Aerial material AI intelligent procurement and optimization method based on big data

Abstract

The invention discloses an aerial material AI intelligent purchase and optimization method based on big data, and belongs to the technical field of big data, and the method comprises the steps: carrying out the learning of historical big data through employing an AI algorithm, so as to obtain an aerial material consumption prediction model with an aerial material consumption prediction capability; the method comprises the following steps: acquiring an aerial material consumption influence factor, identifying the aerial material consumption influence factor by adopting an aerial material consumption prediction model, determining an aerial material demand quantity, automatically optimizing a purchasing strategy according to the aerial material demand quantity and a current aerial material inventory quantity, and automatically generating a purchasing order according to the purchasing strategy, so that the purchasing period can be effectively shortened, and the purchasing efficiency is improved. And historical big data can be effectively utilized, and effective decision support is provided for enterprises.

Description

Technical Field

[0001] The present invention belongs to the field of big data technology, and specifically relates to an aviation material AI intelligent procurement and optimization method based on big data. Background Art

[0002] Aviation material procurement refers to the purchase of aviation equipment and parts required for the maintenance, repair and overhaul (MRO) of aircraft (such as airplanes, helicopters, etc.) and the production of aircraft. Aviation material procurement is an important part of the supply chain management of the aviation industry and has the following characteristics: The quality of aviation materials is directly related to flight safety, so the purchased aviation materials must meet strict safety standards and quality certifications. Aviation materials involve a variety of professional fields, including machinery, electronics, materials science, etc., and procurement personnel need to have corresponding professional knowledge. Aviation material suppliers are spread all over the world, with many supply chain links and complex management. Aviation materials are usually of high value, and the procurement cost has a significant impact on the operating costs of airlines. There are many types of aviation materials, demand forecasting is difficult, and inventory management needs to balance cost and availability.

[0003] With the rapid development of the aviation industry, the importance of aviation material procurement in the operation of aviation enterprises has become increasingly prominent. However, the traditional aviation material procurement method has the following problems: long procurement cycle and low efficiency; low utilization rate of procurement data and inability to provide effective decision-making support for enterprises. Therefore, it is of great significance to study an AI intelligent procurement and optimization method for aviation materials based on big data. Summary of the invention

[0004] The present invention provides an aviation material AI intelligent procurement and optimization method based on big data, which is used to solve technical problems such as long procurement cycle and low utilization rate of procurement data in the prior art.

[0005] An AI intelligent procurement and optimization method for aviation materials based on big data, including: Acquire big data of historical aviation material consumption influencing factors and historical aviation material consumption, and perform data cleaning on the big data of historical aviation material consumption influencing factors and historical aviation material consumption to obtain historical big data; Using AI algorithms to learn the historical big data to obtain an aviation material consumption prediction model with aviation material consumption prediction capability; Obtaining factors influencing aviation material consumption, and using an aviation material consumption prediction model to identify the factors influencing aviation material consumption, and determining aviation material demand; According to the demand for aviation materials and the current inventory of aviation materials, the procurement strategy is automatically optimized, and the purchase order is automatically generated according to the procurement strategy to complete the AI ​​intelligent procurement and optimization of aviation materials.

[0006] Furthermore, the big data of historical aviation material consumption influencing factors and the historical aviation material consumption are obtained, and the big data of historical aviation material consumption influencing factors and the historical aviation material consumption are cleaned to obtain the historical big data, including: Obtain the aircraft type, age, flight time and maintenance cycle of the historical period, and obtain the big data of factors affecting the historical aviation material consumption; at the same time, obtain the historical aviation material consumption of the historical period; The historical aviation material consumption influencing factors big data and the historical aviation material consumption are deduplicated, missing values ​​are removed, and formatted in a standardized manner to obtain historical big data.

