Liquid cooling plate detection method and device suitable for lithium battery module

Abstract

The application discloses a liquid cooling plate detection method suitable for a lithium battery module, relates to the technical field of liquid cooling plates, and provides a flexible optical detection film with a random speckle pattern and a pre-calibrated elastic deformation characteristic, the liquid cooling plate to be detected is positioned on a platform provided with at least three non-collinear positioning reference features with a concave-convex surface facing upward, a special tool is not needed for different specifications, the detection versatility and the rapid change type capability are realized, the actual measurement profile is compared with a design theoretical profile in the same coordinate system under the constraint of the positioning reference features, a cloud picture directly reflecting the distribution of the fitting gap is generated and the pose deviation is quantified, the assembly fitting effect of the liquid cooling plate and the battery module can be comprehensively evaluated, the efficiency bottleneck and the consistency fluctuation of manual detection are eliminated while the measurement accuracy is ensured, and a high-efficiency, accurate and low-cost universal detection scheme is provided for various and batch battery thermal management components.

Description

Technical Field

[0001] This invention relates to the field of liquid cooling plate technology, specifically to a liquid cooling plate testing method applicable to lithium battery modules. Background Technology

[0002] With the rapid development of the new energy vehicle industry, lithium-ion battery modules, as their core power source, directly affect the vehicle's safety and range through their thermal management performance. Since lithium-ion batteries generate a significant amount of heat during charging and discharging, liquid cooling plates are typically attached to the surface of the battery module. Internal circulating coolant removes this heat, maintaining the battery within a suitable temperature range. To improve heat exchange efficiency, the liquid cooling plate needs to be tightly fitted to the battery module surface. However, due to the cell arrangement and structural design, battery modules often have an uneven shape. Therefore, the side of the liquid cooling plate facing the battery must be designed with a corresponding uneven structure to achieve a conformal fit, minimizing air gaps and reducing contact thermal resistance.

[0003] As stated in the published patent CN121453660A, "An Automatic Blue Light Measurement Method for Liquid Cooling Plates for New Energy Vehicle Batteries," "Due to different working scenarios, liquid cooling plates have different shapes and sizes. Existing liquid cooling plate inspections mostly rely on manual labor or fixed-mold measurement equipment. Although manual inspection is flexible, it is extremely inefficient and affected by many operational measurement factors, resulting in large fluctuations in measurement accuracy and making it impossible to guarantee measurement precision. While fixed-mold measurement equipment has high efficiency, when measuring and inspecting liquid cooling plates of different specifications, the tooling is customized, requiring the replacement of the measuring tooling to complete the measurement and inspection of different liquid cooling plates. This results in poor work flexibility and high usage costs, failing to meet the measurement and inspection needs of different liquid cooling plates." The quality of the liquid cooling plate bonding effect, namely the gap distribution between its concave and convex surfaces and the actual assembly state of the battery module, as well as the positional deviation of the liquid cooling plate relative to the design position, are key indicators affecting the battery thermal management performance and need to be accurately tested before leaving the factory.

[0004] However, due to varying working environments, liquid cooling plates come in diverse shapes and sizes, and existing testing methods have significant shortcomings: manual inspection, while highly flexible, is extremely inefficient and subject to operator error, resulting in large fluctuations in measurement accuracy and inconsistent results; while using dedicated fixed-mold measuring equipment can improve efficiency, its tooling is customizable, requiring frequent tooling changes for different specifications of liquid cooling plates, leading to poor flexibility and high operating costs, making it difficult to meet the testing needs of diverse products and varying batch sizes. Therefore, how to achieve rapid, accurate, and versatile testing of the bonding effect and positional deviation of liquid cooling plates of different specifications has become a pressing technical problem to be solved in this field. Summary of the Invention

[0005] To overcome the shortcomings mentioned above, this invention aims to provide a technical solution that can solve the aforementioned problems.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A method for testing liquid cooling plates in lithium battery modules includes the following steps: S1, a flexible optical detection film is provided, the flexible optical detection film having a uniform and known thickness, a random speckle pattern pre-set on its surface, and the elastic deformation characteristics of the flexible optical detection film being pre-calibrated; S2, Position the liquid cooling plate to be tested on the testing platform with its concave and convex surfaces facing upwards for bonding with the lithium battery, and set at least three non-collinear positioning reference features around the liquid cooling plate to be tested on the testing platform. S3, cover the liquid cooling plate to be tested with the flexible optical detection film, and seal the edge of the flexible optical detection film with the detection platform to form a closed cavity to accommodate the liquid cooling plate to be tested between the flexible optical detection film and the detection platform; S4. While maintaining the positioning reference feature fixed relative to the detection platform, a vacuuming operation is performed on the closed cavity. By controlling the pressure difference between the ambient pressure on the side of the flexible optical detection film away from the liquid cooling plate to be tested and the pressure inside the closed cavity, the flexible optical detection film is driven to adaptively adhere to and completely cover the uneven surface of the liquid cooling plate to be tested. S5, simultaneously acquire random speckle patterns on the surface of the flexible optical detection film in the attached state from multiple angles, and calculate the three-dimensional morphology of the outer surface of the flexible optical detection film in the attached state. S6, perform thickness compensation and deformation inversion on the three-dimensional morphology of the outer surface to peel out the actual three-dimensional contour of the convex and concave surface of the liquid cooling plate under test. S7. The actual three-dimensional contour of the concave and convex surface is registered and compared with the theoretical design contour of the liquid cooling plate to be tested in the same coordinate system based on the positioning reference features. A bonding gap distribution cloud map is generated to characterize the local bonding effect, and the positional deviation of the liquid cooling plate to be tested relative to the design posture is calculated.

[0007] A liquid-cooled plate testing device for lithium battery modules includes a testing platform and at least one flexible optical testing film, wherein the testing platform and the flexible optical testing film are operated using the method described above. The detection platform has high flatness and is equipped with an array of mounting holes. The detection platform integrates an adjustable support clamp and a lateral clamping mechanism. The flexible optical detection film is pre-prepared with precision and its elastic deformation characteristics are fully calibrated.

