A liquid cooling temperature field reconstruction method and system for a sodium-ion battery

Abstract

This invention provides a method and system for reconstructing the temperature field of a sodium-ion battery liquid cooling system. It relates to the field of battery thermal management and temperature field reconstruction technology. The reconstruction method includes: acquiring visible light and infrared images of a sodium-ion battery pack and performing spatiotemporal synchronization; performing structural semantic recognition on the visible light images to solve for three-dimensional spatial attitude transformation relationships, and registering them with a preset liquid cooling plate template to generate structural guidance information; extracting the temperature of directly observed points and the surface temperature characteristics of each cell from the infrared images; calculating the initial temperature field of the entire liquid cooling plate based on the structural guidance information; calculating the original residual between the cell surface temperature characteristics and the reference temperature, removing common-mode errors to obtain the relative residual, and converting the relative residual into a local temperature compensation source; using the directly observed area as a fixed standard point, applying the local temperature compensation source to the obscured unknown area for iterative calculation, and outputting the corrected temperature field.

Description

Technical Field

[0001] This application relates to the field of battery thermal management and temperature field reconstruction technology, specifically to a method and system for reconstructing the liquid cooling temperature field of a sodium-ion battery. Background Technology

[0002] With the increasing application of sodium-ion batteries in energy storage devices, low-speed vehicles, construction machinery, and other electrochemical energy storage scenarios, the thermal safety and temperature consistency of sodium-ion battery systems under high-rate charge-discharge and continuous operation conditions have attracted widespread attention. Liquid cooling has the advantages of strong heat exchange capacity, adjustable cooling process, and good temperature uniformity, and is therefore widely used in the thermal management of battery packs. In existing liquid-cooled battery packs, multiple cells are typically arranged closely together, and a liquid cooling plate is placed at the bottom of the cell array. The coolant circulates through the channels inside the liquid cooling plate to remove heat.

[0003] However, mastering the complete and high-precision temperature field distribution inside the liquid cooling plate has always been a significant challenge for the industry. Existing technologies suffer from the following drawbacks: Traditional contact monitoring relies primarily on temperature sensors such as thermistors placed on the cell surface or liquid cooling pipes. This approach not only increases the complexity of the wiring harness and manufacturing costs within the battery pack, but also limits the sensors to a very limited number of discrete points, failing to reflect the overall thermal state and carrying the risk of single-point failure due to aging and detachment. Furthermore, the actual internal structure of a liquid cooling plate typically employs a compact flow channel with serpentine or U-shaped bends. Two pixels spatially close in an image may represent vastly different flow distances in the real fluid channel. Direct two-dimensional smoothing inevitably leads to severe cross-channel thermal short-circuit physical fallacies, failing to accurately reflect the topological law of coolant heat absorption and temperature rise along the flow path, and further failing to accurately separate thermal distortions caused by micro-blockages in the flow channels or individual cell anomalies.

[0004] Therefore, a non-contact temperature field reconstruction method and system is urgently needed. How to accurately reconstruct the complete temperature field of the obscured liquid cooling plate under severe shading constraints, utilizing limited visibility gaps and surface cell temperature, combined with the one-dimensional and two-dimensional real-space topology of the underlying flow channels, is a core engineering challenge that urgently needs to be addressed in this field. Summary of the Invention

[0005] This invention provides a method for reconstructing the liquid cooling temperature field of a sodium-ion battery, the method comprising: Acquire visible light and infrared images of sodium-ion battery packs and perform spatiotemporal synchronization; Structural semantic recognition is performed on visible light images to solve the three-dimensional spatial pose transformation relationship, and the images are registered with a preset liquid cooling plate template to generate structural guidance information. Extract the temperature characteristics of the directly observed points and the surface temperature characteristics of each battery cell from the infrared images; The initial temperature field of the entire liquid cooling plate was calculated based on the structural guidance information. The original residual between the surface temperature characteristics of the battery cell and the reference temperature is calculated. The common-mode error is removed to obtain the relative residual, and the relative residual is transformed into a local temperature compensation source. Using the directly observed area as a fixed standard point, a local temperature compensation source is applied to the shaded unknown area for iterative calculation, and the corrected temperature field is output.

[0006] Structural semantic recognition of visible light images to solve for 3D spatial pose transformation relationships includes: Identify the positioning markers, extract their pixel center coordinates, and denote them as a two-dimensional observation point set. ; Retrieve the preset coordinates of the positioning markers in the three-dimensional world coordinate system, and record them as the three-dimensional reference point set. ,in These are preset absolute depth coordinates; Construct an optimization objective function based on minimizing the reprojection residual, and solve for the rotation matrix from the 3D world coordinate system to the camera coordinate system. Translation vector : in, This indicates that the three-dimensional coordinates are passed through extrinsic parameters. Transformed and via camera intrinsic parameters Normalize the projected 2D pixel coordinates and force the Z-axis component of the translation vector to be positive in order to lock a unique spatial pose.

[0007] Register with the preset liquid cooling plate template to generate structural guidance information, including: Extract the flow channel region from the preset liquid cooling plate template, as well as the standard projected rectangular region of the bottom surface of each cell on the bottom plane, and calculate the intersection area of ​​the standard projected rectangular region and the flow channel region; Set the depth of the flow channel region on the bottom plane and its boundary vertex coordinates to zero, and substitute them into the rotation matrix. Translation vector and camera intrinsic parameter matrix The target mapping coordinates are obtained by solving the established three-dimensional perspective projection equation after homogeneous scaling factor normalization. A flow channel observation mask is generated based on the target mapping coordinates, and the upstream and downstream connectivity order with direction is output simultaneously as structural guidance information.

[0008] The initial temperature field of the entire liquid cooling plate is calculated based on the structural guidance information, including one-dimensional interpolation calculation of the internal fluid temperature based on the flow channel length: Given the set of known fluid temperatures, for any unknown temperature location within the flow channel region... In the set of known fluid temperatures, find the nearest upstream and downstream known points in the direction of water flow to the given location; use a one-dimensional linear interpolation formula to calculate the initial temperature of the coolant inside the flow channel at that location. : in, The path length to the upstream known point. The temperature of the upstream known point, The path length to the known downstream point. The temperature is the temperature at a known point downstream.

