Marine fully-enclosed energy storage battery thermal management system and method

Abstract

The application belongs to the technical field of energy storage and relates to a full-closed energy storage battery thermal management system and method for a ship. The thermal management system comprises an air conditioner installed on a battery cabin of an energy storage battery, an air duct comprising an air duct horizontal section arranged at the top of a battery cluster of the energy storage battery and an air duct vertical section arranged at the first side surface of the battery cluster, an air inlet of the air duct horizontal section being in communication with an air outlet of the air conditioner, the air duct vertical section being provided with a first air outlet at a corresponding interval position of the battery cluster, a plurality of cross-flow fans installed in the air duct vertical section, the plurality of cross-flow fans being arranged at intervals from top to bottom along the air duct vertical section, the cross-flow fans being arranged opposite to the first air outlet, PTC heating sheets arranged on the surface of the battery cluster, a data acquisition unit for generating state quantities by means of battery cluster temperature information, temperature influence factors affecting the battery cluster temperature, air conditioner energy consumption, cross-flow fan energy consumption and PTC heating sheet energy consumption, and a controller configured to control the air conditioner, the cross-flow fans and the PTC heating sheets by means of corresponding control quantities when a reward function obtained by means of a control agent model based on a SAC algorithm is maximized. The application can maintain the battery monomer temperature in the optimal working temperature range by means of active heating or heat dissipation and minimize the system operation energy consumption.

Description

Technical Field

[0001] This invention belongs to the field of marine energy storage technology, and relates to thermal management technology for energy storage batteries. Specifically, it relates to a fully enclosed thermal management system and method for marine energy storage batteries. Background Technology

[0002] In the field of marine energy storage, lithium-ion batteries (such as lithium iron phosphate batteries) are widely used due to their high energy density, long lifespan, and environmental adaptability. However, the marine navigation environment is complex and variable, and battery systems must cope with multiple challenges such as high humidity salt spray corrosion, continuous mechanical vibration, and extreme temperature changes. If the heat generated inside the battery during charging and discharging cannot be dissipated in time, it will lead to uneven temperature distribution within the battery pack, accelerating battery capacity decay and creating a risk of thermal runaway. Therefore, the battery thermal management system has become a core subsystem for ensuring the safe and efficient operation of marine energy storage. The battery thermal management system mainly uses methods such as air cooling, liquid cooling, phase change cooling, and heat pipe cooling to cool the batteries.

[0003] Air cooling is divided into natural air cooling and forced air cooling. Natural cooling is simple but has the worst heat dissipation effect and is mostly used in scenarios with low current. Forced air cooling uses equipment such as fans to generate forced airflow, accelerating convective heat transfer efficiency. It has a simple structure and wide application, but its heat transfer capacity is insufficient when the battery is charged and discharged at high rates.

[0004] Liquid cooling uses a coolant as the medium, leveraging its high specific heat capacity and heat transfer coefficient. Pumps and other equipment propel the coolant through the battery pack to dissipate heat. It is divided into contact and non-contact types. Contact liquid cooling directly immerses the battery in the coolant to remove heat, resulting in very high heat exchange efficiency. However, it also has high safety requirements, posing a risk of leakage, and the cost of safe, non-conductive, non-flammable, and non-explosive coolants is high. Non-contact liquid cooling indirectly dissipates heat from the battery by passing the coolant through heat exchangers such as cold plates. It has high cooling efficiency but suffers from volume expansion and cannot actively dissipate heat.

[0005] Phase change cooling utilizes the phase transition of phase change materials (such as melting, evaporation, and sublimation) to absorb a large amount of heat and maintain battery temperature. It meets the requirements of high-power charging and discharging and has good cooling efficiency, but it is more expensive, and its thermal conductivity decreases significantly after the latent heat capacity of the material is depleted.

[0006] The principle of heat pipe cooling is that the working fluid inside the heat pipe, which is in contact with the battery at the evaporation end, evaporates and absorbs heat, turning into a gaseous state. It then moves along the heat pipe to the cooling end, exchanges heat with the outside environment, liquefies, and flows back to the evaporation end through capillary action. Heat pipe cooling can significantly reduce battery temperature rise and is suitable for high-rate charging and discharging, but its manufacturing cost is high and there is a risk of leakage. Summary of the Invention

[0007] To address the aforementioned problems in existing technologies, this invention provides a fully enclosed thermal management system and method for marine energy storage batteries. It employs forced air cooling, supplemented by enhanced heat exchange via air ducts, to dissipate heat from the energy storage batteries. This system is easy to maintain, lightweight, and maintains the individual battery cell temperature within the optimal operating range under varying external ambient temperatures, solar radiation, and charging / discharging current conditions through active heating or cooling. The temperature difference within the battery cluster does not exceed 5°C, and the system's energy consumption is minimized.

[0008] To achieve the above objectives, a first aspect of the present invention provides a marine fully enclosed energy storage battery thermal management system, comprising: An air conditioner, installed on the battery compartment of the energy storage battery, is used for cooling or heating to dissipate heat or heat the battery clusters of the energy storage battery. The air duct includes a horizontal section located at the top of the battery cluster and a vertical section located on the first side of the battery cluster. The air inlet of the horizontal section is connected to the air outlet of the air conditioner, and the vertical section is provided with a first air outlet at the interval position of the battery cluster. Multiple crossflow fans are installed in the vertical section of the air duct. The multiple crossflow fans are arranged at intervals from top to bottom along the vertical section of the air duct. The crossflow fans are positioned opposite to the first air outlet. PTC heating elements are placed on the surface of the battery clusters to heat them. The data acquisition unit generates state variables for the battery cluster temperature information, temperature influence factors affecting battery cluster temperature, air conditioning energy consumption, crossflow fan energy consumption, and PTC heating element energy consumption. The controller is configured to: obtain control quantities through a control agent model based on the SAC algorithm based on the state quantities generated by the data acquisition unit; and control the air supply temperature and air supply speed of the air conditioner, the PWM signal duty cycle of the crossflow fan, and the DC control voltage of the PTC heating element based on the control quantities corresponding to the maximization of the reward function of the control agent model.

[0009] In some embodiments, the minimum cooling capacity of the air conditioner is:

[0010] in,

[0011]

[0012]

[0013] In the formula, Indicates the safety factor; Indicates the battery's heating power; I Indicates the battery charging and discharging current;R Indicates the battery's DC internal resistance; This indicates the heat exchange capacity between the outer wall of the battery compartment and the outside air. This represents the convective heat transfer coefficient, with units of W / (m³). 2 ·K); A Indicates heat transfer area, unit: m² 2 ; This indicates the temperature difference between the ambient temperature and the temperature inside the battery compartment, in °C. This indicates that the battery absorbs heat; Specific heat capacity of the battery, unit: J / (kg·K); Battery mass, unit: kg; To increase battery temperature, unit: °C; This indicates the charging and discharging time of the battery cluster.

