Method, system, and medium for optimizing an aircraft equipment inventory

Abstract

The application discloses an aircraft spare parts device reserve optimization method, system and medium, through collecting multi-dimensional dynamic data and carrying out fusion cleaning and structured feature extraction, a high-dimensional feature set is generated; a hybrid intelligent prediction model of fusion time series prediction and integrated learning is used to output dynamic demand prediction value and uncertainty measurement; combined with the key grade of the aircraft spare parts and the loss cost of out-of-stock, the optimal inventory control parameter is calculated through the random optimization model; the optimal inventory control parameter and the real-time inventory state are used to generate the inventory operation instruction, and the strategy report with the confidence evaluation and the key influence factor analysis is output. The application realizes multi-dimensional accurate prediction of aircraft spare parts demand and dynamic optimization of inventory strategy, significantly improves the inventory control precision and resource utilization efficiency, effectively reduces the out-of-stock risk and the overstock cost, and enhances the reliability and economy of the aircraft spare parts guarantee.

Description

Technical Field

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

[0002] Inventory management of aviation equipment is a crucial link in the aviation operations support system, directly impacting fleet availability, maintenance efficiency, and overall operating costs. Because aviation material demand is influenced by multiple dynamic factors such as flight plans, component lifespan, supply chain conditions, and maintenance strategies, traditional methods have significant limitations in handling complex uncertainties. This often leads to discrepancies between inventory allocation and actual demand, affecting support effectiveness and resource utilization efficiency. Furthermore, engineering instructions and process changes involved in maintenance plans are typically recorded in unstructured text format, making it difficult for existing technologies to automate the parsing and quantification of such information, further limiting the adaptability and accuracy of forecasting and decision-making models. Therefore, how to achieve accurate forecasting of aviation material demand and dynamic optimization of inventory strategies under the coupling of multiple factors has become a core problem urgently needing to be solved in this field. Summary of the Invention

[0003] In view of the above problems, the present invention provides an optimization method, system and medium for the stockpiling of aviation equipment and components. By integrating multi-source dynamic data and intelligent prediction models, it achieves accurate prediction of aviation material demand and dynamic optimization of inventory strategies, solving the problems of inaccurate prediction and delayed decision-making under the coupling of multiple factors.

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

[0005] Collect multi-dimensional dynamic data of aviation equipment, including historical consumption data of aviation materials, on-wing life monitoring data, fleet operation plan data, supply chain reliability data, and maintenance plan change data;

[0006] Multi-dimensional dynamic data is fused, cleaned, and structured feature extracted to generate a high-dimensional feature set for demand forecasting and inventory optimization. The structured feature extraction is configured to use natural language processing algorithms to automatically identify and quantify the impact of engineering instructions from maintenance plan text.

[0007] The high-dimensional feature set is input into the hybrid intelligent prediction model that integrates time series prediction and ensemble learning, and the dynamic demand prediction values ​​of various aviation materials and the corresponding uncertainty measures are output.

[0008] Based on dynamic demand forecasts and uncertainty measures, combined with predefined criticality levels of aviation materials and stockout loss costs, the optimal inventory control parameters for each aviation material are calculated through a stochastic optimization model. The optimal inventory control parameters include reorder point, economic order quantity, and maximum inventory level.

[0009] Based on the optimal inventory control parameters and real-time inventory status, the system automatically makes decisions and generates inventory operation instructions, including purchase requests, repair suggestions, or disposal of obsolete materials.

[0010] Output inventory operation instructions and an inventory optimization strategy report with confidence level assessment and analysis of key influencing factors.

[0011] Furthermore, multi-dimensional dynamic data is fused, cleaned, and structured feature extracted to generate a high-dimensional feature set for demand forecasting and inventory optimization, including:

[0012] The historical consumption data of aircraft materials is analyzed to distinguish the characteristics of historical consumption patterns. The characteristics of historical consumption patterns include planned replacement events and unplanned replacement events. The first time series features and the first volatility index corresponding to the planned replacement events are extracted, and the second time series features and the second volatility index corresponding to the unplanned replacement events are extracted, which are denoted as the first dimension features.

[0013] The on-wing life monitoring data is used to assess the health of components and generate a remaining life prediction value and reliability decay curve based on actual flight hours and cycle count, which is denoted as the second dimension feature.

[0014] The fleet operation plan data is transformed into a capacity mapping conversion, and the flight schedule is converted into the expected flight hour distribution and aircraft material load coefficient for each aircraft type, which is denoted as the third dimension feature.

[0015] Supply risk assessment is conducted using supply chain reliability data, quantifying the volatility of procurement lead time, supplier on-time delivery rate, and minimum order quantity constraints, which are recorded as the fourth dimension feature.

[0016] Engineering impact analysis was performed on the maintenance plan change data. The engineering instruction text was parsed using natural language processing algorithms to extract the list of affected part numbers, modification scope and execution time window, which were recorded as the fifth dimension feature.

[0017] The first, second, third, fourth, and fifth dimension features are spatiotemporally aligned and standardized to form a high-dimensional feature set that includes historical consumption patterns, component health status, future capacity demand, supply risk profiles, and the impact of engineering changes.

[0018] Furthermore, historical consumption data of aircraft materials is analyzed to identify consumption patterns and distinguish their characteristics. These patterns include planned and unplanned replacement events. The first time-series features and first volatility indicators corresponding to planned replacement events are extracted, as are the second time-series features and second volatility indicators corresponding to unplanned replacement events. These include:

[0019] Based on the maintenance work order type and fault code, historical consumption data is labeled with event type. Scheduled maintenance and life-cycle control replacement are labeled as planned replacement events, while fault reports and unplanned replacements are labeled as unplanned replacement events.

[0020] A seasonal decomposition algorithm is used to extract the trend component, periodic component and residual component of the planned replacement event sequence as the first time series feature, and the coefficient of variation is calculated as the first volatility index.

[0021] An extreme value distribution fitting algorithm is used to extract the burst and intermittent characteristics of unplanned replacement event sequences as second time series features, and its kurtosis and skewness are calculated as second volatility indicators.

[0022] Establish an event association mapping model to associate replacement events with corresponding aircraft serial numbers, component serial numbers, and installation locations, thereby enhancing the interpretability and traceability of features;

[0023] The output contains the first dimension features of both planned and unplanned swap event feature sets.

[0024] Furthermore, the high-dimensional feature set is input into a hybrid intelligent prediction model that integrates time-series prediction and ensemble learning, outputting dynamic demand predictions for various aviation materials and corresponding uncertainty measures, including:

[0025] Based on the characteristics of the first time series, a deterministic demand forecast is generated by using an autoregressive integral moving average model with external regression variables to generate a baseline demand forecast curve.

[0026] For the second time series features, a gradient boosting decision tree algorithm is used to construct a nonlinear regression model to capture the complex mapping relationship between the second time series features and the health status of components, the distribution of transportation demand, and the supply risk profile, and to obtain the calculation results;

[0027] The baseline demand forecast curve and the calculation results of the nonlinear regression model are adaptively weighted and fused, and the weight coefficients are dynamically adjusted according to the consumption pattern characteristics of each aviation material to generate the final dynamic demand forecast value.

[0028] Based on the statistical characteristics of the first volatility index and the second volatility index, and combined with the model prediction error distribution, the quantile regression algorithm is used to calculate the demand fluctuation range corresponding to each prediction time point, and generate an uncertainty measure.

[0029] Furthermore, for aviation materials with a criticality level of high, the Monte Carlo simulation method is introduced to conduct multi-scenario demand extrapolation based on demand fluctuation ranges, enhance the prediction calculation of extreme demand scenarios, and obtain multi-scenario uncertainty measurement.

[0030] Furthermore, for high-criticality aerospace materials, a Monte Carlo simulation method is introduced to perform multi-scenario demand extrapolation based on demand fluctuation ranges, enhancing the prediction calculations for extreme demand scenarios and obtaining multi-scenario uncertainty measures, including:

[0031] Based on the probability distribution characteristics of demand fluctuation range, a multidimensional stochastic demand generation model is constructed, which simultaneously considers the stochastic fluctuation of demand quantity and the stochastic distribution of demand time.

[0032] Set the number of iterations and convergence conditions for the Monte Carlo simulation, and generate a random demand sequence that conforms to the probability distribution characteristics in each iteration;

[0033] The generated random demand sequence is input into the inventory control strategy simulation environment to simulate changes in inventory levels, the frequency of stockout events, and the corresponding cost impact under different demand scenarios.

[0034] The output indicators from each simulation are statistically analyzed, including the distribution of demand extreme values, the distribution of stockout probability, and the distribution of inventory turnover rate.