[0008] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention provides a flexible optical detection film with a random speckle pattern and pre-calibrated elastic deformation characteristics. The liquid-cooled plate to be tested is positioned with its concave and convex surfaces facing upwards on a platform with at least three non-collinear positioning reference features. After sealing the edges of the film to form a closed cavity, a vacuum is applied. The pressure difference drives the flexible film to adaptively and tightly wrap the complex concave and convex surface of the liquid-cooled plate, thus eliminating the need for specialized tooling for different specifications and achieving universality and rapid changeover capability. Subsequently, speckle images of the film surface in the attached state are acquired, and a high-precision three-dimensional morphology of the outer surface of the film is calculated based on digital image correlation. This, combined with the known film thickness, ... The calibrated elastic deformation characteristics and differential pressure data are used for thickness compensation and deformation inversion to accurately peel out the real concave and convex contours of the liquid cooling plate, effectively overcoming the interference of the membrane itself on the measurement. Finally, the measured contour is registered and compared with the theoretical design contour in the same coordinate system constrained by the positioning reference features to generate a cloud map that intuitively reflects the distribution of the bonding gap and quantifies the pose deviation. This allows for a comprehensive evaluation of the assembly bonding effect between the liquid cooling plate and the battery module. While ensuring measurement accuracy, it eliminates the efficiency bottleneck and consistency fluctuations of manual inspection, providing an efficient, accurate, and low-cost universal inspection solution for multi-variety and batch battery thermal management components. Attached Figure Description

[0009] Figure 1 This is a 3D structural diagram of a lithium-ion battery module in a new energy vehicle in the current technology. Figure 2 This is a 3D diagram of the internal structure of a lithium-ion battery module in a new energy vehicle in the current technology. Figure 3 yes Figure 2 A partial view at point A in the middle; Figure 4 This is a flowchart of the operation process S1-S7 in this invention. Detailed Implementation

[0010] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0011] Please see Figure 1-4 A liquid cooling plate testing method applicable to lithium battery modules includes the following steps: S1, a flexible optical detection film is provided, the flexible optical detection film having a uniform and known thickness, and a random speckle pattern is preset on the upper surface of the liquid cooling plate to be tested or the lower surface of the liquid cooling plate to be tested, and the elastic deformation characteristics of the flexible optical detection film are pre-calibrated. S2, Position the liquid cooling plate to be tested on the testing platform with its concave and convex surfaces facing upwards for bonding with the lithium battery, and set at least three non-collinear positioning reference features around the liquid cooling plate to be tested on the testing platform. S3, cover the liquid cooling plate to be tested with the flexible optical detection film, and seal the edge of the flexible optical detection film with the detection platform to form a closed cavity to accommodate the liquid cooling plate to be tested between the flexible optical detection film and the detection platform; S4. While maintaining the positioning reference feature fixed relative to the detection platform, a vacuuming operation is performed on the closed cavity. By controlling the pressure difference between the ambient pressure on the side of the flexible optical detection film away from the liquid cooling plate to be tested and the pressure inside the closed cavity, the flexible optical detection film is driven to adaptively adhere to and completely cover the uneven surface of the liquid cooling plate to be tested. S5. Using a calibrated multi-view stereo vision system, the random speckle pattern on the surface of the flexible optical detection film in the attached state is acquired simultaneously from multiple angles, and the three-dimensional morphology of the outer surface of the flexible optical detection film in the attached state is calculated based on the digital image correlation method. S6. Based on the known thickness, the pre-calibrated elastic deformation characteristics and the pressure difference, the thickness compensation and deformation inversion are performed on the three-dimensional morphology of the outer surface to peel out the actual three-dimensional contour of the uneven surface of the liquid cooling plate under test. S7. The actual three-dimensional contour of the concave and convex surface is registered and compared with the theoretical design contour of the liquid cooling plate to be tested in the same coordinate system based on the positioning reference features. A bonding gap distribution cloud map is generated to characterize the local bonding effect, and the positional deviation of the liquid cooling plate to be tested relative to the design posture is calculated. The reason for using a flexible optical detection film in this invention is that direct visual inspection is indeed impractical in this scenario. The core issue lies in the material properties of the liquid-cooled plate itself and the "adhesion effect" as the detection target; the combination of these two factors renders purely optical methods ineffective. The flexible optical detection film is not superfluous; rather, it addresses three fundamental problems: The first optical problem is that metal surfaces cannot be directly reconstructed in 3D. The main material of liquid-cooled plates is usually aluminum alloy, which, after stamping, chipping, or brazing, has irregular reflections, scratches, and oxidation color differences on the surface. Moreover, the uneven features will form strong highlights and shadow blind spots. Furthermore, stereo vision detection relies on identifying the "texture features" of the same object point in different images. On a smooth, highly reflective, and monotonous metal plate, the camera will lose features in the recesses and make mismatches due to reflections on the slopes, making it impossible to generate dense and accurate 3D point clouds. The transformation role of flexible optical detection films is then demonstrated. The random speckle patterns on the flexible optical detection film are artificial high-contrast optical features, perfectly adapted to digital image correlation methods. At this time, the camera no longer "sees" the liquid-cooled plate, but "sees" the film, transforming the complex metal surface morphology measurement into a mature, high-precision non-contact full-field deformation / morphology measurement. Secondly, there's the issue of measurement mechanics. Because the free state and the bonded state are inconsistent, the customer wants to test the "bonding effect," that is, the gap distribution after the liquid cooling plate and the uneven surface of the lithium battery are in close contact. This is essentially the result of an assembly under stress, not a free form. Direct visual measurement measures the free morphology: if a camera is used to scan an unpressurized liquid cooling plate, the resulting outline is in the free state. However, when the liquid cooling plate is actually working, it is pressed against the battery by external forces, and its thin-walled uneven structure undergoes elastic deformation, resulting in a completely different bonding gap from the free state. Measuring the free morphology makes it difficult to deduce the true bonding effect. This is where the mechanical simulation function of the flexible optical detection film comes into play: the uniform pressure difference created by vacuuming accurately simulates the uniformly distributed loading pressure when the liquid cooling plate is covered by the battery. Under this pressure, the flexible optical detection film forces the liquid cooling plate into a "working bonded state." The measured outer surface morphology of the film at this time, after inversion, yields the bonding outline of the liquid cooling plate under actual stress. This is the true input required to test the bonding effect. Thirdly, there's the issue of compensation and inversion. Even if the surface of the liquid cooling plate is scanned directly using optical methods, only the geometric data of its inner surface (facing the battery) is obtained. Direct vision cannot reveal the precise thickness distribution from the scanned surface to the bonding surface. Flexible optical detection films provide a known and controllable "indirect layer": the film's thickness is known and uniform, and its elastic properties are calibrated. Through a stress-deformation model, the film thickness and local deformation can be precisely "peeled" from the outer surface morphology of the flexible optical detection film, revealing the true contour of the liquid cooling plate's uneven surface and the bonding effect. This is a physically meaningful measurement chain, not a simple visual projection. Figure 1-3As shown, the surface of the lithium battery module has different heights and is uneven. Therefore, in order to bond with the lithium battery pack, the thickness of the liquid cooling plate itself will vary and the surface will be uneven. Thus, the bonding between the liquid cooling plate and the lithium battery pack involves n planes, curved surfaces, bends and corners of different heights. Moreover, some of the surfaces are partially bonded to the lithium battery pack. Directly inferring the bonding situation from the outer surface of the liquid cooling plate in the photograph will result in a huge error and cannot be calibrated. In summary, direct visual inspection is "inaccurate, incorrect, and useless" in this context. Flexible optical inspection films cleverly transform a physical field (metal irregularity under force and bonding) that is difficult to measure directly into a standard optical metrology problem. At the same time, they achieve the triple functions of forming stable optical features, simulating real stress states, and providing a compensable indirect layer. This makes high-precision and highly flexible full-field bonding effect inspection possible. Using a pre-calibrated flexible optical detection film with preset random speckle and elastic deformation characteristics, combined with a vacuum in a closed cavity to form a controllable pressure difference, the detection film adaptively and tightly wraps the complex concave and convex surface of the liquid cooling plate under test. This completely eliminates the need for traditional special customized tooling and fundamentally solves the problem of frequent fixture changes due to the variety of liquid cooling plate specifications. It achieves highly versatile detection of liquid cooling plates of any shape, size and concave and convex structure, reducing changeover time and tooling costs in multi-variety and variable batch production. Meanwhile, the uniform and controllable adhesion force provided by the vacuum pressure difference ensures complete adhesion between the detection membrane and the workpiece, eliminating the gaps and random errors that are unavoidable in manual adhesion, and ensuring high consistency and repeatability of each measurement state; by acquiring speckle images of the membrane surface under the adhesion state and combining them with digital image correlation (DIC) for full-field topography calculation, millions of spatial point cloud data can be acquired in seconds in a non-contact, full-field measurement manner. Compared with manual point-by-point inspection, the measurement speed is increased by orders of magnitude, and the measurement fluctuations caused by operator differences are completely eliminated; The pre-calibrated film thickness and elastic deformation characteristics, through thickness compensation and pressure difference-based deformation inversion of the measured three-dimensional morphology of the outer surface, can accurately peel out the true concave and convex surface contour of the liquid cooling plate under test, effectively eliminating the systematic errors introduced by the thickness of the film itself and the tensile deformation under load, and ensuring that the contour restoration reaches the preset accuracy. Setting at least three non-collinear positioning reference features fixed to the platform on the testing platform not only establishes a stable reference coordinate system for each test, but also enables the measured contour to be accurately registered with the design theoretical contour under the same reference. This automatically generates an intuitive bonding gap distribution cloud map and quantifies the positional deviation, comprehensively evaluating the local bonding effect and global assembly posture, and providing complete and reliable data support for the accurate prediction of battery thermal management performance. This invention provides a flexible optical detection film with a random speckle pattern and pre-calibrated elastic deformation characteristics. The liquid-cooled plate to be tested is positioned with its concave and convex surfaces facing upwards on a platform with at least three non-collinear positioning reference features. After sealing the edges of the film to form a closed cavity, a vacuum is applied. The pressure difference drives the flexible film to adaptively and tightly wrap the complex concave and convex surface of the liquid-cooled plate, thus eliminating the need for specialized tooling for different specifications and achieving universality and rapid changeover capability. Subsequently, speckle images of the film surface in the attached state are acquired, and a high-precision three-dimensional morphology of the outer surface of the film is calculated based on digital image correlation. This, combined with the known film thickness, ... The calibrated elastic deformation characteristics and differential pressure data are used for thickness compensation and deformation inversion to accurately peel out the real concave and convex contours of the liquid cooling plate, effectively overcoming the interference of the membrane itself on the measurement. Finally, the measured contour is registered and compared with the theoretical design contour in the same coordinate system constrained by the positioning reference features to generate a cloud map that intuitively reflects the distribution of the bonding gap and quantifies the pose deviation. This allows for a comprehensive evaluation of the assembly bonding effect between the liquid cooling plate and the battery module. While ensuring measurement accuracy, it eliminates the efficiency bottleneck and consistency fluctuations of manual inspection, providing an efficient, accurate, and low-cost universal inspection solution for multi-variety and batch battery thermal management components.

