A hybrid energy scheduling method and system for environmental perception in polar environment

Abstract

The application provides an energy scheduling method and system for environmental perception in a polar environment, and relates to the technical field of energy management and control. The scheduling method comprises the following steps: acquiring external environment parameters, dynamically setting a dynamic safety thermal interval of a BESS based on the external environment parameters; acquiring internal state parameters of the BESS, wherein the internal state parameters comprise an electrochemical state, a thermal distribution map and a thermal velocity map of a modular battery pack; constructing a time-series thermodynamic map of the BESS, wherein the map aggregates the internal state parameters to a graph node representing the modular battery pack, and comprises physical adjacency relationships of the node and external exposure characteristics; determining an optimal intervention strategy, wherein the optimal intervention strategy is a thermal management action output by reinforcement learning based on the time-series thermodynamic map; measuring a safety schedulable interval of the BESS; and performing energy collaborative scheduling.

Description

Technical Field

[0001] This application relates to the field of energy management and control technology, and more specifically, to a hybrid energy dispatching method and system for extreme environments. Background Technology

[0002] Polar research stations are crucial bases for scientific research in extreme environments. Their operation is highly dependent on a stable and reliable energy supply. Traditionally, polar research stations have relied primarily on fossil fuels, such as diesel generators, to ensure power and heating. However, fossil fuels are expensive to transport, have short resupply windows, and their combustion emissions threaten the fragile polar ecosystems. Therefore, introducing renewable energy sources such as wind and solar power to construct hybrid energy systems combining fossil fuels and renewable energy has become a trend in polar energy development.

[0003] In conventional hybrid energy systems, the core objective of energy dispatch is usually to achieve economic optimization or load supply-demand balance. For example, in grid-connected systems, dispatch strategies aim to maximize the absorption of renewable energy, or to charge during off-peak hours and discharge during peak hours to pursue the lowest possible electricity costs. In conventional islanded microgrids, traditional dispatch also focuses on how to combine different energy sources to meet real-time load demand with the lowest possible fuel consumption, with a focus on whether the power supply is sufficient.

[0004] However, the unique characteristics of the polar environment render traditional energy dispatch objectives, oriented towards economic efficiency or power balance, inapplicable. The polar environment is characterized by extreme low temperatures, strong winds, sudden blizzards, and the alternation of polar day and night. These harsh environmental factors pose a serious threat to the safety of energy equipment itself. For example, key components in renewable energy systems, such as wind turbines and solar photovoltaic panels, face problems such as mechanical icing and material embrittlement.

[0005] Modular battery energy storage systems, in particular, which serve as the core buffer unit of hybrid energy systems, are extremely sensitive to operating temperatures. Even with the use of low-temperature resistant cells, lithium-ion batteries still face safety risks such as reduced electrochemical activity, rapid capacity decay, permanent damage due to lithium plating, and even thermal runaway when operating at extremely low temperatures. Therefore, energy storage systems must be maintained within a specific safe operating temperature range to ensure their operational safety and lifespan.

[0006] Furthermore, in large energy storage cabinets, the heat exchange conditions vary due to the different physical placement of battery modules. Modules closer to the cabinet shell are more susceptible to extreme low temperatures, while central modules are prone to heat accumulation, resulting in a significant temperature gradient within the cabinet. This temperature inhomogeneity directly leads to differences in the electrochemical characteristics of each module, causing circulating currents and inconsistent output characteristics during charging and discharging, exacerbating the "weakest link" effect in module performance, and ultimately shortening the lifespan of the entire energy storage system.

[0007] In summary, traditional energy dispatching methods focus on the balance between electricity supply and demand and economic efficiency, while neglecting the physical safety constraints of equipment in polar environments. In polar scenarios, fossil fuel reserves are usually plentiful, and energy shortages are not the primary concern. The primary concern lies in ensuring the safe, consistent, and long-term operation of core components of green energy sources such as wind and solar power under extreme low temperatures and high wind cooling effects. Therefore, a new energy dispatching method is urgently needed, which no longer prioritizes energy replenishment but rather focuses on ensuring the physical operational safety of energy supply equipment, and then coordinates energy dispatching on this basis. Summary of the Invention

[0008] This invention provides an energy dispatching method based on environmental sensing in polar environments, the method comprising:

[0009] Obtain external environmental parameters and dynamically set the dynamic safety thermal range of BESS based on these parameters;

[0010] Obtain the internal state parameters of BESS, including the electrochemical state, thermal distribution map, and thermal rate map of the modular battery pack;

[0011] A time-series thermodynamic map of BESS is constructed, which aggregates internal state parameters to graph nodes representing modular battery packs and includes the physical adjacency relationships and external exposure features of the nodes.

[0012] The optimal intervention strategy is determined by the thermal management action output through reinforcement learning, which is based on the time-series thermodynamic spectrum.

[0013] Calculate the safe and schedulable range of BESS and perform coordinated energy scheduling.

[0014] The steps for obtaining external environmental parameters include: collecting the instantaneous wind speed in the near field of BESS. and ambient temperature ;based on and Calculate the air-cooling index to characterize the air-cooling effect. :

[0015]

[0016] in, These are calibration coefficients.

[0017] The steps to obtain solar shading prediction parameters from external environmental parameters include:

[0018] Image sequences of the sky captured by an all-weather fisheye camera ,in The image pixel coordinates are used; the image sequence is processed by a Kalman filter to obtain the optimal angular velocity estimate of the occlusion. );

[0019] Calculating the future path of the sun and the front pixel set of the occlusion angular coordinate extrapolation ,

[0020]

[0021] The estimated occlusion time is determined based on collision detection using angular coordinate extrapolation and the future path of the sun. and the expected duration of the obstruction ;

[0022] in, for time The Middle The angular coordinates of each pixel. For future time steps.

