (220). The factor that has the greatest influence on the temperature of the refrigerant passing through the pipeline (210) may be the distance between the cooling plate (120) and the chiller (220). This is because the temperature of the refrigerant rises due to the influence of the ambient temperature as it passes through the pipeline (210). Therefore, the controller (300) can adjust the opening rate of the valve (230) by considering the distance between the chiller (220) and the cooling plate (120) as the most important factor.

[0082] The step (S30) of outputting a control signal to the valve (230) is to output a control signal to the valve (230) corresponding to a value determined by the controller (300) as the opening rate of the valve (230). The valve (230) may be a proportional control solenoid valve (230). The controller (300) may generate a control signal capable of operating the solenoid valve (230) and output it to the valve (230). When the valve (230) receives the control signal, it may operate according to the control signal to change the degree of opening or closing. When the opening rate of the valve (230) changes, the flow rate of the refrigerant supplied to the cooling plate (120) through the pipeline (210) connected to the valve (230) may change. Therefore, the performance of the cooling plate (120) in cooling the battery pack (110) may change. The controller (300) controls each of the plurality of valves (230) to control the temperature of the plurality of battery packs (110) within a target temperature range.

[0083] FIG. 6 is a diagram illustrating the step (S40) of generating a cooling control model (350) according to one embodiment. Refer to FIG. 1 and FIG. 6 together.

[0084] The controller (300) may use a cooling control model (350) to determine the opening rate of the valve (230). A method for cooling the ESS (1) according to one embodiment may further include a step (S40) of generating a cooling control model (350). The step (S40) of generating a cooling control model (350) may be performed in the controller (300). Alternatively, the step (S40) of generating a cooling control model (350) may be performed in a computer device other than the controller (300). A cooling control model (350) generated in a computer device other than the controller (300) may be stored in the memory (320) of the controller (300).

[0085] The step (S40) of generating a cooling control model (350) may be to reinforce state-action pairs in which the temperature of a plurality of battery packs (110), the opening rate of a valve (230) connected to each of a plurality of pipelines (210), and the operating rate of a chiller (220) are in a starting state, and the opening rate of a valve (230) connected to each of a plurality of pipelines (210) and the operating rate of a chiller (220) are in an action.

[0086] The cooling control model (350) is an artificial intelligence model trained through reinforcement learning. The cooling control model (350) may include a neural network structure comprising an input layer, a hidden layer, and an output layer. The cooling control model (350) may be generated by using a Deep Q-Learning algorithm to learn state-action pairs in which the temperature of a plurality of battery packs (110), the opening rate of a valve (230) connected to each of a plurality of pipelines (210), and the operating rate of a chiller (220) are in a starting state, and the opening rate of a valve (230) connected to each of a plurality of pipelines (210) and the operating rate of a chiller (220) are in an action, and by using a Proximal Policy Optimization algorithm to continuously control the opening rate of the valve (230) and the operating rate of the chiller (220).

[0087] The neural network of the Deep Q-Learning algorithm can be trained to select an action when a state is input by learning state-action pairs. When the temperature of a plurality of battery packs (110) as the starting state, the opening rate of a valve (230) connected to each of the plurality of pipelines (210), and the operating rate of a chiller (220) are input, the neural network can select the opening rate of the valve (230) and the operating rate of the chiller (220) as actions. Starting state-action pairs for reinforcement learning can be generated using existing operational data.

[0088] The step (S40) of creating a cooling control model (350) can reinforce the cooling control model with a policy of providing a reward when the temperature of a plurality of battery packs (110) is maintained within a target range, increasing the reward when the power required for the operation of the chiller (220) is reduced, and imposing a penalty when the temperature of the plurality of battery packs (110) is unbalanced or when overcooling or overheating occurs.

[0089] In the process of learning state-action pairs, the selected action can be evaluated using a reward policy. If the selected action maintains the temperature of the battery cell (111) within a target range, a reward can be given for selecting that action. If the selected action reduces the power required for the operation of the chiller (220), a reward can be given. If the selected action causes a temperature imbalance in the battery pack (110), or causes overcooling or overheating, a penalty can be given. The results of performing such action evaluations can be fed back to the neural network of the Deep Q-Learning algorithm that performs action selection.

