Intelligent pricing method and system for aviation equipment and device, and medium

Abstract

The invention discloses an aviation equipment intelligent pricing method and system and a medium, and the method comprises the steps: carrying out the fusion processing and feature extraction of multi-source heterogeneous data, such as aviation material internal transaction data, external market information data, supply chain state data, and regulation event text data, and generating a standardized feature vector; utilizing a natural language model to extract structured event features from the law and regulation event text data; inputting the standardized feature vector into a pre-training price prediction model adopting a mixed architecture of an integrated learning algorithm and a time sequence deep learning algorithm, and outputting a reference price prediction value and a price prediction interval; according to a pricing strategy instruction selected by a user, a final pricing result is generated through a strategy mapping module, and an interpretable price cause analysis report is output, so that the accuracy and adaptability of aerial material pricing are effectively improved, the dynamic response of various pricing strategies is supported, the transparency and interpretability of a decision-making process are enhanced, and the user experience is improved. And scientific and reliable intelligent decision support is provided for aviation equipment pricing.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and specifically to a method, system, and medium for intelligent pricing of aviation equipment and devices. Background Technology

[0002] Pricing management of aviation equipment is a crucial link in aviation operations and the supply chain system, directly impacting the efficiency of equipment turnover and the realization of asset value. Currently, the pricing methods commonly used in the industry rely primarily on historical transaction data, human experience, or simple statistical analysis based on single market conditions, lacking the ability to systematically integrate multi-source information and dynamically respond. Furthermore, due to the complex environment of aviation equipment transactions and the dispersed and diverse forms of data sources, traditional methods struggle to effectively analyze and integrate unstructured information, limiting the timeliness and accuracy of price assessments. In addition, existing pricing mechanisms typically fail to consider the strategic adaptability under different operational objectives of enterprises and cannot provide pricing recommendations with credible explanatory evidence, resulting in a certain degree of lag and uncertainty in aviation equipment management when responding to market changes and internal decision-making needs. Therefore, how to build a more intelligent, reliable, and interpretable pricing support method in the complex and ever-changing aviation equipment transaction environment has become a core issue that urgently needs to be addressed in this field. Summary of the Invention

[0003] In view of the above problems, the present invention provides a method, system and medium for intelligent pricing of aviation equipment and components. By integrating multi-source heterogeneous data and introducing a strategy adaptive mechanism, dynamic and interpretable intelligent pricing is achieved, so as to solve the problems of insufficient data utilization and lack of strategy flexibility in the process of aviation material pricing.

[0004] To achieve the above objectives, in a first aspect, this application provides a smart pricing method for aviation equipment and components, comprising:

[0005] Collect multi-source heterogeneous data, including internal transaction data of aviation materials, external market information data, supply chain status data, and regulatory event text data;

[0006] Multi-source heterogeneous data are fused and processed, and features are extracted to generate standardized feature vectors for model input. Feature extraction includes extracting structured event features from regulatory event text data based on natural language models.

[0007] The standardized feature vectors are input into the pre-trained price prediction model to output the benchmark price prediction value and the corresponding price prediction range. The price prediction model adopts a hybrid architecture that combines ensemble learning algorithm and time series deep learning algorithm.

[0008] Receive pricing strategy instructions selected by the user, including profit maximization strategy or rapid turnover strategy;

[0009] Based on the pricing strategy instructions, the benchmark price forecast, and the price forecast range, the final pricing result is generated through the strategy mapping module.

[0010] Output the final pricing results and an interpretable price cause analysis report.

[0011] Furthermore, multi-source heterogeneous data are fused and feature extracted to generate standardized feature vectors for model input, including:

[0012] Data cleaning and standardization are performed on internal aviation material transaction data to generate a unified format of historical transaction feature sequences.

[0013] Real-time collection and aggregation of external market data to generate dynamic market characteristic indicators;

[0014] Multi-source verification and status coding are performed on supply chain status data to generate supply chain stability feature vectors;

[0015] Entity identification and impact analysis are performed on textual data of regulatory events using natural language models to extract structured event features;

[0016] By fusing historical transaction feature sequences, market dynamic feature indicators, supply chain stability feature vectors, and structured event features into a multi-dimensional feature vector, a standardized feature vector is generated.

[0017] Furthermore, natural language models are used to perform entity recognition and impact analysis on the textual data of regulatory events, extracting structured event features, including:

[0018] The text data of legal events is preprocessed and segmented to generate a sequence of texts to be analyzed.

[0019] The text sequence to be analyzed is input into a pre-trained named entity recognition model to extract key entity information related to aerospace materials and components. The key entity information includes the affected part number, applicable aircraft model range, and compliance requirement type.

[0020] Based on the attention mechanism, the influence of key entity information is quantitatively analyzed to generate event influence weight coefficients, including:

[0021] Construct an attention scoring function to calculate the correlation score between each entity in the key entity information and the reference events in the historical event database;

[0022] The relevance score is converted into attention distribution weights using the softmax normalization function;

[0023] The entity feature vectors are weighted and fused according to the attention distribution weights to generate context-aware entity representations;

[0024] A multilayer perceptron is used to perform a nonlinear transformation on the context-aware entity representation, and a preliminary impact strength estimate is output.

[0025] By combining the event timeliness decay factor to correct the initial impact intensity estimate in the time dimension, the final event impact weight coefficient is generated. The event timeliness decay factor is negatively correlated with the time interval from the event occurrence time to the current time.

[0026] Based on key entity information and event impact weighting coefficients, a time-sensitive event feature matrix is ​​constructed.

[0027] The event feature matrix is ​​matched with the historical event database for similarity to generate the final structured event features.

[0028] Furthermore, the price prediction model is constructed through the following steps:

[0029] An initial training sample set is constructed based on historical aviation material transaction data. The initial training sample set includes historical standardized feature vectors and corresponding historical actual transaction price labels.

[0030] A stratified sampling strategy was adopted to divide the initial training sample set into a training subset and a validation subset. The stratified sampling strategy allocated samples according to the type of aviation materials and components and the frequency of transactions.

[0031] A benchmark price prediction module is constructed using an ensemble learning algorithm. The benchmark price prediction module uses a gradient boosting decision tree architecture to rank feature importance and fit nonlinear relationships.

[0032] A price fluctuation prediction module is constructed using a time-series deep learning algorithm. The price fluctuation prediction module uses an attention mechanism to capture long-term dependencies and cyclical change patterns.

[0033] The benchmark price prediction module and the price fluctuation prediction module are weighted and fused to generate a hybrid prediction model;

[0034] The parameters of the hybrid prediction model are optimized using a training subset, and the prediction performance is verified using a validation subset, resulting in a fully trained price prediction model.

[0035] Furthermore, the standardized feature vectors are input into the pre-trained price prediction model to output the benchmark price prediction value and the corresponding price prediction range, including:

[0036] The benchmark price prediction module performs preliminary price prediction on the standardized feature vector to generate an initial benchmark price prediction value.

[0037] The price fluctuation prediction module analyzes the price fluctuation range of the standardized feature vector to generate price fluctuation interval parameters.

[0038] The initial benchmark price forecast and the price fluctuation range parameter are jointly calibrated based on the quantile regression algorithm to generate the calibrated benchmark price forecast.