[0014] In some embodiments, the dimensions of the air duct satisfy the following constraints:

[0015] In the formula, The height of the air conditioner vent, in cm; Width of the air conditioner vent; The height of the horizontal section of the air duct, in cm; The number of air ducts ; Width of the air inlet in a single horizontal section of an air duct, in cm; Length of the vertical section of the air duct, in cm; Width of the vertical section of the air duct, unit: cm; The length of the crossflow fan is in cm. Width of the crossflow fan, unit: cm.

[0016] In some embodiments, the crossflow fan is mounted on the air duct by a fan mounting member, which is fixed at the first air outlet. The side of the fan mounting member that is fixed to the air duct is provided with a second air outlet that communicates with the first air outlet, and the side of the fan mounting member that is equipped with the crossflow fan is provided with a first air inlet. The air generated by the crossflow fan blows the gas in the air duct into the space of the battery cluster through the first air inlet, the second air outlet, and the first air outlet to dissipate heat or heat the battery cluster.

[0017] In some embodiments, the fan mounting bracket is provided with an air guide plate, which is set at an angle so that the air generated by the crossflow fan is discharged at an angle through the air guide plate and forms convection with the bottom shell of the battery cluster.

[0018] In some embodiments, the battery cluster temperature information includes battery cluster temperature and battery cluster temperature range; temperature influencing factors include ambient temperature, solar radiation received by the battery compartment, and battery cluster charging and discharging current; the state quantity is represented as:

[0019] In the formula, express State quantity at any given time. express The temperature of the battery cluster at any given time, Due to the extreme temperature difference in the battery clusters, for ambient temperature at any time for The battery compartment receives external solar radiation at all times. for The charging and discharging current of the battery cluster at any time, for Air conditioning energy consumption at all times for Crossflow fan energy consumption at all times for Energy consumption of the PTC heating element at any given time; The control quantities include the air conditioner set air supply temperature, air conditioner air supply speed, crossflow fan PWM signal duty cycle, and PTC heating element DC control voltage.

[0020] In some embodiments, the reward function of the control agent model is expressed as:

[0021] In the formula, express The reward function of the control agent model at each time step. express The temperature-related sub-reward function at time intervals, This represents the weighting coefficient of the temperature-related sub-reward function. express The energy consumption sub-reward function at time step. This represents the weighting coefficient of the energy consumption item's sub-reward function. express The time-varying control of the sub-reward function This represents the weighting coefficient of the sub-reward function that controls fluctuations. express The safety item reward function at any given time. This represents the weighting coefficient of the reward function for the safety item.

[0022] In some embodiments, the temperature-related sub-reward function is expressed as:

[0023] In the formula, This indicates a positive reward given when the battery cluster temperature is stable within a set temperature range. >0; Indicates the boundary threshold; This represents the penalty coefficient for violations of the average temperature regulation of the battery cluster. <0; This indicates the penalty coefficient imposed when the temperature difference between battery clusters exceeds 5°C. <0; This indicates the average temperature of the battery cluster; The energy consumption item sub-reward function is expressed as follows:

[0024] In the formula, This represents the penalty coefficient for the energy consumption term. <0, energy consumption items include air conditioning energy consumption. Crossflow fan energy consumption PTC heating element energy consumption ; This represents the penalty coefficient for energy consumption when the air conditioner is switched on or off. <0, Energy consumption for switching on and off the air conditioner; The sub-reward function for the control fluctuation term is expressed as follows:

[0025] In the formula, This represents the penalty coefficient for controlling for fluctuations. <0; Set the change value of the supply air temperature between the previous moment and the current moment; The change value of the air supply speed between the previous moment and the current moment; This represents the change in the duty cycle of the PWM signal for the crossflow fan between the previous and current times. The change in the DC control voltage of the PTC heating element between the previous moment and the current moment; The sub-reward function for the security item is expressed as follows:

[0026] In the formula, This represents the penalty coefficient for incorrect temperature adjustment decisions made by the control agent model. <0; This represents the penalty coefficient when the decision made by the control agent model results in negative pressure in the air duct. <0; This indicates the air conditioner's set airflow temperature. This represents the total change in duct pressure.

[0027] In some embodiments, the method for constructing the reward function of the control agent model is as follows: Set the weight coefficients for the reward function of the security item; Construct a judgment matrix based on the relative importance of the temperature sub-reward function, the energy consumption sub-reward function, and the control fluctuation sub-reward function; Calculate the importance weight coefficients of each sub-reward function in the judgment matrix; The reward function of the control agent model is obtained by linearly weighting each sub-reward function with its corresponding importance weight coefficient.

[0028] In some embodiments, after calculating the importance weights of each sub-reward function in the judgment matrix, a consistency check is performed to verify the rationality of the judgment matrix.

[0029] In some embodiments, the method for verifying the consistency of the judgment matrix is ​​as follows: calculate the consistency ratio using the consistency index and the random consistency index, determine whether the consistency ratio is less than a set threshold, and consider the consistency of the judgment matrix to be reasonable when the consistency ratio is less than the set threshold. The consistency index is expressed as follows:

[0030] In the formula, Indicators of consistency This represents the judgment matrix to be tested. Represents the judgment matrix The corresponding feature vector, Indicates the first The importance weight coefficient of each sub-reward function; The consistency ratio is expressed as:

[0031] In the formula, Indicates the consistency ratio. This represents the random consistency index.

[0032] To achieve the above objectives, a second aspect of the present invention provides a thermal management method for a fully enclosed marine energy storage battery, employing the thermal management system for a fully enclosed marine energy storage battery described in the first aspect of the present invention, characterized in that its steps are as follows: The acquired battery cluster temperature information, temperature influence factors affecting battery cluster temperature, air conditioning energy consumption, crossflow fan energy consumption, and PTC heating element energy consumption will be used to generate state variables. The control quantity is obtained from the state quantity through a control agent model based on the SAC algorithm; The control quantity corresponding to maximizing the reward function of the control agent model controls the air conditioner's air supply temperature and air supply speed, the PWM signal duty cycle of the crossflow fan, and the DC control voltage of the PTC heating element.

[0033] Compared with the prior art, the advantages and positive effects of the present invention are as follows: (1) The marine fully enclosed energy storage battery thermal management system of the present invention adopts forced air cooling and a cooling method of enhanced heat exchange through air ducts to dissipate heat from the energy storage battery. Through the overall architecture of air conditioning outlet + directional air duct guidance + crossflow fan embedded in the air duct, the heat exchange effect is enhanced, forming a progressive airflow enhancement, with high heat exchange efficiency, simple maintenance, and lightweight system. Under the changing external ambient temperature, solar radiation and current charging and discharging conditions, the battery cell temperature is maintained within the optimal operating temperature range (20℃~30℃) through active heating or heat dissipation. The temperature difference of the battery cells inside the energy storage battery cluster does not exceed 5℃, and the system operating energy consumption is minimized.

