Three-dimensional design and quality management system for wind turbine ring forgings based on 5G communication

Abstract

The present invention discloses a three-dimensional design and quality management system for wind power ring forgings based on 5G communication, which belongs to the field of wind power equipment manufacturing technology. The system includes: a modeling module for constructing a parametric three-dimensional design model and outputting design parameters for quality assessment and defect identification; a process data acquisition module for real-time acquisition of process data such as temperature, pressure, and vibration in key structural areas; a defect identification module for fusing image and signal data, and outputting defect type, location, and its sensitivity impact; a quality assessment module for quantitatively scoring product quality based on a scoring function and disturbance response parameters; a feedback control module for identifying dominant deviations and generating control instructions when the scoring is abnormal; and a 5G communication module for realizing high-speed data interaction between modules. The present invention realizes closed-loop control of design, manufacturing, testing, and evaluation, improves the manufacturing quality and efficiency of ring forgings, and is suitable for the manufacturing scenarios of large-scale, high-reliability wind power structural parts.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power equipment manufacturing, and in particular to a three-dimensional design and quality management system for wind power ring forgings based on 5G communication. Background Art

[0002] Wind turbine ring forgings are widely used in key structural components such as wind turbine main shafts, hubs, and flanges. Their large dimensions, complex structures, and high performance requirements place stringent technical standards on manufacturing quality. Currently, as wind turbine equipment evolves toward higher power levels, the complexity and precision requirements of ring forging manufacturing are increasing simultaneously. However, existing technologies for the design and quality management of ring forgings still face significant challenges. Existing designs often rely on static 3D modeling, with design parameters serving only initial structural optimization. This makes it difficult to guide process execution and defect control during actual manufacturing, and lacks a dynamic closed-loop linkage from design to process. Current defect detection relies primarily on post-manufacturing imaging, making it difficult to detect quality fluctuations during the manufacturing process. Furthermore, traditional detection models struggle to integrate and analyze multimodal data from complex structures, resulting in low recognition accuracy and delayed response times. Ring forging quality assessment still relies primarily on manual judgment or post-analysis data collection, failing to quantify the impact of defects, structural performance, and process stability in real time, nor effectively establishing a scoring mechanism to support feedback control. Current systems typically rely on manual adjustments based on experience, lacking a mechanism to automatically determine control paths based on scoring residuals and defect impacts, and thus cannot achieve intelligent closed-loop optimization of design parameters and manufacturing processes. Traditional wired or low-speed wireless communication methods struggle to support the real-time exchange of high-frequency, multi-source process data across multiple modules, hindering the efficiency and responsiveness of each module in the manufacturing process.

[0003] Therefore, it is urgent to establish an intelligent system that supports multi-module collaboration, real-time data perception and feedback regulation, and realize closed-loop control of the entire process from design, manufacturing to quality assessment, so as to effectively improve the manufacturing quality and production efficiency of wind power ring forgings. Summary of the Invention

[0004] The purpose of the present invention is to provide a three-dimensional design and quality management system for wind power ring forgings based on 5G communication, so as to solve the limitations of existing ring forging manufacturing, such as disconnection between design and process, delayed defect detection, non-quantified quality assessment, slow feedback control response and low system communication efficiency, to achieve real-time collaboration and closed-loop control of key links such as design, manufacturing, testing and evaluation, and greatly improve the intelligence level, quality consistency and response efficiency of the manufacturing process.

[0005] In order to achieve the above object, the present invention is implemented through the following technical solutions:

[0006] The 3D design and quality management system for wind power ring forgings based on 5G communication is characterized by including:

[0007] A modeling module is used to build a parametric 3D design model based on the structural requirements and performance specifications of wind turbine ring forgings, and output design parameters that can be used for process evaluation and defect identification;

[0008] A process data acquisition module is used to collect process data corresponding to the key structural areas involved in the design parameters in real time during the ring forging manufacturing process. The process data includes temperature, pressure and vibration information to support defect identification and quality assessment processes;

[0009] A defect recognition module is configured to obtain defect information based on the process data and the image or signal data obtained from post-manufacturing inspection by fusing and analyzing the process data, wherein the defect information includes the location, size, and type of the defect;

[0010] A quality assessment module is used to integrate the design parameters, process data and defect information, build a scoring model, and generate quantifiable ring forging quality scoring results;

[0011] A feedback control module is used to determine the main cause of the score deviation and generate feedback instructions based on the contribution factors and defect sensitivity in the scoring model when the quality score result does not meet the set requirements. The feedback instructions are used to correct design parameters or adjust manufacturing process parameters and are sent to the modeling module or process data acquisition module respectively;

[0012] The 5G communication module is used to establish a high-bandwidth, low-latency data communication connection between the above modules, realizing real-time transmission and remote linkage processing of design parameters, process data, defect information, quality scoring results and feedback instructions.