[0035] Based on the statistical characteristics of the simulation results, the probability and impact of extreme demand scenarios are quantified, and a multi-scenario uncertainty metric is generated and output. The multi-scenario uncertainty metric includes demand forecast intervals at different confidence levels, extreme risk probability assessments, and corresponding inventory buffer recommendations.

[0036] Furthermore, based on dynamic demand forecasts and uncertainty metrics, and combined with predefined criticality levels and stockout loss costs for aircraft materials, the optimal inventory control parameters for each type of aircraft material are calculated using a stochastic optimization model, including:

[0037] Based on the predefined criticality level of aviation materials, the quantitative requirements for the target service level are generated, and the predefined stockout loss cost is transformed into an inventory cost optimization target under service level constraints.

[0038] A baseline demand model is established based on dynamic demand forecasts, and a demand probability distribution function is constructed using uncertainty measures.

[0039] Based on the baseline demand model, a stochastic optimization model is established with the objective of minimizing total expected cost, which includes the sum of the expected values ​​of inventory holding cost, ordering cost, and stockout loss cost.

[0040] The stochastic dynamic programming algorithm is used to solve the stochastic optimization model. Under the premise of considering the uncertainty of demand, the optimal combination of reorder point and maximum inventory level is obtained through iterative calculation.

[0041] For repairable parts categories, the stochastic optimization model is extended to a multi-level inventory system, and the optimal ratio between available parts inventory and repairable parts inventory and their respective reordering strategies are solved.

[0042] Based on the supply risk assessment results, the optimal inventory control parameters are robustly calibrated to ensure that the preset service level requirements can still be met in the event of supply delays or interruptions.

[0043] Furthermore, based on the optimal inventory control parameters and real-time inventory status, automatic decision-making matching is performed to generate inventory operation instructions, including purchase requests, repair suggestions, or disposal of obsolete materials, including:

[0044] Continuously collect data on the current inventory, in-transit orders, number of parts awaiting repair, and recent consumption rate of various aviation materials to obtain real-time inventory status;

[0045] The real-time inventory status is compared with the corresponding reorder point. When the sum of available inventory and in-transit quantity is lower than the reorder point, the purchase requisition generation process is automatically triggered. The purchase quantity of the newly generated purchase requisition is determined based on the economic order quantity and takes into account the minimum order quantity constraint.

[0046] For repairable parts, when the available inventory is below the reorder point and the number of parts to be repaired reaches the repair trigger threshold, a repair suggestion instruction is generated. The repair trigger threshold is dynamically calculated based on the repair cycle and demand forecast.

[0047] Based on recent consumption rates, demand forecast trends, and the impact of engineering changes, identify aviation materials that have been stagnant for a long time and whose future demand is significantly reduced, and generate recommendations for the disposal of obsolete materials.

[0048] The generated inventory operation instructions are prioritized based on factors including the criticality level of the aircraft materials, the urgency of the shortage risk, and the expected time of the shortage.

[0049] The sorted inventory operation instructions and corresponding decision-making data are packaged and output to form executable purchase requisitions, repair work orders, and solutions for handling obsolete materials.

[0050] Furthermore, output inventory operation instructions and an inventory optimization strategy report with confidence level assessment and analysis of key influencing factors, including:

[0051] Based on uncertainty measurement and model prediction error distribution, the decision confidence index corresponding to each inventory operation instruction is calculated and classified and labeled according to the confidence level.

[0052] Attribution analysis was conducted on key influencing factors to identify the main drivers affecting inventory decisions, including demand volatility characteristics, changes in supply reliability, the impact of engineering orders, and fleet capacity adjustments.

[0053] Generate a structured strategy report, including an inventory health assessment matrix, recommendations for adjusting inventory control parameters for each aviation material, estimated cost savings, and risk warning information;

[0054] The inventory operation instructions are linked and mapped with the strategy report, providing each operation instruction with corresponding decision-making basis explanations, influencing factor analysis and alternative plan suggestions;

[0055] Continuous learning is conducted based on historical decision-making performance data to dynamically update the confidence assessment model and the weights of key influencing factors;

[0056] Output an inventory optimization strategy report that includes visual charts and data tables.

[0057] In a second aspect, the present invention also provides an optimization system for the storage of aviation equipment and components, applicable to the method described in the first aspect. The system includes: a data acquisition module, a feature extraction module, a predictive analysis module, an optimization calculation module, a decision generation module, and a report output module. The data acquisition module is configured to collect multi-dimensional dynamic data of aviation equipment. The feature extraction module is configured to perform fusion cleaning and structured feature extraction on the multi-dimensional dynamic data to generate a high-dimensional feature set oriented towards demand forecasting and inventory optimization. The predictive analysis module is configured to input the high-dimensional feature set into a hybrid intelligent prediction model that integrates time-series prediction and ensemble learning, and output dynamic demand forecast values ​​for various aviation materials and their corresponding uncertainty metrics. The optimization calculation module is configured to calculate the optimal inventory control parameters for each aviation material based on the dynamic demand forecast values ​​and uncertainty metrics, combined with predefined criticality levels and stockout loss costs, through a stochastic optimization model. The decision generation module is configured to automatically match decisions based on the optimal inventory control parameters and real-time inventory status, generating inventory operation instructions including purchase requests, repair suggestions, or disposal of obsolete materials. The report output module is configured to output inventory operation instructions and an inventory optimization strategy report with confidence assessment and analysis of key influencing factors.

[0058] 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.

[0059] Unlike existing technologies, the above-mentioned technical solution provides an optimized method, system, and medium for the storage of aviation equipment and components. It generates a high-dimensional feature set by collecting multi-dimensional dynamic data, performing fusion cleaning and structured feature extraction, and utilizing a hybrid intelligent prediction model that integrates time-series prediction and ensemble learning to output dynamic demand forecasts and uncertainty metrics. Combining the criticality level of aviation materials with the cost of stockout losses, it calculates optimal inventory control parameters through a stochastic optimization model. Based on the optimal inventory control parameters and real-time inventory status, it generates inventory operation instructions and outputs a strategy report with confidence assessment and analysis of key influencing factors. This invention achieves multi-dimensional and accurate prediction of aviation material demand and dynamic optimization of inventory strategies, significantly improving inventory control accuracy and resource utilization efficiency, effectively reducing stockout risks and backlog costs, and enhancing the reliability and economy of aviation equipment support.

[0060] 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

[0061] 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.

[0062] In the accompanying drawings of the instruction manual:

[0063] Figure 1 This is a schematic diagram illustrating steps S101 to S106 of the method described in the specific implementation embodiment;

[0064] Figure 2 This is a schematic diagram illustrating steps S201 to S206 of the method described in a specific implementation.

[0065] Figure 3 This is a schematic diagram illustrating steps S301 to S305 of the method described in a specific implementation.

[0066] Figure 4 This is a schematic diagram illustrating steps S401 to S404 of the method described in a specific embodiment;

[0067] Figure 5 This is a schematic diagram of the system described in a specific implementation.

[0068] The reference numerals used in the above figures are explained as follows:

[0069] 1. Optimize the system;

[0070] 11. Data acquisition module;

[0071] 12. Feature extraction module;

[0072] 13. Predictive Analysis Module;

[0073] 14. Optimize the calculation module;

[0074] 15. Decision Generation Module;

[0075] 16. Report output module. Detailed Implementation

[0076] 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.

[0077] 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.

[0078] 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.

[0079] 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.

[0080] 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.

[0081] 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.

[0082] 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.

[0083] 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.

[0084] 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.

[0085] Please see Figure 1 In a first aspect, this embodiment provides an optimization method for the stockpiling of aviation equipment and components, including:

[0086] S101. Collect multi-dimensional dynamic data of aircraft equipment, including historical consumption data of aircraft materials, on-wing life monitoring data, fleet operation plan data, supply chain reliability data, and maintenance plan change data.

[0087] S102. Perform fusion cleaning and structured feature extraction on multi-dimensional dynamic data to generate a high-dimensional feature set for demand forecasting and inventory optimization. The structured feature extraction is configured to use natural language processing algorithms to automatically identify and quantify the impact of engineering instructions from maintenance plan text.

[0088] S103. Input the high-dimensional feature set into the hybrid intelligent prediction model that integrates time series prediction and ensemble learning, and output the dynamic demand prediction values ​​of various aviation materials and the corresponding uncertainty measures.

[0089] S104. Based on dynamic demand forecasts and uncertainty measures, and combined with predefined criticality levels and stockout loss costs of aviation materials, the optimal inventory control parameters for each aviation material are calculated through a stochastic optimization model. The optimal inventory control parameters include reorder point, economic order quantity, and maximum inventory level.