[0009] The initial temperature field of the entire liquid-cooled plate is calculated based on the structural guidance information, including the solid temperature calculation of the solid plate: Based on the calculated initial temperature of the coolant inside the flow channel, a set solid-liquid interface temperature difference constant is added to obtain the temperature of the metal tube wall outside the flow channel. A calculation grid is divided on the bottom surface, and the temperatures of the tube wall grid close to the outer edge of the flow channel and the solid plate grid directly seen through the gap are used as fixed known benchmarks. For the obscured metal grid that is neither in the gap nor in the flow channel, the average value of the metal temperature of its four adjacent grid positions (up, down, left, and right) is used as its initial solid temperature. The internal temperature of the fluid and the solid plate temperature are spliced ​​together to generate an initial temperature field covering the entire liquid cooling plate.

[0010] The calculation of the original residual between the surface temperature characteristics of the battery cell and the reference temperature includes: extracting the standard projected rectangular area corresponding to each battery cell on the bottom surface, averaging the area of ​​the initial temperature field in the area to obtain the reference temperature; and subtracting the extracted infrared observation temperature of the battery cell surface from the reference temperature to obtain the original residual of the corresponding battery cell.

[0011] The common-mode error is removed to obtain the relative residual, which is then transformed into a local temperature compensation source. This process includes: dividing the cells into groups based on their natural heat dissipation conditions, introducing the observation reliability of each cell as a weight, and performing a weighted average calculation on the original residual to obtain the common-mode temperature difference for that group; subtracting the common-mode temperature difference for the same group from the original residual to obtain the relative residual. Introducing thermal coupling coefficient and observation credibility The relative residual is transformed into a local temperature compensation source. : Among them, thermal coupling coefficient The optimization objective is determined by first-order boundary smoothness, which minimizes the temperature gradient jump at the boundary between the hidden area covered by the cell and the adjacent visible gap.

[0012] After converting the relative residual into a local temperature compensation source, the process also includes differentiating the local temperature: for a base plate grid point covered by the battery cell, if the grid point is located in a solid plate area, the injected residual source for that grid point is equal to the local temperature compensation source. If the grid point is located in the flow channel region, then the injected residual source at that grid point is equal to the local temperature compensation source. The product of the liquid heat transfer ratio and the liquid heat transfer ratio, where the liquid heat transfer ratio is greater than or equal to zero and much less than 1.

[0013] Using the directly observed area as a fixed standard point, a local temperature compensation source is applied to the occluded unknown area for iterative calculation, outputting a corrected temperature field. This includes: locking the grid belonging to the visible gap and the liquid cooling plate inlet / outlet area to its actual infrared measured temperature as a fixed standard point; and applying this to all occluded hidden grid points. The following formula is used to iterate multiple times until numerical convergence is achieved: in, Indicates the first Grid points in the next iteration temperature, For point The four directly adjacent grids in the top, bottom, left, and right directions It is a balance parameter that takes values ​​between 0 and 1. For the initial temperature field, To inject residual sources; in iterative calculation of adjacent points When encountering a fixed standard point, its fixed true temperature is directly substituted; when the temperature change rate of all hidden grid points is lower than the convergence threshold, the corrected temperature field is output.

[0014] The present invention also provides a sodium-ion battery liquid-cooled temperature field reconstruction system, the system comprising: Acquisition module: Acquires visible light and infrared images of the sodium-ion battery pack and performs spatiotemporal synchronization; Structural guidance information extraction module: Performs structural semantic recognition on visible light images to solve the three-dimensional spatial posture transformation relationship, and registers it with a preset liquid cooling plate template to generate structural guidance information; Initial temperature field construction module: Extracts the temperature characteristics of the directly observed points and the surface temperature characteristics of each cell from the infrared image; calculates the initial temperature field of the entire liquid cooling plate based on the structural guidance information; Compensation module: Calculates the original residual between the cell surface temperature characteristics and the reference temperature, removes common-mode error to obtain the relative residual, and converts the relative residual into a local temperature compensation source; Correction module: Using the directly observed area as a fixed standard point, the local temperature compensation source is applied to the shaded unknown area for iterative calculation, and the corrected temperature field is output.

[0015] This invention provides a method and system for reconstructing the temperature field of sodium-ion battery liquid cooling. Addressing the challenge of the liquid cooling plate being obscured by the cell array and exhibiting significant visual perspective parallax, this application deeply integrates the guiding role of three-dimensional structural information in the temperature field. By inversely mapping a preset real three-dimensional structural template to a two-dimensional observation field, this method effectively eliminates spatial misalignment caused by visual perspective, dividing the visible gap region segmented by the cells into internal flow channels and a solid base plate. Based on this, this application calculates the temperature of the fluid and solid regions separately according to the flow law of the coolant in the one-dimensional pipeline and the thermal conductivity of the metal in the two-dimensional base plate. Fragmented direct observations are transformed into real boundary constraints. Thus, the system can deduce an initial temperature field benchmark that conforms to the real laws of fluid heat absorption and solid heat conduction. After establishing the structure-guided initial temperature field benchmark, this application further iteratively corrects it based on local spatial residual modes. Since the initial temperature field benchmark objectively reflects the uniform trend of passive heat absorption by the coolant, it is difficult to characterize the non-uniform heating phenomenon caused by the cells. This method treats the difference between the measured infrared temperature at the top of the battery cell and the calculated temperature as an observable feature of the intensity of local heat input. By calculating the original residual between the measured temperature of the battery cell and the baseline predicted temperature, and performing a regionalized common-mode error removal operation, this scheme can obtain a relative residual reflecting local heating or heat dissipation anomalies. This relative residual, as a feature reflecting the thermal distortion of the hidden space, is applied as a local compensation to the unknown area blocked by the battery cell.