[0013] A further improvement of the present invention is that the modeling module includes:

[0014] The model feedback correction unit is used to construct a loss function with the scoring error as the target and update the prediction model parameters in the modeling module when there is a difference between the quality scoring result and the predicted performance value of the modeling module;

[0015] The multi-performance optimization unit is used to establish a set of objective functions for multiple performance indicators based on yield strength, manufacturing energy consumption and material utilization during the modeling stage, and to construct a multi-objective optimization model in combination with preset constraints. The multi-objective optimization model is solved using a non-dominated sorting genetic algorithm to output multiple design parameter solution sets, which are used by the modeling module to generate design parameters.

[0016] A further improvement of the present invention is that the multi-objective optimization model constructed in the modeling module includes the following construction process:

[0017] Establish an objective function group with yield strength, material utilization rate and manufacturing energy consumption as multiple performance objective functions, where each objective function establishes a function expression based on the actual ring forging structural parameters;

[0018] In the process of constructing the objective function, structural constraints are introduced, including the lower limit of the material safety factor, the extreme range of geometric dimensions, and the heat treatment process window, to form a feasible solution space with constraint boundaries;

[0019] Based on the scoring function structure provided by the scoring model, a scoring prediction value calculation module is introduced in the objective function solution stage to pre-estimate the scoring of each design parameter solution. The scoring results are embedded in the objective function set as auxiliary criteria of the optimization model to guide the solution set to converge in the direction that takes into account both performance and score expectations.

[0020] A further improvement of the present invention is that the process data acquisition module includes:

[0021] The sampling adaptive unit is used to dynamically adjust the sensor layout density and sampling frequency according to the key structural areas involved in the design parameters output by the modeling module;

[0022] Anomaly detection unit, which is used to monitor real-time process data based on statistical control methods, identify abnormal changes or drift trends in the data, and activate data redundancy channels when an abnormal state is detected;

[0023] The credibility evaluation unit is used to perform credibility scoring on the process data collection results and mark the corresponding data as a low credibility segment when the score is lower than a preset threshold.

[0024] A further improvement of the present invention is that the defect recognition module includes:

[0025] The modal adaptation fusion unit is used to adjust the fusion ratio of the image recognition model and the signal recognition model according to the structural depth of the area corresponding to the defect information to generate a defect recognition result. The fusion ratio is set according to the depth level of the defect location;

[0026] The defect sensitivity calculation unit is used to calculate the influence coefficient of the defect relative to the target performance based on the identified defect type, the location of the defect in the structure and the corresponding stress area.

[0027] A further improvement of the present invention is that the quality assessment module includes:

[0028] The scoring factor contribution calculation unit is used to calculate the dynamic contribution value of each scoring factor based on multiple quality index inputs and construct a scoring contribution vector. , the expression is:

[0029] ;

[0030] in: For the The score of each quality indicator; Indicates the basic weight of the corresponding quality indicator; The response coefficient indicates the degree to which the indicator is affected by the disturbance; is the sensitivity adjustment parameter, which is used to control the contribution strength of the disturbance term to the score; For the The weight coefficient of a quality indicator indicates the importance of the quality indicator in the overall score. The proportion can be determined through expert experience or historical data analysis. The value range is [0,1], and the sum after normalization is 1; For the The scoring input value of each quality indicator represents the actual measured or calculated technical performance, such as yield strength compliance rate, elastic modulus deviation rate, and defect area ratio; For the The disturbance response coefficient of a quality indicator is used to reflect the sensitivity or variability of the indicator under the current manufacturing disturbance conditions (such as temperature fluctuation, stress concentration or forming fluctuation). It is usually obtained through sensor data analysis or experimental simulation; Enter the number of items for scoring; For the The contribution value of each quality indicator in the total score.

[0031] A further improvement of the present invention is that the quality assessment module further includes:

[0032] Scoring function construction unit, used to select the corresponding scoring function based on the type of each quality indicator , and calculate the total score according to the following formula , the formula is:

[0033] ;

[0034] in: For the Item Input The scoring mapping function used is classified and set according to whether the indicator belongs to the structural category, defect category or energy consumption category;

[0035] Rating normalization unit, used to normalize the rating value Normalize to the interval [0,100]. Normalization methods include maximum and minimum scaling, standard deviation normalization, or quantile mapping, and output standard scoring results.

[0036] A further improvement of the present invention is that the quality assessment module further includes:

[0037] The weight optimization unit is used to train multiple scoring models for different defect level sub-samples, and to calculate the weight vector of each scoring model. Optimize, the optimization goal is to minimize the following loss function, the calculation formula is:

[0038] ;

[0039] in: Indicates in The loss function for training and optimizing the weight vector under the defect samples is to minimize the mean square error between the predicted score and the target score, and introduce a sparsity regularization term to improve the generalization ability of the model; the input is the first The scoring input matrix and target scoring vector of the class training samples are output as the optimal weight vector, so that smallest; For the The total number of class-scored training samples; For the The first defect sample Target score for each sample; Score the corresponding model prediction; For the The weight vector of the class scoring model; is the regularization coefficient, which is used to constrain the sparsity of weights; is a vector of norm;

[0040] Rating feedback unit, used to calculate the rating residual: ; and according to the aggregation trend of the residuals and their source categories, select the corresponding feedback control path between modeling parameters, process parameters or defect detection strategies.

