Method and device for optimizing an internal cooling channel of a gas turbine blade

Abstract

This disclosure provides a method and apparatus for optimizing the internal cooling channel of a gas turbine blade, which can be applied to the field of artificial intelligence technology. The method includes: acquiring an original image characterizing the original design parameters of the internal cooling channel; performing image recognition on the original image to identify information about the cooling ribs, width variations, and turning structures within the internal cooling channel; inputting the cooling rib information, width variation information, and turning structure information into a knowledge base to determine target historical cases corresponding to the original image; the knowledge base includes operating parameters, heat flow field distribution, and prior knowledge indicating the design principles of the internal cooling channel corresponding to the target historical case; and inputting the target historical case and the original image into a pre-trained language model to generate performance information corresponding to the original design parameters and a first response text indicating the optimization parameters and optimization approach for the internal cooling channel.

Description

Technical Field

[0001] This disclosure relates to the field of artificial intelligence technology, and more specifically to a method and apparatus for optimizing the internal cooling channel of a gas turbine blade. Background Technology

[0002] Gas turbine blade cooling design plays a central role in aircraft overall design, directly determining the performance of the gas turbine. However, traditional cooling design still relies heavily on engineers' experience and iterative trial and error. Designers typically adjust geometric parameters based on mission requirements, gradually converging to a feasible solution by incorporating computational fluid dynamics, heat transfer, and interdisciplinary constraints. This process requires designers to possess sufficient professional background and rich practical experience, as well as extensive iterative optimization under uncertainty. For inexperienced engineers, conducting high-quality turbine cooling design in a short period is extremely challenging, and traditional processes are insufficient to meet the high efficiency and reliability requirements of modern gas turbine design. Furthermore, the repetitive and complex nature of the design process itself limits the accumulation and reuse of knowledge, making it difficult to systematize and standardize engineering experience, thus hindering the development of design automation and intelligence. Summary of the Invention

[0003] In view of the above problems, this disclosure provides a method and apparatus for optimizing the internal cooling channel of gas turbine blades.

[0004] According to the first aspect of this disclosure, a method for optimizing the internal cooling channel of a gas turbine blade is provided, comprising: acquiring an original image characterizing the original design parameters of the internal cooling channel; performing image recognition on the original image to identify cooling rib information, width variation information, and turning structure information of the internal cooling channel; inputting the cooling rib information, width variation information, and turning structure information into a knowledge base to determine a target historical case corresponding to the original image, wherein the knowledge base includes operating parameters, heat flow field distribution, and prior knowledge indicating the design principle of the internal cooling channel corresponding to the target historical case; and inputting the target historical case and the original image into a pre-trained language model to generate performance information corresponding to the original design parameters and a first response text indicating the optimization parameters and optimization ideas of the internal cooling channel.

[0005] According to embodiments of this disclosure, the above-mentioned image recognition of the original image to identify the cooling rib information, width variation information, and turning structure information of the internal cooling channel includes: image enhancement of the original image to obtain an enhanced image; extraction of linear geometric features from the enhanced image to obtain a grayscale line map characterizing the structural features of the internal cooling channel of the gas turbine blade; segmentation of the grayscale line map to obtain multiple sub-grayscale line maps; and determination of at least one of the following information based on any one of the sub-grayscale line maps: the cooling rib information, width variation information, and turning structure information of the internal cooling channel.

[0006] According to an embodiment of this disclosure, the grayscale line map includes a grayscale projection line map. The step of extracting linear geometric features from the enhanced image to obtain a grayscale line map characterizing the structural features of the internal cooling channel of the gas turbine blade includes: converting the enhanced image to grayscale to obtain a grayscale image; and performing integral projection on the grayscale values ​​of each pixel in the grayscale image along a preset projection direction to obtain the grayscale projection line map. The grayscale projection line map characterizes the width variation of the internal cooling channel.

[0007] According to an embodiment of this disclosure, the grayscale line map includes a skeleton centerline map. The step of extracting linear geometric features from the enhanced image to obtain a grayscale line map characterizing the structural features of the internal cooling channel of the gas turbine blade includes: performing binarization processing on the enhanced image to obtain a binarized enhanced image; and performing skeletonization processing on the binarized enhanced image to obtain the skeleton centerline map, wherein the skeleton centerline map characterizes the main path of the internal cooling channel.

[0008] According to embodiments of this disclosure, the grayscale line map further includes an edge scan line map. The extraction of linear geometric features from the enhanced image to obtain a grayscale line map characterizing the internal cooling channel structure of the gas turbine blade further includes: using the skeleton centerline in the skeleton centerline map as a guide line; determining a scan line corresponding to each pixel on the guide line to determine multiple scan lines; calculating the gradient value of each pixel on the binarized enhanced image for any one of the multiple scan lines to determine edge pixels; and determining the edge scan line map based on all the edge pixels.

[0009] According to embodiments of this disclosure, the method further includes: obtaining original question text, which indicates the original design parameters of the internal cooling aisle; determining multiple candidate question graphs corresponding to the original question text based on the original question text; determining a target question graph in response to a target user's selection of at least one candidate question graph from the multiple candidate question graphs; searching the knowledge base based on the target question graph and the original question text to obtain multiple candidate historical cases; and inputting the multiple candidate historical cases and the original question text into the pre-trained language model to generate a second response text for the optimization approach.

[0010] According to embodiments of this disclosure, the above-mentioned search of the knowledge base based on the target question graph and the original question text to obtain multiple candidate historical cases includes: performing geometric similarity retrieval in the knowledge base based on the target question graph to obtain multiple candidate retrieval graphs; determining multiple first candidate historical cases based on the multiple candidate retrieval graphs; performing retrieval in the knowledge base based on the original question text to determine multiple second candidate historical cases; and determining multiple candidate historical cases based on the multiple first candidate historical cases and the multiple second candidate historical cases.