[0090] S105. Based on the optimal inventory control parameters and real-time inventory status, automatically make decisions and generate inventory operation instructions including purchase requests, repair suggestions, or disposal of obsolete materials.

[0091] S106. Output inventory operation instructions and an inventory optimization strategy report with confidence assessment and analysis of key influencing factors.

[0092] In step S101, multi-dimensional dynamic data refers to a multi-source heterogeneous data set supporting aircraft material demand forecasting and inventory optimization. This includes: historical aircraft material consumption data extracted from the aircraft material management system, recording the historical requisition and consumption records of various equipment; on-wing life monitoring data acquired through airborne sensors and maintenance systems, characterizing the actual usage time and health status of components; fleet operation plan data from the flight scheduling system, reflecting future flight missions and capacity distribution; supply chain reliability data integrated from the supplier management system, quantifying supply risks such as procurement cycles and delivery stability; and maintenance plan change data from the engineering management department, including engineering instructions and technical notices issued in text form. This data is collected and initially integrated through the enterprise data platform, providing comprehensive input for subsequent analysis.

[0093] In step S102, the fusion cleaning and structured feature extraction refers to the process of aligning, denoising, handling missing values, and standardizing multi-source data. The high-dimensional feature set is a structured dataset extracted using feature engineering methods for prediction and optimization. During its generation, natural language processing algorithms are used to parse the maintenance plan text, identify the equipment scope, execution time, and degree of impact of the engineering instructions, and quantify these into feature variables that can be incorporated into the model. This step improves feature quality and model usability by eliminating data inconsistencies and redundancy.

[0094] Preferably, the step of automatically identifying and quantifying the impact of engineering instructions from the maintenance plan text using natural language processing algorithms specifically includes:

[0095] A named entity recognition model is used to automatically extract key entity information from the unstructured engineering instruction text, including affected part number, modification instruction number, executing fleet, effective date, etc.

[0096] Text classification or sentiment analysis techniques are used to determine the impact level of engineering instructions (such as "mandatory immediate execution", "planned modification", "recommendation for execution"), and these levels are quantified into different weighting coefficients.

[0097] Based on the extracted execution time window and impact level, a time decay function is constructed to quantify the impact intensity of engineering instructions on the demand for aviation materials in different time periods in the future, thereby transforming the abstract text description into structured and quantifiable feature values, which are then input into the prediction model.

[0098] This application addresses the core pain points in the aviation maintenance field, such as inconsistent engineering instruction text formats, numerous technical terms, and time-consuming and error-prone manual interpretation, achieving automated conversion from unstructured text to structured decision data.

[0099] In step S103, the hybrid intelligent prediction model integrating time series forecasting and ensemble learning refers to a prediction algorithm that combines time series analysis methods with machine learning integration strategies. Dynamic demand forecasts characterize the expected demand quantity of aviation materials in various future time periods, while uncertainty metrics are used to quantify the fluctuation range and reliability of the prediction results, expressed, for example, through confidence intervals or probability distributions. This model, by integrating historical patterns and multi-factor correlations, achieves the evolution of demand forecasting from single-point estimation to probabilistic assessment.

[0100] In step S104, the stochastic optimization model is a mathematical programming method for making inventory decisions under uncertain demand conditions. The predefined criticality levels of aircraft materials are based on their importance to flight safety and operational support, while the cost of stockout losses integrates economic and operational impacts such as grounding losses and emergency allocation costs. The optimal inventory control parameters are obtained by minimizing the expected total cost objective, where the reorder point triggers replenishment decisions, the economic order quantity balances ordering and holding costs, and the maximum inventory level sets the upper limit for inventory. This model incorporates predictive uncertainty into the optimization framework, enhancing the robustness and adaptability of the strategy.

[0101] In step S105, automatic decision matching refers to the process of logically comparing and strategically mapping the optimized inventory parameters with the real-time inventory status. The real-time inventory status is dynamically obtained through the inventory management system, including real-time information such as existing inventory, orders in transit, and the quantity of parts awaiting repair. When the inventory level reaches the reorder point or other thresholds, a purchase request is automatically generated to replenish the inventory. For repairable equipment, a repair recommendation is generated based on the quantity of parts awaiting repair and the repair cycle. For obsolete materials, disposal recommendations propose solutions for disposing of or reusing equipment that has not been used for a long time and whose demand is declining. This step achieves real-time response and automated operation of inventory management.

[0102] In step S106, the confidence assessment, based on the uncertainty measure of the prediction model and historical accuracy calculation, reflects the reliability of the decision-making results. Key influencing factor analysis identifies the main dimensions driving inventory changes, such as demand fluctuations, supply delays, or engineering changes. The inventory optimization strategy report integrates the above analysis, outputting actionable instructions and decision-making support to assist managers in final review and execution.

[0103] This embodiment collects multi-dimensional dynamic data, performs fusion cleaning and feature extraction to construct a high-dimensional feature set as the basis for prediction; it uses a hybrid intelligent model that integrates time-series prediction and ensemble learning to output dynamic demand forecasts and uncertainty measures; based on a stochastic optimization model, it calculates optimal inventory control parameters while considering demand uncertainty and the criticality of aviation materials; finally, it automatically generates inventory operation instructions and strategy reports based on real-time inventory status. This embodiment achieves precise and dynamic aviation material inventory management by systematically integrating multi-source data, intelligently predicting demand and uncertainty, and generating inventory strategies based on stochastic optimization. This effectively improves supply chain efficiency, reduces redundant inventory and stockout risks, and enhances the overall supply chain's responsiveness and economic efficiency.

[0104] Please see Figure 2 In some embodiments, multi-dimensional dynamic data is fused, cleaned, and structured feature extracted to generate a high-dimensional feature set for demand forecasting and inventory optimization, including:

[0105] S201. Analyze the consumption patterns of historical aircraft material consumption data, distinguish the characteristics of historical consumption patterns, including planned replacement events and unplanned replacement events, and extract the first time series features and the first volatility index corresponding to the planned replacement events, as well as the second time series features and the second volatility index corresponding to the unplanned replacement events, which are denoted as the first dimension features.

[0106] S202. Perform component health assessment on the on-wing life monitoring data, and generate remaining life prediction values ​​and reliability decay curves based on actual flight hours and cycle counts, denoted as the second dimension feature.

[0107] S203. Perform capacity mapping transformation on fleet operation plan data, and convert the flight schedule into the expected flight hour distribution and aircraft material load coefficient for each aircraft type, which is denoted as the third dimension feature;

[0108] S204. Conduct supply risk assessment on supply chain reliability data, quantify the volatility of procurement lead time, supplier on-time delivery rate and minimum order quantity constraints, and record them as the fourth dimension feature.

[0109] S205. Perform engineering impact analysis on maintenance plan change data, parse engineering instruction text using natural language processing algorithm, extract the list of affected part numbers, modification scope and execution time window, and record them as the fifth dimension feature;

[0110] S206. The first, second, third, fourth, and fifth dimension features are spatiotemporally aligned and standardized to form a high-dimensional feature set that includes historical consumption patterns, component health status, future capacity demand, supply risk profile, and the impact of engineering changes.

[0111] In step S201, consumption pattern analysis is the process of identifying event types and extracting features from historical aircraft material consumption data. Planned replacement events correspond to planned maintenance and replacements, such as periodic inspections or lifespan control replacements; unplanned replacement events include replacements caused by sudden failures. Preferably, the first time series feature can be extracted using time series decomposition methods, such as using a seasonal decomposition algorithm to separate trend, periodic, and residual components; the first volatility index can use statistics such as the coefficient of variation to measure the stability of planned consumption; the second time series feature targets unplanned events, such as extracting explosive and intermittent patterns through extreme value theory; the second volatility index can calculate kurtosis and skewness to characterize the abnormal characteristics of the distribution. This step achieves a refined characterization of consumption behavior, providing highly discriminative input features for subsequent predictions.

[0112] In step S202, the component health assessment is a quantitative analysis of the remaining service life and reliability status of aircraft components based on on-wing life data. The predicted remaining service life can be calculated using a degradation model based on flight hours and cycles, such as fitting a Weibull distribution or proportional hazards model; the reliability decay curve describes the trend of component failure rate over time. This assessment provides dynamic health status input at the component level for demand forecasting.

[0113] In step S203, capacity mapping conversion is the process of transforming flight schedules into quantifiable aircraft material demand drivers. The expected flight hour distribution reflects the planned flight volume of each aircraft type in the future; the aircraft material load factor, based on the correlation between aircraft type and aircraft materials, converts flight hours into the expected usage intensity of the corresponding equipment, for example, by determining the conversion factor through historical data regression or engineering experience. This step transforms the macro-level capacity plan into a basis for equipment-level demand forecasting.