[0041] The scoring function model and weight optimization formula proposed in this paper are not isolated mathematical expressions; rather, they directly guide numerical modeling and parameter feedback in the ring forging scoring and prediction process. All parameters are obtained through actual data collection or structural modeling, and the output results are used to determine the process control path, forming a complete "data acquisition-modeling-scoring-control" closed-loop chain with clear engineering feasibility.

[0042] A further improvement of the present invention is that the feedback control module includes:

[0043] The path selection unit is used to determine the dominant factor that causes the score to drop based on the score input item with the largest proportion in the score factor contribution vector when the quality score value does not meet the preset score threshold, and select the execution order of the feedback control path based on the category of the dominant factor, including:

[0044] If the dominant factor belongs to a modeling parameter item, a feedback instruction for updating the structural parameters or performance model parameters in the modeling module is generated;

[0045] If the dominant factor belongs to the process deviation item, a feedback instruction for adjusting the manufacturing process parameter setting is generated;

[0046] If the dominant factor belongs to the defect detection item, a control instruction for adjusting the quality inspection frequency or inspection conditions is generated;

[0047] The path updating unit is used to record the residual change between the score prediction value and the actual score value after the feedback instruction is executed, and to update the priority selection rule of the feedback path based on the residual change.

[0048] A further improvement of the present invention is that the 5G communication module includes:

[0049] A real-time scheduling strategy unit is used to dynamically adjust the allocation strategy of communication resources based on the priority order of task types and the real-time load status of the current 5G communication link to achieve priority transmission of critical task data. The task data includes design parameters, process data, defect identification information, quality scoring results, and feedback control instruction content;

[0050] The data structure control unit is used to build a unified message structure for different types of data transmission tasks, including adding data source identification, timestamp field, compression flag, retransmission strategy flag and cyclic redundancy check code, to improve the error detection capability and repeated transmission control capability of data packets during transmission, and to achieve reliability and timeliness of remote data interaction.

[0051] The beneficial effects of the present invention are as follows: the present invention constructs a parameterized three-dimensional design model through a modeling module, and outputs key design parameters that can be used for process evaluation and defect identification, so that the design stage and the actual manufacturing process are consistent in data and model-driven. Combined with the process data acquisition module, the process parameters such as temperature, pressure and vibration in the key structural area are acquired in real time, so that the state changes in the manufacturing process can be dynamically perceived and recorded, providing comprehensive data support for subsequent defect identification and quality analysis. The defect recognition module integrates multi-source image and signal data to improve the timeliness and accuracy of defect detection, and accurately locates manufacturing anomalies by identifying the location, size and type of defects. The quality assessment module quantitatively expresses the quality of ring forgings by constructing a scoring model that integrates multi-dimensional inputs of design, process and defects, so that the evaluation results are determinable and comparable. When the scoring results deviate from the set requirements, the feedback control module can determine the dominant cause based on the contribution factors and defect sensitivity in the scoring model, and automatically generate feedback instructions to drive the correction of design parameters or optimization of manufacturing processes, thereby forming a quality control loop with adaptive capabilities. The 5G communication module ensures high-bandwidth, low-latency real-time data transmission and remote linkage between functional modules, breaking down the information barriers between the design and manufacturing systems and significantly improving the system's collaborative efficiency and responsiveness. In summary, this invention not only improves the quality control and intelligence level of the wind power ring forging manufacturing process, but also significantly enhances the system's stability, traceability, and production efficiency, making it suitable for batch manufacturing of high-reliability annular structural parts. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be derived from these drawings without inventive effort. Among them:

[0053] Figure 1 This is a system modular diagram of the present invention. DETAILED DESCRIPTION

[0054] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field are within the scope of protection of the present invention.

[0055] like Figure 1FIG. 1 is an embodiment of the present invention, which provides a 3D design and quality management system for wind power ring forgings based on 5G communication, including:

[0056] (1) Modeling module

[0057] It is used to build a parametric 3D design model based on the structural requirements and performance specifications of wind power ring forgings, and output design parameters that can be used for process evaluation and defect identification;

[0058] In this embodiment, the modeling module includes:

[0059] The model feedback correction unit is used to construct a loss function with the scoring error as the target and update the prediction model parameters in the modeling module when there is a difference between the quality scoring result and the predicted performance value of the modeling module;

[0060] The multi-performance optimization unit is used to establish a set of objective functions for multiple performance indicators based on yield strength, manufacturing energy consumption and material utilization during the modeling stage, and to construct a multi-objective optimization model in combination with preset constraints. The multi-objective optimization model is solved using a non-dominated sorting genetic algorithm to output multiple design parameter solution sets, which are used by the modeling module to generate design parameters.