[0007] Furthermore, an AI algorithm is used to learn the historical big data to obtain an aviation material consumption prediction model with aviation material consumption prediction capability, including: An initial learning model is constructed using an AI algorithm, and parameters of the initial learning model are encoded to obtain multiple different parameter encodings; For each parameter code, after applying the parameters contained in the parameter code to the initial learning model, the historical aviation material consumption influencing factor big data in the historical big data is used as input, and the historical aviation material consumption corresponding to the historical aviation material consumption influencing factor big data is used as output to obtain the loss function value corresponding to the parameter code; Determine the optimal parameter encoding based on the loss function values ​​corresponding to all parameter encodings; Taking the optimal parameter code as the center, the optimal region search is performed on the parameter code to obtain the parameter code after the optimal region search; Performing a full parameter coding cooperative search on the parameter coding after the optimal area search to obtain the parameter coding after the full parameter coding cooperative search; Performing an adaptive full-space search on the parameter coding after the cooperative search of the full-parameter coding to obtain the parameter coding after the adaptive full-space search; Performing a global variation search on the parameter encoding after the adaptive full-space search to obtain the parameter encoding after the global variation search; Repeat the optimal area search, full parameter coding cooperative search, adaptive full space search and global variation search until the learning end condition is met, re-obtain the optimal parameter coding according to the parameter coding after the last optimal area search, and use the re-obtained optimal parameter coding as the final parameter of the initial learning model to obtain an aviation material consumption prediction model with aviation material consumption prediction capability, otherwise return to the step of obtaining the loss function value corresponding to the parameter coding.

[0008] Furthermore, taking the optimal parameter code as the center, an optimal region search is performed on the parameter code to obtain the parameter code after the optimal region search, including: According to all parameter encodings, the parameter center encoding is determined as:

[0009] in, Indicates t The parameter center encoding in the learning process is Indicates t The first m parameter encoding, m =1,2,…,M, where M represents the total number of parameter codes; According to the parameter center code and the optimal parameter code, the parameter code is searched, and the parameter code after the optimal area search is obtained as follows:

[0010]

[0011] in, Indicates the parameter encoding after the optimal area search , Indicates t The optimal parameter encoding in the learning process, represents the optimal area search coefficient, represents a natural constant, represents the first random number between (0,1), and T represents the preset maximum number of learning times.

[0012] Furthermore, the parameter coding after the optimal area search is subjected to a full parameter coding cooperative search to obtain the parameter coding after the full parameter coding cooperative search, including: Obtain the loss function values ​​corresponding to the parameter codes after the optimal area search, and arrange them in ascending order of the loss function values ​​to obtain the sorted parameter codes; For the sorted n Parameter encoding, obtain the spiral search angle and spiral search radius:

[0013]

[0014] in, Indicates n The parameter encodes the corresponding spiral search angle, represents pi, represents the first spiral trajectory control coefficient, Indicates n The parameter encodes the corresponding spiral search radius, n =1,2,…,M, where M represents the total number of parameter codes; represents the second spiral trajectory control coefficient, represents the second random number between (0,1), Represents the third random number between (0,1); According to the spiral search angle and the spiral search radius, the first cooperation coefficient and the second cooperation coefficient are obtained as follows:

[0015]

[0016] in, Indicates n The first cooperation coefficient corresponding to the parameter encoding, Indicates k The parameters encode the corresponding spiral search radius, Indicates k The parameter codes correspond to the spiral search angle, || represents the absolute value, sin represents the sine function, and max represents the maximum value function. Indicates n The second cooperation coefficient corresponding to the parameter encoding, cos represents the cosine function; According to the first cooperation coefficient and the second cooperation coefficient, the parameter coding after the optimal area search is searched, and the parameter coding after the full parameter coding cooperation search is obtained as follows:

[0017] in, Indicates t The first n The parameter encoding after the optimal region search, Indicates the parameter encoding after the full parameter encoding cooperative search , Indicates the sorted n +1 parameter encoding, Indicates t Parameter center encoding during the learning process.