[0039] The uncertainty of the calibrated benchmark price forecast is quantified using confidence intervals to generate price forecast intervals with confidence level indicators.

[0040] The calibrated benchmark price forecast and the price forecast range are output.

[0041] Furthermore, when the pricing strategy instruction is a profit maximization strategy, the final pricing result is generated through the strategy mapping module based on the pricing strategy instruction, the benchmark price forecast, and the price forecast range. This includes:

[0042] Obtain the upper boundary value of the price prediction range as the benchmark reference price;

[0043] The benchmark reference price is dynamically adjusted based on real-time market competition environment data to generate a competition environment adjustment coefficient.

[0044] By combining inventory turnover rate and supply and demand indicators, the profit maximization weighting parameters are calculated.

[0045] The benchmark reference price is weighted and optimized based on the competitive environment adjustment coefficient and the profit maximization weight parameter.

[0046] Output the optimized final pricing result.

[0047] Furthermore, when the pricing strategy instruction is a rapid turnover strategy, the final pricing result is generated through the strategy mapping module based on the pricing strategy instruction, the benchmark price forecast value, and the price forecast range, including:

[0048] Obtain the lower boundary value of the price forecast range as the benchmark reference price;

[0049] An inventory pressure coefficient is generated based on inventory backlog data, and the inventory pressure coefficient is positively correlated with inventory turnover days.

[0050] The price elasticity adjustment parameter is calculated based on the market supply and demand tight balance indicator;

[0051] The inventory pressure coefficient and the price elasticity adjustment parameter are combined to generate the fast turnover discount coefficient.

[0052] The benchmark reference price is discounted based on the fast turnover discount factor to generate the final pricing result.

[0053] Furthermore, the output includes the final pricing results and an interpretable analysis report on the causes of the prices, including:

[0054] Key feature factors influencing this pricing were extracted based on a feature importance analysis algorithm.

[0055] Attribution analysis is performed on key characteristic factors to generate quantitative indicators of the degree of influence of each key characteristic factor on the final pricing result;

[0056] By comparing and analyzing quantitative indicators of the degree of impact with historical aviation material transaction data, an early warning of abnormal price fluctuations is generated.

[0057] Based on key characteristic factors, quantitative indicators of impact, and early warnings of abnormal price fluctuations, an initial report is generated according to the structured report template.

[0058] The final pricing results are integrated with the initial report to output a visual report document containing price values ​​and causal analysis.

[0059] In a second aspect, the present invention also provides an intelligent pricing system for aviation equipment and components, applicable to the method described in the first aspect. The system includes: a multi-source data acquisition module, a data fusion and feature extraction module, a price prediction module, a strategy instruction receiving module, a pricing decision module, and a result output module. The multi-source data acquisition module is configured to acquire multi-source heterogeneous data, including internal aviation material transaction data, external market data, supply chain status data, and regulatory event text data. The data fusion and feature extraction module is configured to perform fusion processing and feature extraction on the multi-source heterogeneous data, generating standardized feature vectors for model input. Feature extraction includes extracting features from regulatory event texts based on a natural language model. This data extracts structured event features; the price prediction module is configured to input standardized feature vectors into a pre-trained price prediction model to output a benchmark price prediction value and the corresponding price prediction range. The price prediction model adopts a hybrid architecture combining ensemble learning algorithms and temporal deep learning algorithms; the strategy instruction receiving module is configured to receive pricing strategy instructions selected by the user, including profit maximization strategies or rapid turnover strategies; the pricing decision module is configured to generate the final pricing result through the strategy mapping module based on the pricing strategy instructions, benchmark price prediction value, and price prediction range; the result output module is configured to output the final pricing result and an interpretable price cause analysis report.

[0060] In a third aspect, the present invention also provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the method described in the first aspect.

[0061] Unlike existing technologies, the above-mentioned technical solution collects heterogeneous data from multiple sources, including internal aircraft material transaction data, external market data, supply chain status data, and regulatory event text data. This data is then fused and processed for feature extraction to generate standardized feature vectors. A natural language model is used to extract structured event features from the regulatory event text data. These standardized feature vectors are then input into a pre-trained price prediction model employing a hybrid architecture of ensemble learning and temporal deep learning algorithms. The model outputs a benchmark price prediction value and a price prediction range. Based on the user's selected pricing strategy, a strategy mapping module generates the final pricing result and outputs an interpretable price causal analysis report. This invention effectively improves the accuracy and adaptability of aircraft material pricing, supports dynamic responses to various pricing strategies, enhances the transparency and interpretability of the decision-making process, and provides scientific and reliable intelligent decision support for aircraft equipment pricing.

[0062] The above description of the invention is merely an overview of the technical solution of this application. In order to enable those skilled in the art to better understand the technical solution of this application and to implement it based on the description and drawings, and to make the above-mentioned objectives and other objectives, features and advantages of this application easier to understand, the following description is provided in conjunction with the specific embodiments and drawings of this application. Attached Figure Description

[0063] The accompanying drawings are only used to illustrate the principles, implementation methods, applications, features, and effects of specific embodiments of the present invention and other related contents, and should not be considered as limitations on this application.

[0064] In the accompanying drawings of the instruction manual:

[0065] Figure 1 This is a schematic diagram illustrating steps S101 to S106 of the method described in the specific implementation embodiment;

[0066] Figure 2 This is a schematic diagram illustrating steps S201 to S205 of the method described in a specific implementation.

[0067] Figure 3 This is a schematic diagram illustrating steps S301 to S305 of the method described in a specific implementation.

[0068] Figure 4 This is a schematic diagram illustrating steps S401 to S406 of the method described in a specific embodiment;

[0069] Figure 5 This is a schematic diagram of the intelligent pricing system described in a specific implementation.

[0070] The reference numerals used in the above figures are explained as follows:

[0071] 1. Intelligent pricing system;

[0072] 11. Multi-source data acquisition module;

[0073] 12. Data fusion and feature extraction module;

[0074] 13. Price prediction module;

[0075] 14. Policy instruction receiving module;

[0076] 15. Pricing Decision Module;

[0077] 16. Results Output Module. Detailed Implementation

[0078] To illustrate the possible application scenarios, technical principles, implementable specific solutions, and achievable objectives and effects of this application in detail, the following description, in conjunction with the listed specific embodiments and accompanying drawings, provides a detailed explanation. The embodiments described herein are merely illustrative of the technical solutions of this application and are therefore intended to limit the scope of protection of this application.

[0079] In this document, the term "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The term "embodiment" appearing in various places throughout the specification does not necessarily refer to the same embodiment, nor does it specifically limit its independence or connection with other embodiments. In principle, in this application, as long as there are no technical contradictions or conflicts, the technical features mentioned in each embodiment can be combined in any way to form corresponding implementable technical solutions.

[0080] Unless otherwise defined, the technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the use of related terms herein is merely for the purpose of describing particular embodiments and is not intended to limit this application.

[0081] In the description of this application, the term "and / or" is used to describe the logical relationship between objects, indicating that three relationships can exist. For example, A and / or B means: A exists, B exists, and A and B exist simultaneously. Additionally, the character " / " in this document generally indicates that the preceding and following objects have an "or" logical relationship.