[0034] (2) The marine fully enclosed energy storage battery thermal management system and method of the present invention generates state quantities based on battery cluster temperature information, temperature influence factors affecting battery cluster temperature, air conditioning energy consumption, crossflow fan energy consumption, and PTC heating element energy consumption. The control quantity is obtained through a control agent model based on SAC algorithm. The air supply temperature and air supply speed of the air conditioner, the duty cycle of the PWM signal of the crossflow fan and the DC control voltage of the PTC heating element are controlled according to the control quantity corresponding to the maximization of the reward function of the control agent model, thereby realizing the pre-control of battery temperature.

[0035] (3) The marine fully enclosed energy storage battery thermal management system and method of the present invention constructs a reward function including temperature, energy consumption, control fluctuation and safety as the reward function of the control agent model based on SAC algorithm, which can accurately obtain the control quantity, realize precise control of the temperature of the energy storage battery and improve the stability of the energy storage battery. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the structure of the marine fully enclosed energy storage battery thermal management system according to an embodiment of the present invention; Figure 2 for Figure 1 Enlarged view of section A; Figure 3 This is a control principle diagram of the marine fully enclosed energy storage battery thermal management system described in an embodiment of the present invention; Figure 4 This is a schematic diagram of the fan mounting component according to an embodiment of the present invention; Figure 5 This is a flowchart illustrating the method for constructing the reward function of the control agent model according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the method for calculating the importance weight coefficients of each sub-reward function in the judgment matrix according to an embodiment of the present invention; Figure 7 This is a schematic flowchart of the thermal management method for a fully enclosed marine energy storage battery according to an embodiment of the present invention; Figure 8 This is a schematic diagram of the Actor network structure described in an embodiment of the present invention; Figure 9 This is a schematic diagram of the Critic network structure described in an embodiment of the present invention; Figure 10 This diagram illustrates a comparison of the average battery cluster temperature obtained by using the marine fully enclosed energy storage battery thermal management method, ON / OFF control strategy, and rule-based level control method of the present invention for thermal management control of the battery thermal management system. Figure 11 This diagram illustrates a comparison of the maximum temperature range of the battery obtained by using the thermal management method for marine fully enclosed energy storage batteries, the ON / OFF control strategy, and the rule-based segmented control method of the present invention for thermal management of the battery thermal management system. Figure 12 This diagram illustrates a comparison of battery compartment temperature results obtained by using the thermal management method for marine fully enclosed energy storage batteries, the ON / OFF control strategy, and the rule-based segmented control method of the present invention for thermal management of the battery thermal management system. Figure 13 This diagram illustrates a comparison of air conditioning power results obtained by using the marine fully enclosed energy storage battery thermal management method, ON / OFF control strategy, and rule-based level control method of the present invention for thermal management control of the battery thermal management system. Figure 14 This diagram illustrates a comparison of total power consumption results obtained by using the marine fully enclosed energy storage battery thermal management method, ON / OFF control strategy, and rule-based level control method of the present invention for thermal management control of the battery thermal management system. Figure 15 This is a comparative diagram showing the quantitative results of the average temperature of the battery cluster and the maximum temperature range of the battery obtained by using the thermal management method of the marine fully enclosed energy storage battery of this invention and the ON / OFF control strategy method for the thermal management system of the battery thermal management system under summer operating conditions. Figure 16 This is a comparative diagram showing the quantitative results of the average temperature of the battery cluster and the maximum temperature range of the battery obtained by using the thermal management method of the marine fully enclosed energy storage battery of this invention and the ON / OFF control strategy method for the thermal management system of the battery thermal management system under winter operating conditions. Figure 17This is a schematic diagram comparing the quantitative results of total power consumption and air conditioning power consumption obtained by using the thermal management method of the marine fully enclosed energy storage battery and the ON / OFF control strategy of the present invention for thermal management of the battery thermal management system under summer operating conditions. Figure 18 This diagram illustrates a comparison of the quantitative results of total power consumption, air conditioning power consumption, and PTC heating element power consumption obtained by using the thermal management method of the marine fully enclosed energy storage battery and the ON / OFF control strategy of this invention for thermal management of the battery thermal management system under winter operating conditions.

[0037] In the diagram, 101 is the air conditioner, 102 is the air duct, 103 is the crossflow fan, 104 is the PTC heating element, 105 is the data acquisition unit, 106 is the controller, 107 is the fan mounting bracket, 1071 is the second air outlet, 1072 is the first air inlet, 1073 is the air guide plate, 201 is the battery compartment, and 202 is the battery cluster. Detailed Implementation

[0038] The present invention will now be described in detail through exemplary embodiments. However, it should be understood that, without further description, elements, structures, and features in one embodiment may be advantageously incorporated into other embodiments.

[0039] A first aspect of the present invention provides a thermal management system for a fully enclosed marine energy storage battery. Figure 1 The diagram shown is a structural schematic of the marine fully enclosed energy storage battery thermal management system according to an embodiment of the present invention.

[0040] See Figures 1 to 3 The ship's fully enclosed energy storage battery thermal management system includes an air conditioner 101, an air duct 102, multiple crossflow fans 103 installed in the air duct 102, a PTC heating element 104, a data acquisition unit 105, and a controller 106.

[0041] Air conditioner 101 is installed on the battery compartment 201 of the energy storage battery for cooling or heating, in order to dissipate heat or heat the battery cluster 202 of the energy storage battery.

[0042] In one embodiment of the present invention, the minimum cooling capacity of the air conditioner 101 is:

[0043] in,

[0044]

[0045]

[0046] In the formula, This represents the safety factor, which typically ranges from 1.2 to 1.5. Indicates the battery's heating power; I Indicates the battery charging and discharging current; R This indicates the battery's DC internal resistance (including ohmic internal resistance and polarization internal resistance, which varies with SOC and battery temperature). This indicates the heat exchange capacity between the outer wall of the battery compartment and the outside air. This represents the convective heat transfer coefficient, with units of W / (m³). 2 ·K); A Indicates heat transfer area, unit: m² 2 ; This indicates the temperature difference between the ambient temperature and the temperature inside the battery compartment, in °C. This indicates that the battery absorbs heat; Specific heat capacity of the battery, unit: J / (kg·K); Battery mass, unit: kg; To increase battery temperature, unit: °C; This indicates the charging and discharging time of the battery cluster.

[0047] It should be noted that the selection of air conditioning mainly considers the cooling and heating capacity required by the entire battery compartment. The cooling load within the battery compartment primarily includes the heat generated during battery charging and discharging, as well as the heat transferred between the external environment and the outer wall of the battery compartment through convection and radiation. The heat generated during battery charging and discharging, along with the heat exchange between the battery compartment and the external environment, constitute the total cooling load (i.e., total heat) of the thermal management system. Part of this total heat becomes the temperature rise of the batteries, while a portion is transferred through air conditioning cooling. The heat transferred through air conditioning cooling is the minimum cooling capacity of the thermal management system.

[0048] For example, the charging and discharging current of the battery cluster is calculated at 0.5C. The internal resistance of a single cell within the battery cluster is 1.3mΩ, the battery heat capacity is 1140J / (Kg·K), the mass of a single cell is 2.62kg, and the peak heat transfer power from the external environment to the battery compartment is calculated to be 2000W. The minimum cooling capacity is approximately 3800W. Therefore, the air conditioner selected is the Guangzhou Mingnuo MN50A / 4A1 embedded energy storage air conditioner, which provides airflow from the top of the energy storage battery and return airflow from the bottom, with a cooling capacity of 4000W and a maximum airflow of 850m³. 3 / h.