[0018] Further, performing an adaptive full-space search on the parameter coding after the full-parameter coding cooperative search to obtain the parameter coding after the adaptive full-space search includes: According to the loss function value of the parameter encoding after the cooperative search of all parameter encodings, the position importance corresponding to each parameter encoding is obtained as follows:

[0019] in, Indicates t The first iThe loss function value corresponding to the parameter encoding after the cooperative search of all parameter encodings, Indicates t The first i The importance of the position of the parameter coding corresponding to the full parameter coding cooperative search, i =1,2,…,M, where M represents the total number of parameter codes; According to the position importance corresponding to the parameter encoding, the adaptive attenuation factor is obtained as:

[0020] in, Indicates t The first i The adaptive attenuation factor corresponding to the parameter coding after the cooperative search of the full parameter coding, Indicates t -1st learning process i The importance of the position of the parameter coding corresponding to the full parameter coding cooperative search, represents pi, and sin represents the sine function; According to the adaptive attenuation factor, the parameter coding after the cooperative search of the full parameter coding is searched, and the parameter coding after the adaptive full space search is obtained as follows:

[0021]

[0022] in, Indicates t The first i The parameter encoding after the cooperative search of all parameter encodings, Indicates the parameter encoding after adaptive full space search , Indicates t +1 optimal cumulative learning code in the learning process, Indicates t The optimal cumulative learning code in the learning process, represents the expansion factor, represents the fourth random number uniformly distributed between (0,1), represents the first optimal learning factor, represents the second optimal learning factor, represents the fifth random number between (0,1), represents the sixth random number between (0,1), Indicates t The optimal parameter encoding in the learning process.

[0023] Furthermore, a global variation search is performed on the parameter encoding after the adaptive full-space search to obtain the parameter encoding after the global variation search, including: Get the normal distribution parameters as:

[0024]

[0025]

[0026] in, represents the first normal distribution parameter, u represents the second normal distribution parameter, It means that the normal distribution parameters follow the mean of 0 and the variance is The normal distribution of represents the first normal distribution factor, and =1; It means that the second normal distribution parameter follows the mean of 0 and the variance is The normal distribution of represents the second normal distribution factor, represents the gamma function, represents a random parameter factor between (0,1), represents pi, and sin represents the sine function; According to the normal distribution parameters, the parameter encoding after the adaptive full space search is searched, and the parameter encoding after the global variation search is obtained as follows:

[0027] in, Indicates t The first j Parameter encoding after an adaptive full-space search, Indicates the parameter encoding after the global mutation search , j =1,2,…,M, where M represents the total number of parameter codes; represents the seventh random number between (0,1), represents the symbolic function, represents the search step length, Representation and parameter encoding Different random parameter encodings.

[0028] Furthermore, the learning end condition is set as: when the total learning times is greater than or equal to a preset maximum learning times, the learning end condition is met.

[0029] Furthermore, according to the demand for aviation materials and the current inventory of aviation materials, the procurement strategy is automatically optimized, and a purchase order is automatically generated according to the procurement strategy, including: Obtain the current inventory of aviation materials stored in advance or entered by staff; Based on the demand for aviation materials, determine the total demand for aviation materials in multiple future cycles, and judge whether the total demand for aviation materials is less than the current inventory of aviation materials. If so, it is determined that the inventory of aviation materials is sufficient and no purchase is required. Otherwise, the total demand for aviation materials is subtracted from the current inventory of aviation materials to obtain the purchase quantity of aviation materials, and the purchase quantity of aviation materials is used as the purchase strategy; According to the procurement strategy, the aviation material procurement volume is allocated to suppliers according to preset rules, and a purchase order is automatically generated.

[0030] Furthermore, after the purchase order is automatically generated according to the purchase strategy, it also includes: sending the purchase order to a preset approval person, and after the approval person revise and approves it, transmitting the purchase order to the supplier to complete the intelligent procurement.

[0031] The present invention provides an AI intelligent procurement and optimization method for aviation materials based on big data. The method first uses an AI algorithm to learn the historical big data to obtain an aviation material consumption prediction model with the ability to predict aviation material consumption, and uses the aviation material consumption prediction model to identify the factors affecting the aviation material consumption and determine the aviation material demand. Finally, according to the aviation material demand and the current aviation material inventory, the procurement strategy is automatically optimized, and a purchase order is automatically generated according to the procurement strategy. This method can not only effectively shorten the procurement cycle, but also effectively utilize historical big data to provide effective decision-making support for enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0033] Figure 1 A flowchart of an AI intelligent procurement and optimization method for aviation materials based on big data is provided in an embodiment of the present invention.