[0061] In a feasible implementation, the multi-objective optimization model constructed in the modeling module includes the following construction process:

[0062] Establish an objective function group with yield strength, material utilization rate and manufacturing energy consumption as multiple performance objective functions, where each objective function establishes a function expression based on the actual ring forging structural parameters;

[0063] In the process of constructing the objective function, structural constraints are introduced, including the lower limit of the material safety factor, the extreme range of geometric dimensions, and the heat treatment process window, to form a feasible solution space with constraint boundaries;

[0064] Based on the scoring function structure provided by the scoring model, a scoring prediction value calculation module is introduced in the objective function solution stage to pre-estimate the scoring of each design parameter solution. The scoring results are embedded in the objective function set as auxiliary criteria of the optimization model to guide the solution set to converge in the direction that takes into account both performance and score expectations.

[0065] Based on this feasible solution space, the modeling module solves a multi-objective optimization model. Using a non-dominated sorting genetic algorithm, it iteratively calculates and screens each design parameter combination for optimal solutions, outputting a set of design parameter solutions that maintains a balance between different performance objectives. This parameter solution set has been proven to guide the modeling module in generating multiple 3D design models that meet structural and performance constraints, supporting subsequent module invocations such as process configuration, defect identification, and quality assessment.

[0066] Once multiple design parameter solution sets are generated, the system selects representative parameter sets based on preset criteria (such as meeting minimum yield strength constraints and heat treatment process boundary feasibility) as the final modeling results, which are then input into the modeling module to produce a 3D model. In this embodiment, the entire modeling process features clear input conditions, solution steps, and output rules, enabling dynamic trade-offs between multiple performance objectives. This improves the design's adaptability to actual manufacturing conditions and enhances the efficiency of the closed-loop quality response for ring forging design.

[0067] (2) Process data acquisition module

[0068] Used to collect process data corresponding to the key structural areas involved in the design parameters in real time during the ring forging manufacturing process, wherein the process data includes temperature, pressure and vibration information, and is used to support defect identification and quality assessment processes;

[0069] In a feasible implementation manner, the process data acquisition module includes:

[0070] The sampling adaptive unit is used to dynamically adjust the sensor layout density and sampling frequency according to the key structural areas involved in the design parameters output by the modeling module;

[0071] Anomaly detection unit, which is used to monitor real-time process data based on statistical control methods, identify abnormal changes or drift trends in the data, and activate data redundancy channels when an abnormal state is detected;

[0072] The credibility evaluation unit is used to perform credibility scoring on the process data collection results and mark the corresponding data as a low credibility segment when the score is lower than a preset threshold.

[0073] In this embodiment, the sampling adaptive unit identifies the key structural areas in the ring forging based on the design parameters output by the modeling module, and dynamically adjusts the sensor layout density and sampling frequency accordingly to ensure higher-precision data collection in areas prone to defects.

[0074] During the manufacturing process, the anomaly detection unit uses statistical control methods to monitor process data in real time, identify abnormal fluctuations or drift trends in parameters such as temperature, pressure or vibration, and automatically enable data redundancy channels when an abnormal state is detected to ensure data continuity and integrity.

[0075] The credibility assessment unit scores the collected data, determining its credibility based on real-time fluctuation characteristics, historical comparison data, and model deviations. Data segments below a preset threshold are marked as low-credibility areas, and their corresponding structural locations and sampling times are recorded, providing a basis for subsequent defect identification and quality assessment.

[0076] Through the above method, the process data acquisition module realizes a collection process with strong structural correlation, high real-time response, and guaranteed data reliability, effectively supporting the downstream modules to accurately judge and respond to the quality status.

[0077] (3) Defect identification module

[0078] for obtaining defect information based on the process data and the image or signal data obtained from post-manufacturing inspection by fusing and analyzing the defect information, wherein the defect information includes the location, size and type of the defect;

[0079] In a feasible implementation, the defect identification module includes:

[0080] The modal adaptation fusion unit is used to adjust the fusion ratio of the image recognition model and the signal recognition model according to the structural depth of the area corresponding to the defect information to generate a defect recognition result. The fusion ratio is set according to the depth level of the defect location;

[0081] The defect sensitivity calculation unit is used to calculate the influence coefficient of the defect relative to the target performance based on the identified defect type, the location of the defect in the structure and the corresponding stress area.

[0082] In this embodiment, the modality adaptive fusion unit divides the target structure into different levels based on the structural depth of the defect area and adjusts the fusion weights of the image recognition model and the signal recognition model accordingly. For example, shallow defects are primarily image-recognized, while deeper defects are weighted more heavily towards signal recognition, thereby improving recognition accuracy and adaptability.

[0083] The defect sensitivity calculation unit assesses the impact of defects on target performance by combining defect type, location, and corresponding stress area. The impact coefficient, derived from historical data or simulation analysis, serves as an important input to the quality assessment module, supporting a quantitative assessment of the defect's impact.

[0084] This module realizes multi-source data fusion and structural correlation recognition, enhances the reliability of defect detection, and provides precise support for subsequent quality assessment and feedback control.