[0082] In this application, terms such as “first” and “second” are used only to distinguish one entity or operation from another, and do not necessarily require or imply any actual quantity, hierarchy or order relationship between these entities or operations.

[0083] Without further limitations, the use of terms such as “comprising,” “including,” “having,” or other similar open-ended expressions in this application is intended to cover non-exclusive inclusion, which does not exclude the presence of additional elements in a process, method, or product that includes the stated elements, such that a process, method, or product that includes a list of elements may include not only those defined elements but also other elements not expressly listed, or elements inherent to such a process, method, or product.

[0084] Similar to the understanding in the Examination Guidelines, in this application, expressions such as "greater than," "less than," and "exceeding" are understood to exclude the stated number; expressions such as "above," "below," and "within" are understood to include the stated number. Furthermore, in the description of the embodiments in this application, "multiple" means two or more (including two), and similar expressions related to "multiple" are also understood in this way, such as "multiple groups" and "multiple times," unless otherwise explicitly specified.

[0085] In the description of the embodiments of this application, the space-related expressions used, such as "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "vertical," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential," indicate the orientation or positional relationship based on the orientation or positional relationship shown in the specific embodiments or drawings. They are only for the purpose of describing the specific embodiments of this application or for the reader's understanding, and do not indicate or imply that the device or component referred to must have a specific position, a specific orientation, or be constructed or operated in a specific orientation. Therefore, they should not be construed as limitations on the embodiments of this application.

[0086] The processor described in the embodiments of this application can be implemented by hardware, firmware, software, or a combination thereof. It can be a circuit, one or more of an application-specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field-programmable gate array (FPGA), a central processing unit (CPU), a controller, a microcontroller, or a microprocessor. It also includes other physical, biological, or chemical structures that can implement the same or equivalent functions as the processors listed above, such as biological neurons, quantum computing units, DNA computing units, etc., so that the processor can execute some or all of the steps in the computer program or method involved in the various embodiments of this application, or any combination of the steps mentioned therein.

[0087] Please see Figure 1 In a first aspect, this embodiment provides a smart pricing method for aviation equipment and components, including:

[0088] S101. Collect multi-source heterogeneous data, including internal transaction data of aviation materials, external market data, supply chain status data, and regulatory event text data;

[0089] S102. Perform fusion processing and feature extraction on multi-source heterogeneous data to generate standardized feature vectors for model input. Feature extraction includes extracting structured event features from regulatory event text data based on natural language models.

[0090] S103. Input the standardized feature vector into the pre-trained price prediction model to output the benchmark price prediction value and the corresponding price prediction range. The price prediction model adopts a hybrid architecture that combines ensemble learning algorithm and time series deep learning algorithm.

[0091] S104. Receive the pricing strategy instruction selected by the user. The pricing strategy instruction includes a profit maximization strategy or a rapid turnover strategy.

[0092] S105. Based on the pricing strategy instructions, the benchmark price forecast value, and the price forecast range, the final pricing result is generated through the strategy mapping module.

[0093] S106. Output the final pricing results and an interpretable price cause analysis report.

[0094] In step S101, the multi-source heterogeneous data refers to aviation material-related data obtained from different sources and formats. This includes internal aviation material transaction data such as historical procurement records, sales prices, and inventory turnover information; external market data covering dynamic information such as market price fluctuations and changes in supply and demand for similar components; supply chain status data involving real-time indicators such as supplier delivery cycles, logistics status, and capacity utilization; and regulatory event text data including announcements or news in text form related to policy and regulatory changes and industry standard updates. This data can be collected through data interfaces or web scraping technology to ensure its comprehensiveness and timeliness, providing multi-dimensional input for subsequent analysis.

[0095] In step S102, the multi-source heterogeneous data is cleaned, aligned, and transformed to generate standardized feature vectors. Specifically, internal aircraft material transaction data undergoes missing value processing and normalization to generate historical transaction feature sequences; external market data is aggregated to generate market dynamic indicators; supply chain status data undergoes multi-source verification and encoding to generate stability feature vectors; and regulatory event text data undergoes entity recognition and semantic analysis using a natural language model to extract structured event features, such as the scope of policy impact and timeliness. Finally, standardized feature vectors are generated through feature concatenation or weighted fusion to ensure the consistency and comparability of model inputs.

[0096] In step S103, the pre-trained price prediction model employs a hybrid architecture of ensemble learning and temporal deep learning algorithms to process static and dynamic information in the feature vectors. Preferably, the ensemble learning algorithm uses a gradient boosting decision tree, and the temporal deep learning algorithm uses a long short-term memory network. The ensemble learning algorithm captures the nonlinear relationships between features, while the temporal deep learning algorithm analyzes price trends and cyclical fluctuations. The outputs of both are weighted and fused to generate a benchmark price prediction value and a price prediction range. This range is calculated using quantile regression or confidence intervals, characterizing the uncertainty range of the prediction results and providing a reliable reference for strategy decision-making.

[0097] In step S104, the pricing strategy instruction is the strategy type selected by the user based on business objectives. The profit maximization strategy focuses on increasing the revenue per transaction, while the rapid turnover strategy prioritizes inventory turnover efficiency. Users submit instructions through a graphical interface or API interface, and the system calls the corresponding strategy mapping logic according to the instruction type to ensure that the pricing process is consistent with business objectives.

[0098] In step S105, the strategy mapping module dynamically adjusts the benchmark price forecast and price forecast range according to the pricing strategy instructions. For the profit maximization strategy, the module prioritizes selecting the upper limit of the price range and fine-tunes it based on market competition parameters; for the fast turnover strategy, the module prioritizes selecting the lower limit of the price range and calculates a discount based on the inventory pressure coefficient. The adjustment process is implemented through a weighted algorithm or rule engine to generate the final pricing result, ensuring the flexibility and adaptability of the strategy.

[0099] In step S106, the Explainable Price Causal Analysis Report is generated based on feature importance analysis and attribution calculations. It extracts key factors influencing pricing (such as market supply and demand, policy events, etc.) and quantifies their contribution. The report presents the final pricing results and causal analysis through visual charts and text descriptions, helping users understand the pricing logic and supporting decision optimization.

[0100] This embodiment collects heterogeneous data from multiple sources and performs standardization processing. It then uses a hybrid model combining ensemble learning and temporal deep learning to predict benchmark prices and fluctuation ranges. Finally, it dynamically adjusts the pricing based on the user-selected pricing strategy to generate the final pricing result and outputs an interpretable causal analysis. This method achieves data-driven intelligent pricing, effectively improving the accuracy and adaptability of price predictions. Simultaneously, it enhances decision-making transparency and credibility through visualized reports, providing scientific and efficient pricing support for aircraft material management.

[0101] Please see Figure 2 In some embodiments, multi-source heterogeneous data are fused and feature extracted to generate standardized feature vectors for model input, including:

[0102] S201. Perform data cleaning and standardization on the internal transaction data of aviation materials to generate a historical transaction feature sequence in a unified format;

[0103] S202. Real-time collection and aggregation of external market data to generate dynamic market characteristic indicators;

[0104] S203. Perform multi-source verification and status coding on the supply chain status data to generate a supply chain stability feature vector;

[0105] S204. Entity identification and impact analysis are performed on the text data of regulatory events using natural language models to extract structured event features;

[0106] S205. Multi-dimensional feature fusion is performed on historical transaction feature sequences, market dynamic feature indicators, supply chain stability feature vectors, and structured event features to generate standardized feature vectors.