[0049] The air duct 102 includes a horizontal section located at the top of the battery cluster 202 and a vertical section located on the second side of the battery cluster 202. The air inlet of the horizontal section is connected to the air outlet of the air conditioner 101, and the vertical section is provided with a first air outlet at intervals corresponding to the battery cluster 202.

[0050] Multiple crossflow fans 103 are arranged at intervals from top to bottom along the vertical section of the air duct, and the crossflow fans 103 are positioned opposite to the first air outlet.

[0051] In one embodiment of the present invention, the crossflow fan 103 is directly mounted on the side wall of the vertical section of the air duct, and the dimensions of the air duct 102 satisfy the following constraints:

[0052] In the formula, The height of the air conditioner vent, in cm; Width of the air conditioner vent; The height of the horizontal section of the air duct, in cm; The number of air ducts ; Width of the air inlet in a single horizontal section of an air duct, in cm; Length of the vertical section of the air duct, in cm; Width of the vertical section of the air duct, unit: cm; The length of the crossflow fan is in cm. Width of crossflow fan and fan mounting hardware, unit: cm.

[0053] For example, the battery cluster consists of two battery racks placed side by side. The air duct adopts a symmetrical design, and the overall air inlet of the air duct is composed of two air ducts. The width of the overall air inlet of the air duct should be greater than the width of the air conditioner outlet, and the height of the horizontal section of the air duct should be greater than the height of the air conditioner outlet, so as to guide the air conditioning airflow to the battery cluster. The length and width of the vertical section of the air duct should be greater than the dimensions of the crossflow fan and fan mounting components to ensure smooth airflow and facilitate equipment installation and commissioning. In other words, the dimensions of the air duct meet the following constraints:

[0054] In one embodiment of the present invention, the crossflow fan 103 is mounted on the air duct 102 via a fan fixing member 107, which is fixed at the first air outlet. The side of the fan fixing member 107 that is fixed to the air duct 102 has a second air outlet 1071 that communicates with the first air outlet. The side of the fan fixing member 107 where the crossflow fan 103 is mounted has a first air inlet 1072. The air generated by the crossflow fan blows the gas in the air duct through the first air inlet 1072, the second air outlet 1071, and the first air outlet into the space of the battery cluster 202 to dissipate heat or heat the battery cluster.

[0055] Specifically, the dimensions of the air duct 102 satisfy the following constraints:

[0056] In the formula, The height of the air conditioner vent, in cm; Width of the air conditioner vent; The height of the horizontal section of the air duct, in cm; The number of air ducts ; Width of the air inlet in a single horizontal section of an air duct, in cm; Length of the vertical section of the air duct, in cm; Width of the vertical section of the air duct, unit: cm; The length of the crossflow fan is in cm. Width of crossflow fan and fan mounting hardware, unit: cm.

[0057] In some embodiments, the fan mounting member 107 is provided with an air guide plate 1073, which is set at an angle so that the air generated by the crossflow fan 103 is discharged at an angle through the air guide plate 1073 and forms convection with the bottom shell of the battery cluster 202.

[0058] In this embodiment, the air guide plate guides the airflow direction of the crossflow fan at an angle, preventing the airflow from being directly vertical or randomly diffused, and instead directing it precisely at a preset angle towards the bottom casing of the battery cluster. On one hand, this directional airflow allows the airflow to more directly impact the casing surface, breaking up any static air layer that may exist at the bottom and accelerating the heat exchange rate between the casing and the airflow. Compared to the scattered airflow without an air guide plate, the convective heat transfer efficiency can be improved by more than 30%. On the other hand, the convection formed by the angled airflow and the bottom casing of the battery cluster creates a spiraling or directional circulating airflow field between the casing and the fan mounting components: after the cold air is obliquely blown towards the casing by the air guide plate and carries away heat, the heated air naturally flows upward due to the density difference, forming a closed loop of "air intake-heat absorption-air exhaust". This circulation can cover the corner areas at the bottom of the battery cluster, avoiding the heat dissipation blind spots caused by airflow rebound in traditional direct blowing methods, keeping the temperature difference at various locations at the bottom within ±2℃, and improving heat dissipation uniformity.

[0059] For example, if the air guide plate is set at a 60° upward angle, the air generated by the crossflow fan will be discharged at a 60° upward angle through the air guide plate, forming forced convection cooling with the bottom shell of the battery cluster.

[0060] PTC heating element 104 is disposed on the surface of battery cluster 202 for heating battery cluster 202.

[0061] The data acquisition unit 105 is used to acquire battery cluster temperature information, temperature influence factors affecting battery cluster temperature, air conditioning energy consumption, crossflow fan energy consumption, and PTC heating element energy consumption, and to generate state quantities from the acquired battery cluster temperature information, temperature influence factors affecting battery cluster temperature, air conditioning energy consumption, crossflow fan energy consumption, and PTC heating element energy consumption.

[0062] In one embodiment of the present invention, the battery cluster temperature information includes the battery cluster temperature and the battery cluster temperature range. Temperature influencing factors include ambient temperature, solar radiation received by the battery compartment, and the battery cluster charging and discharging current. The state quantity is represented as:

[0063] In the formula, express State quantity at any given time. express The temperature of the battery cluster at any given time, Due to the extreme temperature difference in the battery clusters, for ambient temperature at any time for The battery compartment receives external solar radiation at all times. for The charging and discharging current of the battery cluster at any time, for Air conditioning energy consumption at all times for Crossflow fan energy consumption at all times for Energy consumption of the PTC heating element at any given time.

[0064] Specifically, the battery cluster temperature is measured using PT100 temperature sensors installed on each battery pack. Ambient temperature and solar radiation are obtained from meteorological data software (e.g., Meteonorm meteorological data software). The battery cluster temperature range is obtained by subtracting the maximum and minimum temperatures measured by the PT100 temperature sensors. The charging and discharging current is measured by current sensors. Air conditioning energy consumption, cross-flow fan energy consumption, and PTC energy consumption are obtained from the corresponding electricity meters.

[0065] The controller 106 is configured to: obtain control quantities from the state quantities generated by the data acquisition unit 105 through a control agent model 1061 based on the SAC algorithm; and control the air supply temperature and air supply speed of the air conditioner 101, the PWM signal duty cycle of the crossflow fan 103, and the DC control voltage of the PTC heating element 104 according to the control quantities corresponding to maximizing the reward function of the control agent model 1061. The control quantities include the set air supply temperature of the air conditioner, the air supply speed of the air conditioner, the PWM signal duty cycle of the crossflow fan, and the DC control voltage of the PTC heating element. The control quantities are expressed as:

[0066] In the formula, for Set the air supply temperature at all times. for The airflow setting at all times. for The duty cycle of the crossflow fan PWM signal at any given time. for The DC control voltage of the PTC heating element at any given time.