[0023] The steps for dynamically setting the dynamic safety hot zone of BESS include:

[0024] based on , and Calculate the thermal threat factor :

[0025]

[0026] in, This is the lower limit of standard operation. For the air-cooled threat factor, To mask the threat factor, To mask the duration scale factor, To mask the approach attenuation factor;

[0027] based on Calculate the dynamic safety lower limit :

[0028] in, This is the standard operating limit. This is the minimum safe interval width.

[0029] The steps to obtain BESS's internal state parameters include:

[0030] With sampling interval Acquire raw thermal image sequences ,in The image pixel coordinates of the infrared thermal imager; according to generate Thermal distribution diagram at time ; retrieve Thermal distribution diagram at time And calculate the thermal velocity map. :

[0031]

[0032] The steps for constructing the time-series thermodynamic map of BESS include: for each graph node representing a modular battery pack Define a corresponding set of pixel regions ;

[0033] use thermal distribution map and heat rate diagram Perform spatial aggregation to compute nodes. The dynamic characteristics include: average surface temperature. Minimum heating rate .

[0034] The heat management actions output through reinforcement learning are based on maximizing cumulative scalar rewards. strategy Definitely. Defined as:

[0035]

[0036] in, These are the low temperature penalty coefficient, the unevenness penalty coefficient, the intervention cost penalty coefficient, the SOH loss penalty coefficient, and the high temperature penalty coefficient, respectively. for The lowest surface temperature at any given time; Standard deviation; for Average surface temperature at time; For indicator functions; For nodes Performed thermal management actions; This is an idle action; It is a self-heating mechanism; for A person's health status at all times.

[0037] It is generated through a graph neural network, and the computation of a graph neural network includes:

[0038] (a) Perform node feature embedding, embedding the dynamic feature vectors of the nodes. External exposure vector corresponding components splicing :

[0039] in For the 0th layer node of GNN Feature representation;

[0040] (b) to proceed Message propagation in layer graph convolution, at the th The layer updates the hidden state of nodes as follows: :

[0041] in, for In the internal adjacency matrix The set of neighboring nodes defined in the code. and For the first The learnable weight matrix of the layer, It is a non-linear activation function.

[0042] The steps for calculating the safe schedulable range of BESS and performing coordinated energy dispatch include:

[0043] According to strategy Determine the optimal intervention strategy vector Calculation execution Required thermal regulation power consumption ;

[0044] Calculate the upper limit of net discharge power of BESS :

[0045]

[0046] in, This is the set of nodes that can participate in normal scheduling. The set of nodes to be subject to forced rate reduction. As a depreciation factor, for Theoretical maximum safe discharge power; Determine BESS scheduling instructions and fossil energy dispatch instructions :

[0047] in, This represents the net load of the research station.

[0048] This invention also provides an environmental sensing energy dispatching system for polar environments, the system comprising:

[0049] Acquisition module: Acquires external environmental parameters and dynamically sets the dynamic safety hot zone of BESS based on these parameters;

[0050] Analysis module: Acquires internal state parameters of BESS, including the electrochemical state, thermal distribution map, and thermal rate map of the modular battery pack;

[0051] Optimization module: Constructs a time-series thermodynamic map of BESS, which aggregates internal state parameters to graph nodes representing modular battery packs, and includes the physical adjacency relationships and external exposure features of the nodes;

[0052] Decision module: Determines the optimal intervention strategy, which is a thermal management action output through reinforcement learning based on the time-series thermodynamic spectrum;

[0053] Scheduling module: Calculates the safe schedulable range of BESS and performs coordinated energy scheduling.

[0054] This invention addresses the lag problem of traditional fixed safety threshold management strategies. By acquiring local environmental parameters in the near field of the BESS, this invention can anticipate impending thermal threats. Based on these forward-looking external environmental parameters, the system can dynamically set the dynamic safe thermal range of the BESS and proactively raise the dynamic safety lower limit to achieve proactive preheating of the BESS. This allows the BESS to accumulate sufficient thermal inertia before extreme wind-cooling impacts, thereby transforming thermal safety management from a passive response to an active defense.

[0055] This invention addresses the problem of traditional battery management systems' inability to perceive the complex thermal gradients within a Battery Escalator (BESS). By introducing an internal state perception module, this invention utilizes an infrared thermal imager to generate high-resolution thermal distribution and thermal velocity maps. The thermal velocity map, as a crucial transient indicator, can instantly capture the rapid cooling trend of edge battery modules caused by wind cooling effects. This response speed occurs far before significant changes in the absolute temperature of the module, providing the fastest decision-making basis for intervention. This invention also provides an intelligent decision-making mechanism based on spatial context. It constructs a time-series thermodynamic map of the BESS. This map not only aggregates high-resolution thermodynamic data and electrochemical states, such as health status, onto graph nodes representing physical modules, but also represents the physical adjacency relationships and external exposure characteristics between modules. This graph structure enables the thermal safety reinforcement learning agent deployed in the energy management controller to perform true spatial analysis. When making decisions using a graph neural network, the agent considers not only the state of the node itself, such as the minimum thermal velocity, but also the states of its neighbors and whether it is an edge module. Based on this spatial perception capability, the agent can learn an optimal intervention strategy with minimal intervention by optimizing a multi-objective reward function. Finally, this invention clarifies the master-slave relationship between safety and scheduling. The system first calculates thermal regulation power consumption based on the optimal intervention strategy and determines which modules must be derated due to thermal management or heat accumulation, thereby calculating the true safe and dispatchable range of the BESS. Fossil fuels only intervene as balancing units when the safe and dispatchable power of the BESS cannot meet the net load. This method not only ensures the reliability of the polar research station's load but also fundamentally ensures the physical safety, temperature consistency, and long-term operation of the BESS in extreme low-temperature and high-air-cooling environments. Attached Figure Description