[0114] In step S204, the supply risk assessment is a quantitative analysis of supply chain reliability. Procurement lead time volatility can be measured by the standard deviation or coefficient of variation of historical delivery times; supplier on-time delivery rate is based on historical delivery records; and minimum order quantity constraints reflect the minimum batch requirements of suppliers or logistics links. This assessment provides supply-side uncertainty input for inventory optimization.

[0115] In step S205, the engineering impact analysis is the process of extracting structured information from the maintenance plan text using natural language processing technology. The affected part number list identifies the specific equipment involved in the engineering instruction; the modification scope describes the content and extent of the changes; and the execution time window determines the planned implementation period of the engineering instruction. This step transforms unstructured text information into quantifiable features to capture the impact of engineering changes on requirements.

[0116] In step S206, spatiotemporal alignment and standardized stitching are processes that integrate multi-source features in the temporal and spatial dimensions. Temporal alignment ensures that the feature data have consistent validity at the same point in time; spatial alignment refers to establishing mapping relationships between entities such as equipment and models across different data sources. Standardization eliminates dimensional differences between features, and can be achieved using Z-score standardization or min-max scaling. The resulting high-dimensional feature set integrates multi-dimensional information such as consumption patterns, health status, capacity demand, supply risks, and engineering impacts, providing comprehensive and consistent input for the prediction model.

[0117] This embodiment constructs a high-dimensional feature set that comprehensively reflects the influencing factors of aviation material demand by performing refined feature extraction and fusion of multi-source data. This provides a high-quality, multi-dimensional and structured data foundation for subsequent intelligent prediction and optimization decisions, significantly improving the accuracy of demand perception and the adaptability of inventory strategies.

[0118] Please see Figure 3 In some embodiments, historical consumption data of aircraft materials is analyzed to distinguish historical consumption pattern characteristics, including planned replacement events and unplanned replacement events. First time-series features and first volatility indicators are extracted for planned replacement events, and second time-series features and second volatility indicators are extracted for unplanned replacement events, including:

[0119] S301. Based on the maintenance work order type and fault code, the historical consumption data is labeled with event type. Scheduled maintenance and life control component replacement are labeled as planned replacement events, and fault reports and unplanned replacements are labeled as unplanned replacement events.

[0120] S302. The seasonal decomposition algorithm is used to extract the trend component, periodic component and residual component of the planned replacement event sequence as the first time series feature, and the coefficient of variation is calculated as the first volatility index.

[0121] S303. For unplanned replacement event sequences, an extreme value distribution fitting algorithm is used to extract their explosive and intermittent characteristics as second time series features, and their kurtosis and skewness are calculated as second volatility indicators.

[0122] S304. Establish an event association mapping model to associate replacement events with the corresponding aircraft serial number, component serial number, and installation location, thereby enhancing the interpretability and traceability of features.

[0123] S305. Output the first dimension feature, which includes the feature set of planned replacement events and the feature set of unplanned replacement events.

[0124] In step S301, event type labeling refers to the process of classifying historical consumption events based on the type field and fault code in the maintenance work order. Scheduled maintenance and life-sustaining component replacement usually have predetermined execution cycles and clear planning basis, and are labeled as planned replacement events; fault reports and unplanned replacements are caused by sudden failures or accidental damage, and are labeled as unplanned replacement events. This step achieves automatic labeling through rule matching or classification models, laying the foundation for subsequent differential feature extraction.

[0125] In step S302, the seasonal decomposition algorithm is used to analyze regular patterns in the planned replacement event sequence. The trend component reflects the long-term direction of demand changes, the periodic component captures recurring patterns within a fixed period, and the residual component represents random fluctuations that cannot be explained by the trend and periodicity. Preferably, the coefficient of variation is calculated as the ratio of the standard deviation to the mean, used to measure the relative volatility of the sequence, serving as a first volatility indicator. This analysis helps identify the stability and predictability characteristics of planned demand.

[0126] In step S303, the extreme value distribution fitting algorithm is suitable for analyzing abnormal fluctuation patterns in unplanned replacement events. Preferably, burst characteristics are captured by extreme value models such as the generalized Pareto distribution to identify rare high-value events, while intermittent characteristics are identified by zero-inflation models or sequence interval analysis to identify the sparsity and clustering of events. Kurtosis measures the sharpness of the distribution shape, and skewness measures the asymmetry of the distribution, both serving as a second volatility indicator. This step effectively characterizes the suddenness and uncertainty of unplanned demand.

[0127] In step S304, the event association mapping model establishes associations between replacement events and aircraft serial numbers, component serial numbers, and installation locations to achieve event traceability analysis. This model is based on entity identification information in maintenance records and can achieve data linking through relational databases or graph structures, enhancing the interpretability and business relevance of features in practical applications.

[0128] In step S305, the first dimension feature, as the output, integrates the differentiated feature set of planned and unplanned swapping events, providing highly discriminative and interpretable input features for the subsequent prediction model.

[0129] This embodiment achieves differentiated characterization of planned and unplanned demand by refining the classification and feature extraction of aircraft material consumption events, providing an important feature foundation for building a high-precision prediction model and effectively improving the accuracy of demand forecasting and business interpretability.

[0130] Please see Figure 4In some embodiments, a high-dimensional feature set is input into a hybrid intelligent prediction model that integrates time-series prediction and ensemble learning, outputting dynamic demand predictions for various types of aviation materials and corresponding uncertainty measures, including:

[0131] S401. Based on the characteristics of the first time series, a deterministic demand forecast is generated by using an autoregressive integral moving average model with external regression variables to generate a baseline demand forecast curve.

[0132] S402. For the second time series characteristics, a gradient boosting decision tree algorithm is used to construct a nonlinear regression model to capture the complex mapping relationship between the second time series characteristics and the health status of components, the distribution of transportation demand and the supply risk profile, and to obtain the calculation results.

[0133] S403. Adaptively weightedly fuse the baseline demand forecast curve with the calculation results of the nonlinear regression model, and dynamically adjust the weight coefficients according to the consumption pattern characteristics of each aviation material to generate the final dynamic demand forecast value.

[0134] S404. Based on the statistical characteristics of the first volatility index and the second volatility index, and combined with the model prediction error distribution, the quantile regression algorithm is used to calculate the demand fluctuation range corresponding to each prediction time point, and generate an uncertainty measure.

[0135] Furthermore, for aviation materials with a criticality level of high, the Monte Carlo simulation method is introduced to conduct multi-scenario demand extrapolation based on demand fluctuation ranges, enhance the prediction calculation of extreme demand scenarios, and obtain multi-scenario uncertainty measurement.

[0136] In step S401, the autoregressive integral moving average model with external regression variables is a forecasting method that combines the autocorrelation characteristics of time series with external influencing factors. This model introduces external regression variables, such as seasonal characteristics and trend components, into the traditional ARIMA model to model the first-time series characteristics corresponding to planned swap events. The baseline demand forecast curve is generated through this model, reflecting the deterministic demand trend under the influence of external factors, providing a basic reference for subsequent fusion forecasting.

[0137] In step S402, the gradient boosting decision tree algorithm iteratively trains multiple weak learners and combines their prediction results to construct a powerful nonlinear regression model. This model, targeting the second time-series features corresponding to unplanned replacement events, can effectively capture the complex nonlinear relationship between these features and component health status, capacity demand distribution, and supply risk profile. The calculation results characterize the unplanned demand predictions based on multi-dimensional features.

[0138] In step S403, adaptive weighted fusion refers to the process of dynamically adjusting the weights of different model prediction results based on the characteristics of aircraft material consumption patterns. The weight coefficients can be calculated based on factors such as historical prediction accuracy and feature stability. For example, aircraft materials with significant planned characteristics are given higher weights to the ARIMA model, while aircraft materials with predominantly unplanned characteristics have their gradients increased to enhance the contribution of the decision tree model. The final dynamic demand prediction value is generated through weighted fusion, taking into account the characteristics of different consumption patterns.

[0139] The specific strategy of the adaptive weighted fusion is as follows: define a dynamic weight coefficient α for each aerospace component, with a value range of 0≤α≤1;

[0140] For aircraft materials with significant planned replacement characteristics (such as those with a high proportion of planned events and low volatility in their historical consumption), a higher α value (such as α > 0.7) is set to allow the prediction results of the ARIMA model to dominate.

[0141] For aircraft materials with significant unplanned replacement characteristics (such as those with obvious sudden and intermittent characteristics), a lower α value (such as α < 0.3) is set to allow the prediction results of the gradient boosting decision tree model to dominate.