[0012] In this embodiment of the invention, step S1 specifically includes the following steps: A highly elastic, isotropic transparent polymer film is selected as the substrate. Its thickness tolerance is controlled within ±0.01 mm through precision casting or calendering processes. The thickness distribution of the film across the entire field is then non-contactly scanned and calibrated to obtain the known thickness. On the upper or lower surface of the substrate, a random speckle pattern consisting of randomly distributed, size-controllable high-contrast speckles is generated by one of the following processes: microdroplet jetting, vapor deposition combined with mask etching, or laser speckle forming. The speckle quality is evaluated to ensure that it meets the matching requirements of digital image correlation. The prepared speckle pattern film is clamped in a calibration cavity with a known standard geometry. After sealing, a closed calibration cavity is formed. By gradually changing the pressure difference applied to both sides of the film, and simultaneously using a calibrated multi-view stereo vision system to collect the three-dimensional morphology of the film under each pressure difference, the elastic modulus, Poisson's ratio and hyperelastic constitutive model parameters of the film are calculated by inverse finite element analysis, thus completing the pre-calibration of the elastic deformation characteristics. The calibrated flexible optical detection film is subjected to aging treatment under constant temperature and humidity and stress-free conditions to eliminate the internal stress introduced during the preparation and calibration process, thereby obtaining a flexible optical detection film that can be used for testing. In one embodiment, a highly elastic isotropic transparent polymer film substrate is selected, and a precision casting or calendering process is used to strictly control the thickness tolerance within ±0.01mm. Combined with full-field non-contact scanning calibration, accurate thickness distribution is obtained, providing a high-precision data foundation for subsequent thickness compensation, and avoiding systematic errors caused by uneven film thickness to the contour inversion of the measured surface. On the surface of the thin film, high-contrast, size-controllable random speckle patterns are generated by microdroplet jetting, vapor deposition combined with mask etching or laser speckle forming processes. The speckle quality is evaluated to ensure that it meets the matching requirements of digital image correlation method, thereby ensuring that the multi-view stereo vision system can obtain high-density, high-accuracy full-field point cloud data when solving the shape. The prepared speckled film was clamped onto a calibration cavity with a known standard geometry. After sealing, the stress state in actual testing was simulated by gradually changing the pressure difference across the film. Simultaneously, the three-dimensional morphology of the film under various pressure differences was acquired using a calibrated multi-view stereo vision system. The elastic modulus, Poisson's ratio, and hyperelastic constitutive model parameters were then derived using inverse finite element analysis. This enabled precise calibration of the complex nonlinear elastic deformation characteristics of the film, providing a reliable constitutive model for subsequent deformation inversion based on pressure difference. This ensured that the true contour of the workpiece could be accurately extracted from the surface morphology of the deformed film. Finally, the calibrated film was subjected to aging treatment under constant temperature, humidity, and stress-free conditions. This effectively eliminated the internal stress introduced during preparation and calibration, significantly improving the dimensional stability and mechanical repeatability of the flexible optical inspection film in actual testing. This ensured high accuracy in the measurement results of the overall bonding gap distribution and positional deviation.