[0056] Figure 1 This is a flowchart of the energy dispatching based on environmental perception in polar environments according to the present invention. Detailed Implementation

[0057] S1: Deployment of energy dispatching systems in polar environments

[0058] This embodiment discloses an energy dispatching method and system for environmental sensing in polar environments. The system is installed at a polar research station, which is equipped with a fossil fuel power supply module, a renewable energy power supply module, and predetermined loads for scientific research and life support. The renewable energy power supply module includes a photovoltaic array and a wind turbine generator. The system also includes: a modular battery energy storage system (BESS) for storing the electrical energy generated by the renewable energy power supply module; an environmental sensing module for collecting external environmental parameters and weather forecast data of the polar research station; an internal state sensing module for collecting internal operating state parameters of the modular battery energy storage system; and an energy management controller (EMS) connected to the environmental sensing module, the internal state sensing module, the fossil fuel power supply module, the renewable energy power supply module, and the modular battery energy storage system.

[0059] In this embodiment, the system is deployed in the environment of a research station on the Antarctic inland ice sheet, such as Dome A. This environment is characterized by extreme low temperatures, with winter temperatures reaching below -80°C, strong winds, prolonged polar nights and days, and low humidity and low air pressure. The research station's load consists of highly reliable equipment such as life support systems, scientific instruments, and a data center. The fossil fuel power supply module is specifically a diesel generator set with sufficient capacity to independently handle the research station's maximum annual load, and is equipped with a combined heat and power (CHP) module for recovering waste heat for heating.

[0060] Photovoltaic arrays can generate electricity continuously for 24 hours during the polar day, from November to February of the following year, but generate zero electricity during the polar night. Wind turbines mainly rely on polar descent winds, and their power output fluctuates drastically depending on wind speed. The electricity generated by these renewable energy sources (solar and wind power) is input into a modular battery energy storage system (BESS) through an energy storage converter. The BESS is the core protected object of this invention, and its specific form is one or more containerized or rack-mounted energy storage units, which integrate a large number of modular battery packs arranged in a matrix.

[0061] The safety threat faced by this invention does not stem from insufficient power, but rather from thermodynamic instability under extreme conditions. The BESS enclosure is directly or indirectly exposed to the polar atmosphere. When the external environment is -40°C with high-speed descending winds, the enclosure experiences extremely high wind cooling, resulting in a huge convective heat transfer coefficient. This intense localized cooling effect causes the temperature of the edge battery modules located near the outer wall or vents in the energy storage cabinet to drop rapidly and non-linearly, much faster than that of the inner modules. The battery modules located in the center of the energy storage cabinet, being surrounded by other modules, have poor heat dissipation conditions, and during continuous charging or high-power discharging during polar days, internal heat easily accumulates. A complex three-dimensional thermodynamic gradient forms inside the BESS, which is difficult for traditional temperature sensors to accurately detect. If the temperature of the edge modules falls below their lower safety limit, it will lead to irreversible lithium plating and permanent capacity decay; if the temperature of the central module exceeds its upper safety limit, there is a risk of thermal runaway. Given an abundance of fossil fuels, the primary task of energy dispatching in this embodiment is not how to utilize BESS (Battery Energy Storage System) to supplement the power shortage, but rather how to proactively manage the internal thermal state of the BESS to ensure that each battery module operates within a safe temperature range and maintains a high degree of temperature consistency, thereby guaranteeing the physical safety and operational lifespan of the entire energy storage system. The energy dispatching of this invention is a result achieved under the premise of ensuring the aforementioned thermal safety constraints.

[0062] S2 acquires environmental perception and prediction data.

[0063] To address the issue that conventional weather forecasting systems' macroscopic, large-scale meteorological data cannot meet the transient thermal safety management requirements of modular battery energy storage systems (BESS) in polar environments in terms of both temporal and spatial resolution, this embodiment deploys sensing units in the near field of the BESS to acquire local environmental parameters that have a direct and immediate impact on the BESS's heat exchange state. This generates short-term forecast data that predicts the heat input to the BESS, providing constraints for subsequent energy dispatch.

[0064] S2.1: Quantify the near-field heat exchange parameters of BESS, accurately quantifying the local and transient thermodynamic boundary conditions that the BESS enclosure shell experiences in real time, rather than using average data from the weather tower. Due to the presence of the research station complex, building-around flow and funneling effects will form around BESS, resulting in significant differences in local wind speed and heat flux density compared to the open field.

[0065] S2.1.1: Deploy at least one anemometer and temperature sensor in the modular battery energy storage system (BESS) cabinet.

[0066] S2.1.2: The Energy Management System (EMS) collects the instantaneous wind speed measured by the anemometer. and the instantaneous ambient temperature near the BESS cabinet enclosure as measured by a temperature sensor. .

[0067] S2.1.3: EMS based on data acquisition and Calculate the air-cooling index, which characterizes the intensity of convective heat transfer between the BESS cabinet shell and the external environment. :

[0068]

[0069] in, Indicates in The air cooling index at any given time express Ambient air temperature measured in the near field of BESS at all times. express The equivalent wind speed at a height of 10 meters in the near field of BESS. These are the calibration coefficients for the air-cooled index model.

[0070] S2.2: Predicting solar shading events. Traditional solar intensity meters can only measure changes in irradiance that have already occurred. This invention uses vision to predict upcoming solar shading events in advance, which may be caused by fast-moving clouds or polar-specific snow blowing phenomena.