[0142] The dynamic weighting coefficient α is not a fixed value, but is dynamically calculated based on the latest first volatility index and second volatility index (such as the coefficient of variation and the moving average of kurtosis) of the aircraft material, thereby achieving complementary advantages and adaptive adjustment of the model and improving the overall prediction accuracy.

[0143] In step S404, the quantile regression algorithm is used to estimate the conditional distribution at different quantiles, thereby calculating the demand fluctuation range. Based on the statistical characteristics of the first and second volatility indices, combined with the model prediction error distribution, this algorithm can generate demand prediction ranges at different confidence levels as a measure of uncertainty. For aviation materials with a high criticality level, a Monte Carlo simulation method is additionally introduced. By randomly sampling, a large number of possible demand scenarios are generated, and multi-scenario extrapolation is performed based on the demand fluctuation range to enhance the predictive ability for extreme demand events, resulting in a multi-scenario uncertainty measure that includes risk probability assessment.

[0144] This embodiment achieves accurate forecasting of demand for different types of aviation materials by integrating time series forecasting and machine learning methods. Combined with uncertainty quantification technology, it provides comprehensive forecasting information for inventory decisions, especially enhancing the forecasting capability for extreme demand scenarios of key aviation materials, and significantly improving the robustness and reliability of inventory management.

[0145] In some embodiments, for aircraft materials with a high criticality level, a Monte Carlo simulation method is additionally introduced to perform multi-scenario demand extrapolation based on demand fluctuation ranges, enhancing the prediction calculation for extreme demand scenarios and obtaining multi-scenario uncertainty measures, including:

[0146] Based on the probability distribution characteristics of demand fluctuation range, a multidimensional stochastic demand generation model is constructed, which simultaneously considers the stochastic fluctuation of demand quantity and the stochastic distribution of demand time.

[0147] Set the number of iterations and convergence conditions for the Monte Carlo simulation, and generate a random demand sequence that conforms to the probability distribution characteristics in each iteration;

[0148] The generated random demand sequence is input into the inventory control strategy simulation environment to simulate changes in inventory levels, the frequency of stockout events, and the corresponding cost impact under different demand scenarios.

[0149] The output indicators from each simulation are statistically analyzed, including the distribution of demand extreme values, the distribution of stockout probability, and the distribution of inventory turnover rate.

[0150] Based on the statistical characteristics of the simulation results, the probability and impact of extreme demand scenarios are quantified, and a multi-scenario uncertainty metric is generated and output. The multi-scenario uncertainty metric includes demand forecast intervals at different confidence levels, extreme risk probability assessments, and corresponding inventory buffer recommendations.

[0151] In this embodiment, the Monte Carlo simulation method is a numerical calculation method based on random sampling. It generates possible demand scenarios by randomly sampling a large number of samples from the probability distribution characteristics of demand fluctuation ranges. The multidimensional stochastic demand generation model considers both the random fluctuation of demand quantity and the random distribution of demand time. For example, demand quantity can be modeled using a normal distribution or a Poisson distribution, and demand time intervals can be modeled using an exponential distribution, thereby more realistically simulating the uncertainty characteristics of actual demand.

[0152] Preferably, the number of iterations in the Monte Carlo simulation is set to 1,000 to 10,000 to ensure the statistical stability of the simulation results; the convergence condition can be set to the variance of the simulation results being less than a preset threshold or reaching the maximum number of iterations. In each iteration, a random demand sequence is generated based on the probability distribution characteristics of the demand fluctuation range. These sequences reflect the demand changes under different probabilities.

[0153] The generated random demand sequence is input into an inventory control strategy simulation environment, which simulates the operational logic of a real inventory system, including ordering, receiving, and consumption. The simulation allows observation of inventory level changes, the frequency of stockout events, and the corresponding cost impacts under different demand scenarios. These output indicators provide a quantitative basis for risk assessment.

[0154] By analyzing the output metrics from each simulation, the demand extreme value distribution reflects the probability of extreme demand events, the stockout probability distribution measures the risk level of insufficient inventory, and the inventory turnover rate distribution assesses inventory utilization efficiency. Based on these statistical characteristics, the probability and impact of extreme demand scenarios can be quantified.

[0155] The generated multi-scenario uncertainty measure includes demand forecast intervals at different confidence levels, such as 90% or 95% confidence intervals; the extreme risk probability assessment quantifies the likelihood of rare high demand events occurring; and the corresponding inventory buffer recommendations provide inventory adjustment schemes for different risk levels, such as recommendations for setting safety stock levels.

[0156] This embodiment uses the Monte Carlo simulation method to conduct an in-depth analysis of the demand uncertainty of key aviation materials, realizes the quantitative assessment of extreme demand scenarios, provides a more comprehensive risk management basis for inventory decisions of high-risk aviation materials, and significantly improves the inventory system's ability to cope with abnormal demand fluctuations.

[0157] In some embodiments, based on dynamic demand forecasts and uncertainty measures, and combined with predefined criticality levels and stockout loss costs for aviation materials, the optimal inventory control parameters for each aviation material are calculated using a stochastic optimization model, including:

[0158] Based on the predefined criticality level of aviation materials, the quantitative requirements for the target service level are generated, and the predefined stockout loss cost is transformed into an inventory cost optimization target under service level constraints.

[0159] A baseline demand model is established based on dynamic demand forecasts, and a demand probability distribution function is constructed using uncertainty measures.

[0160] Based on the baseline demand model, a stochastic optimization model is established with the objective of minimizing total expected cost, which includes the sum of the expected values ​​of inventory holding cost, ordering cost, and stockout loss cost.

[0161] The stochastic dynamic programming algorithm is used to solve the stochastic optimization model. Under the premise of considering the uncertainty of demand, the optimal combination of reorder point and maximum inventory level is obtained through iterative calculation.

[0162] For repairable parts categories, the stochastic optimization model is extended to a multi-level inventory system, and the optimal ratio between available parts inventory and repairable parts inventory and their respective reordering strategies are solved.

[0163] Based on the supply risk assessment results, the optimal inventory control parameters are robustly calibrated to ensure that the preset service level requirements can still be met in the event of supply delays or interruptions.

[0164] In this embodiment, the quantitative requirement for the target service level is based on the inventory guarantee target set according to the criticality level of the aircraft materials. Preferably, critical aircraft materials require a service level of 99%, and ordinary aircraft materials require a service level of 95%. The stockout loss cost is transformed into an inventory cost optimization target under service level constraints. By quantifying the stockout loss into economic cost, it together with inventory holding cost and ordering cost constitutes the total cost optimization target.

[0165] The baseline demand model is built upon dynamic demand forecasts, reflecting the expected trend of demand; the demand probability distribution function is constructed using uncertainty measures to describe the uncertainty characteristics of demand, and can model demand fluctuations using either a normal distribution or a Poisson distribution. These two models together provide the input basis for stochastic optimization.

[0166] The stochastic optimization model aims to minimize the total expected cost, where inventory holding costs are related to average inventory levels, ordering costs are related to ordering frequency, and stockout loss costs are constrained by service level requirements. This model comprehensively considers various cost factors of the inventory system and seeks the optimal balance under conditions of demand uncertainty.

[0167] Preferably, the function that aims to minimize the total expected cost is expressed by formula (1) as follows:

[0168]

[0169] In formula (1), E is the total expected cost, h is the inventory holding cost, E1 is the expected value of the average inventory level, o is the fixed cost of each order, D is the predicted demand rate, Q is the order quantity (i.e., economic order quantity), b is the stockout loss cost, and E2 is the expected value of the stockout quantity. The stochastic dynamic programming algorithm iteratively solves the above objective function under different inventory states (inventory levels, orders in transit, etc.) and stochastic demand scenarios, thereby obtaining the reorder point and the maximum inventory level. This model directly embeds the uncertainty measure of demand (i.e., the demand probability distribution function) into each decision cycle of the optimization process, rather than the traditional ex-post safety stock adjustment, thus making it more scientific and robust.

[0170] The stochastic dynamic programming algorithm solves the optimization model iteratively, considering the uncertainty of current inventory status and future demand in each decision cycle, and progressively solves to obtain the optimal combination of reorder point and maximum inventory level. This algorithm can effectively handle multi-period stochastic optimization problems.

[0171] For repairable parts, the multi-level inventory system simultaneously considers the collaborative optimization of available parts inventory and parts-in-repair inventory. By establishing a correlation model between the two levels of inventory, the optimal inventory allocation relationship and reordering strategy are solved to improve the inventory utilization efficiency of repairable parts.

[0172] Based on the supply risk assessment results, the optimal inventory control parameters are robustly calibrated, for example by increasing safety stock or adjusting reorder points to address the risk of supply delays, ensuring that the preset service level requirements can still be met in the event of supply anomalies.