[0034] Figure 2 A schematic diagram of a process for learning the historical big data using an AI algorithm provided in an embodiment of the present invention.

[0035] The above drawings have shown clear embodiments of the present invention, which will be described in more detail below. These drawings and text descriptions are not intended to limit the scope of the present invention in any way, but to illustrate the concept of the present invention for those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0036] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0037] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0038] like Figure 1 As shown, an embodiment of the present invention provides an aviation material AI intelligent procurement and optimization method based on big data, including: S11, obtaining big data of historical aviation material consumption influencing factors and historical aviation material consumption, and performing data cleaning on the big data of historical aviation material consumption influencing factors and historical aviation material consumption to obtain historical big data; The big data of historical aviation material consumption influencing factors may be some influencing factors affecting aviation material consumption, and the historical aviation material consumption may be the consumption of a certain type of aviation material. Different types of aviation materials may be purchased and optimized using the method provided in the embodiment of the present invention, thereby shortening the procurement cycle and improving procurement efficiency.

[0039] S12. Use AI (Artificial Intelligence) to learn historical big data to obtain an aviation material consumption prediction model with the ability to predict aviation material consumption; Using an AI algorithm to learn the historical big data may include: using a CNN (Convolutional Neural Networks) algorithm or a BP (Back Propagation) neural network algorithm to learn the historical big data, but it is worth noting that these algorithms are only preferred implementation methods of the embodiments of the present invention, and other AI algorithms may also be selected to learn the historical big data.

[0040] S13, obtaining factors influencing aviation material consumption, and using an aviation material consumption prediction model to identify the factors influencing aviation material consumption, and determining aviation material demand; The data type and data structure of the big data of factors affecting aviation material consumption and historical factors affecting aviation material consumption should be the same, so as to ensure that the aviation material consumption prediction model can correctly identify the data and ultimately realize the prediction of aviation material demand. After the aviation material demand is predicted, procurement can be carried out based on the aviation material demand.

[0041] S14. Automatically optimize the procurement strategy based on the demand for aviation materials and the current inventory of aviation materials, and automatically generate a purchase order based on the procurement strategy to complete AI intelligent procurement and optimization of aviation materials.

[0042] The current inventory of aviation materials refers to the remaining amount of a certain type of aviation materials. Then, the total aviation material demand corresponding to multiple cycles can be determined based on the aviation material demand. When the current inventory of aviation materials is less than the total aviation material demand, procurement is required, thereby optimizing the procurement quantity each time, ensuring the accuracy of the procurement quantity, shortening the procurement cycle, and realizing AI intelligent procurement and optimization of aviation materials.

[0043] The present invention provides an AI intelligent procurement and optimization method for aviation materials based on big data. The method first uses an AI algorithm to learn the historical big data to obtain an aviation material consumption prediction model with the ability to predict aviation material consumption, and uses the aviation material consumption prediction model to identify the factors affecting the aviation material consumption and determine the aviation material demand. Finally, according to the aviation material demand and the current aviation material inventory, the procurement strategy is automatically optimized, and a purchase order is automatically generated according to the procurement strategy. This method can not only effectively shorten the procurement cycle, but also effectively utilize historical big data to provide effective decision-making support for enterprises.

[0044] In an embodiment of the present invention, historical aviation material consumption influencing factor big data and historical aviation material consumption are obtained, and data cleaning is performed on the historical aviation material consumption influencing factor big data and historical aviation material consumption to obtain historical big data, including: Obtain the aircraft type, age, flight time and maintenance cycle of the historical period, and obtain the big data of factors affecting the historical aviation material consumption; at the same time, obtain the historical aviation material consumption of the historical period; It is worth noting that in addition to the above-mentioned aircraft models, aircraft ages, flight times and maintenance cycles, big data of other historical factors affecting aviation material consumption can also be collected to make the forecast of aviation material demand more accurate.

[0045] To obtain the aircraft type, age, flight time and maintenance cycle of the historical period, you can collect data by taking days as the purchase cycle or by taking the maintenance cycle as the purchase cycle.

[0046] The historical aviation material consumption influencing factors big data and the historical aviation material consumption are deduplicated, missing values ​​are removed, and formatted in a standardized manner to obtain historical big data.