[0013] In this embodiment of the invention, step S2 specifically includes the following steps: A detection platform with a high flatness reference surface is provided, wherein an array of mounting holes or T-slots are provided on the reference surface for mounting and fixing adjustable support fixtures and positioning reference feature components; The liquid cooling plate to be tested is placed on the support fixture with its concave and convex surfaces facing upwards, which are used to attach the lithium battery. The support fixture is adjusted so that the liquid cooling plate to be tested is in a horizontal state. The lateral clamping mechanism is used to lock the liquid cooling plate to be tested in the plane from the edge of the non-detection area, so as to achieve a six-degree-of-freedom fully constrained fixation of the liquid cooling plate to be tested relative to the detection platform. At least three non-collinear positioning reference features are installed on the detection platform within the space surrounding the liquid cooling plate to be tested. The positioning reference features are optical target spheres or coded markers with diffuse reflective surfaces and whose center point spatial coordinates can be extracted with high precision. Their bottoms are rigidly connected to the detection platform, and their installation positions are such that the positioning reference features are visible and unobstructed from all perspectives of the multi-view stereo vision system in the subsequent step S5. The multi-view stereo vision system is used to acquire images of the positioning reference features and the reference surface. The three-dimensional coordinates of each positioning reference feature in the coordinate system of the detection platform are calibrated by photogrammetry to establish a global measurement reference. This global measurement reference remains unchanged when different liquid cooling plates to be tested are replaced. The reliable transfer of measurement data to the coordinate system of the design model is achieved only through the positioning reference features. The testing platform, featuring a high-flatness reference surface and arrayed mounting holes or T-slots, along with adjustable support fixtures and lateral clamping mechanisms, can flexibly adapt to liquid cooling plates of different sizes. After leveling the liquid cooling plate with its concave and convex surfaces facing upwards, it is fixed with six degrees of freedom of full constraint, ensuring the absolute stability and repeatability of the workpiece's posture during the testing process. This completely eliminates the interference of rigid body displacement caused by improper clamping on the subsequent calculation of the bonding gap distribution and position deviation. At least three non-collinear optical target spheres or coded markers with diffuse reflective surfaces and high-precision extractable centers are installed around the workpiece on the platform as positioning references, and are ensured to be visible and unobstructed in all viewpoints of the multi-view stereo vision system, thus providing a stable and unified control framework for seamless stitching of multi-angle image data. The vision system is used to acquire positioning reference features and reference surface images. The three-dimensional coordinates of each reference point in the platform coordinate system are accurately calibrated by photogrammetry, and a global measurement reference independent of the workpiece itself is established. When different models of liquid cooling plates are replaced, their coordinate system remains unchanged. All measurement data can be reliably transmitted to their respective design model coordinate systems through this set of permanent positioning references. This not only realizes true multi-variety flexible testing, but also fundamentally ensures the coordinate uniformity and long-term traceability of the test results of different batches and specifications of workpieces. This makes the calculation of position deviation truly reflect the absolute deviation of the workpiece relative to the design posture, rather than the relative difference between different clamping times.

[0014] In this embodiment of the invention, step S3 specifically includes the following steps: On the reference surface of the testing platform, within the predetermined sealing area surrounding the liquid cooling plate to be tested, a continuous elastic sealing groove is pre-processed or installed. A replaceable annular sealing ring is embedded in the elastic sealing groove. The sealing ring is made of fluororubber or silicone rubber, and its cross-sectional height is higher than the reference surface. It is used to generate elastic deformation under pressure to achieve airtight contact. The flexible optical detection film provided in step S1 is laid flat and covered above the liquid cooling plate to be tested and the annular sealing ring in a tension-free and wrinkle-free state. The edge size of the flexible optical detection film exceeds the outer boundary of the annular sealing ring by at least a preset margin. The edge is temporarily pre-fixed by multiple circumferentially distributed clamping positioning pins or adjustable tension frames to prevent slippage or curling during the sealing process. A rigid clamping frame matching the shape of the annular sealing ring is aligned and pressed against the edge of the flexible optical detection film. Several fasteners or quick clamps evenly distributed along the circumference of the rigid clamping frame are used to sequentially press the rigid clamping frame, the edge of the flexible optical detection film, the annular sealing ring, and the detection platform with a preset tightening torque or clamping force, so that the flexible optical detection film, the detection platform, and the annular sealing ring together form the closed cavity that accommodates the liquid cooling plate to be tested. A vacuum pipeline is connected to at least one evacuation port on the testing platform. The evacuation port is located within the projection area of ​​the closed cavity and avoids being directly below the liquid cooling plate to be tested. A vacuum pump and a precision pressure regulating valve are connected externally through the vacuum pipeline, and a high-precision vacuum pressure sensor is connected in series on the pipeline to monitor the absolute pressure value inside the closed cavity in real time. Before the formal test, a pre-test of sealing is performed: the vacuum pipeline is closed, the sealed cavity is evacuated to a preset pre-test pressure and then the pipeline is closed. The pressure rise of the high-precision vacuum pressure sensor is monitored within a preset time. If the pressure rise exceeds the allowable threshold, the clamping force of the rigid clamping frame is adjusted or the annular sealing ring is replaced until the sealing meets the test requirements. An elastic sealing groove with a replaceable fluororubber or silicone rubber annular sealing ring is set around the liquid cooling plate to be tested on the testing platform. The elastic deformation of the sealing ring under pressure provides the first adaptive sealing defense for forming a reliable airtight cavity. Next, the flexible optical detection film is laid flat to cover the film in a tension-free and wrinkle-free state, and temporarily pre-fixed by the edge clamping positioning pins or tensioning frame. This effectively avoids slippage, curling or initial internal stress of the film during the subsequent pressing and sealing process, ensuring the purity and consistency of the initial state of the detection film. Then, a rigid clamping frame matching the shape of the sealing ring is used, and the detection membrane, sealing ring and platform are pressed together in sequence with a preset precise torque by fasteners evenly distributed around the circumference. This uniform and controllable mechanical clamping method, together with the elastic sealing ring, forms a stable and reliable closed cavity, which fundamentally prevents boundary leakage during the vacuuming and pressure holding process. Meanwhile, the vacuum line is connected to the air extraction port on the detection platform located in the projection area of ​​the closed cavity but avoiding the workpiece directly below, and a high-precision vacuum pressure sensor is connected in series for real-time pressure monitoring. This achieves optimization of the air extraction path and precise closed-loop control of the pressure, ensuring that the differential pressure conditions for driving the attachment of the detection membrane are accurate and controllable. Most importantly, the pre-testing of sealing before formal testing forms a quantitative leak detection loop by monitoring the pressure rise during the pressure holding period and comparing it with the allowable threshold. This can promptly detect and eliminate potential problems such as insufficient clamping force or aging and failure of the sealing ring, ensuring that every test is carried out in a tightly sealed airtight environment, thereby eliminating the potential risk of fluctuations in the fit or test failure due to minor leaks.