[0090] A neural network that evaluates actions may utilize a Proximal Policy Optimization algorithm. Evaluating actions using a neural network utilizing the Proximal Policy Optimization algorithm allows for the evaluation of the opening rate of the valve (230) and the operating rate of the chiller (220) corresponding to the action from a continuous perspective. Therefore, when a neural network that selects an action selects an action, it can output an action corresponding to a continuous control operation.

[0091] Reinforcement learning is performed to learn state-action pairs and provide feedback on the results of evaluating selected actions using a reward policy, so that learning can be stopped when the target performance is achieved.

[0092] The step (S20) of determining the opening rate of the valve (230) involves the controller (300) inputting the temperature of the plurality of battery packs (110) that are updated in real time, the opening rate of the valve (230) connected to each of the plurality of pipelines (210), and the operating rate of the chiller (220) into the cooling control model (350), and the opening rate of the valve (230) connected to each of the plurality of pipelines (210) and the operating rate of the chiller (220), which are actions output by the cooling control model (350), can be determined as the opening rate of the valve (230) and the operating rate of the chiller (220) for controlling the valve (230) and the chiller (220).

[0093] When the controller (300) inputs the temperature of a plurality of battery packs (110), the opening rate of a valve (230) connected to each of a plurality of pipelines (210), and the operating rate of a chiller (220) in real time to a learned cooling control model (350), the cooling control model (350) can output the opening rate of a plurality of valves (230) and the operating rate of a chiller (220) for cooling the plurality of battery packs (110) within a target temperature range. The controller (300) can determine the values ​​output by the cooling control model (350) as the opening rate of the valve (230) and the operating rate of the chiller (220).

[0094] In the step (S40) of generating a cooling control model (350), reinforcement learning was performed using a policy that not only provides a reward when the temperature of a plurality of battery packs (110) is maintained within a target range, but also increases the reward when the power required for the operation of the chiller (220) is reduced, and imposes a penalty when the temperature of the plurality of battery packs (110) is unbalanced or when overcooling or overheating occurs. Accordingly, in the step (S20) of determining the opening rate of the valve (230), in order to minimize the power consumption required for cooling while maintaining the temperature of the plurality of battery packs (210) within a target temperature range, the opening rate of the valve (230) and the operating rate of the chiller (220) connected to each of the plurality of pipelines (210) output by the cooling control model (350) can be determined as the opening rate of the valve (230) and the operating rate of the chiller (220) for controlling the valve (230) and the chiller (220). Since the compensation policy includes reducing the power required for the operation of the chiller (220), the power required for the chiller (220) can be minimized by performing the action output by the cooling control model (350).

[0095] The step (S30) of outputting a control signal to the valve (230) may be performed by the controller (300) outputting a control signal to the valve (230) and the chiller (220) according to the opening rate of the valve (230) and the operating rate of the chiller (220) determined by the values ​​output by the cooling control model (350). The controller (300) can efficiently maintain the temperature of the battery pack (110) by performing both the adjustment of the opening rate of the valve (230) and the adjustment of the operating rate of the chiller (220). The operating rate of the chiller (220) may include the performance of the chiller (220) cooling the refrigerant and the pumping performance of the chiller (220) pumping the refrigerant to the cooling plate (120).

[0096] By using the device (10) and method for cooling the ESS described above, the temperature of the multiple battery packs (110) can be maintained uniformly even if the cooling performance of the cooling plate (120) varies depending on the position of the multiple battery packs (110) stored in the container (100). In addition, the device (10) and method for cooling the ESS can uniformly cool the temperature of the multiple battery packs (110) by considering the path of the pipeline (210), such as the distance between the chiller (220) and the cooling plate (120), and increasing the opening rate of the valve (230) when the temperature of the refrigerant supplied to the cooling plate (120) is relatively high, thereby increasing the flow rate of the refrigerant supplied to the cooling plate (120). In addition, the device (10) and method for cooling the ESS can determine the opening rate of the valve (230) and the operating rate of the chiller (220) in real time using a cooling control model (350), which is an artificial intelligence model learned using a reinforcement learning algorithm, and control the valve (230) and the chiller (220) in real time. Since the cooling control model (350) is learned not only to maintain the temperature of a plurality of battery packs (110) within a target temperature range but also to minimize the amount of power required for cooling the battery packs (110), efficient cooling operation is possible.