[0071] S2.2.1: Continuously acquire image sequences of the sky at a constant frame rate using an all-weather fisheye camera deployed at the research station. ,in These are the image pixel coordinates. The calibrated lens projection function for an all-weather fisheye camera. and its inverse function Used to realize pixel coordinates of 2D images With 3D sky angular coordinates (azimuth) Elevation angle Mapping between ).

[0072] S2.2.2: For the acquired image sequence Preprocessing is performed.

[0073] Will Convert from RGB color space to Color space. The polar environment has extremely high surface albedo, resulting in a lower sky brightness (L channel) than surface reflection, and extremely low spatial contrast between clouds / snow / sky. However, in the Lab space... (Red and Green) and The (yellow-blue) chromaticity channel is not sensitive to these disturbances. Clear polar skies exhibit strong negative values ​​(blue) in the b channel, while clouds, fog, and blowing snow exhibit b values ​​close to zero (white / gray).

[0074] S2.2.3: Based on Color channels To segment occlusions, the K-Means clustering algorithm (K=2) is used, employing only pixel-based clustering. Using chroma vectors as features, the sky image is segmented into Sky Clusters. ) and Obscuration Cluster This effectively overcomes lens glare and high-brightness interference caused by the low-angle sunlight during polar days. S2.2.3 outputs a binary mask for the obstruction. ,in area This refers to the identified obstructions.

[0075] S2.2.4: Calculate the translation vector of the shading area .

[0076] The dense optical flow method is used to calculate two consecutive frames. and pixel-level displacement field between , Indicates the time interval between two frames. (This is achieved through...) Masking, calculating the weighted average motion vector of the occluded cluster. As Observations at time:

[0077]

[0078] S2.2.5: Establish a Kalman filter for the motion state of the obstruction.

[0079] Because the optical flow method is susceptible to interference from rapidly changing lighting and the non-rigid deformation caused by snow in polar environments, it leads to... It contains a significant amount of noise. Furthermore, the projection distortion of the fisheye lens causes the pixel velocity of the occluded object on the image plane to decrease. It is not constant. To address this issue, the present invention does not directly use... Instead, it is transformed to the sky angular coordinate system and optimally estimated using a Kalman filter.

[0080] S2.2.5.1 Calculation time Centroid pixels of the region .

[0081] S2.2.5.2 Using the camera inverse projection function Convert it to sky angular coordinates (azimuth). Elevation angle ). As observations of the Kalman filter .

[0082] S2.2.5.3 Define the state vector of the Kalman filter. The position and angular velocity of the obstructing object in the sky angular coordinate system:

[0083]

[0084] in These are the angular velocities of the azimuth and elevation angles, respectively.

[0085] S2.2.5.4 In a short period of time State transition equations and observation equations are established based on uniform angular velocity motion. Prediction and update are performed using standard Kalman filtering. Output optimal state estimate at time step . Includes the optimal angular velocity estimate after noise smoothing. .

[0086] S2.2.6: Collision detection is performed based on optimal state estimation.

[0087] S2.2.6.1 EMS is based on the geographical coordinates (latitude and longitude) of the research station. ) and current time The future path of the sun in the sky was calculated using the Ephemeris astronomical database. ,in For future time steps.

[0088] S2.2.6.2 Identify Obstacle Masks The leading edge pixel set in the direction of solar motion .

[0089] S2.2.6.3 Each pixel in ,use Convert it to sky angular coordinates .

[0090] S2.2.6.4 Using the estimated optimal angular velocity Extrapolate the angular coordinates of all points on the frontier:

[0091]

[0092] S2.2.6.5 Step by step Step, calculate the leading edge of the extrapolated shading. Future path of the sun angular distance between .

[0093] S2.2.6.6 When First time less than the preset threshold At this time This is the expected shading time. .

[0094] S2.2.6.7 Identify Obstacle Masks The set of trailing edge pixels opposite to the direction of motion ,right use Convert it to sky angular coordinates, and interpolate the angular coordinates of all points on the trailing edge to obtain its collision time. The expected duration of the obstruction is... .

[0095] In this embodiment, the present invention solves the problem of mismatch in the spatiotemporal scale of macroscopic meteorological data in polar environments. Wind-cooling index. This allows the EMS to accurately quantify the convective cooling threat faced by the BESS enclosure in real time, rather than relying on a lagging and inaccurate far-field temperature value. Shading prediction parameters This enables the EMS to anticipate the impending interruption of the BESS's photovoltaic heat source and power source within minutes. These two types of high-precision, high-time-response environmental perception and prediction data serve as essential inputs for the energy dispatch controller to formulate proactive thermal safety management strategies and energy dispatch, thereby transforming BESS management from a passive response to an active defense.

[0096] S3 dynamically sets the BESS safety hot zone.

[0097] S3.1: Define the baseline physical security boundary

[0098] EMS loads the inherent physicochemical constraints of the modular battery pack used by BESS from its internal memory. These constraints are static and represent the absolute physical boundaries of BESS operation: This represents the absolute physical lower limit; below this temperature, charging will cause irreversible lithium plating damage. This represents the absolute physical upper limit; exceeding this temperature will trigger the risk of thermal runaway. This indicates the standard operating lower limit, the minimum normal operating temperature to ensure BESS efficiency and lifespan. Standard operating limit: The maximum normal operating temperature to ensure BESS efficiency and lifespan.

[0099] S3.2: Calculate the dynamic safety lower limit

[0100] Based on the predicted external cooling threat, the minimum operating temperature target of BESS is proactively increased, i.e., proactive preheating, in order to increase the thermal inertia of BESS to counteract the impending heat loss.