[0011] According to embodiments of this disclosure, the above-mentioned retrieval of multiple second candidate historical cases in the knowledge base based on the original question text includes: extracting features from the original design parameters to obtain original parameter feature vectors; and using the original parameter feature vectors to search the knowledge base to obtain the multiple second candidate historical cases.

[0012] According to embodiments of this disclosure, the method further includes: determining the thermodynamic characteristics of the internal cooling channel based on the target problem diagram, wherein the thermodynamic feature vector characterizes the cooling performance of the internal cooling channel; searching the knowledge base based on the thermodynamic feature vector to determine at least one third candidate historical case; and determining the plurality of candidate historical cases based on the plurality of first candidate historical cases, the plurality of second candidate historical cases, and the at least one third candidate historical case.

[0013] The second aspect of this disclosure provides a device for optimizing the internal cooling channel of a gas turbine blade, comprising: an acquisition module for acquiring an original image characterizing the original design parameters of the internal cooling channel; an identification module for performing image recognition on the original image to identify information on the cooling ribs, width variation, and turning structure within the internal cooling channel; a first input module for inputting the cooling rib information, width variation, and turning structure information into a knowledge base to determine a target historical case corresponding to the original image, wherein the knowledge base includes operating parameters, heat flow field distribution, and prior knowledge indicating the design principle of the internal cooling channel corresponding to the target historical case; and a second input module for inputting the target historical case and the original image into a pre-trained language model to generate performance information corresponding to the original design parameters and a first response text indicating the optimization parameters and optimization approach of the internal cooling channel.

[0014] A third aspect of this disclosure provides an electronic device comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the method described above.

[0015] A fourth aspect of this disclosure also provides a computer-readable storage medium having a computer program or instructions stored thereon, which, when executed by a processor, implement the steps of the above-described method.

[0016] The fifth aspect of this disclosure also provides a computer program product, including a computer program or instructions that, when executed by a processor, implement the steps of the above-described method.

[0017] According to embodiments of this disclosure, by performing image recognition on the original image, information on the cooling ribs, width variation, and turning structure of the internal cooling channel is obtained. This achieves automated extraction of key geometric features in the internal cooling channel, improving optimization efficiency. Furthermore, the cooling rib information, width variation information, and turning structure information are input into a knowledge base to identify target historical cases. This associates the original image with the target historical cases. The operating parameters, heat flow field distribution, and prior knowledge indicating the design principles of the internal cooling channel stored in the knowledge base corresponding to the target historical cases provide interpretable case support for optimization. Simultaneously, a pre-trained language model is used to integrate the target historical cases and the original image, thereby generating a first response text that integrates performance information, optimization parameters, and optimization ideas. This realizes the construction of an interpretable knowledge system for the internal cooling mechanism of the internal cooling channel. Attached Figure Description

[0018] The foregoing contents, as well as other objects, features, and advantages of this disclosure, will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0019] Figure 1 The illustration schematically shows an application scenario of the gas turbine blade internal cooling channel optimization method and apparatus according to embodiments of the present disclosure;

[0020] Figure 2 A flowchart illustrating a method for optimizing the internal cooling passage of a gas turbine blade according to an embodiment of the present disclosure is shown schematically.

[0021] Figure 3 A schematic diagram of a gas turbine blade internal cooling channel optimization device according to an embodiment of the present disclosure is shown.

[0022] Figure 4 A block diagram schematically illustrates an electronic device suitable for implementing a method for optimizing the internal cooling channels of gas turbine blades according to an embodiment of the present disclosure. Detailed Implementation

[0023] The embodiments of the present disclosure will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the disclosure. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the present disclosure for ease of explanation. However, it will be apparent that one or more embodiments may be practiced without these specific details. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concepts of the present disclosure.

[0024] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit this disclosure. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0025] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0026] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0027] In related technologies, mapping models learn the relationship between cooling channel geometry and heat transfer performance and pressure loss characteristics, enabling tasks such as cooling efficiency prediction, internal flow reconstruction, and cooling structure reverse design. For example, convolutional networks are used to extract channel flow features and predict local heat transfer coefficients; transfer learning is employed to quickly infer cooling effects under different operating conditions; three-dimensional temperature fields are reconstructed through encoder-decoder structures; or generative methods such as generative adversarial networks, variational autoencoders, and diffusion models are used to directly complete the automatic design of cooling channel geometry. These studies demonstrate that deep learning has significant efficiency advantages in complex cooling problems, reducing reliance on costly computational fluid dynamics simulations and providing rapid evaluation methods for cooling structure optimization. However, these data-driven models still face key bottlenecks: because they mainly rely on large-scale sample learning of input-output black-box mappings, they struggle to reveal internal cooling mechanisms, such as secondary flow structures, rib reinforcement mechanisms, and main-sideflow coupling effects, resulting in a lack of interpretability within the models; simultaneously, the limited prior physical knowledge they can incorporate makes it difficult to understand the reasons behind geometric choices from a heat-fluid-structure coupling perspective, as experts do. Therefore, model reliability, generalization, and engineering usability remain constrained. Although methods such as physical information neural networks attempt to alleviate this problem by introducing constraints from energy and momentum equations, overall, current data-driven methods have not yet achieved an understanding of the "knowledge level" of cooling design, let alone generated systematic and interpretable engineering knowledge that can directly guide the design of blade internal cooling structures.

[0028] In multidisciplinary fields, large language models have demonstrated powerful knowledge integration and reasoning capabilities, achieving functions such as structured knowledge acquisition, scientific question answering, and solving physically constrained problems in fields such as biology, materials science, and mechanical engineering. Through supervised fine-tuning, retrieval-enhanced generation, and multi-agent collaboration, their understanding depth and reasoning ability in specialized domains have been significantly improved. However, in the highly complex field of gas turbine blade cooling design, involving heat-fluid coupling, there is still little exploration into how to enable large language models to extract interpretable cooling mechanism knowledge from cooling channel geometric parameters, heat transfer data, and pressure loss characteristics. While current data-driven cooling design methods can achieve rapid heat transfer prediction and performance evaluation, they still lack a mechanism to systematically integrate expert experience, classical enhanced heat transfer laws, and modern deep learning methods, making it difficult to generate engineering knowledge that can guide practical design.