[0016] In the process of compensating for residuals to the hidden region, this method introduces spatial weighting based on observation confidence, effectively suppressing the interference of temperature measurement noise at the field of view edge on the calculation. More importantly, the correction process is always absolutely constrained by structural information, that is, the local temperature of the bottom layer directly observed through the gap is used as an unchangeable absolute constraint node. Under this boundary constraint, the correction process, driven by the initial uniform heating gradient, the visible observation true value as the boundary, and the cell anomaly residual as the driving source, allows the originally discrete temperature compensation amount to transition naturally and smoothly within the metal substrate. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the reconstruction of the liquid cooling temperature field of the sodium-ion battery according to the present invention. Figure 2 Infrared images of sodium-ion batteries acquired in this invention; Figure 3 This is an experimental curve showing the temperature change along the flow path of the present invention; Figure 4This is a comparison chart of the test point temperatures before and after the residual correction of this invention. Detailed Implementation

[0018] This embodiment provides a sodium-ion battery liquid-cooled temperature field reconstruction system, used for visual, non-contact, high-precision reconstruction of the temperature field of the bottom liquid-cooled plate, which is severely obscured by the battery cells, during normal charging, discharging, or cyclic operation of a sodium-ion battery. This embodiment uses a square sodium-ion battery pack installed in an energy storage battery cluster as the application object.

[0019] The temperature field reconstruction system eliminates the need for contact temperature sensors attached to the cell surface or inside the liquid cooling plate. The highly integrated system primarily comprises a sodium-ion battery pack, a multimodal image acquisition device, a synchronous control device, and a data processing device. The sodium-ion battery pack includes a housing, cell assemblies, a bottom liquid cooling plate, a thermally conductive interface layer, and an upper insulating protective assembly. The housing includes a bottom support, circumferential sidewalls, and a top cover. The cell assemblies, housed within the housing, contain forty-eight square sodium-ion cells arranged vertically in a six-row, eight-column configuration. The bottom surface of each cell faces the bottom liquid cooling plate, and the terminals face the top cover. Electrically insulating gaps are provided between adjacent cells. These gaps provide a channel for visual observation from the top, allowing infrared thermal radiation to penetrate from the bottom. A circumferential limiting frame is provided around the cell assemblies to restrict relative displacement of the cells. The liquid cooling plate is a plate-type liquid cooling component located below all the cells, with internally machined closed channels for coolant flow. A thermally conductive interface layer (such as a thermally conductive silicone pad) is placed between the upper surface of the liquid cooling plate and the bottom surface of each battery cell. The liquid cooling plate has one coolant inlet, one coolant outlet, and multiple parallel cooling branches located between them. Each cooling branch includes multiple sequentially connected straight flow channel sections and U-shaped bend flow channel sections. The design parameters of the liquid cooling plate are pre-converted into a preset liquid cooling plate structural template and stored in the data processing device. This template is in an absolute three-dimensional world coordinate system (…). The bottom plane of ) Within this, the flow channel region is defined parametrically. (Including the directional flow channel centerline and the length of the path along the flow path) Solid base plate area Standard projected rectangular area of ​​battery cell and the channel width used for overlap operations. .

[0020] The four corners of the limiting frame are equipped with high-contrast positioning marks. These positioning marks are located at absolute depth coordinates independent of the cell height. The limiting frame also has directional markings to distinguish between the coolant inlet and outlet sides, preventing errors during image reconstruction. Topological inversion. Simultaneously, the coolant inlet and outlet connectors are partially exposed to the visible range of the image acquisition device, forming unobstructed inlet and outlet temperature reference zones.

[0021] The multimodal image acquisition device includes a top image acquisition component and a side image acquisition component. The top image acquisition component is located inside the top cover of the enclosure, providing a top-down view of the battery cell assembly. Due to the compact internal space, it operates in a short focal length, wide field-of-view observation environment. The side image acquisition components are located within the equipment installation area inside the long side wall of the enclosure. Each image acquisition component contains a dual-spectrum image acquisition unit, namely a strictly calibrated and fixed visible light camera and a long-wave infrared camera. The visible light camera uses a high-resolution area array image sensor to clearly acquire the geometric features of the top contour, positioning marks, orientation marks, and visible gap boundaries of the battery cell; the long-wave infrared camera uses an uncooled focal plane detector to penetrate dark areas that are indistinguishable by visible light and acquire the infrared observation temperature of the top surface of the battery cell. The temperature of the inlet and outlet reference areas of the liquid cooling plate and the infrared temperature of the bottom layer obtained through the insulation gap.

[0022] The synchronization control device includes a unified clock module and a hardware trigger module. The hardware trigger module sends synchronization pulse signals to each visible light camera and long-wave infrared camera according to a preset sampling frame rate. Upon receiving image data, the data processing device assigns an absolutely unified trigger timestamp to the multi-view image sequence based on the unified clock module. This fundamentally eliminates time-domain misalignment caused by bus transmission queuing in multimodal data.

[0023] The data processing device uses an industrial-grade computer, including a processor, memory, and image processing unit. The memory pre-stores the intrinsic parameter matrices of each camera. The data processing device executes a preset reconstruction program to sequentially complete multimodal data spatiotemporal registration, three-dimensional spatial attitude analysis and distortion-free hidden channel mapping, forward deduction of the initial temperature field based on one-dimensional / two-dimensional topology, and inverse smoothing correction of spatial mode separation based on local cell temperature residuals and observation reliability. Finally, it outputs the high-precision physical temperature field of the deeply occluded liquid cooling plate using a purely visual, non-contact method.

[0024] Based on the provided sodium-ion battery liquid-cooled temperature field reconstruction system, this embodiment provides a method for reconstructing the liquid-cooled temperature field of a sodium-ion battery. This method includes four steps: data acquisition, forward temperature calculation, reverse residual correction, and verification output.