[0085] (4) Quality Assessment Module

[0086] It is used to integrate the design parameters, process data and defect information, build a scoring model, and generate quantifiable ring forging quality scoring results;

[0087] In a feasible implementation manner, the quality assessment module includes:

[0088] The scoring factor contribution calculation unit is used to calculate the dynamic contribution value of each scoring factor based on multiple quality index inputs and construct a scoring contribution vector. , the expression is:

[0089] ;

[0090] in: For the The score of each quality indicator; Indicates the basic weight of the corresponding quality indicator; The response coefficient indicates the degree to which the indicator is affected by the disturbance; is the sensitivity adjustment parameter, which is used to control the contribution strength of the disturbance term to the score; For the The weight coefficient of a quality indicator indicates the importance of the quality indicator in the overall score. The proportion can be determined through expert experience or historical data analysis. The value range is [0,1], and the sum after normalization is 1; For the The scoring input value of each quality indicator represents the actual measured or calculated technical performance, such as yield strength compliance rate, elastic modulus deviation rate, and defect area ratio; For the The disturbance response coefficient of a quality indicator is used to reflect the sensitivity or variability of the indicator under the current manufacturing disturbance conditions (such as temperature fluctuation, stress concentration or forming fluctuation). It is usually obtained through sensor data analysis or experimental simulation; Enter the number of items for scoring; For the The contribution value of each quality indicator in the total score.

[0091] In a feasible implementation manner, the quality assessment module further includes:

[0092] Scoring function construction unit, used to select the corresponding scoring function based on the type of each quality indicator , and calculate the total score according to the following formula , the formula is:

[0093] ;

[0094] in: For the Item Input The scoring mapping function used is classified and set according to whether the indicator belongs to the structural category, defect category or energy consumption category;

[0095] Rating normalization unit, used to normalize the rating value Normalize to the interval [0,100]. Normalization methods include maximum and minimum scaling, standard deviation normalization, or quantile mapping, and output standard scoring results.

[0096] In a feasible implementation manner, the quality assessment module further includes:

[0097] The weight optimization unit is used to train multiple scoring models for different defect level sub-samples, and to calculate the weight vector of each scoring model. Optimize, the optimization goal is to minimize the following loss function, the calculation formula is:

[0098] ;

[0099] in: Indicates in The loss function for training and optimizing the weight vector under the defect samples is to minimize the mean square error between the predicted score and the target score, and introduce a sparsity regularization term to improve the generalization ability of the model; the input is the first The scoring input matrix and target scoring vector of the class training samples are output as the optimal weight vector, so that smallest; For the The total number of class-scored training samples; For the The first defect sample Target score for each sample; Score the corresponding model prediction; For the The weight vector of the class scoring model; is the regularization coefficient, which is used to constrain the sparsity of weights; is a vector of norm;

[0100] Rating feedback unit, used to calculate the rating residual: ; and according to the aggregation trend of the residuals and their source categories, select the corresponding feedback control path between modeling parameters, process parameters or defect detection strategies.

[0101] The scoring function model and weight optimization formula proposed in this paper are not isolated mathematical expressions; rather, they directly guide numerical modeling and parameter feedback in the ring forging scoring and prediction process. All parameters are obtained through actual data collection or structural modeling, and the output results are used to determine the process control path, forming a complete "data acquisition-modeling-scoring-control" closed-loop chain with clear engineering feasibility.

[0102] In this embodiment, the scoring model in the quality assessment module adopts a modular integrated structure. Each scoring sub-model uses actual manufacturing data as input and is integrated and deployed after offline training. Specifically, it includes:

[0103] Scoring function implementation:

[0104] The scoring function is implemented using different mathematical mapping models according to the category to which the quality indicator belongs, specifically including:

[0105] For structural indicators (such as yield strength compliance rate), a linear interval mapping function is used;

[0106] For defect indicators (such as defect area ratio), a logarithmic decay function or a logistic S-type function is used;

[0107] For energy consumption indicators (such as unit forming energy consumption), the piecewise minimum function is used.

[0108] The above function forms are determined by fitting manufacturing history samples and verified by minimum mean square error, and have engineering adaptability and repeatability.

[0109] Scoring contribution value and weight vector training process:

[0110] The scoring factor contribution values ​​and multi-submodel weight vectors are trained using supervised learning. The training samples are historical real-world manufacturing data from the modeling process, covering various defect levels and process disturbance states. Minimization of the mean squared residual is used as the objective function, and L1 / L2 regularization terms are introduced to constrain weight sparsity.

[0111] The optimization process uses the Adam adaptive gradient descent algorithm for batch updates. The training results are evaluated on the validation set to evaluate the fitting accuracy, residual distribution and robustness, and then the model solidification parameters are determined.

[0112] Physical constraint embedding mechanism:

[0113] During the scoring calculation and optimization process, to prevent the purely data-driven model from deviating from physical reality, the system introduces the following physical boundary constraints during the model training phase:

[0114] Strength ratings are limited to the permitted fluctuation range of national material standards;

[0115] Defect scoring must meet the coupling weight adjustment logic with key structural areas;

[0116] Energy consumption indicators must not exceed the theoretical minimum value of the equipment and the lower limit of operational safety.

[0117] These constraints are embedded into the model objective function in the form of soft constraint functions or used as pruning conditions to control the effectiveness of the optimization results.