[0029] In view of this, embodiments of the present disclosure provide a method for optimizing the internal cooling channel of a gas turbine blade. The method involves acquiring an original image characterizing the original design parameters of the internal cooling channel; performing image recognition on the original image to identify information about the cooling ribs, width variations, and turning structures within the internal cooling channel; inputting the cooling rib information, width variations, and turning structure information into a knowledge base to determine target historical cases corresponding to the original image. The knowledge base includes operating parameters, heat flow field distribution, and prior knowledge indicating the design principles of the internal cooling channel corresponding to the target historical case; and inputting the target historical case and the original image into a pre-trained language model to generate performance information corresponding to the original design parameters and a first response text indicating the optimization parameters and optimization approach for the internal cooling channel.

[0030] Figure 1 The illustration shows an application scenario of the gas turbine blade internal cooling channel optimization method and apparatus according to an embodiment of the present disclosure.

[0031] like Figure 1 As shown, application scenario 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 serves as a medium for providing a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0032] Users can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 via the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social media platform software, etc. (for example only).

[0033] The first terminal device 101, the second terminal device 102, and the third terminal device 103 can be various electronic devices with displays and support web browsing, including but not limited to smartphones, tablets, laptops, and desktop computers.

[0034] Server 105 can be a server that provides various services, such as a backend management server that supports websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103 (this is just an example). The backend management server can analyze and process data such as received user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices.

[0035] It should be noted that the gas turbine blade internal cooling channel optimization method provided in this embodiment can generally be executed by server 105. Correspondingly, the gas turbine blade internal cooling channel optimization device provided in this embodiment can generally be installed in server 105. The gas turbine blade internal cooling channel optimization method provided in this embodiment can also be executed by a server or server cluster that is different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105. Correspondingly, the gas turbine blade internal cooling channel optimization device provided in this embodiment can also be installed in a server or server cluster that is different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105.

[0036] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0037] The following will be based on Figure 1 The described scene, through Figure 2 The method for optimizing the internal cooling channel of gas turbine blades according to embodiments of this disclosure will be described in detail.

[0038] Figure 2 A flowchart illustrating a method for optimizing the internal cooling channel of a gas turbine blade according to an embodiment of the present disclosure is shown.

[0039] like Figure 2 As shown, the gas turbine blade internal cooling channel optimization method of this embodiment includes operations S210 to S240.

[0040] In operation S210, the original image characterizing the original design parameters of the internal cooling channel is obtained.

[0041] According to embodiments of this disclosure, the aforementioned original images can be endoscopic photographs, X-ray images, slice scans, thermal images, or CFD visualization projections. (Image quality must be guaranteed; all the aforementioned images must cover typical structures and contain sufficient structural diversity.)

[0042] According to embodiments of this disclosure, the original image can also be preprocessed to improve image quality, facilitating subsequent image recognition operations. Preprocessing may include, for example, operations such as denoising, grayscale conversion, normalization, and edge enhancement.

[0043] During operation S220, image recognition is performed on the original image to identify information such as the cooling ribs of the internal cooling channel, the width variation information within the internal cooling channel, and the turning structure information within the internal cooling channel.

[0044] According to embodiments of this disclosure, image recognition can be performed using models such as 1D-CNN (One-Dimensional Convolutional Neural Network), LSTM (Long Short-Term Memory) / GRU (Gated Recurrent Unit), Transformer, and hybrid structures like CNN-Transformer (Convolutional Neural Network – Transformer). Among these, 1D-CNN excels at handling local structures, such as rib spacing and periodic perturbation structures; LSTM can capture sequence trends, such as channel thickness variations; and Transformer is suitable for long sequences and complex topology recognition, and is well-suited for cooling channels with diverse structural forms. All of the above models are pre-trained, and the training data undergoes data augmentation processing (such as noise addition, translation, blurring, and brightness changes) during model training. Data augmentation improves the model's robustness to endoscopic noise and angular deviations. Furthermore, supervised learning, semi-supervised training, or contrastive learning can be used to enhance the model's generalization ability.

[0045] According to embodiments of this disclosure, after image recognition, category labels, structural location markers, size change curves, or fault detection results of each sub-region in the original image can be obtained, and the above recognition results can be further mapped to geometric parameters that can be used in engineering (such as cooling rib information, width change information in the internal cooling channel, turning structure information in the internal cooling channel, rib height information, spacing information, channel width information, serpentine turning angle information, blockage location information, etc.).

[0046] According to embodiments of this disclosure, the structure of the internal cooling channel can also be reconstructed based on the image recognition results using parametric geometric models (such as NURBS (Non-Uniform Rational B-Splines) curves or topological graphs), thereby transforming the one-dimensional recognition results into a two-dimensional or three-dimensional geometric model for further analysis or numerical simulation.

[0047] According to embodiments of this disclosure, before performing image recognition, it is necessary to clarify the objectives of the recognition task, such as identifying the location and number of cooling ribs, determining channel width changes, identifying serpentine bend structures, detecting blockages or erosion, etc., and then preparing image data containing different types of cooling structures according to the recognition task.

[0048] In operation S230, cooling rib information, width change information, and turning structure information are input into the knowledge base to determine the target historical case corresponding to the original image.

[0049] The knowledge base includes operating parameters, heat flow field distribution, and prior knowledge indicating the design principles of the internal cooling aisle, corresponding to the target historical cases.

[0050] According to embodiments of this disclosure, the aforementioned knowledge base can be a three-layer architecture, comprising a first sub-knowledge base, a second sub-knowledge base, and a third sub-knowledge base. The first sub-knowledge base stores quantifiable parameters such as geometry, operating conditions, and materials; the second sub-knowledge base stores data such as temperature fields, flow fields, and heat transfer distributions; and the third sub-knowledge base is a text-based knowledge base that can store empirical rules, literature conclusions, and mechanistic explanations (i.e., prior knowledge) in a vector database format. This multi-layered knowledge base effectively improves the efficiency of precise retrieval and facilitates integration with language models to generate explanatory content.