[0025] Specifically, the first step is to retrieve synchronized data from multiple sources: The system acquires visible light and infrared images of the sodium-ion battery pack, simultaneously obtains the corresponding battery and liquid cooling system operating parameters, and performs multimodal spatiotemporal synchronization.

[0026] During the charging, discharging, or continuous operation of sodium-ion battery packs, the changes in the battery's thermal state are closely related to its real-time electrical parameters (such as current and voltage) and liquid cooling system parameters (such as flow rate and voltage drop). Because the image acquisition device transmits data via an image acquisition interface, while the battery management system and liquid cooling operating parameter acquisition device transmit data via the controller local area network and communication bus, there are transmission queuing delays and sampling period differences in the multi-source data. Furthermore, the visible light camera and long-wave infrared camera within the dual-spectrum image acquisition unit have inherent parallax and resolution differences. Step S110 eliminates the asynchronous differences in the time dimension and the coordinate deviations in the spatial dimension of the aforementioned multi-source data, specifically including the following steps: Obtain multi-source raw data sequences based on a unified clock reference.

[0027] Through the hardware trigger module of the synchronization control device, according to the preset sampling period, in this embodiment, the corresponding image acquisition frame rate is ten frames per second, sending synchronization pulse signals to the top image acquisition component and the side image acquisition component to acquire the original visible light image and the original infrared image. When the data processing device receives the image data, it assigns a trigger timestamp under a unified time base to each frame of the image based on the unified clock module.

[0028] Simultaneously, the battery management system and the liquid cooling operating parameter acquisition device continuously upload operating parameter messages according to their respective underlying logic. These messages preferentially include total current, individual cell voltages, measurement data from the twelve cell temperature sensors, as well as total coolant flow rate, inlet pressure, outlet pressure, coolant inlet temperature, and outlet temperature. When a message arrives at the communication interface or controller area network interface of the data processing device, the data processing device immediately assigns a corresponding arrival timestamp to each operating parameter via a unified clock module. Thus, image data and operating parameter data are unified into the same clock domain, forming multi-source raw data with an absolute time sequence.

[0029] Perform synchronization of operating parameters based on the image sampling time reference.

[0030] To ensure that the thermal boundary conditions and infrared temperature observation results are synchronized during reconstruction, the data processing device uses the trigger timestamp of the original image data. As a reference time. For any continuously changing operating parameter. The data processing device searches for the reference time within the sampling time series in this clock domain. Two adjacent valid arrival times are denoted as preceding times. and subsequent time , making .

[0031] The data processing device uses a linear interpolation algorithm to calculate the reference time. Corresponding synchronous operation parameters The calculation formula is: This invention resamples asynchronous parameters to the image reference time with minimal assumptions and no phase delay, avoiding the computational delay introduced by complex digital filtering algorithms during the transient heat generation stage of sodium-ion batteries, and ensuring the synchronous alignment of transient characteristics of multi-source physical quantities. This calculation is repeated for all operating parameters to generate a set of synchronous operating parameters without delay deviation.

[0032] Perform bispectral image spatial registration.

[0033] Raw visible light images have high spatial resolution and can clearly reflect structural geometry; raw infrared images contain continuous surface temperature distribution, but are limited by the resolution of the focal plane detector and lack edges.

[0034] The data processing device calls the camera intrinsic parameter matrix and distortion coefficients pre-stored in the memory, and processes them respectively. The raw visible light image and raw infrared image at each time point are subjected to distortion correction processing. For each dual-spectrum image acquisition unit, the calibrated extrinsic parameter matrix, including the rotation matrix, is called to transform the coordinate system from the long-wave infrared camera coordinate system to the visible light camera coordinate system. Translation vector .

[0035] Because the side image acquisition component simultaneously covers the side surface of the battery cell and the exposed edge of the liquid cooling plate below the battery cell—that is, the inlet temperature reference area and the outlet temperature reference area—within the same field of view, there is a significant depth abrupt change between the two. Based on the liquid cooling plate structure template, the data processing device extracts the battery cell observation area mask and the liquid cooling plate exposed area mask from the infrared image, and assigns preset depth constants for the battery cell side surface and the liquid cooling plate exposed area to them respectively, constructing a partitioned depth map. .

[0036] Using the visible light camera coordinate system as a reference, the coordinates of the target pixels in the original infrared image after distortion correction are... Mapped coordinates converted to the visible light image coordinate system The conversion relationship is as follows: in, This is the intrinsic parameter matrix of a visible light camera. This is the intrinsic parameter matrix of a long-wave infrared camera. This represents the depth constant corresponding to the target pixel within its corresponding partition mask. After calculating the right-hand 3D column vector using the matrix multiplication method described above, the third component of this column vector is taken as the scale factor. And divide the first two components by respectively To achieve homogeneous coordinate normalization, we obtain the final planar mapped coordinates. .

[0037] Using the mapped coordinate points as anchor points, a bilinear interpolation algorithm is used to resample the infrared temperature values ​​at the pixel level, so that the size of the generated registered infrared temperature matrix is ​​perfectly aligned with the visible light image array.

[0038] The final output corresponds to the absolute time. A multimodal synchronization dataset. This dataset includes: a set of synchronization operation parameters, as well as scale- and viewpoint-consistent registered visible light images and registered infrared temperature images.

[0039] Based on the provided system and the output multimodal synchronization dataset, this embodiment provides a detailed explanation of the steps in the temperature field reconstruction method, namely, parsing spatial attitude and mapping hidden flow channels.

[0040] In energy storage battery packs, to save internal space and increase energy density, the top image acquisition component, installed inside the casing cover, is typically very close to the top of the battery cells, creating a short focal length, wide field-of-view observation environment. This structure results in significant perspective retraction and interlayer parallax effects.