[0107] In step S201, internal transaction data for aviation materials refers to the enterprise's internal procurement, sales, and inventory records. Data cleaning and standardization processes include removing outliers, filling in missing data, and converting transaction data from different sources into a unified time series format. Preferably, data cleaning employs outlier detection and missing value imputation methods, and standardization is achieved through Z-score normalization or max-min scaling to ensure consistent data scale. Historical transaction feature sequences are generated by extracting key indicators such as transaction prices, transaction volumes, and turnover cycles. These sequences reflect the historical market behavior and value patterns of aviation materials, providing fundamental time series characteristics for price forecasting.

[0108] In step S202, external market data includes competitor pricing, market demand fluctuations, and macroeconomic indicators. This data is collected in real-time via API interfaces or web crawlers. Aggregation calculation refers to weighted averaging or indexing of multi-source market data. Market dynamic characteristic indicators may include price volatility, supply-demand ratio, and trend momentum, used to capture the impact of changes in the external market environment on aircraft material prices.

[0109] In step S203, the supply chain status data involves information such as supplier reliability, logistics timeliness, and inventory levels. Multi-source verification cross-validates the consistency of different data sources, and status coding transforms qualitative indicators (such as delivery delay levels) into numerical features. The supply chain stability feature vector is generated through principal component analysis or weighted fusion and is used to quantify the potential impact of supply chain risks on pricing.

[0110] In step S204, the regulatory event text data includes unstructured texts such as policy documents and industry notices. A natural language model uses a pre-trained BERT or similar architecture for entity recognition and sentiment analysis to extract structured event features such as policy type, scope of impact, and urgency. Impact analysis uses an attention mechanism to calculate the correlation between the event and aviation materials, ensuring that key policy changes are effectively captured and quantified.

[0111] In step S205, multidimensional feature fusion is achieved through feature concatenation, weighted averaging, or a neural network fusion layer, integrating historical transaction feature sequences, market dynamic feature indicators, supply chain stability feature vectors, and structured event features into a unified, standardized feature vector. This vector retains the complementary information from multiple data sources, eliminates dimensional differences, and provides high-quality input for subsequent price prediction models.

[0112] This embodiment transforms multi-source heterogeneous data into standardized feature vectors through systematic data cleaning, feature extraction, and fusion processing, significantly improving data quality and model input consistency, and laying a reliable foundation for accurate price prediction and strategy decision-making.

[0113] Please see Figure 3In some embodiments, the text data of regulatory events is analyzed for entity recognition and impact using natural language models to extract structured event features, including:

[0114] S301. Preprocess and segment the text data of regulatory events to generate a text sequence to be analyzed;

[0115] S302. Input the text sequence to be analyzed into a pre-trained named entity recognition model to extract key entity information involving aviation materials and components. The key entity information includes the affected part number, applicable aircraft model range, and compliance requirement type.

[0116] S303. Based on the attention mechanism, perform quantitative analysis of the impact of key entity information to generate event impact weight coefficients, including:

[0117] Construct an attention scoring function to calculate the correlation score between each entity in the key entity information and the reference events in the historical event database;

[0118] The relevance score is converted into attention distribution weights using the softmax normalization function;

[0119] The entity feature vectors are weighted and fused according to the attention distribution weights to generate context-aware entity representations;

[0120] A multilayer perceptron is used to perform a nonlinear transformation on the context-aware entity representation, and a preliminary impact strength estimate is output.

[0121] By combining the event timeliness decay factor to correct the initial impact intensity estimate in the time dimension, the final event impact weight coefficient is generated. The event timeliness decay factor is negatively correlated with the time interval from the event occurrence time to the current time.

[0122] S304. Construct a time-sensitive event feature matrix based on key entity information and event impact weight coefficients;

[0123] S305. Perform similarity matching between the event feature matrix and the historical event database to generate the final structured event features.

[0124] In step S301, preprocessing and word segmentation refer to standardizing the format of the regulatory event text data, removing irrelevant characters, and segmenting the continuous text into a sequence of lexical units using a word segmentation tool. The text sequence to be analyzed is further optimized through part-of-speech tagging and stop word filtering to ensure that the input quality meets the requirements of the natural language processing model, providing a structured text foundation for subsequent entity recognition.

[0125] In step S302, the pre-trained named entity recognition model uses a Transformer-based model (such as BERT) to identify proprietary entities in the text through sequence labeling technology. Key entity information includes affected part numbers (such as aircraft component numbers), applicable aircraft model ranges (such as aircraft model series), and compliance requirement types (such as mandatory certification or usage restrictions). These entities are extracted and structured by the label sequences output by the model to accurately represent the core elements of regulatory events.

[0126] In step S303, the attention mechanism quantifies the degree of influence by calculating the semantic relevance between entities and historical events. The attention scoring function typically employs dot product or additive attention to calculate the similarity score between the entity vector and reference event vectors in the historical event database. Softmax normalization converts the score into a weight distribution, ensuring the total weights equal to 1. The weighted fusion of context-aware entity representations captures the dependencies between entities. A multilayer perceptron introduces nonlinear transformations to enhance feature representation capabilities. The event timeliness decay factor is calculated based on a time decay function (such as exponential decay) to ensure that recent events receive higher weights. The final generated event influence weight coefficient comprehensively reflects both the semantic importance and temporal relevance of the entity.

[0127] In step S304, the event feature matrix is ​​constructed by combining key entity information with corresponding event impact weight coefficients in rows and columns. Rows in the matrix represent different entities, and columns represent weight coefficients and timeliness annotations. This matrix preserves the multidimensional features of the event in a structured form, facilitating subsequent similarity matching and feature fusion.

[0128] In step S305, similarity matching uses cosine similarity or Euclidean distance to calculate the closeness between the event feature matrix and samples in the historical event database. The most similar historical event features are selected as a reference to generate the final structured event features. This process is achieved through a retrieval-comparison mechanism to ensure that the feature representation of new events conforms to historical patterns while possessing uniqueness.

[0129] This embodiment combines natural language processing with attention mechanisms to achieve accurate parsing and impact quantification of regulatory event texts. The generated structured event features have both semantic accuracy and timeliness sensitivity, providing reliable external event factor inputs for price prediction.

[0130] Please see Figure 4 In some embodiments, the price prediction model is constructed through the following steps:

[0131] S401. Construct an initial training sample set based on historical aviation material transaction data. The initial training sample set includes historical standardized feature vectors and corresponding historical actual transaction price labels.

[0132] S402. The initial training sample set is divided into a training subset and a validation subset using a stratified sampling strategy. The stratified sampling strategy allocates samples according to the type of aviation materials and components and the frequency of transactions.

[0133] S403. A benchmark price prediction module is constructed through an ensemble learning algorithm. The benchmark price prediction module uses a gradient boosting decision tree architecture to rank feature importance and fit nonlinear relationships.

[0134] S404. A price fluctuation prediction module is constructed using a time-series deep learning algorithm. The price fluctuation prediction module uses an attention mechanism to capture long-term dependencies and cyclical change patterns.