[0015] In this embodiment of the invention, step S4 specifically includes the following steps: Ensure that the mechanical connection between the positioning reference feature and the detection platform is locked to maintain its spatial position in the global measurement reference constant throughout the vacuuming process; Start the vacuum pump connected to the sealed cavity, and adjust the pumping rate through an electric proportional valve or a precision needle valve so that the absolute pressure in the sealed cavity starts from the current ambient pressure and decreases steadily at a preset low rate, thereby gradually establishing a uniform pressure difference from the ambient pressure to the cavity on both the inside and outside of the flexible optical detection film. Based on the elastic modulus and bending stiffness of the flexible optical detection film pre-calibrated in step S1, and combined with the minimum radius of curvature or maximum aspect ratio geometric features of the uneven surface of the liquid cooling plate to be tested, the minimum critical pressure difference required for the flexible optical detection film to be completely attached to the uneven surface is pre-calculated and set as the target pressure difference. A graded pressurization control strategy is adopted to drive the pressure difference to increase towards the target pressure difference: In the first stage, the flexible optical detection film completes the initial coating of the macroscopic contour of the liquid cooling plate under test with a first pressure difference increment; in at least one subsequent stage, the pressure difference is gradually increased with progressively decreasing pressure difference increments, so that the local areas of the flexible optical detection film successively yield to the bending resistance and gradually fill the grooves, depressions and sharp inner corners of the liquid cooling plate under test until the pressure difference reaches the target pressure difference; During the graded pressurization process, the multi-view stereo vision system continuously acquires the speckle pattern on the surface of the flexible optical detection film at a predetermined frame rate and calculates its full-field three-dimensional morphology in real time, and calculates the root mean square rate of change of the morphology between adjacent frames; when the root mean square rate of change drops below a preset convergence threshold, it is determined that the flexible optical detection film has adaptively adhered to and completely covered the uneven surface of the liquid cooling plate under test. After determining that the enclosure is complete, the absolute pressure in the sealed cavity is dynamically maintained at the pressure set point corresponding to the target pressure difference through the closed-loop servo control system composed of the electro-proportional valve and the high-precision vacuum pressure sensor, so as to offset the pressure drift caused by system micro-leakage or temperature fluctuations and ensure the constant pressure difference during the execution of subsequent steps. After the vacuum pump is started, the pressure is steadily reduced at a preset low rate through an electric proportional valve or a precision needle valve, so that a uniform and controllable pressure difference is gradually established on the inside and outside of the flexible optical detection film, avoiding the impact of sudden pressure changes on the film bonding shape and the generation of initial wrinkles. Based on this, the minimum critical pressure difference is accurately calculated by combining the minimum radius of curvature or the maximum aspect ratio of the liquid cooling plate under test with the pre-calibrated elastic modulus and bending stiffness, providing a scientific basis for setting the target pressure and ensuring the precise matching of the bonding force. By adopting a graded pressure control strategy, the initial coverage of the macroscopic contour is achieved with a large pressure differential increment, and then the film is gradually made to yield to the bending resistance with a progressively decreasing increment, filling the grooves, depressions and the bottom of sharp inner corners in sequence. This effectively avoids the phenomenon of local overstretching, tearing or bridging of the film that may be caused by a single high-pressure bonding, and achieves adaptive bonding to complex uneven surfaces without dead angles. It should be noted that in very rare exceptional cases, if the liquid cooling plate to be tested has a particularly deep groove (which is not usually the case, because the liquid cooling plate is generally not very thick), then this excessively deep groove needs to be measured manually, because the flexible optical detection film cannot possibly fit this excessively deep groove. In other words, when the depth of the groove exceeds the deformation limit of the flexible optical detection film, this groove needs to be excluded from step S4. During this process, a multi-view stereo vision system is used to continuously acquire speckle patterns at a predetermined frame rate and calculate the full-field three-dimensional topography in real time. By calculating the root mean square rate of change of the topography and comparing it with the convergence threshold, an objective and quantitative basis for determining the complete adhesion state is provided, eliminating the uncertainty of subjective human experience judgment. Once the adhesion is determined to be complete, a closed-loop servo system composed of an electric proportional valve and a high-precision vacuum pressure sensor is immediately used to dynamically maintain the target pressure, actively compensate for pressure drift caused by micro-leakage or temperature fluctuations, and ensure the constant pressure difference during subsequent high-precision topography acquisition.

[0016] In this embodiment of the invention, step S5 specifically includes the following steps: A calibrated multi-view stereo vision system is provided, comprising at least three industrial cameras fixedly mounted above the inspection platform at different spatial angles. Each industrial camera is equipped with a telecentric lens or a low-distortion fixed-focus lens and is configured with a coaxial light source or a ring light source capable of generating high-contrast illumination for random speckle patterns on the flexible optical inspection film. The intrinsic parameters, extrinsic parameters, and distortion coefficients of the multi-view stereo vision system are pre-calibrated using a precision calibration plate. While maintaining the target pressure difference constant, all the industrial cameras are triggered by the external trigger synchronization controller to synchronously acquire speckle pattern images of the surface of the flexible optical detection film in the attached state at the same time and with the same exposure parameters. Each camera acquires at least one frame of speckle image corresponding to its viewing angle and transmits the speckle images of each viewing angle to the computing processing unit. The computational processing unit divides the speckle images from each viewpoint into multiple overlapping virtual sub-region grids. For each virtual sub-region, using one viewpoint image as a reference and the other viewpoint images as targets, it performs stereo matching based on digital image correlation: by optimizing the normalized cross-correlation function or the zero-mean normalized minimum distance sum of squares function, it finds the corresponding image point coordinates of the center point of each virtual sub-region in all viewpoint images with sub-pixel accuracy. The optimization process uses the inverse combination Gauss-Newton algorithm or surface fitting method to achieve high-precision matching (as described in the published patent "Real-time Optical Extensometer Measurement Method Based on Digital Image Correlation" with publication number CN105258642B). Based on the pre-calibrated intrinsic and extrinsic parameters of the multi-view stereo vision system, multi-view triangulation reconstruction is performed on the coordinates of the image points of each successfully matched virtual sub-region to obtain the three-dimensional spatial coordinates of the center point of each virtual sub-region under the global measurement reference, thereby generating an original three-dimensional point cloud that reflects the overall appearance of the outer surface under the attached state of the flexible optical detection film. The original three-dimensional point cloud is sequentially subjected to outlier removal, smoothing filtering based on moving least squares, and uniform resampling processing that adaptively densifies or thins the local point cloud density. Based on the prior knowledge of the continuity of the flexible optical detection film surface, the tiny holes caused by local reflection or occlusion are filled in, resulting in a complete and defect-free three-dimensional morphology of the outer surface that can be used for subsequent thickness compensation and deformation inversion. At least three industrial cameras, fixed at different spatial angles and pre-calibrated with internal and external parameters, are equipped with telecentric lenses or low-distortion fixed-focus lenses to optically suppress perspective errors and image distortion. They are also equipped with coaxial or ring light sources to provide high-contrast illumination for the random speckle pattern on the surface of the inspection film. This combination of multi-view redundant imaging and high-quality optical design ensures the clarity, consistency and spatial resolution of the image data required for subsequent 3D reconstruction from a hardware perspective. Under constant pressure difference, nanosecond-level synchronous exposure acquisition of all cameras is achieved through external trigger synchronization controller, "freezing" the instantaneous attachment shape of the entire detection film surface at the same moment, completely eliminating the positional misalignment between multi-view images caused by mechanical vibration or environmental disturbance, and laying a stable data foundation for stereo matching. Subsequently, the computing unit divides the images from each viewpoint into virtual sub-region grids. Using digital image correlation as the core, it optimizes functions such as normalized cross-correlation or zero-mean normalized minimum distance sum of squares, and uses inverse combination of Gauss-Newton algorithm or surface fitting method to accurately match the image point coordinates of each sub-region in all viewpoints with sub-pixel accuracy. This full-field point-by-point matching strategy can obtain tens of thousands or even millions of high-precision corresponding point pairs, far exceeding the information density of traditional single-point or sparse feature matching. Based on this, multi-view triangulation reconstruction is performed on the matched point pairs using the calibrated system parameters to generate a dense original 3D point cloud that fully expresses the overall appearance of the membrane surface. Finally, by sequentially performing outlier removal, moving least squares smoothing filtering, adaptive uniform resampling, and micro-hole filling based on surface continuity prior on the original point cloud, sensing noise is effectively eliminated, point cloud quality is optimized, and local data loss is repaired. The final output is a complete and defect-free 3D morphology, which provides a high-quality input guarantee for subsequent thickness compensation and deformation inversion based on accurate data, thereby restoring the true contour of the liquid cooling plate.