[0047] Deduplication can be achieved by retaining only one copy of the data when there are repeated big data of historical aviation material consumption influencing factors and corresponding historical aviation material consumption.

[0048] Removing missing values ​​can be: when data of a certain period is missing, the data with missing values ​​can be removed.

[0049] Standard format processing can be: for data with inconsistent lengths, zeros can be filled in front of the data to make the data more regular and easier to identify.

[0050] It is worth noting that, in addition to the above data cleaning methods, other data cleaning methods can also be used to make the data easier to identify and more accurate, such as normalizing the data.

[0051] like Figure 2 As shown, the AI ​​algorithm is used to learn the historical big data to obtain an aviation material consumption prediction model with aviation material consumption prediction capability, including: S21. constructing an initial learning model using an AI algorithm, and encoding parameters of the initial learning model to obtain a plurality of different parameter encodings; For example, when the parameters of the initial learning model are weight parameters, they can be randomly initialized between the upper and lower limits of the weight parameters, or the weight parameters can be initialized using a chaotic mapping algorithm. After the parameters are initialized, the parameters can be encoded into vectors to obtain parameter encodings, and then multiple different parameter encodings can be obtained.

[0052] S22, for each parameter code, after applying the parameters contained in the parameter code to the initial learning model, using the historical aviation material consumption influencing factor big data in the historical big data as input, using the historical aviation material consumption corresponding to the historical aviation material consumption influencing factor big data as output, and obtaining the loss function value corresponding to the parameter code; For example, the loss function value corresponding to each parameter encoding can be obtained according to the cross entropy loss function or the root mean square loss function.

[0053] S23, determining the optimal parameter encoding according to the loss function values ​​corresponding to all parameter encodings; S24, taking the optimal parameter code as the center, performing an optimal region search on the parameter code to obtain the parameter code after the optimal region search; S25, performing a full parameter coding cooperative search on the parameter coding after the optimal area search, to obtain the parameter coding after the full parameter coding cooperative search; S26, performing an adaptive full-space search on the parameter coding after the full-parameter coding cooperative search to obtain the parameter coding after the adaptive full-space search; S27, performing a global variation search on the parameter coding after the adaptive full-space search to obtain the parameter coding after the global variation search; S28. Repeat the optimal area search, full parameter coding cooperative search, adaptive full space search and global variation search until the learning end condition is met, re-obtain the optimal parameter coding according to the parameter coding after the last optimal area search, and use the re-obtained optimal parameter coding as the final parameter of the initial learning model to obtain an aviation material consumption prediction model with aviation material consumption prediction capability. Otherwise, return to the step of obtaining the loss function value corresponding to the parameter coding.

[0054] The prior art often uses gradient descent or genetic algorithms to optimize parameters. Although parameter optimization can be achieved, the parameter optimization effect is poor and it is easy to fall into local optimality, resulting in inaccurate final prediction of aviation material demand. The learning algorithm provided in the embodiment of the present invention can effectively analyze the data relationship between the big data of historical aviation material consumption influencing factors and the historical aviation material consumption, thereby improving the prediction accuracy of consumables demand.

[0055] In the embodiment of the present invention, taking the optimal parameter coding as the center, performing an optimal region search on the parameter coding to obtain the parameter coding after the optimal region search includes: According to all parameter encodings, the parameter center encoding is determined as:

[0056] in, Indicates t The parameter center encoding in the learning process is Indicates t The first m parameter encoding, m =1,2,…,M, where M represents the total number of parameter codes; According to the parameter center code and the optimal parameter code, the parameter code is searched, and the parameter code after the optimal area search is obtained as follows:

[0057]

[0058] in, Indicates the parameter encoding after the optimal area search , Indicates t The optimal parameter encoding in the learning process, represents the optimal area search coefficient, represents a natural constant, represents the first random number between (0,1), and T represents the preset maximum number of learning times.

[0059] The embodiment of the present invention performs an optimal area search for parameter codes, allowing all individuals to search around the optimal position, while incorporating the overall position of all parameter codes (i.e., parameter center code), which can effectively ensure the search efficiency of the algorithm and improve the search speed of the algorithm.