[0067] In one embodiment of the present invention, the reward function of the control agent model is expressed as:

[0068] In the formula, express The reward function of the control agent model at each time step. express The temperature-related sub-reward function at time intervals, This represents the weighting coefficient of the temperature-related sub-reward function. express The energy consumption sub-reward function at time step. This represents the weighting coefficient of the energy consumption item's sub-reward function. express The time-varying control of the sub-reward function This represents the weighting coefficient of the sub-reward function that controls fluctuations. express The safety item reward function at any given time. This represents the weighting coefficient of the reward function for the safety item.

[0069] In one embodiment of the present invention, the temperature sub-reward function is expressed as:

[0070] In the formula, This indicates a positive reward given when the battery cluster temperature is stable within a set temperature range. >0; Indicate (supplement its meaning); This represents the penalty coefficient for violations of the average temperature regulation of the battery cluster. <0; This indicates the penalty coefficient imposed when the temperature difference between battery clusters exceeds 5°C. <0; This indicates the average temperature of the battery cluster.

[0071] The temperature-related sub-reward function measures the deviation of the average battery cluster temperature from the target temperature range. A positive reward is given when the average battery cluster temperature remains within the appropriate range; a negative reward is given when the average battery cluster temperature deviates from the normal range, with a greater penalty for larger deviations. A penalty is imposed when the maximum temperature difference within the battery cluster exceeds 5°C.

[0072] In one embodiment of the present invention, the energy consumption item sub-reward function is expressed as:

[0073] In the formula, This represents the penalty coefficient for the energy consumption term. <0, energy consumption items include air conditioning energy consumption. Crossflow fan energy consumption PTC heating element energy consumption ; This represents the penalty coefficient for energy consumption when the air conditioner is switched on or off. <0, Energy consumption for switching on and off the air conditioner.

[0074] It should be noted that minimizing energy consumption is one of the control requirements of the thermal management system described in this invention. Therefore, the energy consumption item reward function is designed using a penalty coefficient method, where the smaller the energy consumption, the smaller the penalty.

[0075] In one embodiment of the present invention, the control fluctuation term sub-reward function is expressed as:

[0076] In the formula, This represents the penalty coefficient for controlling for fluctuations. <0; Set the change value of the supply air temperature between the previous moment and the current moment; The change value of the air supply speed between the previous moment and the current moment; This represents the change in the duty cycle of the PWM signal for the crossflow fan between the previous and current times. This represents the change in the DC control voltage of the PTC heating element between the previous and current times.

[0077] Since the control quantity is not prone to frequent fluctuations during the adjustment process, a penalty for the control fluctuation item is set, and a sub-reward function for the control fluctuation item is designed to implement the penalty for the control fluctuation item.

[0078] In one embodiment of the present invention, the safety item reward function is expressed as:

[0079] In the formula, This represents the penalty coefficient for incorrect temperature adjustment decisions made by the control agent model. <0; This represents the penalty coefficient when the decision made by the control agent model results in negative pressure in the air duct. <0; This indicates the air conditioner's set airflow temperature. This represents the total change in duct pressure.

[0080] To avoid generating obviously incorrect or unsafe control actions, this embodiment of the invention designs a safety item sub-reward function, which ensures that the control agent model actively avoids incorrect or unsafe control actions.

[0081] In some embodiments, see Figure 4 The method for constructing the reward function of the control agent model is as follows: S1. Set the weight coefficient of the reward function for the security item.

[0082] For the reward function of the safety item, a larger penalty coefficient is set independently to ensure that the control strategy actively avoids risks during the training process of the control agent model.

[0083] S2. Construct a judgment matrix based on the relative importance of the temperature sub-reward function, energy consumption sub-reward function, and control fluctuation sub-reward function.

[0084] Specifically, the importance of each sub-reward function is evaluated by constructing a judgment matrix through expert scoring.

[0085] For example, temperature is the most important, followed by energy consumption, and control fluctuation is the least important. Importance is represented on a 1-9 scale: 1 for equal importance, 3 for slightly important, 5 for important, 7 for very important, and 9 for absolutely important. Temperature is slightly more important than energy consumption, so it's set to 3. Temperature is more important than control fluctuation, so it's set to 5. Energy consumption is slightly more important than control fluctuation, so it's set to 3. Using the letter A to represent the temperature sub-reward function, the letter B to represent the energy consumption sub-reward function, and the letter C to represent the control fluctuation sub-reward function, a judgment matrix is ​​constructed based on the relative importance of these sub-reward functions, as shown below:

[0086] In the formula, This represents the judgment matrix.

[0087] It should be noted that the higher the relative importance value, the greater the preference for that sub-reward function.

[0088] S3. Calculate the importance weight coefficient of each sub-reward function in the judgment matrix.

[0089] Specifically, see Figure 5 The method for calculating the importance weight coefficients of each sub-reward function in the judgment matrix is ​​as follows: S31, Normalized Judgment Matrix The normalized matrix is ​​obtained.

[0090] Specifically, the method for normalizing the judgment matrix is ​​as follows: [The judgment matrix is ​​then normalized.] The normalized matrix is ​​obtained by dividing each column element by the sum of the column elements.

[0091] S32. Calculate the average value of each row of the normalized matrix to obtain the importance weight coefficient of each sub-reward function. Indicates the first The importance weight coefficients of each sub-reward function are calculated using the following formula:

[0092] In the formula, Indicates the number of elements. Indicates the first Line 1 The normalized matrix elements of the column, Indicates the first Line 1 The normalized matrix elements of the column.

[0093] In one embodiment of the present invention, after calculating the importance weight coefficients of each sub-reward function in the judgment matrix, the rationality of the judgment matrix is ​​verified by a consistency check. This consistency check ensures the reliability of the importance weight coefficient calculation results and avoids decision-making biases caused by logical contradictions in expert subjective judgments.

[0094] In some embodiments, the method for verifying the consistency of the judgment matrix is ​​as follows: calculate the consistency ratio using the consistency index and the random consistency index, determine whether the consistency ratio is less than a set threshold, and consider the consistency of the judgment matrix to be reasonable when the consistency ratio is less than the set threshold.

[0095] The consistency index is expressed as follows:

[0096] In the formula, Indicators of consistency This represents the judgment matrix to be tested. Represents the judgment matrix The corresponding feature vector; The consistency ratio is expressed as:

[0097] In the formula, Indicates the consistency ratio. This represents the random consistency index.

[0098] For example, setting the threshold to 0.1, the calculated consistency ratio is: If the consistency of the judgment matrix is ​​considered reasonable, then the consistency of the judgment matrix is ​​considered reasonable.

[0099] S4. Linearly weight each sub-reward function with its corresponding importance weight coefficient to obtain the reward function of the control agent model.

[0100] For example, the importance weight coefficient of the security item reward function is set to 20.