[0051] In operation S240, the target historical cases and original images are input into a pre-trained language model to generate performance information corresponding to the original design parameters and a first response text indicating the optimization parameters and optimization ideas for the internal cooling channel.

[0052] According to embodiments of this disclosure, the aforementioned performance information may include, for example, the performance impact of rib angle, channel width, and air film pore parameters on the internal cooling channel. The first response text may include the temperature field and pressure loss corresponding to different internal cooling channel parameters (geometric schemes), and further automatically generate optimization parameters based on design objectives, as well as optimization ideas corresponding to the generation process.

[0053] For example, the original image is a CT scan image of the internal cooling channel of a certain type of gas turbine blade, showing: serpentine channel structure, rectangular cooling ribs arranged at a 45° angle, channel width gradually changing from 3mm at the inlet to 2mm at the outlet, and a 180° sharp turn structure. The matched target historical case ID-2021-GL-089 has the following parameters: operating conditions (inlet temperature 1600K, outlet pressure 1.5MPa, cooling air mass flow rate 0.05kg / s), corresponding heat flow field distribution (maximum wall temperature 1420K, heat transfer coefficient distribution cloud map), and design principle (enhanced design that balances heat transfer and flow resistance). The aforementioned CT scan images and target historical cases can then be input into a pre-trained language model. The prompt message is: "You are an aero-engine cooling system design expert. Based on the provided internal cooling channel image features and historical similar cases, analyze the current design performance and provide optimization suggestions." The language model will output: Based on the similarity matching (92%) of case ID-2021-GL-089 and the comparison analysis of operating parameters, the predicted performance of the current design at an inlet temperature of 1650K is shown in Table 1 below.

[0054] Table 1

[0055]

[0056] Key Risk Identification: The inlet temperature is 50K higher than in historical cases, but the channel contraction ratio (1.5) and cooling fin density have not been adjusted accordingly, resulting in insufficient heat transfer capacity in the leading edge region. The predicted wall temperature exceeds the safe threshold (>1450K). Optimization parameters include adjusting the cooling fin inclination angle of the inlet section (first 30% of the flow channel) from 45° to 60°, and reducing the fin spacing from 8 times the fin length to 5 times the fin length. The optimization strategy is to address the increased heat load caused by the 50K increase in inlet temperature by implementing a leading edge local enhanced heat transfer design; increasing the fin inclination angle to promote secondary flow development and shortening the fin spacing to increase the disturbance frequency are expected to reduce the maximum wall temperature by 40-60K, thus resolving the over-temperature risk.

[0057] According to embodiments of this disclosure, by performing image recognition on the original image, information on the cooling ribs, width variation, and turning structure of the internal cooling channel is obtained. This achieves automated extraction of key geometric features in the internal cooling channel, improving optimization efficiency. Furthermore, the cooling rib information, width variation information, and turning structure information are input into a knowledge base to identify target historical cases. This associates the original image with the target historical cases. The operating parameters, heat flow field distribution, and prior knowledge indicating the design principles of the internal cooling channel stored in the knowledge base corresponding to the target historical cases provide interpretable case support for optimization. Simultaneously, a pre-trained language model is used to integrate the target historical cases and the original image, thereby generating a first response text that integrates performance information, optimization parameters, and optimization ideas. This realizes the construction of an interpretable knowledge system for the internal cooling mechanism of the internal cooling channel.

[0058] According to embodiments of this disclosure, the above-mentioned image recognition of the original image to identify the cooling rib information, width variation information, and turning structure information of the internal cooling channel includes: image enhancement of the original image to obtain an enhanced image; extraction of linear geometric features from the enhanced image to obtain a grayscale line map characterizing the structural features of the internal cooling channel of the gas turbine blade; segmentation of the grayscale line map to obtain multiple sub-grayscale line maps; and determination of at least one of the following information based on any one of the sub-grayscale line maps: cooling rib information of the internal cooling channel, width variation information, and turning structure information.

[0059] According to embodiments of this disclosure, image enhancement is an operation that improves image clarity and feature recognizability by adjusting image contrast, brightness, or employing filtering techniques; linear geometric features are features that characterize the linear geometric structure of the internal cooling channel identified from the enhanced image, and may include edge contours, center lines, and geometric orientations; grayscale line maps are grayscale images generated based on linear geometric features, wherein the structural contour of the internal cooling channel is represented in the form of grayscale lines, which can reflect the overall geometric shape of the internal cooling channel; sub-grayscale line maps are local images obtained by segmenting the grayscale line maps, each sub-grayscale line map corresponding to a specific segment of the internal cooling channel, used for refined structural analysis.

[0060] According to embodiments of this disclosure, by enhancing the original image, the edge blurring caused by metal artifacts and noise interference in the scanned image is at least partially mitigated, providing a high signal-to-noise ratio data foundation for subsequent extraction of linear geometric features to obtain grayscale line maps. Furthermore, the linear geometric features are extracted from the enhanced image to obtain grayscale line maps, the three-dimensional internal cooling channel structure is abstracted into a two-dimensional topological structure, and the grayscale line maps are segmented to obtain sub-grayscale line maps. This allows for the determination of at least one of the following information: cooling rib information of the internal cooling channel, width variation information within the internal cooling channel, and turning structure information within the internal cooling channel. This enables automatic identification of design parameters in the internal cooling channel and improves optimization efficiency.

[0061] According to an embodiment of this disclosure, the grayscale line map includes a grayscale projection line map. Extracting linear geometric features from an enhanced image to obtain a grayscale line map characterizing the structural features of the internal cooling channel of a gas turbine blade includes: converting the enhanced image to grayscale to obtain a grayscale map; performing integral projection on the grayscale values ​​of each pixel in the grayscale map along a preset projection direction to obtain a grayscale projection line map, wherein the grayscale projection line map characterizes the width variation of the internal cooling channel.