[0101] S3.2.1: EMS Real-time Calculation Momentary thermal threat factors The thermal threat factor consists of two parts: the current threat of wind-cooling effect and the anticipated threat of photovoltaic heat source disruption.

[0102]

[0103] First item This indicates that the stronger the air-cooling effect, the lower the air-cooling index is compared to the standard operating limit. The lower the value, the greater the threat factor. and The first term represents the expected shading time and duration. The second term quantifies the anticipated interruption threat of the photovoltaic heat source (i.e., charging heat generation). The closer the shading event (…), the greater the expected interruption threat. And the longer the duration ( The larger the value, the greater the threat level, indicating that BESS is about to lose an important source of heat. , These are the wind-cooling threat coefficient and the shading threat coefficient, used to calibrate the increase in target temperature caused by the wind-cooling effect and the increase in target temperature caused by photovoltaic interruption, respectively. The occlusion duration scale factor is used for normalization. . : Masking proximity attenuation factor, used to define The timeliness of the threat.

[0104] S3.2.2: EMS based on Calculate the lower limit of dynamic safe temperature :

[0105]

[0106] in for The dynamic safety temperature limit at any given time, this value is The sum of the thermal threat factor. The minimum safe zone width is preferably set to 5°C to ensure... No more than And maintain necessary control dead zones.

[0107] S3.3: Output dynamic safety thermal range;

[0108] The calculated The interval output serves as a safety constraint target for thermal scheduling.

[0109] This embodiment transforms the thermal management of BESS from passively adhering to static thresholds to actively adapting to dynamic constraints in the polar environment. By utilizing... and The data enabled the preheating of the BESS to counteract the threat of air cooling; the dynamic safe thermal range output in this embodiment ensures that all subsequent energy dispatch decisions are made within the safe boundary adapted to the current environment.

[0110] S4: Acquire BESS infrared images and perform thermal velocity analysis;

[0111] Next, this embodiment generates a high-resolution thermal distribution map and a thermal velocity map by fusing electrochemical data from the battery management system (BMS) and time-series infrared thermal imaging data. This thermal velocity map quantifies the rate of temperature change within the BESS and is a critical internal state input for safety management and energy dispatch.

[0112] In this embodiment, the method specifically includes the following steps:

[0113] S4.1: Acquire BESS electrochemical state data;

[0114] EMS uses the BESS's built-in battery management system to query the integrated battery management system within BESS at preset intervals. A modular battery pack. For the first One battery pack ( EMS acquires and stores its current state of charge. and health status .

[0115] S4.2: Generate a heat map;

[0116] The internal status sensing module includes at least one infrared thermal imager installed inside the BESS cabinet. The field of view (FOV) of this thermal imager covers the entire area. The surface of a modular battery pack.

[0117] EMS controls the thermal imager to maintain a constant frame rate. Continue acquiring infrared radiation images inside BESS. This frame rate... The corresponding sampling interval is .

[0118] This step generates the original thermal image sequence. ,in These are the pixel coordinates of the infrared thermal imager. This is the current sampling time. A thermodynamic distribution map is generated based on the radiation-temperature conversion.

[0119] S4.3: Generate a heat map; retrieve the current time... and at the previous sampling time Generated thermal distribution map and .

[0120] Perform pixel-level subtraction on the two images and divide by the sampling interval. To calculate the thermal velocity map :

[0121]

[0122] S5 constructs a time-series thermodynamic map ;

[0123] This embodiment uses data fusion and graph modeling to aggregate high-resolution pixel data to the corresponding physical battery modules, establish physical adjacency relationships and heat conduction paths between modules, and finally output a time-series thermodynamic map. .

[0124] S5.1: Define the static structure of the thermodynamic spectrum

[0125] Construct and store the static structure of the graph, which consists of a set of nodes. and adjacency matrix and external exposure vector The structure is used to characterize the physical topology inside the BESS cabinet.

[0126] S5.1.1: Define the node set

[0127] Node set of the graph Integrated from BESS Defined by a modular battery pack:

[0128]

[0129] Among them, nodes Corresponding to the A modular battery pack.

[0130] S5.1.2: Define the internal adjacency matrix

[0131] Internal adjacency matrix Used to describe the internal heat conduction paths between nodes (battery packs).

[0132]

[0133] in For matrix The Line 1 Column elements.

[0134] S5.1.3: Define the external exposure vector

[0135] To characterize the difference between the edge battery modules and the center battery modules, EMS stores a External exposure vector of dimension .

[0136]

[0137] in For vectors The One element, It is a key static feature for quantifying the threat of wind-cooling effects.

[0138] S5.2: Establish pixel-node mapping relationship

[0139] for Each node in Generate a corresponding set of pixel regions :

[0140]

[0141] Mapping relationship It is solidified and stored in EMS.

[0142] S5.3: Dynamic characteristics of computing nodes

[0143] S5.3.1: Aggregate pixel-level thermal data

[0144] Using mapping Regarding the thermal distribution map and heat rate diagram Perform spatial aggregation.

[0145] (a) Calculation average surface temperature Average temperature is the fundamental criterion for determining whether the module meets the set dynamic safety lower limit. Average temperature indicates whether the module is in a safe preheating state or under threat of low temperature in the current environment.

[0146]

[0147] (b) Calculation Average surface thermal velocity In polar regions, a module's average temperature may temporarily remain within a safe range, but if its average thermal rate is a large negative value, it indicates that it is losing heat rapidly and overall. This provides an early warning for the EMS (Electronic Management System), allowing it to monitor the module's average temperature. Not yet We intervened to keep the temperature up before it fell below the warning line.