[0060] In the embodiment of the present invention, the parameter coding after the optimal area search is subjected to a full parameter coding cooperative search to obtain the parameter coding after the full parameter coding cooperative search, including: Obtain the loss function values ​​corresponding to the parameter codes after the optimal area search, and arrange them in ascending order of the loss function values ​​to obtain the sorted parameter codes; For the sorted n Parameter encoding, obtain the spiral search angle and spiral search radius:

[0061]

[0062] in, Indicates n The parameter encodes the corresponding spiral search angle, represents pi, represents the first spiral trajectory control coefficient, Indicates n The parameters encode the corresponding spiral search radius, n =1,2,…,M, where M represents the total number of parameter codes; represents the second spiral trajectory control coefficient, represents the second random number between (0,1), Represents the third random number between (0,1); According to the spiral search angle and the spiral search radius, the first cooperation coefficient and the second cooperation coefficient are obtained as follows:

[0063]

[0064] in, Indicates n The first cooperation coefficient corresponding to the parameter encoding, Indicates k The parameters encode the corresponding spiral search radius, Indicates k The parameter codes correspond to the spiral search angle, || represents the absolute value, sin represents the sine function, and max represents the maximum value function. Indicates n The second cooperation coefficient corresponding to the parameter encoding, cos represents the cosine function; According to the first cooperation coefficient and the second cooperation coefficient, the parameter coding after the optimal area search is searched, and the parameter coding after the full parameter coding cooperation search is obtained as follows:

[0065] in, Indicates t The first n The parameter encoding after the optimal region search, Indicates the parameter encoding after the full parameter encoding cooperative search , Indicates the sorted n +1 parameter encoding, Indicates t Parameter center encoding during the learning process.

[0066] The embodiment of the present invention performs a full-parameter coding cooperative search on the parameter coding after the optimal area search, which can effectively realize the cooperative search of all parameter codings, form a vortex search in the solution space, and learn the overall position of all parameter codings at the same time, which can not only improve the solution space search rate but also ensure the algorithm search speed.

[0067] In an embodiment of the present invention, performing an adaptive full-space search on parameter coding after cooperative search of full-parameter coding to obtain parameter coding after adaptive full-space search includes: According to the loss function value of the parameter encoding after the cooperative search of all parameter encodings, the position importance corresponding to each parameter encoding is obtained as follows:

[0068] in, Indicates t The first i The loss function value corresponding to the parameter encoding after the cooperative search of all parameter encodings, Indicates t The first i The importance of the position of the parameter coding corresponding to the full parameter coding cooperative search, i =1,2,…,M, where M represents the total number of parameter codes; According to the position importance corresponding to the parameter encoding, the adaptive attenuation factor is obtained as:

[0069] in, Indicates t The first i The adaptive attenuation factor corresponding to the parameter coding after the cooperative search of the full parameter coding, Indicates t -1st learning process i The importance of the position of the parameter coding corresponding to the full parameter coding cooperative search, represents pi, and sin represents the sine function; According to the adaptive attenuation factor, the parameter coding after the cooperative search of the full parameter coding is searched, and the parameter coding after the adaptive full space search is obtained as follows:

[0070]

[0071] in, Indicates t The first i The parameter encoding after the cooperative search of all parameter encodings, Indicates the parameter encoding after adaptive full space search , Indicates t +1 optimal cumulative learning code in the learning process, Indicates t The optimal cumulative learning code in the learning process, represents the expansion factor, represents the fourth random number uniformly distributed between (0,1), represents the first optimal learning factor, represents the second optimal learning factor, represents the fifth random number between (0,1), represents the sixth random number between (0,1), Indicates t The optimal parameter encoding in the learning process.

[0072] The embodiment of the present invention performs an adaptive full-space search on the parameter coding after the cooperative search of all parameter coding. When the position importance fluctuates violently, the parameter coding is transformed; when the position importance fluctuates gently, the search degree is increased to enhance the global search capability of the algorithm, thereby ensuring coding diversity and improving recognition accuracy.