[0062] According to embodiments of this disclosure, integral projection is an operation that sums the grayscale values ​​of all pixels in each row (or column) along a defined direction (preset projection direction) in the image. The preset projection direction can be, for example, a direction perpendicular to the extension direction of the cooling ribs or along the center line of the main channel. Setting the aforementioned preset projection direction allows the generated grayscale projection line map to sensitively and directly reflect changes in the channel width.

[0063] According to the embodiments of this disclosure, by converting the enhanced image to grayscale to obtain a grayscale image, the interference of color information in the enhanced image on the recognition of geometric structural features is reduced, and the subsequent computational complexity is reduced. Furthermore, by performing integral projection on the grayscale values ​​of each pixel in the grayscale image along a preset projection direction to obtain a grayscale projection line map, the spatial distribution of the channel width is transformed into the curve change of the grayscale integral value, thereby realizing the automated extraction of the width change of the internal cooling channel and improving the extraction efficiency.

[0064] According to an embodiment of this disclosure, the grayscale line map includes a skeleton centerline map. Extracting linear geometric features from the enhanced image to obtain a grayscale line map characterizing the structural features of the internal cooling channel of the gas turbine blade includes: performing binarization processing on the enhanced image to obtain a binarized enhanced image; and performing skeletonization processing on the binarized enhanced image to obtain a skeleton centerline map, wherein the skeleton centerline map characterizes the main path of the internal cooling channel.

[0065] According to embodiments of this disclosure, skeletonization is a series of image morphological operations that iteratively erode a target object (foreground) in a binary image until its width is only a single pixel, while preserving its original topological connectivity and basic geometry. Skeletonization can extract the central axis of the target object (i.e., the inner cold channel), and the resulting skeleton centerline map retains the object's main orientation, branches, endpoints, and other topological information.

[0066] According to the embodiments of this disclosure, by performing binarization processing on the enhanced image, the continuous image is converted into a binary state representation, providing a discrete binary input that meets the topological requirements for subsequent skeletonization processing. Then, by performing skeletonization processing on the binarized enhanced image, a skeleton centerline map is obtained, which can completely preserve the path branch structure, connection relationship and spatial direction of the internal cooling channel.

[0067] According to embodiments of this disclosure, the grayscale line map further includes an edge scan line map. Linear geometric features are extracted from the enhanced image to obtain a grayscale line map characterizing the structural features of the internal cooling channel of the gas turbine blade. The grayscale line map also includes: using the skeleton centerline in the skeleton centerline map as a guide line; determining a scan line corresponding to each pixel on the guide line to determine multiple scan lines; calculating the gradient value of each pixel on the binarized enhanced image for any one of the multiple scan lines to determine edge pixels; and determining an edge scan line map based on all edge pixels, whereby the edge scan line map characterizes the channel contour of the internal cooling channel.

[0068] According to embodiments of this disclosure, the scan line is a straight line defined perpendicular to the tangent direction of the guide line at each pixel point on the guide line. In an ideal binary image, the pixel values ​​in the inner cold channel region are uniform, while significant gradient extrema are generated at the boundary between the channel and the background. Therefore, edge pixels can be determined by calculating the gradient value (pixels with gradient values ​​greater than a preset threshold are selected as edge pixels).

[0069] According to embodiments of this disclosure, different grayscale line maps can be obtained through different methods, so different processing methods can be selected according to the structural characteristics of the internal cooling channel.

[0070] According to the embodiments of this disclosure, by using the skeleton centerline as a guide line to determine multiple scan lines, automatic sampling path planning in the cross-sectional direction of the internal cooling channel is realized. Then, gradient values ​​are calculated based on the scan lines to determine edge pixels, which at least partially reduces the size measurement error caused by boundary jaggedness in the binarized image. Furthermore, the edge pixels are connected in series according to the scan line sequence to form a continuous contour line, generating an edge scan line map, thus realizing efficient extraction of the channel contour.

[0071] According to embodiments of this disclosure, combining a knowledge base with image recognition enables rapid mapping from image features to structural design parameters. The knowledge base can provide a large number of correspondences between known geometry, heat transfer, pressure loss, and reinforced structures, while image recognition can extract key one-dimensional features such as channel centerline, rib spacing, and width variation from endoscopic images, slice images, or projection images. The identified features can be used as retrieval vectors to match the most similar historical cases in the knowledge base, thereby obtaining the corresponding heat transfer coefficient, pressure loss, and structural mechanism explanation, realizing bidirectional reasoning of "finding structure from images and finding performance from structure".

[0072] According to embodiments of this disclosure, the method further includes: obtaining original question text, which indicates the original design parameters of the internal cooling channel; determining multiple candidate question graphs corresponding to the original question text based on the original question text; determining a target question graph in response to the target user's selection of at least one candidate question graph from the multiple candidate question graphs; searching a knowledge base based on the target question graph and the original question text to obtain multiple candidate historical cases; and inputting the multiple candidate historical cases and the original question text into a pre-trained language model to generate a second response text for optimizing the approach.

[0073] According to embodiments of this disclosure, the original problem text is input by the user in natural language form. It can describe the design requirements or problems of the internal cooling channel to be optimized, and can also be used to supplement abstract design intentions or design constraints that are not fully expressed in the image (e.g., "focus on strengthening the leading edge cooling", "pressure drop needs to be controlled below 10 kPa").

[0074] According to embodiments of this disclosure, for computation and reasoning, the cooling channel data in the knowledge base needs to be parameterized and regularized. Parameterization involves storing the cooling channel's geometric dimensions, rib structure, orifice diameter, flow parameters, etc., as parameters in the knowledge base (using orthogonal parameterization, NURBS curves, etc.). Regularization involves writing empirical knowledge as structured logic and storing it in the knowledge base (e.g., "increased rib angle → enhanced heat transfer but increased pressure loss"). Simultaneously, by associating numerical values ​​with text, both numerical and textual data can be retrieved simultaneously when querying the knowledge base, thus providing a complete context for the analysis and decision-making of the pre-trained language model.