[0148]

[0149] (c) Calculation minimum surface temperature The lowest surface temperature is the actual temperature of the weakest link on the module and is the ultimate safety baseline for determining whether irreversible lithium plating damage has occurred.

[0150]

[0151] (d) Calculation minimum heating rate Minimum thermal velocity is the most direct, instantaneous representation of the air-cooling effect. When a strong blast of cold air hits the BESS cabinet, the minimum temperature may take several minutes to drop; however, the minimum thermal velocity will immediately appear as a large negative value. It indicates that the threat has occurred and the cooling rate at a certain point has become out of control. This allows the system to achieve the fastest transient response, well before the average and minimum temperatures of the modules change significantly.

[0152]

[0153] in For set The number of pixels in the image.

[0154] S5.3.2: Combination Node Feature Vector

[0155] The aggregated pixel thermal data in S5.3.1 is combined with the electrochemical features obtained to form a node. exist Complete feature vector at time step :

[0156]

[0157] The vector Includes nodes The average thermal state, instantaneous thermal trend, local cold spots, instantaneous cooling threat, power consumption, and health status.

[0158] S5.4: Output time-series thermodynamic map

[0159] Combining static structure and dynamic node features, in Construct and output complete time-series thermodynamic maps at all times. :

[0160]

[0161] in It is the set of dynamic feature vectors of all nodes.

[0162] This embodiment performs spatialized data fusion and topological structural modeling. The thermodynamic spectrum output by this embodiment Within a unified data structure, the detailed status of each battery module is simultaneously represented. ) and the physical connection relationship between modules ( More importantly, this embodiment exposes vectors externally. and minimum heating rate These characteristics, combined with other features, integrate the issue of marginal extreme cold and wind chill threats in polar environments into the atlas. This will enable intelligent decision-making based on spatial context, rather than blindly adjusting isolated nodes.

[0163] S6 deploys a thermally safe reinforcement learning agent;

[0164] S6.1: Define the reinforcement learning environment. To deploy a reinforcement learning agent, we first construct a Markov decision process environment, which defines the agent's state space, action space, and reward function.

[0165] S6.1.1: Define the state space ;

[0166] intelligent agents in state of time Defined as in Time-series thermodynamic maps output at each time step :

[0167]

[0168] S6.1.2: Define the action space ;

[0169] intelligent agents in Moment of action Defined as a A composite action vector of dimension, where Total number of modular battery packs in BESS:

[0170]

[0171] in, In order to be in Time for the first Nodes The thermal management actions performed. From a discrete set of actions Select from:

[0172]

[0173] This indicates that the system is idle and does not perform any active thermal management; this is the default option for minimal intervention. This indicates external heating, activating the module. The associated thermal management unit preferably utilizes a low-power positive temperature coefficient heating element for heating. This indicates that the self-heating module is controlled by EMS. It performs charge-discharge cycles at minute rates and generates its own heat using its internal ohmic heat. This refers to cooling, as the central battery module may accumulate heat during continuous charging in the polar day, despite the predominantly low temperatures of the polar environment. Preferably, cooling is achieved by immediately and forcibly reducing the charge / discharge rate of the module to lower its temperature.

[0174] S6.1.3: Define the reward function ;

[0175] according to state of time and the actions performed The result state of time Calculate a scalar reward The reward function is designed as a multi-objective optimization to guide the agent to achieve the core objective of this invention:

[0176]

[0177] The first penalty is low temperature. This is the low-temperature penalty coefficient. This penalty applies to any node. minimum surface temperature Falling below the defined dynamic safety lower bound This behavior is the highest priority security constraint of this invention.

[0178] The second uneven penalty: This represents the non-uniform penalty coefficient. for Average surface temperature of all nodes at time 1 The standard deviation. This penalty applies to the thermal gradient within BESS, guiding the agent to achieve temperature uniformity. The third intervention cost penalty: This represents the intervention cost coefficient. This is an indicator function. This item applies to any non-idle... The action is penalized to achieve the goal of minimal intervention. The fourth item is the SOH depletion penalty: This is the SOH loss penalty coefficient. This factor primarily penalizes the impact on the health status. Lower (i.e.) Larger module execution Actions to protect older battery modules. Fifth high-temperature penalty: This is the high-temperature penalty coefficient. The average temperature of the penalty center module. Exceed .

[0179] S6.2: Construct a policy network based on a graph neural network (GNN);

[0180] To process the graph structure states defined in S6.1.1 Agent policy network A graph neural network architecture is employed. This architecture can learn the complex relationships between node features and graph topology, enabling spatial analysis.

[0181] S6.2.1: Node feature embedding;

[0182] Each node The static and dynamic features are fused. Dynamic feature vector of nodes at time step External exposure vector of the node Corresponding components in Concatenation:

[0183]

[0184] in This represents vector concatenation. The node of the 0th layer (i.e., the input layer) of the GNN. The feature representation of a node is achieved by concatenating its dynamic feature vector with its external exposure vector, ensuring that the agent can distinguish between edge modules and center modules in the first layer.

[0185] S6.2.2: Graph convolution message propagation;

[0186] use The graph convolution operation of the layer allows node information to be transmitted along the defined internal adjacency matrix. To spread. In the... Layers, nodes Hidden state By aggregating itself in the first Layer state and all its neighbors status Update:

[0187]

[0188] in, In order to be in Defined in The set of neighboring nodes. and For the first The weight matrix used in the layer to update itself and aggregate neighbors. This is a non-linear activation function. (Through...) Layer iteration, Includes All information within its neighborhood. This allows the agent to make decisions... When taking action, the neighbor's interests were already considered. The state, for example, neighbors It is an edge module and is rapidly cooling down, thus enabling spatial analysis.