[0101] Based on the relative importance of the reward functions for temperature, energy consumption, and control fluctuation, a judgment matrix is ​​constructed as follows:

[0102] Normalized judgment matrix The normalized matrix is ​​obtained. The average value of each row of the normalized matrix is ​​calculated to obtain the importance weight coefficients of each sub-reward function [0.55 0.25 0.2]. T .

[0103] The reward function of the control agent model is expressed as:

[0104] By designing importance weight coefficients for the sub-items under each sub-reward function using the above-described method of constructing the reward function, the reward weight coefficients for the average temperature of the battery cluster and the temperature range of the battery cluster in the temperature sub-reward function can be obtained as follows: In the energy consumption sub-reward function, the units for air conditioning energy consumption, crossflow fan energy consumption, PTC energy consumption, and air conditioning switch energy consumption are all the same. Based on the peak power consumption ratio of each device, the importance weight coefficients of each sub-item in the energy consumption sub-reward function can be obtained after normalization. In the reward function for controlling fluctuation items, the normalized importance weight coefficients for the following items are set: air supply temperature fluctuation item, air supply speed fluctuation item, crossflow fan PWM signal fluctuation item, and PTC control voltage fluctuation item. Then the reward functions for the temperature term, energy consumption term, and control fluctuation term are expressed as follows:

[0105] See Figure 6 According to a second aspect of the present invention, a thermal management method for a marine fully enclosed energy storage battery is provided, the steps of which are as follows: S1. Generate state variables by acquiring battery cluster temperature information, temperature influence factors affecting battery cluster temperature, air conditioning energy consumption, crossflow fan energy consumption, and PTC heating element energy consumption.

[0106] Specifically, the battery cluster temperature information includes the battery cluster temperature and the battery cluster temperature range. Temperature influencing factors include ambient temperature, solar radiation received by the battery compartment, and the battery cluster charging and discharging current. The state variables are represented as follows:

[0107] In the formula, express State quantity at any given time. express The temperature of the battery cluster at any given time, Due to the extreme temperature difference in the battery clusters, for ambient temperature at any time for The battery compartment receives external solar radiation at all times. for The charging and discharging current of the battery cluster at any time, for Air conditioning energy consumption at all times for Crossflow fan energy consumption at all times for Energy consumption of the PTC heating element at any given time.

[0108] S2. Based on the state variables, the control variables are obtained through a control agent model based on the SAC algorithm.

[0109] Specifically, the control quantity is expressed as:

[0110] In the formula, for Set the air supply temperature at all times. for The airflow setting at all times. for The duty cycle of the crossflow fan PWM signal at any given time. for The DC control voltage of the PTC heating element at any given time.

[0111] Specifically, the control agent model based on the SAC algorithm includes one Actor network and two independent Critic networks. The two Critic networks have the same structure but different parameters, and the parameters of the target Critic network are gradually updated to approach those of the main Critic network.

[0112] Specifically, in the embodiments of this application, the Actor network structure is as follows: Figure 7As shown, the input layer of the Actor network has an 8-dimensional state variable. The hidden layer initially performs preliminary feature space mapping through a 128-unit fully connected neural network (i.e., a fully connected layer); a nonlinear transformation is achieved through the ReLU activation function to alleviate the gradient vanishing problem; then, high-dimensional spatial distribution features are extracted through the fully connected neural network, and the ReLU activation function layer serves as the output time-series feature basis vector for the LSTM layer; the LSTM layer identifies the periodicity of temperature changes and predicts temperature change trends; finally, the mean and standard deviation are output according to the Gaussian sampling strategy, both of which have a 4-dimensional output dimension. The mean provides the baseline operation, and the standard deviation adjusts the randomness, thereby achieving precise control and safe exploration of the action space (i.e., the control variable).

[0113] Specifically, in the embodiments of this application, the Critic network structure is as follows: Figure 8 As shown, the input layers of the Critic network are an 8-dimensional state observation vector (i.e., state variables) and a 4-dimensional action vector (i.e., control variables). The state vector passes through a fully connected neural network in the hidden layers (i.e., a fully connected layer) to extract features, mapping the high-dimensional state space to a low-dimensional feature space and eliminating dimensional differences. The action vector passes through a fully connected neural network in the hidden layers to encode mixed action features, enhancing the continuity of policy evaluation. The state and action features are then concatenated along the feature dimensions through dimensional concatenation. After nonlinear mapping using the ReLU activation function, the high-order feature abstraction layer of the fully connected neural network extracts deep correlation features of the state-action interaction. The ReLU function layer then prepares static feature basis vectors for the LSTM layer. Finally, the LSTM temporal modeling layer outputs the Q-value to evaluate the long-term expected return of the current state-action pair, guiding the policy update of the Actor network.

[0114] The SAC-based control agent model employs an Actor-Critic interactive optimization framework with dual Critic networks. By introducing entropy regularization of the policy, it balances the exploration and utilization of the action space (i.e., control variables) by the control agent model, thereby improving the model's robustness and learning efficiency from samples.

[0115] It's important to note that entropy is used to describe the disorder or information uncertainty of a system. In the SAC algorithm of the control agent model, the entropy of the policy represents the disorder of different actions output by the policy for a given state s. For example, for a random variable... The entropy is defined as follows:

[0116] In the formula, Represents random variables entropy, The formula function representing the mathematical expectation. Represents random variables The corresponding probability distribution.

[0117] The optimization objective of the SAC algorithm is not only to maximize the cumulative reward, but also to maximize the policy entropy. The formula for its objective function is as follows:

[0118] In the formula, Represent the objective function; Representation strategy; Indicate Discount factor for each moment; express The reward function at each time step; This represents the temperature coefficient, used to control the weight of entropy; express Time-state quantity The strategy entropy.

[0119] It should be noted that the greater the entropy, the stronger the randomness of the strategy, which encourages the control agent model to explore more actions and avoids premature convergence to a local optimum.

[0120] The SAC algorithm automatically adjusts... (For example, the Lagrange multiplier method) Dynamic equilibrium exploration and utilization make the policy entropy approach the preset target entropy. Value, target entropy It is usually a negative value of the action space dimension, and its calculation formula is as follows:

[0121] In the formula, This indicates the adjusted temperature coefficient. This indicates calculating the entropy of the corresponding variable. Represents the policy entropy.

[0122] Within the framework of maximum entropy, state value V Function and Action Value Q The function is then rewritten as follows:

[0123]

[0124] In the formula, The state-value function, calculated by the target network, includes the Q-value of the next state and the policy entropy. This represents the entropy of the action policy at the current moment. This represents the action value reward for the current state-action pair. This indicates the strategy for the current action corresponding to the current state. This indicates the reward for the current action. Indicates the discount factor. The entropy represents the state at the next moment. Indicates the state at the next moment. It represents the state value of the state at the next moment.

[0125] The Actor network (policy network) in the SAC algorithm outputs the probability distribution of actions (such as the mean and variance of a Gaussian distribution). Through the reparameterization trick—that is, sampling noise that follows a standard normal distribution—actions are generated. The calculation formula is as follows:

[0126] In the formula, Indicates an action, This represents the mean. Indicates sampling noise. Indicates variance.