[0073] In the embodiment of the present invention, a global variation search is performed on the parameter coding after the adaptive full space search to obtain the parameter coding after the global variation search, including: Get the normal distribution parameters as:

[0074]

[0075]

[0076] in, represents the first normal distribution parameter, u represents the second normal distribution parameter, It means that the normal distribution parameters follow the mean of 0 and the variance is The normal distribution of represents the first normal distribution factor, and =1; It means that the second normal distribution parameter follows the mean of 0 and the variance is The normal distribution of represents the second normal distribution factor, represents the gamma function, represents a random parameter factor between (0,1), represents pi, and sin represents the sine function; According to the normal distribution parameters, the parameter encoding after the adaptive full space search is searched, and the parameter encoding after the global variation search is obtained as follows:

[0077] in, Indicates t The first j Parameter encoding after an adaptive full-space search, Indicates the parameter encoding after the global mutation search , j =1,2,…,M, where M represents the total number of parameter codes; represents the seventh random number between (0,1), represents the symbolic function, represents the search step length, Representation and parameter encoding Different random parameter encodings.

[0078] The embodiment of the present invention performs a global variation search on the parameter encoding after the adaptive full-space search, which can effectively prevent the algorithm from falling into a local optimum. In order to ensure the search speed of the algorithm, a greedy algorithm or an annealing simulation algorithm can also be used to control the global variation search, thereby ensuring the search speed of the algorithm.

[0079] In the embodiment of the present invention, the learning end condition is set as: when the total learning times is greater than or equal to the preset maximum learning times, the learning end condition is met.

[0080] In an embodiment of the present invention, according to the demand for aviation materials and the current inventory of aviation materials, the procurement strategy is automatically optimized, and a procurement order is automatically generated according to the procurement strategy, including: Obtain the current inventory of aviation materials stored in advance or entered by staff; Based on the demand for aviation materials, determine the total demand for aviation materials in multiple future cycles, and judge whether the total demand for aviation materials is less than the current inventory of aviation materials. If so, it is determined that the inventory of aviation materials is sufficient and no purchase is required. Otherwise, the total demand for aviation materials is subtracted from the current inventory of aviation materials to obtain the purchase quantity of aviation materials, and the purchase quantity of aviation materials is used as the purchase strategy; According to the procurement strategy, the aviation material procurement volume is allocated to suppliers according to preset rules, and a purchase order is automatically generated.

[0081] For example, when the big data of historical factors affecting aviation material consumption is collected on a daily basis, the preset aviation material guarantee period can be used as the basis (i.e., the aviation material inventory needs to guarantee the number of days of aviation material consumption), and the aviation material demand can be multiplied by the aviation material guarantee period, so as to obtain the total aviation material demand corresponding to multiple future periods (i.e., the total aviation material demand for multiple days in the future). When the current aviation material inventory is less than the total aviation material demand corresponding to multiple future periods, procurement is required.

[0082] The purchase volume of aviation materials can be allocated to suppliers according to preset rules: Suppliers can be allocated according to a preset purchase ratio. For example, if there are supplier A and supplier B, the purchase ratio between supplier A and supplier B can be set to 1:1. Then, when purchasing, half should be purchased from supplier A and half from supplier B.

[0083] The technical solution of the embodiment of the present invention can realize automatic optimization of procurement strategies and automatic procurement, thereby effectively avoiding the inaccuracy problem caused by manual procurement.

[0084] In an embodiment of the present invention, after the purchase order is automatically generated according to the purchase strategy, it also includes: sending the purchase order to a preset approval person, and after the approval person revise and approves it, transmitting the purchase order to the supplier to complete intelligent procurement.

[0085] The purchase order is sent to a preset approver, who can make certain revisions to the purchase and then approve it, or it can be approved directly.

[0086] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0087] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0088] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0089] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0090] A person of ordinary skill in the art can understand that all or part of the steps in realizing the above-mentioned facts and methods can be completed by instructing the relevant hardware through a program, and the program involved or the program described can be stored in a computer-readable storage medium. When the program is executed, it includes the following steps: At this time, the corresponding method steps are derived, and the storage medium can be ROM / RAM, a disk, an optical disk, etc.