[0041] Assuming the height of the battery cell is The distance between the camera's optical core and the top of the battery cell is For a battery cell located at the edge of the camera's field of view, let its edge field of view angle be... The center of its top surface will be misaligned with the center of its bottom surface in the image. The pixel coordinates of the top surface will be significantly different from those of the bottom surface in space. The projection deviation. If only the visible light image of the top of the battery cell is used to apply a two-dimensional planar mapping to locate the bottom liquid cooling plate, this... Errors will cause the model to misjudge the flow channel as being on the right when it is actually on the left, thus rendering the reconstructed temperature field invalid. Therefore, the observed two-dimensional image is restored to the true three-dimensional space to eliminate the depth parallax caused by thickness.

[0042] Specifically, the data processing device performs the following steps: Structural semantic recognition is performed on visible light images to establish cell array features and determine the inlet and outlet directions of coolant, and the three-dimensional spatial attitude transformation relationship of the current battery pack is solved.

[0043] A three-dimensional coordinate system is established with the upper surface of the liquid cooling plate at the bottom of the battery cell assembly as the zero potential energy surface. The unit of length in this coordinate system is millimeters. According to the battery pack specifications, the absolute depth coordinates of the positioning mark pre-installed on the limiting frame are as follows: , Approximate to, but independent of, the standard height of the battery cell.

[0044] Call the current time The registered visible light image (whose coordinate system is denoted as ) Using an image edge detection algorithm, the location markers are identified, and their pixel center coordinates are extracted and denoted as a two-dimensional observation point set. At the same time, retrieve this The preset real coordinates of each positioning marker in the three-dimensional world coordinate system are denoted as the three-dimensional reference point set. .

[0045] Call the calibrated visible light camera intrinsic parameter matrix For the aforementioned 3D-2D corresponding point pairs, a PnP pose optimization objective function based on minimizing reprojection residuals is constructed, and the rotation matrix from the 3D world coordinate system to the camera coordinate system is solved. Translation vector : Among them, the projection operator This indicates that the three-dimensional coordinates are passed through extrinsic parameters. Transformed and via internal parameters Normalized projected 2D pixel coordinates. Since all reference points are coplanar. The planar PnP solution geometrically has a true solution and a flipped mirror solution that exits the plane. To resolve this ambiguity, the Z-axis component of the translation vector, i.e., the target depth, must be positive. This allows you to pinpoint the unique and correct spatial pose.

[0046] In addition, pre-defined orientation markers in the registered visible light image are identified and verified to correspond to the coolant inlet sidewall defined in the coordinate system, thus eliminating in-plane interference caused by battery pack repositioning. Potential risks of topological inversion.

[0047] Based on the preset liquid cooling plate template, registration is performed based on the spatial attitude to generate structural guidance information including the flow channel region, flow direction, upstream and downstream topology, and cell mapping relationship.

[0048] The liquid cooling plate is shielded by closely packed battery cells. The data processing unit retrieves a preset liquid cooling plate structure template matching the current model from memory; this template is defined in the coordinate system. Within the bottom plane. For the cooling channels inside the liquid cooling plate, the template uses a directional parametric curve with flow direction. Describe the centerline of each flow channel section, where the parameters are... ( This represents the path length of the coolant along the central axis after flowing in from the inlet. Based on this, the upstream and downstream connectivity topology is established: on the same branch... The smaller point is upstream. The larger point is downstream.

[0049] In a preferred embodiment, for a flow channel width of The Each flow channel section, in Generating a distortion-free profile of the actual flow channel in a plane : (1) Straight line segment: perpendicularly offset to both sides of the central axis The distance.

[0050] (2) Bending section ( U-shaped bend: Set the inner and outer corner radii according to the bend center, and introduce a non-negative constraint on the center bending radius. Ensure that the difference between the inner and outer radii is equal to This generates a smooth, non-self-intersecting semi-circular sector region.

[0051] Extract the first Line 1 The bottom surface of the battery cell is Standard projected rectangular area on a plane The data processing device calculates the area of ​​intersection between the two. : Area to be found ( () serves as the geometric weight boundary for heat distribution.

[0052] The flow channel area on the bottom plane and the coordinates of its boundary vertices Set depth Substituting into the established three-dimensional perspective projection equation: To characterize the homogeneous scaling factor of depth in the camera coordinate system ( After left-side normalization, the target mapping coordinates are obtained. .

[0053] Therefore, the bottom contour is reverse-mapped into the two-dimensional observation image coordinate system to generate a flow channel observation mask. It also outputs the upstream and downstream connectivity order in a specific direction.

[0054] Next, this embodiment will provide a detailed explanation of temperature feature extraction and initial temperature field calculation in the temperature field reconstruction method.

[0055] This embodiment utilizes the obtained structural guidance information to determine whether the temperature seen through the gap belongs to the internal flow channel or the solid base plate. Then, for the flow channel region, the water temperature is calculated according to the order in which the coolant flows along the pipe; for the solid base plate region, the pipe wall metal temperature is first calculated, and then the initial temperature distribution of the entire liquid cooling plate is obtained based on the temperature of the observable area.

[0056] Specifically, the infrared images are mapped to the battery pack coordinate system, the temperature of the direct observation point of the liquid cooling plate and the surface temperature characteristics of each cell are extracted, and the observation credibility is assigned to each region.

[0057] Call the collected time The registered infrared image. Obtain the first... Line 1 The average pixel temperature of the infrared observation area at the top of the battery cell is taken as the infrared observation temperature of the cell surface. ( Because infrared camera lenses are prone to temperature measurement deviations at the edges, the data processing device calculates the distance from the center pixel of the battery cell to the center pixel of the camera image. ( The observation reliability of the battery cell was calculated using the Gaussian attenuation formula. : The field-of-view attenuation constant is expressed in pixels. The extracted cell observation temperature... Its reliability is used for fine correction of local temperature, and it is not involved in the calculation of the initial temperature of the liquid cooling plate in this step.

[0058] Next, based on the complementary regions of the bottom mask of the battery cell, the gap regions not covered by the battery cell are identified in the registered infrared image. The infrared pixel temperatures within these gap regions, along with the reference temperatures of the inlet and outlet of the liquid cooling plate, are extracted as the directly observed discrete temperatures of the liquid cooling plate.