[0127] The Critic network uses two independent Q-networks (Q1 and Q2) to estimate the state-action value function, taking the minimum value as the update target to alleviate the overestimation problem of Q-value. The formula for calculating the target Q-value is as follows:

[0128] In the formula, Indicates the target value. Indicates a reward. Indicates the discount factor. This represents the Q-value of the state-action pair at the next moment. This represents the strategy for the state-action pair at the next moment.

[0129] The target network is trained stably using a soft update mechanism, where its parameters are updated slowly, following the main network's updates. The calculation formula is as follows:

[0130] In the formula, This represents the updated set of target network parameters. Indicates the soft update coefficient. This represents the current set of target network parameters.

[0131] The SAC-based control agent model deeply integrates the maximum entropy principle with the Actor network (policy network) - Critic network (evaluation network) framework, forming an entropy-regularized reinforcement learning mode. This model introduces entropy as a quantitative indicator of the model's exploration capability into the policy gradient method. By dynamically balancing exploration and utilization through an adaptive temperature factor, it enhances policy diversity while maximizing cumulative rewards, effectively solving the problem of traditional algorithms easily getting trapped in local optima in complex control tasks.

[0132] For example, when training a control agent model based on the SAC algorithm, the hyperparameters of the control agent model are set as follows: the sampling time is set to 120s, the environment simulation time is 86400s, and the number of random interaction steps per round is set to 720. The discount factor γ is set to 0.99, indicating a greater emphasis on obtaining long-term rewards and guiding the control agent model to obtain the globally optimal policy. The experience pool size M is 10. 6 The size B of the mini-batch sampling dataset N is 256, which is larger than the default value, helping the agent model to obtain temperature changes over a longer time scale for iterative strategy optimization. The sequence length is set to 128, which can cover a longer simulation time, allowing the LSTM neural network layer to better identify time-series relationships. The learning rate of the Actor network in the SAC algorithm affects the stability and convergence speed of policy updates; considering the characteristics of the battery thermal management system, the parameter is set to 0.001. The learning rate of the Critic network affects the fitting accuracy of the Q-value function; too high a rate can lead to Q-value overestimation, while too low a rate can cause value assessment lag. Battery temperature changes have inertia and delay, requiring the Critic network to quickly identify relationships; the parameter is set to 0.003, slightly higher than the Actor network learning rate. The entropy temperature coefficient is a key parameter used by the SAC algorithm to balance exploration and exploitation; the initial value is set to 0.2, and the target entropy coefficient is the negative of the action space dimension, -4.

[0133] S3. Control the air supply temperature and air supply speed of the air conditioner, the duty cycle of the PWM signal of the crossflow fan, and the DC control voltage of the PTC heating element according to the control quantity corresponding to the maximization of the reward function of the control agent model.

[0134] The effectiveness of the above-described thermal management method for marine fully enclosed energy storage batteries in managing energy storage batteries will be explained below with reference to specific experiments.

[0135] Experiment 1: A thermal model of the battery thermal management system was established in the Simulink platform. Under typical summer operating conditions, the thermal management control of the battery thermal management system was carried out using the thermal management method of the marine fully enclosed energy storage battery of this invention (hereinafter referred to as: DRL method), the ON / OFF control strategy method (hereinafter referred to as: ON / OFF method), and the rule-based step-by-step control method (hereinafter referred to as: step-by-step method). The simulation results are as follows: Figures 9 to 13 As shown in Table 1, the quantitative data comparison of the control indicators is presented.

[0136] Table 1

[0137] Depend on Figures 9 to 11 As shown, in terms of temperature control, the segmented method has a slower temperature response than the ON / OFF method, but a smaller overshoot ratio for battery cooling, and a smaller maximum temperature range for the battery. Compared to the ON / OFF and segmented methods, the DRL method of this invention produces a more stable average temperature curve for the battery cluster, the smallest maximum temperature range for the battery, and less temperature fluctuation in the battery compartment compared to the ON / OFF and segmented methods. Figure 12 It can be seen that, regarding air conditioner power fluctuation, the DRL method of this invention is superior to the gear-based method, while the ON / OFF method is the worst. Figure 13 It can be seen that, in terms of system power consumption, the gear-level method uses a single air volume and operates less frequently, which is better than the ON / OFF method; although the DRL method of this invention operates the least frequently, it operates for a longer time, and its total power consumption is better than the ON / OFF method but higher than the gear-level method.

[0138] Experiment 2: A thermal model of the battery thermal management system was established in the Simulink platform. Under typical summer and winter operating conditions, the DRL method and ON / OFF method of this invention were used for thermal management control of the battery thermal management system. The quantitative data of the control indicators using the DRL method of this invention are shown in Table 2. The comparison results of the quantitative data of the control indicators using the DRL method and the ON / OFF method of this invention are as follows: Figures 14 to 17 As shown.

[0139] Table 2

[0140] Regarding temperature control: see Figure 14 Under summer operating conditions, the DRL method of this invention controls the average temperature variance of the battery cluster to be 0.11℃, which is less than the 0.31℃ of the ON / OFF method, representing a reduction of 83%. The maximum temperature range of the battery is 0.25℃, which is less than the 0.43℃ of the ON / OFF method. See also Figure 15Under winter operating conditions, the average temperature variance of the battery cluster is 10.51℃, which is less than 11.83℃ under the ON / OFF method, and the maximum temperature range of the battery is 2.1℃, which is less than 3.1℃ under the ON / OFF method.

[0141] Regarding system energy consumption: Under summer operating conditions, see [reference needed]. Figure 16 The power consumption of the DRL method in this invention is 2.74 kWh, which is better than the 2.84 kWh of the ON / OFF method, reducing system energy consumption by 5%. The frequency of air conditioning operation is reduced from 24 to 4, a reduction of 83%. For winter operating conditions, see [link / reference]. Figure 17 The total power consumption of the DRL method in this invention is 7.76 kWh, which is better than the 8.7 kWh of the ON / OFF method, reducing system energy consumption by 11%; PTC power consumption is reduced from 3.6 kWh to 1.7 kWh, a reduction of 53%. The frequency of air conditioning operation is reduced from 34 to 2, a reduction of 94%.

[0142] The above embodiments are used to explain the present invention, but not to limit the present invention. Any modifications and changes made to the present invention within the spirit and scope of the claims shall fall within the protection scope of the present invention.

Claims

1. A marine fully enclosed energy storage battery thermal management system, characterized in that, include: An air conditioner, installed on the battery compartment of the energy storage battery, is used for cooling or heating to dissipate heat or heat the battery clusters of the energy storage battery. The air duct includes a horizontal section located at the top of the battery cluster and a vertical section located on the first side of the battery cluster. The air inlet of the horizontal section is connected to the air outlet of the air conditioner, and the vertical section is provided with a first air outlet at the interval position of the battery cluster. Multiple crossflow fans are installed in the vertical section of the air duct. The multiple crossflow fans are arranged at intervals from top to bottom along the vertical section of the air duct. The crossflow fans are positioned opposite to the first air outlet. PTC heating elements are placed on the surface of the battery clusters to heat them. The data acquisition unit generates state variables for the battery cluster temperature information, temperature influence factors affecting battery cluster temperature, air conditioning energy consumption, crossflow fan energy consumption, and PTC heating element energy consumption. The controller is configured to: obtain control quantities through a control agent model based on the SAC algorithm based on the state quantities generated by the data acquisition unit; and control the air supply temperature and air supply speed of the air conditioner, the PWM signal duty cycle of the crossflow fan, and the DC control voltage of the PTC heating element based on the control quantities corresponding to the maximization of the reward function of the control agent model.