[0135] S405. Weighted fusion of the benchmark price prediction module and the price fluctuation prediction module to generate a hybrid prediction model;

[0136] S406. Optimize the parameters of the hybrid prediction model using the training subset and verify the prediction performance using the validation subset to obtain the trained price prediction model.

[0137] In step S401, historical aircraft material transaction data refers to multi-source heterogeneous data that has undergone cleaning and standardization. The initial training sample set is constructed by pairing historical standardized feature vectors with actual transaction price labels at corresponding time points. The actual transaction price labels serve as ground truth labels for supervised learning, ensuring that the model learns accurate price mapping relationships and providing a high-quality labeled data foundation for model training.

[0138] In step S402, the stratified sampling strategy divides the samples according to the type of aircraft parts (such as engine components, avionics equipment, etc.) and the frequency of transactions (such as high-frequency traded parts, low-frequency scarce parts), ensuring that the training subset and the validation subset maintain consistency in the sample distribution of different categories and frequencies. This sampling method avoids data bias and improves the model's generalization ability and validation reliability.

[0139] In step S403, the ensemble learning algorithm is implemented using a gradient boosting decision tree (such as XGBoost or LightGBM). This architecture trains weak classifiers through multiple rounds of iteration and weighted combinations to gradually reduce prediction errors. Feature importance ranking is achieved by calculating the information gain or Gini coefficient during feature splitting, and nonlinear relationship fitting is completed through the combined decision boundary of the tree structure, which can effectively capture the complex mapping relationship between aircraft material prices and multi-source features.

[0140] In step S404, the time-series deep learning algorithm employs an attention-based model (such as a Transformer or LSTM-attention hybrid model). The attention mechanism calculates the weight distribution of historical time steps, highlighting the impact of key time points to capture long-term dependencies and cyclical fluctuations. This module specifically handles trend and seasonal changes in price time series, compensating for the shortcomings of static models in time-series forecasting.

[0141] In step S405, weighted fusion is achieved through learnable weight parameters or static weight allocation (such as dynamic adjustment based on module performance), integrating the point prediction results output by the benchmark price prediction module with the fluctuation range information output by the price fluctuation prediction module. The hybrid prediction model combines the stability of ensemble learning with the dynamic capture capability of temporal deep learning, forming a complementary prediction architecture.

[0142] In step S406, parameter optimization employs a gradient descent algorithm (such as Adam or SGD) to minimize the prediction error loss function. A validation subset is used to monitor model overfitting, and an early stopping strategy is used to determine the optimal training epochs. Prediction performance validation quantifies model accuracy using metrics such as mean absolute error and root mean square error, ensuring that the trained model possesses reliable prediction capabilities.

[0143] This embodiment constructs a price prediction model that combines accuracy, stability, and temporal adaptability through a hybrid architecture design of ensemble learning and temporal deep learning, combined with hierarchical sampling and weighted fusion strategies, providing a core computational foundation for intelligent pricing of aviation materials.

[0144] In some embodiments, a standardized feature vector is input into a pre-trained price prediction model to output a baseline price prediction value and the corresponding price prediction range, including:

[0145] The benchmark price prediction module performs preliminary price prediction on the standardized feature vector to generate an initial benchmark price prediction value.

[0146] The price fluctuation prediction module analyzes the price fluctuation range of the standardized feature vector to generate price fluctuation interval parameters.

[0147] The initial benchmark price forecast and the price fluctuation range parameter are jointly calibrated based on the quantile regression algorithm to generate the calibrated benchmark price forecast.

[0148] The uncertainty of the calibrated benchmark price forecast is quantified using confidence intervals to generate price forecast intervals with confidence level indicators.

[0149] The calibrated benchmark price forecast and the price forecast range are output.

[0150] In this embodiment, the benchmark price prediction module employs ensemble learning methods such as gradient boosting decision trees to perform nonlinear mapping on standardized feature vectors, generating initial benchmark price predictions. These predictions reflect the base price level under ideal market conditions, providing a reference benchmark for subsequent calibration.

[0151] The price fluctuation prediction module uses a time-series deep learning algorithm to analyze historical price fluctuation patterns and market uncertainties, and generates price fluctuation range parameters, including quantitative indicators such as fluctuation amplitude and direction probability, to characterize the possible range of price changes.

[0152] Quantile regression algorithm achieves statistical calibration of the predicted value by jointly optimizing the initial benchmark price forecast and the price fluctuation range parameters. This method minimizes the loss function of different quantiles, making the predicted value more consistent with the actual distribution characteristics and generating a more accurate calibrated benchmark price forecast.

[0153] The confidence interval is calculated using statistical inference methods. Based on the historical prediction error distribution and sample size, the uncertainty range of the calibrated predicted value is determined. Preferably, the confidence level is expressed as a percentage, such as a 95% confidence interval, to provide decision-makers with a quantitative reference for risk.

[0154] The final output of the calibrated benchmark price forecast and price forecast range not only retains the accuracy of the benchmark forecast, but also quantifies market uncertainty through statistical methods, providing a forecast basis that balances accuracy and reliability for subsequent strategy decisions.

[0155] This embodiment achieves statistical optimization and uncertainty quantification of price forecasts by combining quantile regression and confidence interval estimation. This results in forecasts that possess both the accuracy of point estimation and the reliability of interval estimation, significantly improving the scientific nature of pricing decisions and risk control capabilities.

[0156] In some embodiments, when the pricing strategy instruction is a profit maximization strategy, the final pricing result is generated through the strategy mapping module based on the pricing strategy instruction, the benchmark price forecast, and the price forecast range, including:

[0157] Obtain the upper boundary value of the price prediction range as the benchmark reference price;

[0158] The benchmark reference price is dynamically adjusted based on real-time market competition environment data to generate a competition environment adjustment coefficient.

[0159] By combining inventory turnover rate and supply and demand indicators, the profit maximization weighting parameters are calculated.

[0160] The benchmark reference price is weighted and optimized based on the competitive environment adjustment coefficient and the profit maximization weight parameter.

[0161] Output the optimized final pricing result.

[0162] In this embodiment, when the pricing strategy instruction is a profit maximization strategy, the strategy mapping module first obtains the upper boundary value of the price prediction range as a benchmark reference price. This upper boundary value reflects the highest expected price level under favorable market conditions, providing a pricing basis for the profit maximization strategy.

[0163] Real-time market competition environment data includes information such as competitor pricing, market supply and demand dynamics, and promotional activities, which are acquired in real time through a data interface. The competition environment adjustment coefficient is calculated using a logistic function or a linear scaling method. Specifically, the market competition intensity can be mapped to an adjustment range of 0.8-1.2 using the sigmoid function. When market competition is intense, the coefficient tends to approach 0.8, and vice versa, it tends to approach 1.2, thereby enabling dynamic adjustment of the benchmark price according to the market environment.

[0164] The profit maximization weighting parameter is calculated by combining inventory turnover rate and supply-demand relationship indicators. Inventory turnover rate reflects the speed of inventory turnover, while supply-demand relationship indicators characterize the market supply-demand equilibrium. A weighted average method or principal component analysis can be used to integrate these two indicators into a single weighting parameter; a higher value indicates a greater tendency to adopt a high-price strategy.