[0017] In this embodiment of the invention, step S6 specifically includes the following steps: From the generated three-dimensional morphology of the outer surface, extract the point cloud fragment of the outer surface of the membrane located outside the edge of the liquid cooling plate to be tested and directly attached to the reference surface area of ​​the detection platform. Obtain a virtual reference plane by least squares plane fitting. This virtual reference plane characterizes the outer surface height reference when the flexible optical detection membrane is attached to the flat rigid surface under the target pressure difference. Each sampling point of the three-dimensional topography of the outer surface is offset inward along its local normal direction by a known thickness value to obtain a preliminary thickness compensation surface; for regions with a radius of curvature greater than a preset flatness threshold, the offset result is directly used as the liquid cooling plate profile of the region; for concave corners, edges, and narrow slits with a radius of curvature less than or equal to the preset flatness threshold, they are marked as regions to be inverted. For the region to be inverted, the pre-calibrated elastic modulus, Poisson's ratio, and hyperelastic constitutive model parameters are used as the material properties of the flexible optical detection film. The maintained target pressure difference is used as the uniformly distributed surface load boundary condition. The actual geometric shape of the three-dimensional morphology segment of the outer surface corresponding to the region to be inverted is used as the initial configuration to construct a finite element inversion model. In the finite element inversion model, loads are applied in the opposite direction along the normal direction of each node on the initial configuration, causing each node to gradually move away from the initial configuration and bounce inward in the normal direction. At the same time, the static equilibrium between the uniformly distributed surface load and the internal stress of the membrane is used as the iterative convergence criterion to solve the neutral layer morphology of the flexible optical detection membrane in the state of no external force attachment. The obtained neutral layer morphology is offset inward along its local normal direction by half of the known thickness value to obtain the true uneven surface contour fragment of the region to be inverted. The contour obtained by direct offset is spliced ​​and fused with the contour fragment obtained by inversion. Weighted smooth transition is applied in the splicing boundary area to eliminate splicing step difference and form a complete initial three-dimensional contour of the liquid cooling plate under test. The coordinates of each sampling point on the initial three-dimensional contour of the liquid cooling plate to be tested are transformed from the relative coordinate system with the virtual reference plane as the reference to the global measurement reference with the positioning reference feature established in step S2 as the reference, so as to obtain the actual concave and convex surface three-dimensional contour in the same coordinate system as the design theoretical contour in step S7. A virtual reference plane is fitted by a point cloud fragment of the outer surface of the membrane attached to the flat reference surface of the testing platform. This provides a unified and accurate height reference for subsequent thickness compensation and eliminates the influence of the membrane's attachment state in the platform area on the zero height point. Next, the entire morphology is offset along the normal direction with a known thickness to obtain a preliminary compensation surface, and the curvature of the region is judged: flat regions have low bending stress when the membrane is attached, and the workpiece contour can be accurately characterized by direct offset, which greatly simplifies the calculation; while sharp feature regions such as concave corners, edges and narrow slits will cause local bridging or non-uniform stretching of the membrane due to bending stiffness, so they are marked as regions to be inverted. For these regions to be inverted, based on the pre-calibrated elastic modulus, Poisson's ratio and hyperelastic constitutive model, and combined with the constant target pressure difference during detection as the load, a finite element inversion model is constructed. By applying the load in reverse, the membrane nodes gradually rebound inward under static equilibrium constraints, and the neutral layer morphology of the membrane when it is not attached by external force is accurately solved. This step mechanically removes the local tensile and bending deformations generated when the membrane is attached. Because the flexible optical detection film cannot perfectly fit the sharp concave corners and narrow slits of the liquid cooling plate, it will act like a tiny bridge across the gap, forming a small arc with a radius larger than the actual concave corner—this is the "bridging effect." Simply subtracting the thickness at this point will introduce an underestimation of the depth. We eliminate this error by using the pre-calibrated elastic deformation characteristics of the flexible optical detection film: First, a model is established: the flexible optical detection film is calibrated as a hyperelastic thin shell with known elastic modulus, Poisson's ratio, and bending stiffness; the input "load"—the pressure difference (ambient pressure minus the internal pressure)—is unique and known. The driving force is uniform and equal everywhere; then the inverse solution is obtained: starting from the contour of the outer surface of the flexible optical detection film with "bridging" error obtained by measurement, a uniformly distributed pressure difference is applied in the opposite direction along the normal direction of each point of the outer contour, and the neutral layer morphology of the film before peeling is calculated using finite element or inverse finite element algorithms; thus obtaining the true contour: this calculated neutral layer morphology, offset inward by half the film thickness, is the true concave and convex surface contour of the liquid cooling plate that we finally require, including the depth of those sharp concave corners covered by "bridging"; thus effectively restoring the detailed geometry of sharp features that are easily "smoothed" away by the attached film; The "inside" of offsetting inward by half the film thickness refers to starting from the "neutral layer" and pointing towards the side of the workpiece. Why half the thickness? Because the total thickness of the film is known (e.g., 0.4 mm). If we consider the film as being composed of two 0.2 mm layers, with the neutral layer located exactly in the middle, then offsetting outward (away from the workpiece direction) from the neutral layer by half the thickness (0.2 mm) yields the outer surface of the film (which is exactly what the camera measures). Offset inward (toward the workpiece direction) from the neutral layer by half the thickness (0.2 mm) yields the inner surface of the film. This inner surface is the interface where the film adheres to the liquid cooling plate, which is the true outline of the workpiece we want. Subsequently, a weighted smooth transition was used to stitch and fuse the contour segments of the directly offset region with the contour segments of the inverted region, eliminating the step difference that may occur at the boundary due to different algorithms, and ensuring the continuity and smoothness of the overall contour. Finally, the reconstructed complete contour was transformed from the virtual reference plane coordinate system to the global measurement reference system established by the permanent positioning reference features, achieving precise spatial alignment with the design theoretical contour. This provided high-precision three-dimensional data with unified coordinates and clear physical meaning for the final fitting gap calculation and position deviation analysis.