2. The marine fully enclosed energy storage battery thermal management system as described in claim 1, characterized in that, The minimum cooling capacity of the air conditioner is: in, In the formula, Indicates the safety factor; Indicates the battery's heating power; I Indicates the battery charging and discharging current; R Indicates the battery's DC internal resistance; This indicates the heat exchange capacity between the outer wall of the battery compartment and the outside air. This represents the convective heat transfer coefficient, with units of W / (m³). 2 ·K); A Indicates heat transfer area, unit: m² 2 ; This indicates the temperature difference between the ambient temperature and the temperature inside the battery compartment, in °C. This indicates that the battery absorbs heat; Specific heat capacity of the battery, unit: J / (kg·K); Battery mass, unit: kg; To increase battery temperature, unit: °C; This indicates the charging and discharging time of the battery cluster.

3. The marine fully enclosed energy storage battery thermal management system as described in claim 1, characterized in that, The dimensions of the air duct satisfy the following constraints: In the formula, The height of the air conditioner vent, in cm; Width of the air conditioner vent; The height of the horizontal section of the air duct, in cm; The number of air ducts ; Width of the air inlet in a single horizontal section of an air duct, in cm; Length of the vertical section of the air duct, in cm; Width of the vertical section of the air duct, unit: cm; The length of the crossflow fan is in cm. Width of the crossflow fan, unit: cm.

4. The marine fully enclosed energy storage battery thermal management system as described in claim 1, characterized in that, The crossflow fan is mounted on the air duct via a fan mounting bracket, which is fixed at the first air outlet. The side of the fan mounting bracket that is fixed to the air duct has a second air outlet that communicates with the first air outlet, and the side of the fan mounting bracket that mounts the crossflow fan has a first air inlet. The air generated by the crossflow fan blows the gas in the air duct through the first air inlet, the second air outlet, and the first air outlet into the space of the battery cluster to dissipate heat or heat the battery cluster.

5. The marine fully enclosed energy storage battery thermal management system as described in claim 4, characterized in that, The fan mounting bracket is equipped with an air guide plate, which is set at an angle so that the air generated by the crossflow fan is discharged at an angle through the air guide plate and forms convection with the bottom shell of the battery cluster.

6. The marine fully enclosed energy storage battery thermal management system as described in claim 1, characterized in that, The battery cluster temperature information includes the battery cluster temperature and the battery cluster temperature range. Temperature influencing factors include ambient temperature, solar radiation received by the battery compartment, and battery cluster charging / discharging current. The state variables are represented as follows: In the formula, express The state quantity at any given time. express The temperature of the battery cluster at any given time, Due to the extreme temperature difference in the battery clusters, for ambient temperature at any time for The battery compartment receives external solar radiation at all times. for The charging and discharging current of the battery cluster at any time, for Air conditioning energy consumption at all times for Crossflow fan energy consumption at all times for Energy consumption of the PTC heating element at any given time; The control quantities include the air conditioner set air supply temperature, air conditioner air supply speed, crossflow fan PWM signal duty cycle, and PTC heating element DC control voltage.

7. The marine fully enclosed energy storage battery thermal management system as described in claim 6, characterized in that, The reward function of the control agent model is expressed as: In the formula, express The reward function of the control agent model at each time step. express The temperature-related sub-reward function at time intervals, This represents the weighting coefficient of the temperature-related sub-reward function. express The energy consumption sub-reward function at time step. This represents the weighting coefficient of the energy consumption item's sub-reward function. express The time-varying control fluctuation term sub-reward function, This represents the weighting coefficient of the sub-reward function that controls fluctuations. express The safety item reward function at any given time. This represents the weighting coefficient of the reward function for the safety item.

8. The marine fully enclosed energy storage battery thermal management system as described in claim 7, characterized in that, The temperature term sub-reward function is expressed as follows: In the formula, This indicates a positive reward given when the battery cluster temperature is stable within a set temperature range. >0; Indicates the boundary threshold; This represents the penalty coefficient for violations of the average temperature regulation of the battery cluster. <0; This indicates the penalty coefficient imposed when the temperature difference between battery clusters exceeds 5°C. <0; This indicates the average temperature of the battery cluster; The energy consumption item sub-reward function is expressed as follows: In the formula, This represents the penalty coefficient for the energy consumption item. <0, energy consumption items include air conditioning energy consumption. Crossflow fan energy consumption PTC heating element energy consumption ; This represents the penalty coefficient for energy consumption when the air conditioner is switched on or off. <0, Energy consumption for switching on and off the air conditioner; The sub-reward function for the control fluctuation term is expressed as follows: In the formula, This represents the penalty coefficient for controlling for fluctuations. <0; Set the change value of the supply air temperature between the previous moment and the current moment; The change value of the air supply speed between the previous moment and the current moment; This represents the change in the duty cycle of the PWM signal for the crossflow fan between the previous and current times. The change in the DC control voltage of the PTC heating element between the previous moment and the current moment; The sub-reward function for the security item is expressed as follows: In the formula, This represents the penalty coefficient for incorrect temperature adjustment decisions made by the control agent model. <0; This represents the penalty coefficient when the decision made by the control agent model results in negative pressure in the air duct. <0; This indicates the air conditioner's set airflow temperature. This represents the total change in duct pressure.

9. The marine fully enclosed energy storage battery thermal management system as described in claim 7, characterized in that, The method for constructing the reward function of the control agent model is as follows: Set the weight coefficients for the reward function of the security item; Construct a judgment matrix based on the relative importance of the temperature sub-reward function, the energy consumption sub-reward function, and the control fluctuation sub-reward function; Calculate the importance weight coefficients of each sub-reward function in the judgment matrix; The reward function of the control agent model is obtained by linearly weighting each sub-reward function with its corresponding importance weight coefficient.

10. A thermal management method for a marine fully enclosed energy storage battery, employing the marine fully enclosed energy storage battery thermal management system as described in any one of claims 1 to 9, characterized in that, The steps are as follows: The acquired battery cluster temperature information, temperature influence factors affecting battery cluster temperature, air conditioning energy consumption, crossflow fan energy consumption, and PTC heating element energy consumption will be used to generate state variables. The control quantity is obtained from the state quantity through a control agent model based on the SAC algorithm; The control quantity corresponding to maximizing the reward function of the control agent model controls the air conditioner's air supply temperature and air supply speed, the PWM signal duty cycle of the crossflow fan, and the DC control voltage of the PTC heating element.

Images