[0173] This embodiment achieves accurate calculation of aircraft material inventory control parameters by establishing a stochastic optimization model that considers demand uncertainty and supply risk. In particular, it proposes a multi-level inventory optimization strategy for repairable parts and enhances the anti-interference capability of the inventory system through robust calibration, significantly improving the economy and reliability of inventory management.

[0174] In some embodiments, automatic decision matching is performed based on optimal inventory control parameters and real-time inventory status to generate inventory operation instructions, including purchase requests, repair suggestions, or disposal of obsolete materials, including:

[0175] Continuously collect data on the current inventory, in-transit orders, number of parts awaiting repair, and recent consumption rate of various aviation materials to obtain real-time inventory status;

[0176] The real-time inventory status is compared with the corresponding reorder point. When the sum of available inventory and in-transit quantity is lower than the reorder point, the purchase requisition generation process is automatically triggered. The purchase quantity of the newly generated purchase requisition is determined based on the economic order quantity and takes into account the minimum order quantity constraint.

[0177] For repairable parts, when the available inventory is below the reorder point and the number of parts to be repaired reaches the repair trigger threshold, a repair suggestion instruction is generated. The repair trigger threshold is dynamically calculated based on the repair cycle and demand forecast.

[0178] Based on recent consumption rates, demand forecast trends, and the impact of engineering changes, identify aviation materials that have been stagnant for a long time and whose future demand is significantly reduced, and generate recommendations for the disposal of obsolete materials.

[0179] The generated inventory operation instructions are prioritized based on factors including the criticality level of the aircraft materials, the urgency of the shortage risk, and the expected time of the shortage.

[0180] The sorted inventory operation instructions and corresponding decision-making data are packaged and output to form executable purchase requisitions, repair work orders, and solutions for handling obsolete materials.

[0181] In this embodiment, real-time inventory status refers to the latest inventory information of various aviation materials continuously collected through the inventory management system. This data is updated in real time through the system interface, providing accurate current status information for inventory decisions.

[0182] Reorder point comparison refers to comparing the sum of available inventory and in-transit inventory in real-time with a preset reorder point. When this sum is lower than the reorder point, it indicates that the inventory level has reached a critical point requiring replenishment, and the system automatically triggers the purchase requisition generation process. The purchase quantity is determined based on the economic order quantity, while also considering the supplier's minimum order quantity constraints to ensure the economy and feasibility of the procurement decision.

[0183] For repairable parts, the repair trigger threshold is a critical value dynamically calculated based on repair cycle and demand forecast. When available inventory falls below the reorder point and the number of parts awaiting repair reaches this threshold, it indicates that a repair decision can effectively replenish inventory, and the system generates a repair recommendation instruction. This threshold is dynamically adjusted according to repair capabilities and the urgency of demand.

[0184] Obsolete material identification identifies aerospace materials that have shown no change in activity for a long time and whose future demand is significantly reduced by analyzing indicators such as recent consumption rates, demand forecast trends, and the impact of engineering changes. Based on preset obsolete material identification rules, the system generates disposal suggestions, such as price reductions, allocation, or scrapping, to avoid inventory backlog.

[0185] Prioritization is based on factors such as the criticality level of aircraft materials, the urgency of the shortage risk, and the expected time of the shortage, ranking various operational instructions according to their importance. Highly critical aircraft materials and operations with urgent shortage risks are processed first to ensure timely response to critical support needs.

[0186] The final output packages the sorted operation instructions with the corresponding decision-making data to form executable standardized documents, including purchase requisition forms, repair work orders, and obsolete material handling plans, providing complete information support for subsequent execution.

[0187] This embodiment achieves automated generation and priority management of inventory operations through real-time inventory status monitoring and intelligent decision matching. In particular, it proposes a dynamic repair triggering mechanism for repairable parts and establishes a process for identifying and handling obsolete materials, which significantly improves the timeliness, accuracy and efficiency of inventory management.

[0188] In some embodiments, the system outputs inventory operation instructions and an inventory optimization strategy report with confidence level assessment and analysis of key influencing factors, including:

[0189] Based on uncertainty measurement and model prediction error distribution, the decision confidence index corresponding to each inventory operation instruction is calculated and classified and labeled according to the confidence level.

[0190] Attribution analysis was conducted on key influencing factors to identify the main drivers affecting inventory decisions, including demand volatility characteristics, changes in supply reliability, the impact of engineering orders, and fleet capacity adjustments.

[0191] Generate a structured strategy report, including an inventory health assessment matrix, recommendations for adjusting inventory control parameters for each aviation material, estimated cost savings, and risk warning information;

[0192] The inventory operation instructions are linked and mapped with the strategy report, providing each operation instruction with corresponding decision-making basis explanations, influencing factor analysis and alternative plan suggestions;

[0193] Continuous learning is conducted based on historical decision-making performance data to dynamically update the confidence assessment model and the weights of key influencing factors;

[0194] Output an inventory optimization strategy report that includes visual charts and data tables.

[0195] In this embodiment, the decision confidence index is used to measure the credibility of inventory operation instructions. It quantifies the fluctuation range and error probability of the prediction results through statistical methods and classifies them into three confidence levels: high, medium, and low, to provide a reliability reference for decision execution.

[0196] Preferably, the calculation of the decision confidence index includes the following steps:

[0197] For procurement requisition instructions, the confidence level is primarily calculated based on a combination of uncertainty measures in demand forecasting (such as the width of the forecast interval) and supply risk assessments (such as supplier on-time delivery rates). For example, when the forecast interval is wide and supplier reliability is low, the confidence level of the generated procurement instruction decreases.

[0198] For repair recommendations, the confidence level is further considered in terms of the stability of the repair cycle and the historical repair rate of the parts to be repaired.

[0199] The confidence level of recommendations for handling obsolete materials is mainly judged based on the quantitative results of the impact of engineering changes and the significance of the trend of demand forecast changes.

[0200] This confidence assessment method is closely coupled with multi-dimensional dynamic data in the field of aircraft material inventory management, forming a domain-specific assessment system.

[0201] Key influencing factor attribution analysis identifies the main drivers affecting inventory decisions through feature importance analysis and causal inference methods. Specifically, demand volatility reflects the degree of uncertainty in demand, changes in supply reliability characterize the stability of the supply chain, engineering directives quantify demand changes resulting from technological shifts, and fleet capacity adjustments reflect the impact of operational plans on demand. This analysis helps to understand the key motivations behind decisions.

[0202] The structured strategy report is a comprehensive output document integrating various analytical results. It includes an inventory health assessment matrix for overall status evaluation, inventory control parameter adjustment suggestions for optimization direction, an estimated cost savings to quantify improvement benefits, and risk warning information to highlight potential problems. The report uses a standardized template to ensure information completeness and readability.

[0203] Correlation mapping refers to establishing a correspondence between inventory operation instructions and strategy reports, providing decision-making basis and explanations for each instruction, clarifying the rationale for its formulation, analyzing influencing factors to clarify key driving factors, and suggesting alternative execution paths. This mapping enhances the transparency and operability of decision-making.

[0204] The continuous learning mechanism is based on historical decision-making performance data. It uses machine learning methods to dynamically update the confidence assessment model and the weights of key influencing factors, enabling the system to adapt to environmental changes and continuously improve decision quality.

[0205] The final inventory optimization strategy report uses a combination of visual charts and data tables to intuitively display the analysis results and recommended solutions, supporting managers in decision review and implementation.

[0206] This embodiment establishes a complete decision output and interpretation mechanism, which not only generates executable inventory operation instructions, but also provides comprehensive decision-making basis and influencing factor analysis, enhancing the transparency and credibility of decision-making. At the same time, it continuously improves the system's adaptability and decision quality through a continuous learning mechanism.

[0207] Please see Figure 5In a second aspect, this embodiment also provides an optimization system 1 for the storage of aviation equipment and components, applicable to the method described in the first aspect. The system includes: a data acquisition module 11, a feature extraction module 12, a predictive analysis module 13, an optimization calculation module 14, a decision generation module 15, and a report output module 16. The data acquisition module 11 is configured to collect multi-dimensional dynamic data of aviation equipment; the feature extraction module 12 is configured to perform fusion cleaning and structured feature extraction on the multi-dimensional dynamic data to generate a high-dimensional feature set oriented towards demand forecasting and inventory optimization; the predictive analysis module 13 is configured to input the high-dimensional feature set into a hybrid system that integrates time-series prediction and ensemble learning. The intelligent prediction model outputs dynamic demand forecasts for various types of aviation materials and their corresponding uncertainty metrics. The optimization calculation module 14 is configured to calculate the optimal inventory control parameters for each type of aviation material based on the dynamic demand forecasts and uncertainty metrics, combined with predefined criticality levels and stockout loss costs, using a stochastic optimization model. The decision generation module 15 is configured to automatically match optimal inventory control parameters with real-time inventory status to generate inventory operation instructions, including purchase requests, repair suggestions, or disposal of obsolete materials. The report output module 16 is configured to output inventory operation instructions and an inventory optimization strategy report with confidence assessment and analysis of key influencing factors.