[0118] In this embodiment of the present invention, the quality assessment module constructs a multi-level scoring system comprising disturbance response factors, indicator weights, and scoring functions. This not only achieves dynamic quantification of multiple quality indicators but also reflects the impact of the manufacturing process on the stability of each indicator. Unlike traditional quality assessment methods that rely on single process parameters or statistical thresholds, the scoring factor contribution calculation mechanism in this embodiment can finely distinguish the proportion of each quality indicator's contribution to score deviation, addressing the shortcomings of existing technologies, such as the lack of structured modeling support for scoring and the inability to quantify contribution relationships.

[0119] By introducing a scoring function construction unit, the system can select corresponding function forms based on the different attributes of the indicators (structural, defect, or energy consumption), making the scoring system both highly discriminatory and sensitive, significantly improving the scoring model's ability to reflect actual performance differences. A normalization unit maps the score values ​​to the range [0,100], facilitating cross-indicator and cross-process comparisons and analysis, and enhancing the model's intuitiveness and versatility for engineering decision-making.

[0120] Furthermore, the weight optimization unit implemented in this embodiment trains multiple scoring sub-models for sub-samples with varying defect levels. It adaptively adjusts the weight vectors of each model by minimizing the loss function, effectively reducing the residual error between the predicted and target scores. Supported by weight sparsity control and a residual trend aggregation feedback mechanism, the scoring results automatically drive the model to adjust the scoring path, improving overall scoring accuracy and the precision of feedback decisions.

[0121] Compared with the existing scoring system that relies solely on expert experience to set weights, has a fixed scoring function, and lacks a reversible correction mechanism, the present invention constructs an evolvable and adaptive quality assessment architecture through parameter driving, data optimization and error feedback, which has higher model adaptability, process responsiveness and quality control value.

[0122] (5) Feedback control module

[0123] When the quality score result does not meet the set requirements, based on the contribution factors and defect sensitivity in the scoring model, determine the main cause of the score deviation and generate feedback instructions, which are used to correct design parameters or adjust manufacturing process parameters and send them to the modeling module or process data acquisition module respectively;

[0124] In a feasible implementation, the feedback control module includes:

[0125] The path selection unit is used to determine the dominant factor that causes the score to drop based on the score input item with the largest proportion in the score factor contribution vector when the quality score value does not meet the preset score threshold, and select the execution order of the feedback control path based on the category of the dominant factor, including:

[0126] If the dominant factor belongs to a modeling parameter item, a feedback instruction for updating the structural parameters or performance model parameters in the modeling module is generated;

[0127] If the dominant factor belongs to the process deviation item, a feedback instruction for adjusting the manufacturing process parameter setting is generated;

[0128] If the dominant factor belongs to the defect detection item, a control instruction for adjusting the quality inspection frequency or inspection conditions is generated;

[0129] The path updating unit is used to record the residual change between the score prediction value and the actual score value after the feedback instruction is executed, and to update the priority selection rule of the feedback path based on the residual change.

[0130] In this embodiment, the path selection unit identifies the dominant factor causing the score deviation based on the contribution factor vector when the score result falls short of the target, and selects the corresponding feedback path based on the category. If it is a modeling parameter item, instructions are generated to adjust the structural design or model parameters. If it is a process deviation item, feedback is sent to the process acquisition module to adjust the manufacturing parameters. If it is a defect detection error, the detection frequency or identification threshold is updated to ensure accurate response to the problem.

[0131] The path update unit records the changes in the score residuals before and after the feedback execution, and dynamically adjusts the priority of the subsequent feedback path according to the residual convergence, forming a closed-loop control strategy for system adaptive optimization.

[0132] Compared with traditional fixed feedback logic, this embodiment is based on a scoring-driven judgment mechanism, combined with dynamic path management, which makes feedback control targeted, flexible and evolvable, significantly improving the system's response efficiency and control accuracy to quality fluctuations.

[0133] (6) 5G communication module

[0134] It is used to establish a high-bandwidth, low-latency data communication connection between the above modules, and realize the real-time transmission and remote linkage processing of design parameters, process data, defect information, quality scoring results and feedback instructions.

[0135] In a feasible implementation, the 5G communication module includes:

[0136] A real-time scheduling strategy unit is used to dynamically adjust the allocation strategy of communication resources based on the priority order of task types and the real-time load status of the current 5G communication link to achieve priority transmission of critical task data. The task data includes design parameters, process data, defect identification information, quality scoring results, and feedback control instruction content;

[0137] The data structure control unit is used to build a unified message structure for different types of data transmission tasks, including adding data source identification, timestamp field, compression flag, retransmission strategy flag and cyclic redundancy check code, to improve the error detection capability and repeated transmission control capability of data packets during transmission, and to achieve reliability and timeliness of remote data interaction.

[0138] In this embodiment, the real-time scheduling strategy unit dynamically allocates communication resources based on the importance of each module's tasks (e.g., quality feedback is more important than ordinary data collection) and the current bandwidth utilization status of the 5G link, thereby enabling priority scheduling and stable transmission of key data such as design parameters and quality scoring results, and ensuring the real-time and continuity of key control data in the manufacturing process.