[0075] According to embodiments of this disclosure, multiple candidate question images corresponding to the original question text can be determined from a knowledge base or other databases based on the original question text provided by the user, and displayed to the target user. The target user selects one of these as the target question image based on the original question text, and further determines multiple candidate historical cases through the target question image and the original question text (the target question image and the original question text can be combined into a joint retrieval vector to perform similarity matching in the knowledge base). The pre-trained language model generates a second response text by comprehensively analyzing the operating parameters, heat flow field distribution, and prior knowledge indicating the design principle of the internal cooling channel in multiple candidate historical cases, and ensures that the generated second response text conforms to the constraints of the original question text.

[0076] According to embodiments of this disclosure, the knowledge base can be data-driven, rule-driven, or fusion-type. Its function—whether design-oriented, simulation-oriented, or intelligent optimization-oriented—can be determined based on different needs (e.g., reusing cooling channel design experience, supporting heat transfer and pressure loss performance prediction, assisting topology or geometric parameter optimization, improving design interpretability and decision quality, etc.). Furthermore, the knowledge related to the internal flow cooling channels of the blades mainly comes from: channel topology and geometric parameters, such as rib angles, serpentine channels, needle ribs, and vortex structures; different operating conditions and material parameters, including cold gas mass flow rate, temperature ratio, wall thickness, and coating characteristics; CFD or CHT simulation results, including heat transfer coefficients, temperature fields, and pressure loss distribution; various experimental data, such as thermal imaging, pressure measurements, and visualized flow experiments; literature and expert experience.

[0077] According to embodiments of this disclosure, multiple candidate question graphs are determined by the original question, and the plain text description is converted into a visual graph, providing an intuitive display for the target user. This lowers the technical threshold for non-professionals to build accurate geometric queries. Furthermore, based on multimodal retrieval of the target question graph and the original question text, the retrieval scope is expanded, making the generated second response text more accurate.

[0078] According to embodiments of this disclosure, the above-mentioned search of the knowledge base based on the target question graph and the original question text to obtain multiple candidate historical cases includes: performing geometric similarity retrieval in the knowledge base based on the target question graph to obtain multiple candidate retrieval graphs; determining multiple first candidate historical cases based on the multiple candidate retrieval graphs; performing retrieval in the knowledge base based on the original question text to determine multiple second candidate historical cases; and determining multiple candidate historical cases based on the multiple first candidate historical cases and the multiple second candidate historical cases.

[0079] According to embodiments of this disclosure, the core capability of the knowledge base is to find the most similar cases. It can simultaneously find first candidate historical cases that are similar to the target question graph in the knowledge base, and retrieve multiple second candidate historical cases corresponding to the original design parameters from the knowledge base using the original question text. Furthermore, if duplicate cases exist between the first and second candidate historical cases, they can be deduplicated to reduce the processing burden on the subsequent pre-trained language model. It can also simultaneously perform geometric similarity retrieval (e.g., parameter distance, point cloud distance, graph structure similarity) and thermal feature similarity retrieval (e.g., heat transfer coefficient and pressure loss feature vector based on sensitivity analysis), thus forming a retrieval system that integrates geometric and aerodynamic channels, thereby providing highly relevant data support for the subsequent generation of response text by the large language model.

[0080] According to embodiments of this disclosure, knowledge can also be injected into a small model through knowledge distillation for heat transfer coefficient or temperature prediction, thereby partially replacing time-consuming CFD simulations.

[0081] According to embodiments of this disclosure, the above-mentioned method of retrieving multiple second candidate historical cases from a knowledge base based on the original question text includes: extracting features from the original design parameters to obtain original parameter feature vectors; and using the original parameter feature vectors to search the knowledge base to obtain multiple second candidate historical cases.

[0082] For example, suppose the original problem text is "The current high-pressure turbine blade trailing edge slot internal cooling channel adopts 9 circular discrete film gas holes, with an outlet Mach number of 0.4. The design requirements are that the wall temperature should not exceed 1100K under the condition of a combustion chamber outlet temperature of 1800K, and the proportion of cold gas flow should be less than 2%. The current simulation shows that there are local hot spots on the wall." Then, the feature words and numerical relationships can be extracted into a high-dimensional original parameter feature vector. At the same time, the similarity between the original parameter feature vector and the feature vector of each historical case in the knowledge base is calculated, and the top few historical cases with the higher similarity are selected as the second candidate historical cases.

[0083] According to embodiments of this disclosure, the method further includes: determining the thermodynamic characteristics of the internal cooling channel based on the target problem diagram, wherein the thermodynamic feature vector characterizes the cooling performance of the internal cooling channel; searching the knowledge base based on the thermodynamic feature vector to determine at least one third candidate historical case; and determining multiple candidate historical cases based on multiple first candidate historical cases, multiple second candidate historical cases, and at least one third candidate historical case.

[0084] For example, the original problem text is: "The current mid-chord chamber has a rectangular channel containing three rows of staggered ribs. CFD results show that in the latter 40% of the flow direction, although the overall flow velocity increases, the local heat transfer coefficient downstream of the ribs decreases, leading to an abnormal wall temperature gradient in this region. The goal is to improve the cooling uniformity in this region without significantly increasing the flow rate and pressure drop." The target problem diagram is a cross-sectional schematic of the mid-chord chamber. The image shows that the channel has a rectangular cross-section and contains three rows of staggered rectangular ribs, with a slight convergence angle in the latter half of the channel. Three independent retrieval processes can be initiated simultaneously. The first is graph retrieval based on the target question graph, extracting structural features ("mid-chord chamber", "rectangular cross-section straight channel", "containing three rows of staggered rectangular ribs", "channel converging at the rear"), and determining multiple first candidate historical cases through graph retrieval. The second is retrieval based on the original question text, extracting key parameters from the original question ("rectangular ribbed channel", "rear region", "heat transfer coefficient decrease", "anomaly wall temperature gradient", "pressure drop / flow constraint"), and determining multiple second candidate historical cases through vector retrieval. The third is retrieval based on thermodynamic features, that is, determining thermodynamic features (region: rear section of channel; feature: flow rate increases but local heat transfer coefficient decreases) from the target question graph, searching for at least one third candidate historical case with similar thermodynamic features in the knowledge base, and then deduplicating the first, second, and third candidate historical cases to obtain multiple candidate historical cases, and finally generating a second response text based on the candidate historical cases and the original question text.