[0165] The weighted optimization calculation is expressed by formula (1) as follows:

[0166] P = P1 × (1 + α × β);

[0167] In formula (1), P represents the final pricing result when the pricing strategy instruction is a profit-maximizing strategy, P1 represents the benchmark reference price when the pricing strategy instruction is a profit-maximizing strategy, α is the competitive environment adjustment coefficient, and β is the profit-maximizing weight parameter. This formula achieves dual adjustment through the product of coefficients, taking into account both the external competitive environment and the internal inventory and supply and demand situation.

[0168] The optimized final pricing result is output through the strategy mapping module. This result is based on the benchmark reference price and comprehensively optimized for market competition and profit targets, ensuring that market competitiveness is maintained while maximizing profits.

[0169] This embodiment achieves an intelligent pricing strategy based on the upper bound of the price prediction range through a dual adjustment mechanism of competitive environment adjustment coefficient and profit maximization weight parameter. This enables the pricing results to fully utilize favorable market conditions and adapt to the real-time competitive environment, effectively improving the profit level of aviation material sales.

[0170] In some embodiments, when the pricing strategy instruction is a fast turnover strategy, the final pricing result is generated through the strategy mapping module based on the pricing strategy instruction, the benchmark price forecast, and the price forecast range, including:

[0171] Obtain the lower boundary value of the price forecast range as the benchmark reference price;

[0172] An inventory pressure coefficient is generated based on inventory backlog data, and the inventory pressure coefficient is positively correlated with inventory turnover days.

[0173] The price elasticity adjustment parameter is calculated based on the market supply and demand tight balance indicator;

[0174] The inventory pressure coefficient and the price elasticity adjustment parameter are combined to generate the fast turnover discount coefficient.

[0175] The benchmark reference price is discounted based on the fast turnover discount factor to generate the final pricing result.

[0176] In this embodiment, when the pricing strategy instruction is a fast turnover strategy, the strategy mapping module first obtains the lower boundary value of the price prediction range as a benchmark reference price. This lower boundary value reflects the lowest expected price level under unfavorable market conditions, providing a pricing basis for the fast turnover strategy.

[0177] The inventory pressure coefficient is generated based on data on inventory backlog levels. Specifically, it can be calculated by the ratio of inventory turnover days to a preset threshold, using a linear or exponential function to map the inventory turnover days to a coefficient value between 0 and 1. The longer the inventory turnover days, the higher the inventory pressure coefficient, indicating a more severe inventory backlog and requiring greater price discounts to stimulate sales.

[0178] The price elasticity adjustment parameter is calculated based on the market supply and demand balance indicator, which can be derived from parameters such as the supply-demand ratio and the availability of substitutes. Specifically, it can be calculated using the demand price elasticity coefficient or an elasticity function fitted from market research data. The larger the parameter value, the more sensitive the market demand is to price changes, and the more suitable a larger discount range is.

[0179] The composite calculation uses formula (2) to generate the fast turnover discount factor, as follows:

[0180] δ = γ × η;

[0181] In formula (2), δ is the rapid turnover discount coefficient, γ is the inventory pressure coefficient, and η is the price elasticity adjustment parameter. This formula achieves dual adjustment by multiplying the coefficients, taking into account both internal inventory pressure and the elasticity of market demand.

[0182] The discount calculation is expressed by formula (3) as follows:

[0183] P′=P2×(1-δ);

[0184] In formula (3), P′ represents the final pricing result when the pricing strategy instruction is a fast turnover strategy, and P2 represents the benchmark reference price when the pricing strategy instruction is a fast turnover strategy. This calculation method ensures that an appropriate discount is given on the basis of the benchmark price to promote the fast turnover of inventory.

[0185] The final pricing result is output through the strategy mapping module. While ensuring basic returns, reasonable price discounts are used to accelerate inventory turnover and optimize capital utilization efficiency.

[0186] This embodiment achieves an intelligent discount strategy based on the lower bound of the price forecast range through a dual adjustment mechanism of inventory pressure coefficient and price elasticity adjustment parameter. This enables the pricing results to effectively promote inventory turnover while maintaining a reasonable profit level, thereby improving the efficiency and effectiveness of aviation material inventory management.

[0187] In some embodiments, the output includes the final pricing result and an interpretable price cause analysis report, including:

[0188] Key feature factors influencing this pricing were extracted based on a feature importance analysis algorithm.

[0189] Attribution analysis is performed on key characteristic factors to generate quantitative indicators of the degree of influence of each key characteristic factor on the final pricing result;

[0190] By comparing and analyzing quantitative indicators of the degree of impact with historical aviation material transaction data, an early warning of abnormal price fluctuations is generated.

[0191] Based on key characteristic factors, quantitative indicators of impact, and early warnings of abnormal price fluctuations, an initial report is generated according to the structured report template.

[0192] The final pricing results are integrated with the initial report to output a visual report document containing price values ​​and causal analysis.

[0193] In this embodiment, the feature importance analysis algorithm employs interpretable artificial intelligence technologies such as SHAP or LIME to extract key feature factors influencing the current pricing by analyzing the contribution of each input feature in the price prediction model. These factors may include core variables such as market supply and demand indicators, the impact of policy events, and inventory status, used to reveal the main driving factors of pricing decisions.

[0194] Attribution analysis quantifies key characteristic factors to generate indicators of the degree of influence of each factor on the final pricing outcome. Specifically, it can use percentage contribution or absolute impact values, calculating the strength of the influence by comparing the deviation of characteristic values ​​with benchmark values, thus providing data support for the price formation mechanism.

[0195] Price fluctuation anomaly warnings are generated by comparing and analyzing quantitative indicators of impact with historical aviation material trading data. When the impact of a certain feature significantly exceeds the historical fluctuation range, the system automatically marks the anomaly and provides a warning explanation to help users identify potential market risks or data anomalies.

[0196] The structured report template organizes content according to causal logic, integrating key characteristic factors, quantitative indicators of impact, and early warning prompts into a unified report framework. The initial report uses a standardized chapter structure and data visualization elements to ensure the standardization and readability of information presentation.

[0197] The fusion processing integrates the final pricing results with the initial report, generating a visual report document through data embedding and correlation analysis. This document includes both specific price figures and detailed causal analysis, using charts and graphs to intuitively display the pricing results and their formation mechanisms.

[0198] This embodiment achieves transparency in the pricing process and visualization of the decision-making basis through a systematic interpretability analysis framework. It provides accurate pricing results and fully presents the logical chain of price formation, significantly improving the credibility and operability of pricing decisions.