[0139] The data structure control unit uniformly constructs standardized message structures for different types of data transmission tasks (such as sensor data, scoring results, and control instructions). By adding fields such as data source identification, timestamp, compression flag, and cyclic redundancy check code, it improves the integrity verification and anti-interference capabilities of data packets during transmission. It also supports retransmission policy control to prevent network jitter or packet loss from disrupting the manufacturing process.

[0140] Compared with low-speed link methods such as serial communication or Wi-Fi used in existing industrial sites, the high-bandwidth, low-latency communication architecture built based on 5G communication in this embodiment can achieve millisecond-level response of module-level data and remote intelligent linkage, providing a solid underlying guarantee for the system's real-time feedback control and distributed decision-making, reflecting a substantial improvement in the communication capabilities of the manufacturing system.

[0141] In summary, the present invention establishes a highly efficient collaborative mechanism among design parameters, process data, defect identification, quality assessment, and feedback control by constructing a three-dimensional design and quality management system for wind power ring forgings based on 5G communications, thus achieving closed-loop control of the entire process from modeling-driven to scoring feedback. The system boasts several technical advantages, including precise data acquisition, intelligent evaluation models, adaptive feedback paths, and high-speed communication response. It overcomes technical bottlenecks in existing ring forging manufacturing, such as the separation of design and quality control, detection lag, and slow response. This invention is particularly suitable for large-scale wind power structural parts with high requirements for manufacturing consistency and real-time performance.

[0142] In the description of this specification, the reference terms "one embodiment," "some embodiments," "example," "specific example," or "some examples" mean that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials, or characteristics described may be combined in any appropriate manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification, as well as features of different embodiments or examples, unless they are mutually inconsistent.

[0143] Any process or method description in the flowchart or otherwise described herein can be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process. The scope of the preferred embodiments of the present application includes additional implementations in which the functions may be performed in a different order than shown or discussed, including in a substantially simultaneous manner or in a reverse order depending on the functions involved.

[0144] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any person skilled in the art can easily conceive of various modifications or substitutions within the technical scope disclosed in this application, and such modifications or substitutions should be included within the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A three-dimensional design and quality management system for wind power ring forgings based on 5G communication, characterized by: include: A modeling module is used to build a parametric 3D design model based on the structural requirements and performance specifications of wind turbine ring forgings, and output design parameters that can be used for process evaluation and defect identification; A process data acquisition module is used to collect process data corresponding to the key structural areas involved in the design parameters in real time during the ring forging manufacturing process. The process data includes temperature, pressure and vibration information to support defect identification and quality assessment processes; A defect recognition module is configured to obtain defect information based on the process data and the image or signal data obtained from post-manufacturing inspection by fusing and analyzing the process data, wherein the defect information includes the location, size, and type of the defect; A quality assessment module is used to integrate the design parameters, process data and defect information, build a scoring model, and generate quantifiable ring forging quality scoring results; The scoring model in the quality assessment module adopts a modular integrated structure. Each scoring sub-model uses actual manufacturing data as input and is integrated and deployed after offline training, including: Scoring function implementation method: The scoring function is implemented using different mathematical mapping models according to the category to which the quality indicator belongs, specifically including: For structural indicators, a linear interval mapping function is used; For defect indicators, a logarithmic decay function or a logistic S-type function is used; For energy consumption indicators, the piecewise minimum function is used; A feedback control module is used to determine the main cause of the score deviation and generate feedback instructions based on the contribution factors and defect sensitivity in the scoring model when the quality score result does not meet the set requirements. The feedback instructions are used to correct design parameters or adjust manufacturing process parameters and are sent to the modeling module or process data acquisition module respectively; The 5G communication module is used to establish a high-bandwidth, low-latency data communication connection between the above modules, realizing real-time transmission and remote linkage processing of design parameters, process data, defect information, quality scoring results and feedback instructions.

2. The 5G communication-based three-dimensional design and quality management system for wind power ring forgings according to claim 1 is characterized in that: The modeling module includes: The model feedback correction unit is used to construct a loss function with the scoring error as the target and update the prediction model parameters in the modeling module when there is a difference between the quality scoring result and the predicted performance value of the modeling module; The multi-performance optimization unit is used to establish a target function group of multiple performance indicators based on yield strength, manufacturing energy consumption and material utilization rate during the modeling stage, and to build a multi-objective optimization model in combination with preset constraints. The multi-objective optimization model is solved using a non-dominated sorting genetic algorithm and outputs a design parameter solution set, which is used by the modeling module to generate design parameters.

3. The 5G communication-based three-dimensional design and quality management system for wind power ring forgings according to claim 2 is characterized in that: The multi-objective optimization model constructed in the modeling module includes the following construction process: Establish an objective function group with yield strength, material utilization rate and manufacturing energy consumption as multiple performance objective functions, where each objective function establishes a function expression based on the actual ring forging structural parameters; In the process of constructing the objective function, structural constraints are introduced, including the lower limit of the material safety factor, the extreme range of geometric dimensions, and the heat treatment process window, to form a feasible solution space with constraint boundaries; Based on the scoring function structure provided by the scoring model, a scoring prediction value calculation module is introduced in the objective function solution stage to pre-estimate the scoring of each set of candidate solutions for design parameters. The scoring results are embedded in the objective function set as auxiliary criteria of the optimization model to guide the solution set to converge in a direction that takes into account both performance and scoring expectations.