[0018] In this embodiment of the invention, step S7 specifically includes the following steps: Obtain the design theoretical profile data of the liquid cooling plate to be tested. The design theoretical profile is a three-dimensional digital model or dense point cloud of the uneven surface of the liquid cooling plate used to fit the lithium battery in an ideal assembly state. The design theoretical profile predefines at least three non-collinear theoretical reference feature points that correspond one-to-one with the positioning reference features in step S2. Using the three-dimensional coordinates of each positioning reference feature obtained in step S2 under the global measurement reference as a fixed reference, the best fitting registration based on point set rigid body transformation is performed with the theoretical reference feature points in the design theoretical profile. The rotation matrix and translation vector of the design theoretical profile are transformed from the design coordinate system to the global measurement reference by the singular value decomposition method or the iterative nearest point algorithm, so as to achieve unconstrained and accurate alignment between the design theoretical profile and the actual concave and convex surface three-dimensional profile obtained in step S6 under the same coordinate system. In the same coordinate system after alignment, traverse each sampling point on the three-dimensional contour of the actual concave and convex surface. Draw a ray through the point along the local normal direction of the design theoretical contour in the neighborhood of the point. Calculate the intersection point of the ray and the design theoretical contour. Define the signed Euclidean distance from the sampling point to the intersection point as the local fitting gap of the point. When the sampling point is located inside the fitting side of the design theoretical contour, the gap is defined as a negative value to indicate an interference fit. When it is located outside the fitting side, it is defined as a positive value to indicate a clearance fit. The three-dimensional coordinates of all sampling points on the actual concave-convex surface three-dimensional contour and the corresponding local fitting gap value are mapped onto the two-dimensional UV unfolding plane of the design theoretical contour through the surface parameterization method. After interpolation filling and color mapping processing, a fitting gap distribution cloud map is generated, and the maximum gap value and its position, the boundary of the interference area and the distribution of the actual contact area are marked on the cloud map. Based on the registration posture of the actual three-dimensional contour of the concave and convex surface relative to the design theoretical contour, the positional deviation of the liquid cooling plate under test relative to the design posture is calculated: Euler angle decomposition is performed on the rotation matrix obtained in step S7 to obtain the rotational deviation angles around the three coordinate axes; the translation vector is directly used as the translational deviation components along the three coordinate axes to obtain the six-degree-of-freedom positional deviation data used to evaluate the assembly positioning accuracy of the liquid cooling plate. The measured three-dimensional contour of the liquid cooling plate under test is precisely compared with the design theoretical model to quantitatively evaluate its bonding effect and positioning accuracy. First, the theoretical profile data of the liquid cooling plate is acquired, and theoretical reference feature points corresponding one-to-one with the global positioning reference features are predefined, providing a clear reference correspondence for subsequent model alignment. Next, using the three-dimensional coordinates of the positioning reference features that have been highly calibrated in step S2 under the global measurement reference as a fixed reference system, the theoretical profile of the design is accurately registered to the global measurement reference where the actual profile is located by means of best fitting rigid body transformation through singular value decomposition or iterative nearest point algorithm. This process realizes the unconstrained accurate alignment of the two data models based on permanent reference features in the same coordinate system, ensuring from the root that the comparison results reflect the real assembly deviation, rather than coordinate system differences or alignment errors. Subsequently, under the same aligned coordinate system, each sampling point on the actual contour is traversed, and the signed Euclidean distance along the local normal direction of the theoretical design contour is calculated as the local fitting gap. Negative values ​​represent interference fit and positive values ​​represent gap fit. This method of calculating the normal distance with positive and negative signs can comprehensively and intuitively reflect the continuous fitting state from interference compression to gap separation, overcoming the limitation that taking only absolute values ​​cannot distinguish the fitting mode. Subsequently, the gap data in three-dimensional space is mapped to the two-dimensional UV unfolding plane corresponding to the design model through surface parametric method. The gap distribution cloud map is generated by interpolation and color mapping, and key features such as the maximum gap position, interference area boundary and actual contact area are automatically marked. The complex three-dimensional spatial relationship is transformed into a clear two-dimensional engineering interpretation chart, which improves the readability and practicality of the test results and enables quality inspectors to quickly locate defect areas. Finally, Euler angle decomposition was performed on the registered rotation matrix to obtain the three-axis rotational deviation angles, and the translation vector was directly taken as the three-axis translational deviation components, thereby obtaining six-degree-of-freedom positional deviation data for evaluating the assembly positioning accuracy of the liquid cooling plate. The pose error was decomposed into six clear and measurable parameters, providing direct data guidance for the adjustment and optimization of the assembly process, and thoroughly realizing a comprehensive and quantitative evaluation of the processing and positioning quality of the liquid cooling plate.

[0019] A liquid-cooled plate testing device for lithium battery modules includes a testing platform and at least one flexible optical testing film, wherein the testing platform and the flexible optical testing film are operated using the method described above. A complete hardware system is formed by organically combining a detection platform with high flatness, arrayed mounting holes or T-slots and integrated adjustable support fixtures, lateral clamping mechanisms and at least three non-collinear permanent positioning reference features with at least one pre-prepared flexible optical detection film with fully calibrated elastic deformation characteristics. The operation of this system strictly follows the above method. The core functions such as high-precision mechanical positioning, adaptive vacuum bonding, multi-view stereo vision acquisition and intelligent data processing are modularly integrated into one unit. The universal fixtures and replaceable annular sealing rings on the testing platform can flexibly adapt to liquid cooling plates of any shape and size and with concave and convex structures, without the need for any special customized tooling, which improves the ability to quickly change over multiple varieties and batches of testing tasks and the utilization rate of equipment. The permanent positioning reference features on the inspection platform constitute a stable global measurement reference independent of the workpiece. Combined with the high-precision vacuum closed-loop control system and the multi-view stereo vision synchronous acquisition unit, it ensures that the coordinate system is consistent, the fit is consistent and the topographic data is complete for each inspection, which fundamentally guarantees the high repeatability and long-term traceability of the measurement results. As a complete technical entity, this device can automatically complete the entire inspection process from workpiece clamping, vacuum adaptive coating, full-field acquisition of three-dimensional morphology, decoupling and inversion of thickness and deformation to the final generation of bonding gap cloud map and calculation of six-degree-of-freedom pose deviation. It effectively replaces traditional inspection methods that rely on manual operation or special molds, and provides an efficient, accurate, universal and low-cost industrial solution for the full inspection of battery thermal management components before they leave the factory.

[0020] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for detecting liquid-cooled plates in lithium battery modules, characterized in that, Includes the following steps: S1, a flexible optical detection film is provided, the flexible optical detection film having a uniform and known thickness, a random speckle pattern pre-set on its surface, and the elastic deformation characteristics of the flexible optical detection film being pre-calibrated; S2, Position the liquid cooling plate to be tested on the testing platform with its concave and convex surfaces facing upwards for bonding with the lithium battery, and set at least three non-collinear positioning reference features around the liquid cooling plate to be tested on the testing platform. S3, cover the liquid cooling plate to be tested with the flexible optical detection film, and seal the edge of the flexible optical detection film with the detection platform to form a closed cavity to accommodate the liquid cooling plate to be tested between the flexible optical detection film and the detection platform; S4. While maintaining the positioning reference feature fixed relative to the detection platform, a vacuuming operation is performed on the closed cavity. By controlling the pressure difference between the ambient pressure on the side of the flexible optical detection film away from the liquid cooling plate to be tested and the pressure inside the closed cavity, the flexible optical detection film is driven to adaptively adhere to and completely cover the uneven surface of the liquid cooling plate to be tested. S5, simultaneously acquire random speckle patterns on the surface of the flexible optical detection film in the attached state from multiple angles, and calculate the three-dimensional morphology of the outer surface of the flexible optical detection film in the attached state. S6, perform thickness compensation and deformation inversion on the three-dimensional morphology of the outer surface to peel out the actual three-dimensional contour of the convex and concave surface of the liquid cooling plate under test. S7. The actual three-dimensional contour of the concave and convex surface is registered and compared with the theoretical design contour of the liquid cooling plate to be tested in the same coordinate system based on the positioning reference features. A bonding gap distribution cloud map is generated to characterize the local bonding effect, and the positional deviation of the liquid cooling plate to be tested relative to the design posture is calculated.