[0097] The present disclosure has been described in detail through specific embodiments. The description above is merely an example of applying the principles of the present disclosure, and other configurations may be further included without departing from the scope of the present invention. Explanation of the symbols

[0098] 1: ESS 10: Device for cooling the ESS 100: Container 110: Battery pack 111: Battery cell 112: Temperature sensor 113: Pack BMS 120: Cooling plate 210: Pipeline 220: Chiller 230: Valve 300: Controller 310: Processor 320: Memory 330: Input / Output Interface 340: Communication interface 350: Cooling control model

Claims

Claim 1 A device for cooling an ESS, comprising: a plurality of cooling plates for cooling each of a plurality of battery packs stored in a container of the ESS; a plurality of temperature sensors for measuring the temperature of each of the plurality of battery packs; a plurality of pipelines for inputting refrigerant to the plurality of cooling plates; a chiller for supplying refrigerant to the cooling plates through the pipelines; a valve connected to each of the plurality of pipelines for controlling the flow rate of refrigerant supplied from the chiller to the cooling plates; and a controller for controlling the opening rate of the valves by considering the paths of the plurality of pipelines in order to maintain the temperature of the plurality of battery packs measured using the temperature sensors uniformly. Claim 2 A device for cooling an ESS according to claim 1, wherein the path of the pipeline includes one or more of the location of the battery pack, the pipe size of the pipeline, the distance of the pipeline from the chiller to the cooling plate, the branch pipe structure of the pipeline, and the ambient temperature of the pipeline, and wherein the controller controls the opening rate of a valve connected to a pipeline where the temperature of the refrigerant reaching the cooling plate is relatively high to be relatively large, and controls the opening rate of a valve connected to a pipeline where the temperature of the refrigerant reaching the cooling plate is relatively low to be relatively small, taking into account the path of the pipeline. Claim 3 A device for cooling an ESS according to claim 1, wherein the controller controls the opening rate of a valve connected to a pipeline supplying refrigerant to a cooling plate connected to a battery pack far from the chiller to be greater than the opening rate of a valve connected to a pipeline supplying refrigerant to a cooling plate connected to a battery pack close to the chiller. Claim 4 A device for cooling an ESS according to claim 1, wherein the valve is a solenoid valve whose opening rate is controlled by a control signal output from the controller. Claim 5 A device for cooling an ESS according to claim 1, wherein the controller inputs the temperature of the plurality of battery packs, which is updated in real time, the opening rate of the valve connected to each of the plurality of pipelines, and the operating rate of the chiller as a starting state to a cooling control model, and controls the plurality of valves and chillers using the opening rate of the valve connected to each of the plurality of pipelines and the operating rate of the chiller output by the cooling control model as an action, and the cooling control model is a reinforcement-learned artificial intelligence model that, upon receiving the temperature of the plurality of battery packs, the opening rate of the valve connected to each of the plurality of pipelines, and the operating rate of the chiller, outputs the opening rate of the valve connected to each of the plurality of pipelines and the operating rate of the chiller to maintain the temperature of the plurality of battery packs within a target temperature range. Claim 6 An apparatus for cooling an ESS according to claim 5, wherein the cooling control model utilizes a Deep Q-Learning algorithm to learn state-action pairs in which the temperature of the plurality of battery packs, the opening rate of the valve connected to each of the plurality of pipelines, and the operating rate of the chiller are in a starting state, and the opening rate of the valve connected to each of the plurality of pipelines and the operating rate of the chiller are in a behavior, and utilizes a Proximal Policy Optimization algorithm to continuously control the opening rate of the valve and the operating rate of the chiller. Claim 7 In claim 5, the cooling control model is an artificial intelligence model that has been reinforced using a policy that provides a reward when the temperature of the plurality of battery packs is maintained within a target range, increases the reward when the power required for the operation of the chiller is reduced, and imposes a penalty when the temperature of the plurality of battery packs is unbalanced or when overcooling or overheating occurs, and the controller controls the valve and the chiller according to the opening rate of the valve connected to each of the plurality of pipelines and the operating rate of the chiller output