[0208] In this embodiment, the optimized system 1 for aviation equipment and components reserves achieves intelligent management of the entire process through modular design. The data acquisition module 11 is responsible for acquiring multi-dimensional dynamic data from various business systems in real time, providing a data foundation for subsequent analysis. Preferably, the multi-dimensional dynamic data includes airworthiness compliance data and dual-source demand data. Airworthiness compliance data includes EO / AD instruction information (mandatory replacement cycle, compliance deadline) and serial number traceability information (unique identifier, validity period) corresponding to the aircraft parts number. Dual-source demand data includes fleet scheduled maintenance plan data (planned replacement trigger conditions, such as flight hours / cycles) and historical fault repair data (non-planned demand basis, such as fault frequency, fault part number). The feature extraction module 12 cleans and performs feature engineering processing on the raw data. A high-quality feature set is generated. The preferred high-quality feature set includes airworthiness compliance features and dual-source demand features. Airworthiness compliance features include, for example, the remaining time until the mandatory replacement deadline (EO / AD) for aircraft materials (quantifying compliance urgency). Dual-source demand features include planned demand features and unplanned demand features. Planned demand features include, for example, the correlation between flight hours / cycles and replacement probability for aircraft materials. Unplanned demand features include, for example, the quantile of historical failure frequencies for aircraft materials. The predictive analysis module 13 uses a fusion machine learning algorithm to output accurate demand forecasts and uncertainty assessments. The predictive analysis module 13 adjusts the prediction logic, changing the original "single prediction" to "dual-source demand split prediction." "Predictions" – using "fleet scheduled maintenance plans + lifespan correlation characteristics" to predict planned demand, and using "historical failure data + failure frequency characteristics" to predict unplanned demand, ultimately merging them into total demand; when supplementing uncertainty assessment, "demand fluctuation assessment caused by sudden EO / AD directives" is added, covering the most core predictive risks of aircraft materials; the optimization calculation module 14 generates the optimal inventory strategy based on the prediction results and business constraints through a stochastic optimization algorithm. The preferred business constraints include airworthiness compliance constraints and emergency support constraints. In the airworthiness compliance constraints, the optimization objective must prioritize meeting the "minimum inventory level required by EO / AD directives"; in the emergency support constraints, "AOG emergency response" is set. The system uses thresholds (e.g., "airport warehouse materials must meet the retrieval requirements within 1 hour for over 95% of AOG scenarios") to ensure the optimized inventory strategy has emergency response capabilities. The decision generation module 15 matches the optimized strategy with the real-time inventory status and automatically generates operable instructions. Preferably, operable instructions include airworthiness compliance instructions and emergency allocation instructions. In the airworthiness compliance instructions, when the materials are "less than 3 months away from the EO / AD deadline and the inventory is insufficient", a "compliance preparation reminder instruction" is automatically generated. In the emergency allocation instructions, when an AOG event is triggered, a basic allocation instruction of "prioritizing the retrieval of airport warehouse → regional warehouse materials" is automatically generated (clearly defining the retrieval priority and not involving complex logistics linkages).The report output module 16 integrates the analysis results and decision-making basis to generate a complete strategy report. Preferably, the strategy report includes an airworthiness compliance summary and an emergency support summary. The airworthiness compliance summary clarifies "the inventory quantity and compliance rate covered by the EO / AD directive"; the emergency support summary clarifies "the inventory's satisfaction rate for AOG scenarios and the expected response time".

[0209] This system achieves fully automated processing from data acquisition to decision output through the collaborative operation of its various modules. The data acquisition module 11 ensures the comprehensiveness and real-time nature of the data; the feature extraction module 12 improves data quality; the predictive analysis module 13 provides accurate demand insights; the optimization calculation module 14 generates scientific inventory strategies; the decision generation module 15 implements the strategies; and the report output module 16 provides decision support. Ultimately, the system achieves accurate prediction, dynamic optimization, and intelligent decision-making for aircraft material inventory, significantly improving inventory management efficiency and service support levels.

[0210] 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.

[0211] 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.

[0212] By adopting the above technical solutions, the present invention differs from the prior art and has the following beneficial effects:

[0213] Through multi-dimensional dynamic data collection and fusion cleaning, a high-dimensional feature set for demand forecasting and inventory optimization was constructed. Specifically, a natural language processing algorithm was employed to automatically identify and quantify the impact of engineering instructions from maintenance plan texts, addressing the challenge of handling unstructured text data in traditional methods. A hybrid intelligent forecasting model integrating time-series forecasting and ensemble learning not only outputs dynamic demand forecasts but also provides corresponding uncertainty metrics, representing an evolution from single-point estimation to probabilistic forecasting. Based on a stochastic optimization model, optimal inventory control parameters were calculated considering demand uncertainty and the criticality of aircraft materials. Inventory operation instructions were automatically generated through real-time inventory status matching, ultimately outputting a strategy report with confidence assessments and analyses of key influencing factors. These technical solutions effectively address the problems of inaccurate forecasting and decision-making lag in existing technologies under multi-factor coupling conditions. They achieve accurate forecasting of aircraft material demand and dynamic optimization of inventory strategies, significantly improving inventory management accuracy and resource utilization efficiency, reducing stockout risks and backlog costs, and enhancing the reliability and economy of aircraft equipment support.

[0214] 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. An optimization method for the storage of aviation equipment and components, characterized in that, include: Collect multi-dimensional dynamic data of aviation equipment, including historical consumption data of aviation materials, on-wing life monitoring data, fleet operation plan data, supply chain reliability data, and maintenance plan change data; The multi-dimensional dynamic data is fused, cleaned, and structured feature extracted to generate a high-dimensional feature set for demand forecasting and inventory optimization. The structured feature extraction is configured to use a natural language processing algorithm to automatically identify and quantify the impact of engineering instructions from the maintenance plan text. The high-dimensional feature set includes a first-dimensional feature. The process of fusing, cleaning, and extracting structured features from the multi-dimensional dynamic data to generate a high-dimensional feature set for demand forecasting and inventory optimization includes: analyzing the consumption patterns of the historical consumption data of aviation materials, distinguishing historical consumption pattern features, which include planned replacement events and unplanned replacement events, and extracting the first time series features and the first volatility index corresponding to the planned replacement events, as well as the second time series features and the second volatility index corresponding to the unplanned replacement events, denoted as the first dimension feature; The high-dimensional feature set is input into a hybrid intelligent prediction model that integrates time-series prediction and ensemble learning, outputting dynamic demand predictions for various types of aviation materials and their corresponding uncertainty metrics. This step specifically includes: for the first time-series features, a deterministic demand prediction is performed using an autoregressive integral moving average model with external regression variables to generate a baseline demand prediction curve; for the second time-series features, a gradient boosting decision tree algorithm is used to construct a nonlinear regression model to capture the complex mapping relationship between the second time-series features and component health status, capacity demand distribution, and supply risk profile, obtaining calculation results; the baseline demand prediction curve and the calculation results of the nonlinear regression model are adaptively weighted and fused, with the weight coefficients dynamically adjusted according to the consumption pattern characteristics of each aviation material, generating the final dynamic demand prediction value; based on the statistical characteristics of the first and second volatility indicators, combined with the model prediction error distribution, a quantile regression algorithm is used to calculate the demand fluctuation range corresponding to each prediction time point, generating the uncertainty metric; and for aviation materials with a high criticality level, a Monte Carlo simulation method is additionally introduced to perform multi-scenario demand extrapolation based on the demand fluctuation range, strengthening the prediction calculation for extreme demand scenarios, and obtaining a multi-scenario uncertainty metric. Based on the dynamic demand forecast and uncertainty measurement, combined with the predefined criticality level of aviation materials and the cost of stockout losses, the optimal inventory control parameters for each aviation material are calculated through a stochastic optimization model. The optimal inventory control parameters include reorder point, economic order quantity and maximum inventory level. Based on the optimal inventory control parameters and real-time inventory status, automatic decision matching is performed to generate inventory operation instructions, including purchase requests, repair suggestions, or disposal of obsolete materials. Output the inventory operation instructions and an inventory optimization strategy report with confidence assessment and analysis of key influencing factors.