[0199] Please see Figure 5In a second aspect, this embodiment also provides an intelligent pricing system 1 for aviation equipment and components, applicable to the method described in the first aspect. The system includes: a multi-source data acquisition module 11, a data fusion and feature extraction module 12, a price prediction module 13, a strategy instruction receiving module 14, a pricing decision module 15, and a result output module 16. The multi-source data acquisition module 11 is configured to collect multi-source heterogeneous data, including internal aviation material transaction data, external market data, supply chain status data, and regulatory event text data, as well as aviation material lifecycle data (service years, maintenance frequency, remaining life) and airworthiness certification data. (Airworthiness certification status, OEM authorization type), AOG emergency demand data (emergency level, response time); the data fusion and feature extraction module 12 is configured to perform fusion processing and feature extraction on multi-source heterogeneous data, generating standardized feature vectors for model input. Feature extraction includes extracting structured event features from regulatory event text data based on natural language models, extracting "remaining life index" and "maintenance health" features from aircraft material lifecycle data, extracting "airworthiness certification level" and "OEM authorization weight" features from airworthiness certification data, and extracting "emergency demand coefficient" features from AOG emergency demand data; The price prediction module 13 is configured to input standardized feature vectors into a pre-trained price prediction model to output a benchmark price prediction value and the corresponding price prediction range. The price prediction model adopts a hybrid architecture combining ensemble learning algorithms and temporal deep learning algorithms, and introduces aviation material industry characteristic weight factors (including OEM supplier monopoly weight, AOG emergency coefficient weight, and airline maintenance plan demand fluctuation weight) during model training to adapt to the monopolistic, urgent, and cyclical characteristics of the aviation material supply chain. The strategy instruction receiving module 14 is configured to receive pricing strategy instructions selected by the user, including profit maximization. The pricing strategy includes a rapid turnover strategy, an airworthiness compliance priority strategy (ensuring transparency of airworthiness costs associated with pricing), and a customer tiered negotiation strategy (adjusting prices based on airline cooperation years / purchase volume). The pricing decision module 15 is configured to generate the final pricing result through the strategy mapping module based on the pricing strategy instructions, benchmark price forecasts, and price forecast ranges. The result output module 16 is configured to output the final pricing result and an interpretable price cause analysis report, which includes an airworthiness compliance cost ratio analysis (such as the proportion of certification fees and compliance warehousing fees in pricing) and a material lifecycle cost-effectiveness analysis (such as the cost of remaining flight hours corresponding to the unit price).

[0200] In this embodiment, the intelligent pricing system 1 for aviation equipment and components acquires real-time internal transaction data, external market data, supply chain status data, and regulatory event text data through the multi-source data acquisition module 11, providing a comprehensive data foundation for pricing analysis. The data fusion and feature extraction module 12 cleans, transforms, and extracts features from the multi-source heterogeneous data, generating standardized feature vectors. The natural language processing component specifically parses structured event features in the regulatory texts. The price prediction module 13 adopts a hybrid architecture combining ensemble learning and temporal deep learning, outputting a benchmark price prediction value and prediction range based on the standardized feature vectors. The strategy instruction receiving module 14 provides a human-computer interaction interface, receiving user-selected profit maximization or rapid turnover strategy instructions. The pricing decision module 15 generates the final price based on user instructions and prediction results using a strategy mapping algorithm. The result output module 16 integrates the pricing results with interpretability causal analysis into a visual report, supporting decision tracing.

[0201] This system automates the entire process of data acquisition, feature extraction, price prediction, strategy decision-making, and result output through a modular architecture. The collaborative operation of each module ensures the scientific nature and accuracy of the pricing process, while enhancing decision-making transparency through interpretability analysis, providing a complete and systematic solution for intelligent pricing of aviation materials.

[0202] In a third aspect, this embodiment also provides a computer-readable storage medium storing computer program instructions thereon, which, when executed by a processor, implement the method described in the first aspect.

[0203] The computer program involved in this embodiment can be stored in a computer device readable storage medium, which includes, but is not limited to, disks, magnetic tapes, magnetic cards, floppy disks, flash memory, optical disks, optical cards, read-only memory (ROM), random access memory (RAM), erasable programmable ROM (EPROM), and electrically erasable programmable ROM (EEPROM), etc. It also includes other biological, physical, or chemical structures capable of performing similar or equivalent functions to the storage media listed above, such as DNA, RNA, proteins, and other units with information storage capabilities. In specific embodiments, the storage medium involved can be one of the above-mentioned media types or a combination of the above media types. In different embodiments, the computer program involved in the embodiment can be centrally stored in a single medium or distributed across multiple media. The memory containing the computer device readable storage medium can be non-volatile memory or random access memory. These computer device readable storage media can be built into the device or connected to the device involved in the embodiment as an external device or part of an external device. In some embodiments, the memory having a computer device readable storage medium is deployed locally; in other embodiments, the memory may be deployed remotely from the processor, for example, as a network-attached memory accessed via RF circuitry or an external port and a communication network, wherein the communication network may be the Internet, one or more intranets, a local area network (LAN), a wide area network (WLAN), a storage area network (SAN), or a suitable combination thereof, as long as computer device access to the memory is enabled. Furthermore, the computer program involved in the embodiments may be stored in plaintext / ciphertext form, or it may be designed as training data, integrated and recombined through model training and implicitly stored in the parameter states of a deep neural network or other machine learning model.

[0204] By adopting the above technical solutions, this invention differs from existing technologies and possesses the following beneficial effects: Through multi-source heterogeneous data fusion and intelligent analysis, it effectively solves the problems of insufficient data utilization, poor strategy adaptability, and low decision-making transparency in aviation material pricing. Specifically, by integrating multi-source information such as internal transaction data, external market conditions, supply chain status, and regulatory event texts, and using natural language processing technology to extract structured event features from unstructured text, it achieves comprehensive data coverage and deep feature mining. Utilizing a hybrid prediction model combining ensemble learning and temporal deep learning, it captures the complex nonlinear relationships between features and analyzes the temporal fluctuation patterns of prices, outputting a benchmark price prediction value and prediction range that combines accuracy and reliability. Furthermore, through a strategy mapping module, it supports dynamic adaptation between profit maximization and rapid turnover strategies, and intelligently adjusts prices by combining multi-dimensional parameters such as market competition and inventory status, ensuring a high degree of alignment between pricing strategies and business objectives. Finally, through an interpretable analysis framework, it generates a causal report, fully presenting the pricing logic and key influencing factors, significantly improving the scientific nature, adaptability, and transparency of pricing decisions. Compared to traditional methods, this invention achieves intelligent processing of the entire process from data acquisition to strategy output, providing an efficient and reliable solution for aircraft material pricing.

[0205] Finally, it should be noted that although the above embodiments have been described in the text and drawings of this application, this should not limit the scope of patent protection of this application. Any technical solutions that are based on the essential concept of this application and utilize the content described in the text and drawings of this application, resulting in equivalent structural or procedural substitutions or modifications, as well as the direct or indirect application of the technical solutions of the above embodiments to other related technical fields, are all included within the scope of patent protection of this application.

Claims

1. A smart pricing method for aviation equipment and components, characterized in that, include: Collect multi-source heterogeneous data, including internal transaction data of aviation materials, external market information data, supply chain status data, and regulatory event text data; The multi-source heterogeneous data is fused and feature extracted to generate a standardized feature vector for model input. The feature extraction includes extracting structured event features from the regulatory event text data based on a natural language model. The standardized feature vector is input into the pre-trained price prediction model to output the benchmark price prediction value and the corresponding price prediction range. The price prediction model adopts a hybrid architecture that combines ensemble learning algorithm and temporal deep learning algorithm. Receive pricing strategy instructions selected by the user, including a profit maximization strategy or a rapid turnover strategy; Based on the pricing strategy instruction and the benchmark price forecast and price forecast range, the final pricing result is generated through the strategy mapping module. Output the final pricing result and an interpretable price cause analysis report.