2. The liquid cooling plate detection method for lithium battery modules according to claim 1, characterized in that: Step S1 specifically includes the following steps: A highly elastic, isotropic polymer film was selected as the substrate, and the thickness distribution of the film across the entire field was scanned and calibrated to obtain the known thickness. The random speckle pattern is generated on the surface of the substrate; The prepared speckle pattern film is clamped in a calibration cavity and sealed to form a calibration cavity. The pre-calibration of the elastic deformation characteristics is completed by gradually changing the pressure difference applied to both sides of the film. The calibrated flexible optical detection film is left to stand for aging treatment to eliminate internal stress, resulting in a flexible optical detection film that can be used for testing.

3. The liquid cooling plate detection method for lithium battery modules according to claim 2, characterized in that, Step S2 specifically includes the following steps: Provide a testing platform with a high flatness reference surface; Place the liquid-cooled plate to be tested on the testing platform; At least three non-collinear positioning reference features are installed on the testing platform within the space surrounding the liquid cooling plate to be tested; Images of the positioning reference features and the reference surface are acquired, and the three-dimensional coordinates of each positioning reference feature in the coordinate system of the detection platform are calibrated by photogrammetry to establish a global measurement reference.

4. The liquid cooling plate detection method for lithium battery modules according to claim 3, characterized in that, Step S3 specifically includes the following steps: On the reference surface of the testing platform, within the predetermined sealing area surrounding the liquid cooling plate to be tested, a continuous elastic sealing groove is pre-machined or installed, and a replaceable annular sealing ring is embedded in the elastic sealing groove. The flexible optical detection film provided in step S1 is laid flat and covered over the liquid cooling plate to be tested and the annular sealing ring in a tension-free and wrinkle-free state. A rigid clamping frame matching the shape of the annular sealing ring is aligned and pressed against the edge of the flexible optical detection film, so that the flexible optical detection film, the detection platform and the annular sealing ring together form the closed cavity for accommodating the liquid cooling plate to be tested; A vacuum pipeline is connected to at least one evacuation port on the detection platform. A vacuum pump and a precision pressure regulating valve are connected to the vacuum pipeline, and a high-precision vacuum pressure sensor is connected in series on the pipeline to monitor the absolute pressure value inside the closed cavity in real time. Before the formal testing, a pre-test for airtightness is performed.

5. The liquid cooling plate detection method for lithium battery modules according to claim 4, characterized in that, Step S4 specifically includes the following steps: Start the vacuum pump connected to the sealed cavity to gradually establish a uniform pressure difference from the ambient pressure to the cavity on both the inside and outside of the flexible optical detection film; The minimum critical pressure difference required for the flexible optical detection film to fully adhere to the deepest part of the uneven surface is calculated in advance and set as the target pressure difference. A graded pressurization control strategy is used to drive the pressure difference to increase toward the target pressure difference; During the graded pressurization process, the speckle pattern on the surface of the flexible optical detection film is acquired and its full-field three-dimensional morphology is calculated in real time. The root mean square rate of change of the morphology between adjacent frames is calculated. When the root mean square rate of change drops below the preset convergence threshold, it is determined that the flexible optical detection film has adaptively adhered to and completely covered the uneven surface of the liquid cooling plate under test. After determining that the cavity is fully enclosed, the absolute pressure inside the sealed cavity is dynamically maintained at the pressure set point corresponding to the target pressure difference.

6. The liquid cooling plate detection method for lithium battery modules according to claim 5, characterized in that, Step S5 specifically includes the following steps: Provides a calibrated multi-view stereo vision system; Acquire speckle pattern images of the surface of the flexible optical detection film in the attached state; Perform stereo matching based on digital image correlation. A three-dimensional point cloud reflecting the overall appearance of the outer surface under the attached state of the flexible optical detection film is generated to obtain the three-dimensional morphology of the outer surface.

7. The liquid cooling plate detection method for lithium battery modules according to claim 6, characterized in that, Step S6 specifically includes the following steps: From the three-dimensional morphology of the outer surface, extract the point cloud fragments of the outer surface of the membrane located outside the edge of the liquid cooling plate to be tested and directly attached to the reference surface area of ​​the detection platform to obtain a virtual reference plane; Each sampling point of the three-dimensional shape of the outer surface is offset inward along its local normal direction by a known thickness value to obtain a preliminary thickness compensation surface; for regions with a radius of curvature greater than a preset flatness threshold, the offset result is directly used as the liquid cooling plate profile of the region. Concave corners, edges, and narrow slits with a radius of curvature less than or equal to the preset flatness threshold are marked as regions to be inverted. A finite element inversion model is constructed for the region to be inverted; In the finite element inversion model, the morphology of the neutral layer of the flexible optical detection film under the state of no external force attachment is solved; The obtained neutral layer morphology is offset inward along its local normal direction by half of the known thickness value to obtain the true uneven surface contour fragment of the region to be inverted. The contour obtained by direct offset is spliced ​​and fused with the contour fragment obtained by inversion to obtain the complete initial three-dimensional contour of the liquid cooling plate under test. The initial three-dimensional profile of the liquid cooling plate under test is converted to a global measurement reference to obtain the actual three-dimensional profile of the uneven surface.

8. The method for detecting liquid-cooled plates in lithium battery modules according to claim 7, characterized in that, Step S7 specifically includes the following steps: Obtain the theoretical profile data of the liquid cooling plate under test; Align the theoretical design profile with the actual three-dimensional profile of the uneven surface in the same coordinate system; Under the same aligned coordinate system, calculate the fitting gap value between the theoretical design profile and the actual three-dimensional profile of the concave and convex surface; All fitting gap values ​​are mapped onto the two-dimensional UV unfolding plane of the design theoretical contour using a surface parametric method. After interpolation filling and color mapping, a fitting gap distribution cloud map is generated, and the maximum gap value and its location, the boundary of the interference area, and the distribution of the actual contact area are marked on the cloud map. Based on the registration posture of the actual three-dimensional contour of the uneven surface relative to the design theoretical contour, the positional deviation of the liquid cooling plate under test relative to the design posture is calculated.

9. A liquid-cooled plate testing device suitable for lithium battery modules, characterized in that, It includes a detection platform and at least one flexible optical detection film, wherein the detection platform and the flexible optical detection film are operated in accordance with the method described in any one of claims 1-8; The detection platform has high flatness and is equipped with an array of mounting holes. The detection platform integrates an adjustable support clamp and a lateral clamping mechanism. The flexible optical detection film is pre-prepared with precision and its elastic deformation characteristics are fully calibrated.

Images