2. The optimization method for the storage of aviation equipment and components according to claim 1, characterized in that, The multi-dimensional dynamic data is fused, cleaned, and structured feature extracted to generate a high-dimensional feature set for demand forecasting and inventory optimization, and the method further includes: The on-wing life monitoring data is used to assess the component health and generate a remaining life prediction value and reliability decay curve based on actual flight hours and cycle count, which is denoted as the second dimension feature. The fleet operation plan data is transformed by capacity mapping, and the flight schedule is converted into the expected flight hour distribution and aviation material load coefficient for each aircraft type, which is denoted as the third dimension feature. The supply chain reliability data is used to conduct a supply risk assessment, quantifying the volatility of procurement lead time, supplier on-time delivery rate, and minimum order quantity constraints, which are denoted as the fourth dimension feature. An engineering impact analysis was performed on the maintenance plan change data. The engineering instruction text was parsed using a natural language processing algorithm to extract the list of affected part numbers, modification scope, and execution time window, which were recorded as the fifth dimension feature. The first, second, third, fourth, and fifth dimension features are spatiotemporally aligned and standardized to form a high-dimensional feature set that includes historical consumption patterns, component health status, future capacity demand, supply risk profiles, and the impact of engineering changes.

3. The optimization method for the storage of aviation equipment and components according to claim 2, characterized in that, The historical consumption data of the aircraft materials is analyzed to identify consumption pattern characteristics, including planned replacement events and unplanned replacement events. First time-series features and first volatility indicators are extracted for planned replacement events, and second time-series features and second volatility indicators are extracted for unplanned replacement events, including: Based on the maintenance work order type and fault code, historical consumption data is labeled with event type. Scheduled maintenance and life-cycle control replacement are labeled as planned replacement events, while fault reports and unplanned replacements are labeled as unplanned replacement events. A seasonal decomposition algorithm is used to extract the trend component, periodic component and residual component of the planned replacement event sequence as the first time series feature, and the coefficient of variation is calculated as the first volatility index. An extreme value distribution fitting algorithm is used to extract the burst and intermittent characteristics of unplanned replacement event sequences as second time series features, and its kurtosis and skewness are calculated as second volatility indicators. Establish an event association mapping model to associate replacement events with corresponding aircraft serial numbers, component serial numbers, and installation locations, thereby enhancing the interpretability and traceability of features; The output contains the first dimension feature, which includes the feature set of planned replacement events and the feature set of unplanned replacement events.

4. The optimization method for the storage of aviation equipment and components according to claim 1, characterized in that, For aviation materials classified as high-criticality, a Monte Carlo simulation method is additionally introduced to perform multi-scenario demand extrapolation based on the aforementioned demand fluctuation range. This enhances the prediction calculations for extreme demand scenarios and yields multi-scenario uncertainty measures, including: Based on the probability distribution characteristics of the demand fluctuation range, a multidimensional stochastic demand generation model is constructed, which simultaneously considers the stochastic fluctuation of demand quantity and the stochastic distribution of demand time. Set the number of iterations and convergence conditions for the Monte Carlo simulation, and generate a random demand sequence that conforms to the probability distribution characteristics in each iteration; The generated random demand sequence is input into the inventory control strategy simulation environment to simulate changes in inventory levels, the frequency of stockout events, and the corresponding cost impact under different demand scenarios. The output indicators from each simulation are statistically analyzed, including the distribution of demand extreme values, the distribution of stockout probability, and the distribution of inventory turnover rate. Based on the statistical characteristics of the simulation results, the probability and impact of extreme demand scenarios are quantified, and a multi-scenario uncertainty metric is generated and output. The multi-scenario uncertainty metric includes demand forecast intervals at different confidence levels, extreme risk probability assessments, and corresponding inventory buffer recommendations.

5. The optimization method for the storage of aviation equipment and components according to claim 1, characterized in that, Based on the dynamic demand forecasts and uncertainty metrics, and combined with predefined criticality levels and stockout loss costs for aviation materials, the optimal inventory control parameters for each type of aviation material are calculated using a stochastic optimization model, including: Based on the predefined criticality level of the aircraft materials, the quantitative requirements for the target service level are generated, and the predefined stockout loss cost is transformed into an inventory cost optimization target under service level constraints. A baseline demand model is established based on the dynamic demand forecast, and a demand probability distribution function is constructed using the uncertainty measure. Based on the baseline demand model, a stochastic optimization model is established with the objective of minimizing total expected cost, which includes the sum of the expected values ​​of inventory holding cost, ordering cost, and stockout loss cost. The stochastic optimization model is solved using a stochastic dynamic programming algorithm. Under the premise of considering the uncertainty of demand, the optimal combination of reorder point and maximum inventory level is obtained through iterative calculation. For the repairable parts category, the stochastic optimization model is extended to a multi-level inventory system, and the optimal ratio between the available parts inventory and the repairable parts inventory and their respective reordering strategies are solved. Based on the supply risk assessment results, the optimal inventory control parameters are robustly calibrated to ensure that the preset service level requirements can still be met in the event of supply delays or interruptions.

6. The optimization method for the storage of aviation equipment and components according to claim 1, characterized in that, Based on the optimal inventory control parameters and real-time inventory status, automatic decision-making matching is performed to generate inventory operation instructions, including purchase requests, repair suggestions, or disposal of obsolete materials, including: Continuously collect data on the current inventory, in-transit orders, number of parts awaiting repair, and recent consumption rate of various aviation materials to obtain real-time inventory status; The real-time inventory status is compared with the corresponding reorder point. When the sum of available inventory and in-transit inventory is lower than the reorder point, the purchase requisition generation process is automatically triggered. The purchase quantity of the newly generated purchase requisition is determined based on the economic order quantity and takes into account the minimum order quantity constraint. For repairable parts, when the available inventory is lower than the reorder point and the number of parts to be repaired reaches the repair trigger threshold, a repair suggestion instruction is generated. The repair trigger threshold is dynamically calculated based on the repair cycle and demand forecast. Based on recent consumption rates, demand forecast trends, and the impact of engineering changes, identify aviation materials that have been stagnant for a long time and whose future demand is significantly reduced, and generate recommendations for the disposal of obsolete materials. The generated inventory operation instructions are prioritized based on factors including the criticality level of the aircraft materials, the urgency of the shortage risk, and the expected time of the shortage. The sorted inventory operation instructions and corresponding decision-making data are packaged and output to form executable purchase requisitions, repair work orders, and solutions for handling obsolete materials.

7. The method for optimizing the storage of aviation equipment and components according to claim 1, characterized in that, Output the aforementioned inventory operation instructions and an inventory optimization strategy report with confidence level assessment and analysis of key influencing factors, including: Based on the uncertainty measure and model prediction error distribution, the decision confidence index corresponding to each inventory operation instruction is calculated, and the instructions are classified and labeled according to the confidence level. Attribution analysis was performed on the key influencing factors to identify the main drivers affecting inventory decisions, including demand volatility characteristics, changes in supply reliability, the impact of engineering orders, and fleet capacity adjustments. Generate a structured strategy report, including an inventory health assessment matrix, recommendations for adjusting inventory control parameters for each aviation material, estimated cost savings, and risk warning information; The inventory operation instructions are linked and mapped with the strategy report, providing each operation instruction with corresponding decision-making basis explanations, influencing factor analysis and alternative plan suggestions; Continuous learning is conducted based on historical decision-making performance data to dynamically update the confidence assessment model and the weights of key influencing factors; Output an inventory optimization strategy report that includes visual charts and data tables.

8. An optimization system for the storage of aviation equipment and components, characterized in that, The system applicable to the method of any one of claims 1 to 7, the system comprising: The data acquisition module is configured to collect multi-dimensional dynamic data of aircraft equipment; The feature extraction module is configured to perform fusion cleaning and structured feature extraction on the multi-dimensional dynamic data to generate a high-dimensional feature set for demand forecasting and inventory optimization. The predictive analysis module is configured to input the high-dimensional feature set into a hybrid intelligent prediction model that integrates time series prediction and ensemble learning, and output the dynamic demand prediction values ​​of various aviation materials and the corresponding uncertainty measures. The optimization calculation module is configured to calculate the optimal inventory control parameters for each type of aircraft material based on the dynamic demand forecast and uncertainty measure, combined with the predefined criticality level of aircraft materials and the cost of stockout losses, through a stochastic optimization model. The decision generation module is configured to automatically match the optimal inventory control parameters with the real-time inventory status to generate inventory operation instructions, including purchase requests, repair suggestions, or disposal of obsolete materials. The report output module is configured to output the inventory operation instructions and an inventory optimization strategy report with confidence assessment and analysis of key influencing factors.

9. 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 7.

Images