[0189] S6.2.3: Policy network output;

[0190] exist At time, embed the final node output by the GNN. ( The inputs are respectively fed into an output policy head composed of a multilayer perceptron (MLP) to generate... The probability distribution of each action:

[0191]

[0192] For intelligent agents in Time to node The final action strategy.

[0193] This embodiment deploys a reinforcement learning agent based on a graph neural network (GNN). This agent can autonomously learn the complex spatial heat conduction patterns within the BESS (Body Estimated Energy System) and external environmental threats in the polar environment. By optimizing a multi-objective reward function, the agent can generate an optimal minimal intervention strategy. Instead of blindly adjusting all modules, it can accurately identify which key nodes require what actions at what times, maintaining the entire BESS map within the defined dynamic safety thermal range at the lowest cost.

[0194] S7 implements security intervention and coordinated energy dispatch.

[0195] After obtaining the output of the policy network This embodiment transforms the probabilistic strategy output by the intelligent agent into deterministic thermal management actions; and based on the constraints imposed by these safety actions, calculates the safe schedulable power of the BESS at the current moment; based on this schedulable power, coordinates the scheduling of renewable energy, BESS, and fossil fuels to meet the load of the research station, thereby realizing the safe scheduling of this invention.

[0196] S7.1: Output the optimal intervention strategy

[0197] exist At any moment, from the strategy network Get the All The probability distribution of actions at each node is used. A greedy strategy is employed to select the action with the highest probability and output it deterministically. Optimal intervention action at the right time :

[0198]

[0199] in In order to be in Time to node Selected deterministic thermal management actions. EMS will handle all of them. Combined into Optimal intervention strategy vector at time step .

[0200] S7.2: Calculate the dispatchable energy of BESS

[0201] Based on Calculate the net power that BESS can safely provide or absorb under strict adherence to this thermal safety intervention strategy.

[0202] S7.2.1: Calculate thermal regulation power consumption

[0203] Calculation execution Internal power consumption required for all active heating actions :

[0204]

[0205] in, for Total power consumption used for thermal management at any given time. for The rated electrical power consumed by the heating action. for The calibrated average power consumption during micro-rate charge-discharge cycles.

[0206] S7.2.2: Calculate module-level safe power limits

[0207] According to the obtained In and Status, query the battery safety work area lookup table stored in EMS. To determine each module Instantaneous charge and discharge power limits.

[0208]

[0209]

[0210] in, and A pre-calibrated SOA lookup table is provided for BESS manufacturers, which clearly defines the safe discharge / charge rates at different temperatures and states of charge. and :for exist The theoretical maximum safe discharge / charge power at any given time.

[0211] S7.2.3: Calculate the net dispatchable power range of BESS

[0212] Calculate the upper limit of net discharge power that BESS as a whole can provide to external loads. and the upper limit of net charging power that can be absorbed from an external power source .

[0213] (a) Net discharge power limit :

[0214]

[0215] (b) Net charging power limit :

[0216]

[0217] in: This is the set of nodes that can participate in normal scheduling. To execute This refers to the set of nodes that are forcibly reduced in value.

[0218] for The preferred reduction factor for the action. (that is, 10% power is allowed). The thermally regulated power consumption is calculated for S, which is considered as the internal load of the BESS and subtracted from its maximum discharge capacity.

[0219] implement Nodes with this characteristic are not included in any of the above sets because their power is already occupied by internal thermal cycling and cannot be scheduled externally. This step outputs BESS in... The safe schedulable interval at any time .

[0220] S7.3: Perform coordinated energy dispatch

[0221] Based on safety constraints, the energy of the entire research station is coordinated and scheduled to meet the predetermined load. In this case, the energy scheduling is the result of thermal safety constraints.

[0222] S7.3.1: Calculate Net Load

[0223] Real-time data collection of total load at the research station Real-time total output of renewable energy power supply module :

[0224]

[0225] in This indicates that supply cannot meet demand. This indicates an energy surplus.

[0226] S7.3.2: Determine BESS scheduling instructions

[0227] Prioritize using BESS to respond to the payload within a safe and schedulable range:

[0228]

[0229] The function is a median-limiting operation, if If it falls within the safe zone, then BESS responded completely. If If the discharge limit is exceeded, then BESS discharges at maximum safe power. If it falls below the charging limit, then BESS charges at maximum safe power.

[0230] S7.3.3: Determine fossil fuel dispatch instructions

[0231] Fossil fuel power modules serve as the final, reliable balancing unit to compensate for the power shortfall that BESS cannot meet under safety constraints.

[0232]

[0233] Only when Exceeded At that time, BESS was limited due to thermal safety. Only when the value is positive will the diesel engine start or increase power.

[0234] This embodiment also provides an energy dispatching system based on environmental sensing in polar environments, the system comprising:

[0235] Acquisition module: Acquires external environmental parameters and dynamically sets the dynamic safety hot zone of BESS based on these parameters;

[0236] Analysis module: Acquires internal state parameters of BESS, including the electrochemical state, thermal distribution map, and thermal rate map of the modular battery pack;

[0237] Optimization module: Constructs a time-series thermodynamic map of BESS, which aggregates internal state parameters to graph nodes representing modular battery packs, and includes the physical adjacency relationships and external exposure features of the nodes;

[0238] Decision module: Determines the optimal intervention strategy, which is a thermal management action output through reinforcement learning based on the time-series thermodynamic spectrum;

[0239] Scheduling module: Calculates the safe schedulable range of BESS and performs coordinated energy scheduling.

[0240] This embodiment also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described energy scheduling method for environmental perception in a polar environment.

[0241] This embodiment also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described energy scheduling method for environmental perception in a polar environment.