[0085] Based on the above-mentioned method for optimizing the internal cooling channels of gas turbine blades, this disclosure also provides a device for optimizing the internal cooling channels of gas turbine blades. The following will be combined with... Figure 3 The device is described in detail.

[0086] Figure 3 A schematic block diagram of a gas turbine blade internal cooling channel optimization device according to an embodiment of the present disclosure is shown.

[0087] like Figure 3 As shown, the gas turbine blade internal cooling channel optimization device 300 of this embodiment includes an acquisition module 310, an identification module 320, a first input module 330, and a second input module 340.

[0088] The acquisition module 310 is used to acquire original images characterizing the original design parameters of the internal cooling channel. In one embodiment, the acquisition module 310 can be used to perform the operation S210 described above, which will not be repeated here.

[0089] The recognition module 320 is used to perform image recognition on the original image to identify information such as the cooling ribs of the internal cooling channel, the width variation information within the internal cooling channel, and the turning structure information within the internal cooling channel. In one embodiment, the recognition module 320 can be used to perform the operation S220 described above, which will not be repeated here.

[0090] The first input module 330 is used to input cooling rib information, width change information, and turning structure information into a knowledge base to determine the target historical case corresponding to the original image. The knowledge base includes operating parameters, heat flow field distribution, and prior knowledge indicating the design principle of the internal cooling channel corresponding to the target historical case. In one embodiment, the first input module 330 can be used to perform the operation S230 described above, which will not be repeated here.

[0091] The second input module 340 is used to input the target historical cases and the original image into the pre-trained language model to generate performance information corresponding to the original design parameters and response text indicating the internal cooling channel optimization parameters and optimization ideas. In one embodiment, the second input module 340 can be used to perform the operation S240 described above, which will not be repeated here.

[0092] According to embodiments of this disclosure, any plurality of modules among the acquisition module 310, identification module 320, first input module 330, and second input module 340 may be combined into one module, or any one of these modules may be split into multiple modules. Alternatively, at least part of the functionality of one or more of these modules may be combined with at least part of the functionality of other modules and implemented in one module. According to embodiments of this disclosure, at least one of the acquisition module 310, identification module 320, first input module 330, and second input module 340 may be at least partially implemented as hardware circuitry, such as a field-programmable gate array (FPGA), a programmable logic array (PLA), a system-on-a-chip, a system-on-a-substrate, a system-on-package, an application-specific integrated circuit (ASIC), or implemented in hardware or firmware by any other reasonable means of integrating or packaging the circuitry, or implemented in any one of the three implementation methods of software, hardware, and firmware, or in a suitable combination of any of these. Alternatively, at least one of the acquisition module 310, the identification module 320, the first input module 330, and the second input module 340 can be at least partially implemented as a computer program module, which can perform corresponding functions when the computer program module is run.

[0093] Figure 4 A block diagram schematically illustrates an electronic device suitable for implementing a method for optimizing the internal cooling channels of gas turbine blades according to an embodiment of the present disclosure.

[0094] like Figure 4As shown, an electronic device 400 according to an embodiment of the present disclosure includes a processor 401, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 402 or a program loaded from a storage portion 408 into a random access memory (RAM) 403. The processor 401 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 401 may also include onboard memory for caching purposes. The processor 401 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0095] RAM 403 stores various programs and data required for the operation of electronic device 400. Processor 401, ROM 402, and RAM 403 are interconnected via bus 404. Processor 401 performs various operations of the method flow according to embodiments of the present disclosure by executing programs in ROM 402 and / or RAM 403. It should be noted that the programs may also be stored in one or more memories other than ROM 402 and RAM 403. Processor 401 may also perform various operations of the method flow according to embodiments of the present disclosure by executing programs stored in said one or more memories.

[0096] According to embodiments of this disclosure, the electronic device 400 may further include an input / output (I / O) interface 405, which is also connected to a bus 404. The electronic device 400 may also include one or more of the following components connected to the input / output (I / O) interface 405: an input section 406 including a keyboard, mouse, etc.; an output section 407 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 408 including a hard disk, etc.; and a communication section 409 including a network interface card such as a LAN card, modem, etc. The communication section 409 performs communication processing via a network such as the Internet. A drive 410 is also connected to the input / output (I / O) interface 405 as needed. A removable medium 411, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 410 as needed so that computer programs read from it can be installed into the storage section 408 as needed.

[0097] This disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs that, when executed, implement the method according to the embodiments of this disclosure.

[0098] According to embodiments of this disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, such as including, but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this disclosure, the computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of this disclosure, the computer-readable storage medium may include ROM 402 and / or RAM 403 and / or one or more memories other than ROM 402 and RAM 403 described above.

[0099] Embodiments of this disclosure also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code is used to enable the computer system to implement the gas turbine blade internal cooling channel optimization method provided in embodiments of this disclosure.

[0100] When the computer program is executed by the processor 401, it performs the functions defined in the system / apparatus of this disclosure embodiments. According to embodiments of this disclosure, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0101] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and downloaded and installed via communication section 409, and / or installed from removable medium 411. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0102] In such an embodiment, the computer program can be downloaded and installed from a network via communication section 409, and / or installed from removable medium 411. When the computer program is executed by processor 401, it performs the functions defined in the system of this disclosure embodiment. According to embodiments of this disclosure, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0103] According to embodiments of this disclosure, program code for executing the computer programs provided in embodiments of this disclosure can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can execute entirely on a user's computing device, partially on a user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0104] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0105] Those skilled in the art will understand that the features described in the various embodiments of this disclosure can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in this disclosure. In particular, the features described in the various embodiments of this disclosure can be combined and / or combined in various ways without departing from the spirit and teachings of this disclosure. All such combinations and / or combinations fall within the scope of this disclosure.