by the cooling control model in order to maintain the temperature of the plurality of battery packs within the target temperature range and minimize the power consumption required for cooling. Claim 8 A method for cooling an ESS, comprising: a step in which a plurality of temperature sensors installed on each of a plurality of battery packs of the ESS measure the temperature of the battery packs; a step in which a controller determines the opening rate of a valve connected to a pipeline by considering the path of a pipeline that supplies a refrigerant to a cooling plate cooling the battery packs to maintain the temperature of the plurality of battery packs uniformly, based on the temperature of the battery packs received from the temperature sensors; and a step in which the controller outputs a control signal to the valve according to the opening rate of the valve. Claim 9 A method for cooling an ESS according to claim 8, wherein the path of the pipeline includes one or more of the location of the battery pack, the pipe size of the pipeline, the distance the pipeline passes from the chiller to the cooling plate, and the branch pipe structure of the pipeline, and the step of determining the opening rate of the valve is such that the controller, considering the path of the pipeline, determines the opening rate of the valve connected to the pipeline where the temperature of the refrigerant reaching the cooling plate is relatively high to be relatively large, and determines the opening rate of the valve connected to the pipeline where the temperature of the refrigerant reaching the cooling plate is relatively low to be relatively small. Claim 10 A method for cooling an ESS according to claim 8, wherein the step of determining the opening rate of the valve is such that the controller determines the opening rate of the valve connected to the pipeline supplying refrigerant to a cooling plate connected to a battery pack far from the chiller to be greater than the opening rate of the valve connected to the pipeline supplying refrigerant to a cooling plate connected to a battery pack close to the chiller. Claim 11 A method for cooling an ESS according to claim 8, further comprising the step of generating a cooling control model by reinforcement learning state-action pairs in which the temperature of the plurality of battery packs, the opening rate of the valve connected to each of the plurality of pipelines, and the operating rate of the chiller are in a starting state, and the opening rate of the valve connected to each of the plurality of pipelines and the operating rate of the chiller are in a action, wherein the step of determining the opening rate of the valve inputs the temperature of the plurality of battery packs, the opening rate of the valve connected to each of the plurality of pipelines, and the operating rate of the chiller, which are updated in real time by the controller, into the cooling control model, and determines the opening rate of the valve connected to each of the plurality of pipelines and the operating rate of the chiller, which are actions output by the cooling control model, as they are as the opening rate of the valve and the operating rate of the chiller for controlling the valve and the chiller. Claim 12 A method for cooling an ESS according to claim 11, wherein the step of generating the cooling control model utilizes a Deep Q-Learning algorithm to learn state-action pairs in which the temperature of the plurality of battery packs, the opening rate of the valve connected to each of the plurality of pipelines, and the operating rate of the chiller are in a starting state, and the opening rate of the valve connected to each of the plurality of pipelines and the operating rate of the chiller are in a behavior, and utilizes a Proximal Policy Optimization algorithm to continuously control the opening rate of the valve and the operating rate of the chiller. Claim 13 A method for cooling an ESS according to claim 11, wherein the step of generating the cooling control model involves reinforcing the cooling control model with a policy of providing a reward when the temperature of the plurality of battery packs is maintained within a target range, increasing the reward when the power required for the operation of the chiller is reduced, and imposing a penalty when the temperature of the plurality of battery packs is unbalanced or when overcooling or overheating occurs; the step of determining the valve opening rate involves determining the valve opening rate and chiller operating rate connected to each of the plurality of pipelines output by the cooling control model as the valve opening rate and chiller operating rate for controlling the valve and chiller in order to minimize the power consumed for cooling while maintaining the temperature of the plurality of battery packs within the target temperature range; and the step of outputting a control signal to the valve involves the controller performing an operation of outputting a control signal to the valve and chiller according to the determined valve opening rate and chiller operating rate.

Images