[0059] The temperature field calculation units are divided according to the structural guidance information, and known temperature reference points are set according to the location of the gap regions.

[0060] Using the established inverse transformation formula for spatial perspective projection, the gap regions on the image are... Mapped to the base plane of the world coordinate system ( On the surface, the gap region in space is obtained. and its temperature distribution.

[0061] Retrieve structural guidance information, which clearly delineates the flow channel area. and solid board area Perform region overlap detection on the bottom plane: Extract known temperature points in the flow channel: Calculate the gap region. With the Slotted channel area The overlapping portion. If overlap exists, extract the average temperature of the overlapping portion and determine the actual flow path length of this location from the coolant inlet based on the flow channel centerline. The temperature of this gap is thus identified as a known temperature reference point for this flow channel. Together with the known temperatures at the inlet and outlet, they constitute a one-dimensional set of known temperatures for the fluid. .

[0062] Calculate the gap region Solid board area The overlapping portion. For the gap temperature falling on the solid aluminum plate, it is preserved in the corresponding coordinates. At this point, serving as a known temperature reference point for the solid plate, a set of known temperatures for two-dimensional solids is formed. .

[0063] The initial temperature field of the entire liquid cooling plate was calculated: After identifying the known temperature points, the data processing device calculates the initial physical temperature field of the entire area of ​​the liquid cooling plate that is blocked in two steps: Step 1: One-dimensional interpolation calculation of fluid internal temperature based on channel length: For the channel region Any location within that has an unknown temperature The data processing device is based on the known temperature set of the fluid. In the context of water flow, find the nearest known upstream point (path length) that is closest to the current location in the direction of water flow. ,temperature ) and downstream known points (path length) ,temperature The initial temperature of the coolant inside the flow channel at this location is calculated using a one-dimensional linear interpolation formula. : Step 2: Solid Plate Temperature Calculation: The flow channel contains a flowing liquid, while the exterior is a solid metal plate. During the heat transfer process, a layer of resistance exists between the metal tube wall and the internal liquid due to convection heat transfer, causing the metal tube wall temperature to be slightly higher than the liquid temperature, rather than equal to the liquid temperature. Therefore, when calculating the boundary temperature of the solid plate, the liquid temperature is not used directly, but rather the liquid temperature calculated in Step 1 is used instead. Based on this, a predetermined solid-liquid interface temperature difference constant is added. ( This allows us to obtain the temperature of the metal tube wall outside the flow channel. : Subsequently, the bottom surface is divided into sections with spacing of... The computational grids are used. Within these grids, the temperatures of two types of grids are used as fixed, known references: one type is the pipe wall grid immediately adjacent to the outer edge of the flow channel, whose temperature is fixed using the calculated pipe wall temperature. Another type is a solid plate mesh that is directly visible through the gaps, and its temperature is fixed using a set. Infrared temperature observation.

[0064] For other obstructed metal meshes that are neither within the gap nor within the flow channel Calculate its initial solid temperature That is, the solid temperature at a certain location is equal to the average of the metal temperatures at the four adjacent grid locations above, below, left, and right.

[0065] Finally, the internal fluid temperature calculated in the first step and the external solid plate temperature calculated in the second step are combined to generate the initial temperature field covering the entire liquid cooling plate. .

[0066] Based on the aforementioned observation data and initial simulation results, this embodiment provides a detailed explanation of the steps in the temperature field reconstruction method based on the correction of the temperature field in the cell temperature space.

[0067] The generated initial temperature field This study only considers the passive heating of the coolant along the flow path and the passive heat conduction of the base plate metal, meaning that the heat injected into the liquid cooling plate by each cell is spatially uniform. However, within the battery pack, heat is primarily conducted vertically within the cells. The temperature of a hidden area of ​​the liquid cooling plate (directly below the cell) is determined by the heat flow injected by the cell directly above and the heat flow carried away by the coolant in that area. Therefore, the infrared observation temperature at the top of the cell... Essentially, this is an observable characteristic of the intensity of local heat input at that location. This invention calculates the difference between the measured temperature of the battery cell and the expected temperature of the frame, removes the local temperature distortion term caused by uneven heating, i.e., the residual, and performs a fine correction on the temperature of the shielded liquid cooling plate.

[0068] Calculate the temperature residual between the cell and the baseline and remove common-mode error.

[0069] The data processing unit calculates the average base plate temperature for each cell region under ideal reference conditions. (For the first...) Line 1 Extract the standard projected rectangular area on the bottom surface of each cell in the array. (Area is denoted as) , The reference temperature is obtained by averaging the initial base plate temperature within the region using area integration. : The surface infrared temperature of the extracted battery cell was observed. The original residual is obtained by subtracting the reference temperature. : The original residual includes a constant temperature difference term caused by the absolute thermal resistance inside the cell. To eliminate the interference of this unobtainable, unknown contact thermal resistance, the cells are classified according to their natural heat dissipation conditions, preferably into central cell groups and peripheral cell groups. For each group... Introducing observation credibility Perform weighted calculations to determine the common-mode temperature difference in the region. : Subtract the common-mode temperature difference in the same region to obtain the relative residual. : Subtract regional common mode back, This indicates the amount of heat generated by the cell relative to the surrounding average level, reflected in temperature. It eliminates the need for complex absolute thermal resistance calibration, and regionalized common-mode elimination prevents edge cells with good natural heat dissipation from being misjudged as cold sources.

[0070] The aforementioned relative residual is transformed into a local temperature compensation source applied to the liquid-cooled plate. Considering potential noise interference in edge field-of-view infrared thermometry, an observation reliability based on optical attenuation is introduced. Perform weighted modulation: in, The thermal coupling coefficient represents the proportion of temperature distortion at the top of the cell that is transmitted to the base plate. It is not a fixed empirical value, but is determined with first-order boundary smoothness as the optimization objective, that is, adjusting... This makes the hidden area covered by the battery cell and the adjacent visible gap... At the boundary line, the temperature gradient jump is minimal.