[0106] The embodiments of this disclosure have been described above. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of this disclosure. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of this disclosure, and all such substitutions and modifications should fall within the scope of this disclosure.

Claims

1. A method of gas turbine vane internal cooling channel optimization, characterized by, The method includes: Obtain the original image characterizing the original design parameters of the internal cooling channel; Image recognition is performed on the original image to identify the cooling rib information of the internal cooling channel, the width change information of the internal cooling channel, and the turning structure information of the internal cooling channel; The cooling rib information, the width change information, and the turning structure information are input into a knowledge base to determine the target historical case corresponding to the original image. The knowledge base includes the operating parameters, heat flow field distribution, and prior knowledge indicating the design principle of the internal cooling channel corresponding to the target historical case. The target historical case and the original image are input into a pre-trained language model to generate performance information corresponding to the original design parameters and a first response text indicating the optimization parameters and optimization ideas of the internal cooling channel.

2. The method of claim 1, wherein, The image recognition process on the original image identifies information about the cooling ribs of the internal cooling channel, the width variation within the internal cooling channel, and the turning structure within the internal cooling channel, including: The original image is enhanced to obtain an enhanced image; Linear geometric features are extracted from the enhanced image to obtain a grayscale line map characterizing the internal cooling channel structure of the gas turbine blade; The grayscale line image is segmented to obtain multiple sub-grayscale line images; For any one of the multiple sub-grayscale line maps, based on the sub-grayscale line map, at least one of the following information is determined, including: cooling rib information of the internal cooling channel, width variation information within the internal cooling channel, and turning structure information within the internal cooling channel.

3. The method of claim 2, wherein, The grayscale line map includes a grayscale projection line map. Extracting linear geometric features from the enhanced image to obtain a grayscale line map characterizing the internal cooling channel structure of the gas turbine blade includes: The enhanced image is converted to grayscale to obtain a grayscale image; The grayscale values ​​of each pixel in the grayscale image are integrally projected along a preset projection direction to obtain the grayscale projection line map, which represents the width variation of the internal cooling channel.

4. The method of claim 2, wherein, The grayscale line map includes a skeleton centerline map. Extracting linear geometric features from the enhanced image to obtain a grayscale line map characterizing the internal cooling channel structure of the gas turbine blade includes: The enhanced image is binarized to obtain a binarized enhanced image; The binarized enhanced image is skeletonized to obtain the skeleton centerline map, which represents the main path of the internal cooling channel.

5. The method of claim 3, wherein, The grayscale line map also includes an edge scan line map. The step of extracting linear geometric features from the enhanced image to obtain a grayscale line map characterizing the internal cooling channel structure of the gas turbine blade further includes: Using the skeleton centerline in the skeleton centerline diagram as the guide line, for each pixel on the guide line, a scan line corresponding to the pixel is determined to determine multiple scan lines. For any one of the plurality of scan lines, calculate the gradient value of each pixel on the binarized enhanced image of the scan line to determine the edge pixels; The edge scan line map is determined based on all the edge pixels, and the edge scan line map represents the channel profile of the internal cooling channel.

6. The method of claim 1, wherein, The method further includes: Obtain the original problem text, which indicates the original design parameters of the internal cooling aisle section; Based on the original question text, multiple candidate question images corresponding to the original question text are determined; In response to the target user's selection of at least one of the multiple candidate problem graphs, a target problem graph is determined; Based on the target problem graph and the original problem text, the knowledge base is searched to obtain multiple candidate historical cases; The multiple candidate historical cases and the original question text are input into the pre-trained language model to generate the second response text of the optimization idea.

7. The method of claim 6, wherein, The process involves searching the knowledge base based on the target question graph and the original question text to obtain multiple candidate historical cases, including: Based on the target problem graph, a geometric similarity search is performed in the knowledge base to obtain multiple candidate search graphs; Based on the multiple candidate retrieval maps, multiple first candidate historical cases are determined; Based on the original question text, a search is performed in the knowledge base to identify multiple second candidate historical cases; The plurality of candidate historical cases are determined based on the plurality of first candidate historical cases and the plurality of second candidate historical cases.

8. The method according to claim 7, characterized in that, The process involves searching the knowledge base based on the original question text to identify multiple second candidate historical cases, including: Feature extraction is performed on the original design parameters to obtain the original parameter feature vector; The knowledge base is searched using the original parameter feature vector to obtain the multiple second candidate historical cases.

9. The method according to claim 8, characterized in that, The method further includes: The thermodynamic characteristics of the internal cooling channel are determined based on the target problem diagram, and the thermodynamic characteristic vector characterizes the cooling performance of the internal cooling channel. The knowledge base is searched based on the thermodynamic feature vector to identify at least one third candidate historical case; The plurality of candidate historical cases are determined based on the plurality of first candidate historical cases, the plurality of second candidate historical cases, and the at least one third candidate historical case.

10. A device for optimizing the internal cooling channel of a gas turbine blade, characterized in that, The device includes: The acquisition module is used to acquire original images that characterize the original design parameters of the internal cooling channel; The recognition module is used to perform image recognition on the original image to identify the cooling rib information of the internal cooling channel, the width change information of the internal cooling channel, and the turning structure information of the internal cooling channel; The first input module is used to input the cooling rib information, the width change information and the turning structure information into the knowledge base to determine the target historical case corresponding to the original image. The knowledge base includes the operating parameters, heat flow field distribution and prior knowledge indicating the design principle of the internal cooling channel corresponding to the target historical case. The second input module is used to input the target historical case and the original image into a pre-trained language model to generate performance information corresponding to the original design parameters and a first response text indicating the optimization parameters and optimization ideas of the internal cooling channel.

Images