4. The 5G communication-based three-dimensional design and quality management system for wind power ring forgings according to claim 1 is characterized in that: The process data acquisition module includes: The sampling adaptive unit is used to dynamically adjust the sensor layout density and sampling frequency according to the key structural areas involved in the design parameters output by the modeling module; Anomaly detection unit, which is used to monitor real-time process data based on statistical control methods, identify abnormal changes or drift trends in the data, and activate data redundancy channels when an abnormal state is detected; The credibility evaluation unit is used to perform credibility scoring on the process data collection results and mark the corresponding data as a low credibility segment when the score is lower than a preset threshold.

5. The 3D design and quality management system for wind power ring forgings based on 5G communication according to claim 1 is characterized in that: The defect identification module includes: The modal adaptation fusion unit is used to adjust the fusion ratio of the image recognition model and the signal recognition model according to the structural depth of the area corresponding to the defect information to generate a defect recognition result. The fusion ratio is set according to the depth level of the defect location; The defect sensitivity calculation unit is used to calculate the influence coefficient of the defect relative to the target performance based on the identified defect type, the location of the defect in the structure and the corresponding stress area.

6. The 5G communication-based three-dimensional design and quality management system for wind power ring forgings according to claim 1 is characterized in that: The quality assessment module includes: The scoring factor contribution calculation unit is used to calculate the scoring contribution value of each scoring factor based on multiple quality index inputs. , the expression is: ； in: For the The score of each quality indicator; Indicates the basic weight of the corresponding quality indicator; The response coefficient indicates the degree to which the quality index is affected by the disturbance; is the sensitivity adjustment parameter; For the The weight coefficient of each quality indicator; For the The scoring input values ​​of the quality indicators; For the Disturbance response coefficient of each quality indicator; Enter the number of items for scoring; For the The contribution value of each quality indicator in the total score.

7. The 5G communication-based three-dimensional design and quality management system for wind power ring forgings according to claim 6 is characterized in that: The quality assessment module also includes: Scoring function construction unit, used to select the corresponding scoring function based on the type of each quality indicator , and calculate the total score according to the following formula , the formula is: ； in: For the Item Input The scoring mapping function used is classified and set according to whether the indicator belongs to the structural category, defect category or energy consumption category; Rating normalization unit, used to normalize the rating value Normalize to the interval [0,100]. Normalization methods include scaling, standard deviation normalization, or quantile mapping, and output standard scoring results.

8. The 5G communication-based three-dimensional design and quality management system for wind power generation ring forgings according to claim 7 is characterized in that: The quality assessment module also includes: The weight optimization unit is used to train multiple scoring models for different defect level sub-samples, and to calculate the weight vector of each scoring model. Optimize, the optimization goal is to minimize the following loss function, the calculation formula is: ； in: Indicates in Loss function for training and optimizing weight vectors under defect-like samples; For the The total number of class-scored training samples; For the The first defect sample target score of each training sample; Score model predictions; For the The weight vector of the class scoring model; is the regularization coefficient; is a vector of norm; Rating feedback unit, used to calculate the rating residual: ; and according to the aggregation trend of the score residuals and their source categories, the corresponding feedback control path is selected between the modeling parameters, process parameters or defect detection strategies.

9. The 5G communication-based three-dimensional design and quality management system for wind power ring forgings according to claim 1 is characterized in that: The feedback control module includes: A path selection unit is configured to determine the dominant factor causing the quality score to decline based on the score input item with the largest proportion in the score factor contribution vector when the quality score value does not meet the preset score threshold, and select the execution order of the feedback control path based on the category of the dominant factor, including: If the dominant factor belongs to a modeling parameter item, a feedback instruction for updating the structural parameters or performance model parameters in the modeling module is generated; If the dominant factor belongs to the process deviation item, a feedback instruction for adjusting the manufacturing process parameter setting is generated; If the dominant factor belongs to the defect detection item, a control instruction for adjusting the quality inspection frequency or inspection conditions is generated; The path updating unit is used to record the residual change between the score prediction value and the actual score value after the feedback instruction is executed, and to update the priority selection rules of the feedback control path based on the residual change.

10. The 5G communication-based wind power generation ring forging three-dimensional design and quality management system according to claim 1 is characterized in that: The 5G communication module includes: A real-time scheduling strategy unit is used to dynamically adjust the allocation strategy of communication resources based on the priority order of task types and the real-time load status of the current 5G communication link. Task data includes design parameters, process data, defect identification information, quality scoring results, and feedback control instruction content; The data structure control unit is used to build a unified message structure for different types of data transmission tasks, including adding data source identification, timestamp field, compression flag, retransmission strategy flag and cyclic redundancy check code, to improve the error detection capability and repeated transmission control capability of data packets during transmission, and to achieve reliability and timeliness of remote data interaction.

Images