[0071] Within the same cell footprint, the heat retention effect differs between a solid base plate and a flowing channel. The solid base plate accumulates heat, while the flowing channel continuously carries heat away with the coolant. Therefore, based on structural design, residual sources... Differentiated treatment: For battery cells Covered base plate grid points :like (Solid plate area), then the injection residual source of this mesh. ;like (Flow channel region), then the injection residual source of this grid. ; in For liquid heat transfer ratio ( This avoids repeatedly adding heat that should be carried away by the fluid to the local temperature.

[0072] The absolute constraint is set to only correct occluded unknown regions; directly observed regions are considered ground truth and must be used as fixed standard points. In the bottom mesh, these will belong to the visible gaps. The grid in the inlet and outlet areas of the liquid cooling plate is locked to the actual temperature measured by infrared radiation.

[0073] For all occluded hidden grid points The following formula is used to iterate multiple times until numerical convergence is achieved: in For point For the four directly adjacent grids (top, bottom, left, and right), for grid points located at the outermost edge of the liquid cooling plate, out-of-domain neighbors are removed according to the zero heat flux boundary condition, and the number of remaining neighbors is renormalized.

[0074] Its value ranges from 0 to 1, and it serves as a balancing parameter. When Approaching At this point, the formula degenerates into a pure four-neighborhood source-free smoothing. However, adding the subsequent source term... This indicates that the initial uniform temperature skeleton plus the local distortion residual is used as the traction source to find a constrained steady-state equilibrium under the diffusion of the four neighboring regions.

[0075] Iterative calculation of adjacent points When encountering the aforementioned locked visible gap grid, directly substitute its fixed true temperature.

[0076] When the rate of temperature change of all hidden meshes is below the convergence threshold, output the corrected temperature field. .

[0077] The method of this embodiment outputs a temperature field that is absolutely equal to the measured true value at the visible point; in the hidden area where it is blocked, it retains the trend of passive heating of the fluid along the flow path, and also incorporates the thermal distortion caused by individual differences in the battery cells and interface heat transfer in a localized manner, and the entire field is smooth and continuous without jumps.

[0078] Finally, it should be noted that those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc. The above computer program embodiments can achieve the same or similar effects as any of the corresponding foregoing method embodiments.

[0079] Furthermore, typically, the devices and equipment disclosed in the embodiments of this invention can be various electronic terminal devices, such as mobile phones, personal digital assistants (PDAs), tablet computers (PADs), smart TVs, etc., or they can be large terminal devices, such as servers. Therefore, the scope of protection disclosed in the embodiments of this invention should not be limited to a specific type of device or equipment. The client disclosed in the embodiments of this invention can be applied to any of the above-mentioned electronic terminal devices in the form of electronic hardware, computer software, or a combination of both. Furthermore, the method disclosed in the embodiments of the present invention can also be implemented as a computer program executed by a CPU, which may be stored in a computer-readable storage medium. When the computer program is executed by the CPU, it performs the functions defined in the method disclosed in the embodiments of the present invention. Furthermore, the above-described method steps and system units can also be implemented using a controller and a computer-readable storage medium for storing a computer program that enables the controller to perform the functions of the above-described steps or units. Furthermore, it should be understood that the computer-readable storage medium (e.g., memory) described herein can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. By way of example, and not limitation, non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM), which can act as external cache memory. By way of example, and not limitation, RAM may be available in various forms, such as synchronous RAM (DRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct Rambus RAM (DRRAM). The storage devices disclosed herein are intended to include, but are not limited to, these and other suitable types of memory. The above are exemplary embodiments disclosed in this invention. However, it should be noted that various changes and modifications can be made without departing from the scope of the embodiments of this invention as defined by the claims. The functions, steps, and / or actions of the methods according to the disclosed embodiments described herein do not need to be performed in any marked order. Furthermore, although the elements disclosed in the embodiments of this invention may be described or claimed individually, they may be understood as multiple unless explicitly limited to a singular number. In this specification, similar or identical parts between various embodiments can be referred to mutually, and each embodiment focuses on describing the differences from other embodiments. In particular, for the embodiments described later, the description is relatively simple, and relevant parts can be referred to the descriptions of the foregoing embodiments. The above descriptions are merely specific embodiments of this application, but the protection scope 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 protection scope of this application. Therefore, the protection scope of this application should be determined by the protection scope of the claims.

Claims

1. A method for reconstructing the liquid-cooled temperature field of a sodium-ion battery, characterized in that, The method includes: Acquire visible light and infrared images of sodium-ion battery packs and perform spatiotemporal synchronization; Structural semantic recognition is performed on visible light images to solve the three-dimensional spatial pose transformation relationship, and the images are registered with a preset liquid cooling plate template to generate structural guidance information. Extract the temperature characteristics of the directly observed points and the surface temperature characteristics of each battery cell from the infrared images; The initial temperature field of the entire liquid cooling plate was calculated based on the structural guidance information. The original residual between the surface temperature characteristics of the battery cell and the reference temperature is calculated. The common-mode error is removed to obtain the relative residual, and the relative residual is transformed into a local temperature compensation source. Using the directly observed area as a fixed standard point, a local temperature compensation source is applied to the shaded unknown area for iterative calculation, and the corrected temperature field is output.

2. The method according to claim 1, characterized in that, Structural semantic recognition of visible light images to solve for 3D spatial pose transformation relationships includes: Identify the positioning markers, extract their pixel center coordinates, and denote them as a two-dimensional observation point set. ; Retrieve the preset coordinates of the positioning markers in the three-dimensional world coordinate system, and record them as the three-dimensional reference point set. ,in These are preset absolute depth coordinates; Construct an optimization objective function based on minimizing the reprojection residual, and solve for the rotation matrix from the 3D world coordinate system to the camera coordinate system. Translation vector : ； in, This indicates that the three-dimensional coordinates are passed through extrinsic parameters. Transformed and via camera intrinsic parameters Normalize the projected 2D pixel coordinates and force the Z-axis component of the translation vector to be positive in order to lock a unique spatial pose.