[0242] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0243] In this specification, the same or similar parts between the various embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the descriptions of the embodiments described later are relatively simple, and relevant parts can be referred to the descriptions of the foregoing embodiments.

[0244] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for energy scheduling for environmental perception in polar environment, the method comprising: The method comprises the following steps: ​ acquiring external environment parameters, and dynamically setting a dynamic safe thermal range of the BESS based on the external environment parameters; acquiring internal state parameters of the BESS, the internal state parameters comprising an electrochemical state, a thermal distribution map and a thermal velocity map of the modular battery pack; constructing a time-series thermodynamic map of the BESS, the map aggregating the internal state parameters to graph nodes representing the modular battery pack and comprising physical adjacency relationships of the nodes and external exposure characteristics; determining an optimal intervention strategy, the optimal intervention strategy being a thermal management action output by reinforcement learning based on the time-series thermodynamic map; measuring a safe dispatchable range of the BESS, and performing energy collaborative dispatching; The step of acquiring the external environment parameter comprises: collecting the instantaneous wind speed of the BESS near field and the ambient temperature ; based on and , calculating a wind cooling index for characterizing the wind cooling effect wherein is a calibration coefficient; the step of acquiring solar shading prediction parameters in the external environment parameters comprises: Acquiring image sequences of the sky by all-weather fisheye camera wherein are image pixel coordinates; processing the image sequences by a Kalman filter to obtain an optimal angular velocity estimate of the occluder ); Computing future path of the sun And occluder fronto-pixel set Angular coordinate extrapolation , and based on angular coordinate extrapolation and collision detection with the future path of the sun, determine an estimated obscuration time and an estimated obscuration duration ; wherein, is the moment the middle of the angular coordinates of the pixel, is the future time step.

2. The method for energy scheduling with environmental perception in polar environment according to claim 1, characterized in that, the step of dynamically setting the dynamic safe thermal range of the BESS comprises: Based on , and calculating the thermal threat factor : wherein, is a standard lower operating limit, is a wind chill threat coefficient, is a shelter threat coefficient, is a shelter duration scale factor, is a shelter approach decay factor; based on computing a dynamic safety lower bound : wherein, is the standard upper operating limit, is the minimum safety zone width.

3. The method for energy scheduling with environmental perception in polar environment according to claim 1, characterized in that, the step of acquiring the internal state parameters of the BESS comprises: at a sampling interval acquiring a sequence of raw thermal images wherein are image pixel coordinates of the infrared thermal imager; according to generating a thermal power distribution map at a time instant ; retrieving a thermal power distribution map at a time instant , and calculating a thermal velocity map : 。 4. The method for energy scheduling with environmental awareness in polar regions as claimed in claim 3, wherein, The steps of constructing the time-thermodynamic map of the BESS include: defining a corresponding pixel region set for each graph node representing the modular battery pack defining a corresponding pixel region set ; Utilizing a thermal profile and a thermal velocity profile spatially aggregated to compute dynamic characteristics of the node , the dynamic characteristics including: average surface temperature , minimum thermal velocity .

5. The method for energy scheduling with environmental perception in polar environment according to claim 4, characterized in that, The thermal management actions output by the reinforcement learning are based on maximizing a cumulative scalar reward policy determined, defined as: wherein, respectively a low temperature penalty coefficient, a non-uniformity penalty coefficient, an intervention cost penalty coefficient, a SOH loss penalty coefficient and a high temperature penalty coefficient; is the minimum surface temperature at the time instant; is a standard deviation; is the average surface temperature at the time instant; is an indicator function; is a thermal management action performed by the node ; is an idle action; is a self-heating action; is the state of health at the time instant.

6. The method for energy scheduling with environmental perception in polar environment according to claim 5, characterized in that, are generated by a graph neural network, computation of the graph neural network comprising: (a) performing node feature embedding to concatenate the dynamic feature vector of a node with the corresponding component of the external exposure vector :​​ wherein is a feature representation of the GNN layer-0 node . (b) performing layer graph convolution message passing, in the first layer updates the node hidden state : wherein, is the internal adjacency matrix defined in the set of neighbor nodes, and is the learnable weight matrix of the layer, is a non-linear activation function.

7. The method for energy scheduling with environmental perception in polar environment according to claim 6, characterized in that, the step of measuring the safe dispatchable range of the BESS and performing the energy collaborative dispatching comprises: According to the policy determining an optimal intervention policy vector computing execution required thermal regulation power consumption ; Calculating a net discharge power upper limit for a BESS : wherein, is a set of nodes that can participate in normal dispatching, is a set of nodes that perform forced derating, is a derating factor, is the theoretical maximum safe discharge power; determine the BESS dispatching instruction and the fossil energy dispatching instruction : wherein, is the net load for the science station.

8. An environmental perception-based energy scheduling system in polar environment, which executes an environmental perception-based energy scheduling method according to claim 1, characterized in that, The system comprises: a collection module configured to acquire external environment parameters, and dynamically set a dynamic safe thermal range of the BESS based on the external environment parameters; an analysis module configured to acquire internal state parameters of the BESS, the internal state parameters comprising an electrochemical state, a thermal distribution map and a thermal velocity map of the modular battery pack; an optimization module configured to construct a time-series thermodynamic map of the BESS, the map aggregating the internal state parameters to graph nodes representing the modular battery pack and comprising physical adjacency relationships of the nodes and external exposure characteristics; a decision module configured to determine an optimal intervention strategy, the optimal intervention strategy being a thermal management action output by reinforcement learning based on the time-series thermodynamic map; a dispatching module configured to measure a safe dispatchable range of the BESS, and perform energy collaborative dispatching.

Images