2. The intelligent pricing method for aviation equipment and components according to claim 1, characterized in that, The multi-source heterogeneous data is fused and its features are extracted to generate a standardized feature vector for model input, including: The internal transaction data of the aircraft materials is cleaned and standardized to generate a historical transaction feature sequence in a unified format. The external market data is collected and aggregated in real time to generate dynamic market characteristic indicators; The supply chain status data is subjected to multi-source verification and status encoding to generate a supply chain stability feature vector; The text data of the regulatory events is analyzed for entity recognition and impact using a natural language model to extract the features of the structured events; The standardized feature vector is generated by multi-dimensional feature fusion of the historical transaction feature sequence, market dynamic feature indicators, supply chain stability feature vector, and structured event features.

3. The intelligent pricing method for aviation equipment and components according to claim 2, characterized in that, The text data of the aforementioned regulatory events is analyzed for entity recognition and impact using a natural language model to extract the features of the structured events, including: The regulatory event text data is preprocessed and segmented to generate a text sequence to be analyzed; The text sequence to be analyzed is input into a pre-trained named entity recognition model to extract key entity information related to aerospace components. The key entity information includes the affected part number, applicable aircraft model range, and compliance requirement type. Based on an attention mechanism, the influence of the key entity information is quantitatively analyzed to generate event influence weight coefficients, including: Construct an attention scoring function to calculate the correlation score between each entity in the key entity information and the reference events in the historical event database; The correlation score is converted into attention distribution weights using the softmax normalization function; The entity feature vectors are weighted and fused according to the attention distribution weights to generate a context-aware entity representation; The context-aware entity representation is nonlinearly transformed using a multilayer perceptron to output a preliminary estimate of the impact intensity. The preliminary impact strength estimate is corrected in the time dimension by combining the event timeliness decay factor to generate the final event impact weight coefficient. The event timeliness decay factor is negatively correlated with the time interval from the event occurrence time to the current time. Based on the key entity information and event impact weight coefficients, a time-sensitive event feature matrix is ​​constructed. The event feature matrix is ​​matched with the historical event database to generate the final structured event features.

4. The intelligent pricing method for aviation equipment and components according to claim 1, characterized in that, The price prediction model is constructed through the following steps: An initial training sample set is constructed based on historical aviation material transaction data. The initial training sample set includes historical standardized feature vectors and corresponding historical actual transaction price labels. The initial training sample set is divided into a training subset and a validation subset using a stratified sampling strategy. The stratified sampling strategy allocates samples according to the type of aviation materials and components and the frequency of transactions. A benchmark price prediction module is constructed using an ensemble learning algorithm. The benchmark price prediction module employs a gradient boosting decision tree architecture to rank feature importance and fit nonlinear relationships. A price fluctuation prediction module is constructed using a time-series deep learning algorithm. The price fluctuation prediction module uses an attention mechanism to capture long-term dependencies and cyclical change patterns. The benchmark price prediction module and the price fluctuation prediction module are weighted and fused to generate a hybrid prediction model; The parameters of the hybrid prediction model are optimized using the training subset, and the prediction performance is verified using the validation subset, thus obtaining the trained price prediction model.

5. The intelligent pricing method for aviation equipment and components according to claim 4, characterized in that, The standardized feature vector is input into a pre-trained price prediction model to output a baseline price prediction value and the corresponding price prediction range, including: The benchmark price prediction module performs a preliminary price prediction on the standardized feature vector to generate an initial benchmark price prediction value. The price fluctuation prediction module analyzes the price fluctuation range of the standardized feature vector to generate price fluctuation interval parameters. The initial benchmark price forecast and the price fluctuation range parameter are jointly calibrated based on the quantile regression algorithm to generate the calibrated benchmark price forecast. The uncertainty of the calibrated benchmark price forecast is quantified using confidence intervals to generate a price forecast interval with a confidence level identifier. The calibrated benchmark price forecast and the price forecast range are output.

6. The intelligent pricing method for aviation equipment and components according to claim 1, characterized in that, When the pricing strategy instruction is a profit maximization strategy; Based on the pricing strategy instruction and the benchmark price forecast and price forecast range, the final pricing result is generated through the strategy mapping module, including: Obtain the upper boundary value of the price prediction range as the benchmark reference price; The benchmark reference price is dynamically adjusted based on real-time market competition environment data to generate a competition environment adjustment coefficient; By combining inventory turnover rate and supply and demand indicators, the profit maximization weighting parameters are calculated. The benchmark reference price is calculated using a weighted optimization based on the competitive environment adjustment coefficient and the profit maximization weight parameter. Output the optimized final pricing result.

7. The intelligent pricing method for aviation equipment and components according to claim 1, characterized in that, When the pricing strategy instruction is a fast turnover strategy; Based on the pricing strategy instruction and the benchmark price forecast and price forecast range, the final pricing result is generated through the strategy mapping module, including: Obtain the lower boundary value of the price prediction range as the benchmark reference price; An inventory pressure coefficient is generated based on inventory backlog data, and the inventory pressure coefficient is positively correlated with inventory turnover days. The price elasticity adjustment parameter is calculated based on the market supply and demand tight balance indicator; The inventory pressure coefficient and the price elasticity adjustment parameter are combined to generate a fast turnover discount coefficient. The benchmark reference price is discounted based on the fast turnover discount factor to generate the final pricing result.

8. The intelligent pricing method for aviation equipment and components according to claim 1, characterized in that, Output the final pricing result and an interpretable price cause analysis report, including: Key feature factors influencing this pricing were extracted based on a feature importance analysis algorithm. Attribution analysis was performed on the key feature factors to generate quantitative indicators of the degree of influence of each key feature factor on the final pricing result; By comparing and analyzing the quantitative indicators of the degree of impact with historical aviation material transaction data, an early warning of abnormal price fluctuations is generated. Based on the key characteristic factors, quantitative indicators of impact, and abnormal price fluctuation warnings, an initial report is generated according to the structured report template. The final pricing result is then merged with the initial report to output a visual report document containing price values ​​and causal analysis.

9. An intelligent pricing system for aviation equipment and components, characterized in that, The system applicable to the method of any one of claims 1 to 8, the system comprising: The multi-source data acquisition module is configured to collect multi-source heterogeneous data, including internal transaction data of aviation materials, external market information data, supply chain status data, and regulatory event text data. The data fusion and feature extraction module is configured to perform fusion processing and feature extraction on the multi-source heterogeneous data to generate a standardized feature vector for model input. The feature extraction includes extracting structured event features from the regulatory event text data based on a natural language model. The price prediction module is configured to input the standardized feature vector into a pre-trained price prediction model to output a benchmark price prediction value and the corresponding price prediction range. The price prediction model adopts a hybrid architecture that combines an ensemble learning algorithm with a temporal deep learning algorithm. The strategy instruction receiving module is configured to receive pricing strategy instructions selected by the user, including a profit maximization strategy or a rapid turnover strategy. The pricing decision module is configured to generate the final pricing result through the strategy mapping module based on the pricing strategy instruction, the benchmark price prediction value, and the price prediction range. The results output module is configured to output the final pricing results and an interpretable price cause analysis report.

10. A computer-readable storage medium storing computer program instructions thereon, characterized in that, The computer program instructions, when executed by a processor, implement the method as described in any one of claims 1 to 8.

Images