3. The method according to claim 2, characterized in that, Register with the preset liquid cooling plate template to generate structural guidance information, including: Extract the flow channel region from the preset liquid cooling plate template, as well as the standard projected rectangular region of the bottom surface of each cell on the bottom plane, and calculate the intersection area of ​​the standard projected rectangular region and the flow channel region; Set the depth of the flow channel region on the bottom plane and its boundary vertex coordinates to zero, and substitute them into the rotation matrix. Translation vector and camera intrinsic parameter matrix The target mapping coordinates are obtained by solving the established three-dimensional perspective projection equation after homogeneous scaling factor normalization. A flow channel observation mask is generated based on the target mapping coordinates, and the upstream and downstream connectivity order with direction is output simultaneously as structural guidance information.

4. The method according to claim 1, characterized in that, The initial temperature field of the entire liquid cooling plate is calculated based on the structural guidance information, including one-dimensional interpolation calculation of the internal fluid temperature based on the flow channel length: Given the set of known fluid temperatures, for any unknown temperature location within the flow channel region... In the set of known fluid temperatures, find the nearest upstream and downstream known points in the direction of water flow to that location; The initial temperature of the coolant inside the flow channel at this location is calculated using a one-dimensional linear interpolation formula. : ； in, The path length to the upstream known point. The temperature of the upstream known point, The path length to the known downstream point. The temperature is the temperature at a known point downstream.

5. The method according to claim 4, characterized in that, The initial temperature field of the entire liquid-cooled plate is calculated based on the structural guidance information, including the solid temperature calculation of the solid plate: Based on the calculated initial temperature of the coolant inside the flow channel, a set solid-liquid interface temperature difference constant is added to obtain the temperature of the metal tube wall outside the flow channel. A calculation grid is divided on the bottom surface, and the temperatures of the tube wall grid close to the outer edge of the flow channel and the solid plate grid directly seen through the gap are used as fixed known benchmarks. For the obscured metal grid that is neither in the gap nor in the flow channel, the average value of the metal temperature of its four adjacent grid positions (up, down, left, and right) is used as its initial solid temperature. The internal temperature of the fluid and the solid plate temperature are spliced ​​together to generate an initial temperature field covering the entire liquid cooling plate.

6. The method according to claim 5, characterized in that, The calculation of the original residual between the surface temperature characteristics of the battery cell and the reference temperature includes: extracting the standard projected rectangular area corresponding to each battery cell on the bottom surface, averaging the area of ​​the initial temperature field in the area to obtain the reference temperature; and subtracting the extracted infrared observation temperature of the battery cell surface from the reference temperature to obtain the original residual of the corresponding battery cell.

7. The method according to claim 6, characterized in that, The common-mode error is removed to obtain the relative residual, which is then transformed into a local temperature compensation source. This process includes: dividing the cells into groups based on their natural heat dissipation conditions, introducing the observation reliability of each cell as a weight, and performing a weighted average calculation on the original residual to obtain the common-mode temperature difference for that group; subtracting the common-mode temperature difference for the same group from the original residual to obtain the relative residual. Introducing thermal coupling coefficient and observation credibility The relative residual is transformed into a local temperature compensation source. : ； Among them, thermal coupling coefficient The optimization objective is determined by first-order boundary smoothness, which minimizes the temperature gradient jump at the boundary between the hidden area covered by the cell and the adjacent visible gap.

8. The method according to claim 7, characterized in that, After converting the relative residual into a local temperature compensation source, the process also includes differentiating the local temperature: for a base plate grid point covered by the battery cell, if the grid point is located in a solid plate area, the injected residual source for that grid point is equal to the local temperature compensation source. If the grid point is located in the flow channel region, then the injected residual source at that grid point is equal to the local temperature compensation source. The product of the liquid heat transfer ratio and the liquid heat transfer ratio, where the liquid heat transfer ratio is greater than or equal to zero and much less than 1.

9. The method according to claim 8, characterized in that, Using the directly observed area as a fixed standard point, a local temperature compensation source is applied to the occluded unknown area for iterative calculation, outputting a corrected temperature field. This includes: locking the grid belonging to the visible gap and the liquid cooling plate inlet / outlet area to its actual infrared measured temperature as a fixed standard point; and applying this to all occluded hidden grid points. The following formula is used to iterate multiple times until numerical convergence is achieved: ； in, Indicates the first Grid points in the next iteration temperature, For point The four directly adjacent grids in the top, bottom, left, and right directions It is a balance parameter that takes values ​​between 0 and 1. For the initial temperature field, To inject residual sources; in iterative calculation of adjacent points When encountering a fixed standard point, its fixed true temperature is directly substituted; when the temperature change rate of all hidden grid points is lower than the convergence threshold, the corrected temperature field is output.

10. A sodium-ion battery liquid-cooled temperature field reconstruction system, characterized in that, The system includes: Acquisition module: Acquires visible light and infrared images of the sodium-ion battery pack and performs spatiotemporal synchronization; Structural guidance information extraction module: Performs structural semantic recognition on visible light images to solve the three-dimensional spatial posture transformation relationship, and registers it with a preset liquid cooling plate template to generate structural guidance information; Initial temperature field construction module: Extracts the temperature characteristics of the directly observed points and the surface temperature characteristics of each cell from the infrared image; calculates the initial temperature field of the entire liquid cooling plate based on the structural guidance information; Compensation module: Calculates the original residual between the cell surface temperature characteristics and the reference temperature, removes common-mode error to obtain the relative residual, and converts the relative residual into a local temperature compensation source; Correction module: Using the directly observed area as a fixed standard point, the local temperature compensation source is applied to the shaded unknown area for iterative calculation, and the corrected temperature field is output.

Images