[0091] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for intelligent procurement and optimization of aviation materials based on big data, characterized in that: include: Acquire big data of historical aviation material consumption influencing factors and historical aviation material consumption, and perform data cleaning on the big data of historical aviation material consumption influencing factors and historical aviation material consumption to obtain historical big data; Using AI algorithms to learn the historical big data to obtain an aviation material consumption prediction model with aviation material consumption prediction capability; Obtaining factors influencing aviation material consumption, and using an aviation material consumption prediction model to identify the factors influencing aviation material consumption, and determining aviation material demand; According to the demand for aviation materials and the current inventory of aviation materials, the procurement strategy is automatically optimized, and the purchase order is automatically generated according to the procurement strategy to complete the AI ​​intelligent procurement and optimization of aviation materials.

2. The AI ​​intelligent procurement and optimization method for aviation materials based on big data according to claim 1 is characterized in that: Obtaining big data of factors influencing historical aviation material consumption and historical aviation material consumption, and performing data cleaning on the big data of factors influencing historical aviation material consumption and historical aviation material consumption to obtain historical big data, including: Obtain the aircraft type, age, flight time and maintenance cycle of the historical period, and obtain the big data of factors affecting the historical aviation material consumption; at the same time, obtain the historical aviation material consumption of the historical period; The historical aviation material consumption influencing factors big data and the historical aviation material consumption are deduplicated, missing values ​​are removed, and formatted in a standardized manner to obtain historical big data.

3. The AI ​​intelligent procurement and optimization method for aviation materials based on big data according to claim 1 is characterized in that: The AI ​​algorithm is used to learn the historical big data to obtain an aviation material consumption prediction model with aviation material consumption prediction capability, including: An initial learning model is constructed using an AI algorithm, and parameters of the initial learning model are encoded to obtain multiple different parameter encodings; For each parameter code, after applying the parameters contained in the parameter code to the initial learning model, the historical aviation material consumption influencing factor big data in the historical big data is used as input, and the historical aviation material consumption corresponding to the historical aviation material consumption influencing factor big data is used as output to obtain the loss function value corresponding to the parameter code; Determine the optimal parameter encoding based on the loss function values ​​corresponding to all parameter encodings; Taking the optimal parameter code as the center, the optimal region search is performed on the parameter code to obtain the parameter code after the optimal region search; Performing a full parameter coding cooperative search on the parameter coding after the optimal area search to obtain the parameter coding after the full parameter coding cooperative search; Performing an adaptive full-space search on the parameter coding after the cooperative search of the full-parameter coding to obtain the parameter coding after the adaptive full-space search; Performing a global variation search on the parameter encoding after the adaptive full-space search to obtain the parameter encoding after the global variation search; Repeat the optimal area search, full parameter coding cooperative search, adaptive full space search and global variation search until the learning end condition is met, re-obtain the optimal parameter coding according to the parameter coding after the last optimal area search, and use the re-obtained optimal parameter coding as the final parameter of the initial learning model to obtain an aviation material consumption prediction model with aviation material consumption prediction capability, otherwise return to the step of obtaining the loss function value corresponding to the parameter coding.

4. The AI ​​intelligent procurement and optimization method for aviation materials based on big data according to claim 3 is characterized in that: Centering on the optimal parameter code, the optimal region search is performed on the parameter code to obtain the parameter code after the optimal region search, including: According to all parameter encodings, the parameter center encoding is determined as: in, Indicates t The parameter center encoding in the learning process is Indicates t The first m parameter encoding, m =1,2,…,M, where M represents the total number of parameter codes; According to the parameter center code and the optimal parameter code, the parameter code is searched, and the parameter code after the optimal area search is obtained as follows: in, Indicates the parameter encoding after the optimal area search , Indicates t The optimal parameter encoding in the learning process, represents the optimal area search coefficient, represents a natural constant, represents the first random number between (0,1), and T represents the preset maximum number of learning times.

5. The AI ​​intelligent procurement and optimization method for aviation materials based on big data according to claim 4 is characterized in that: The parameter coding after the optimal area search is subjected to a full parameter coding cooperative search to obtain the parameter coding after the full parameter coding cooperative search, including: Obtain the loss function values ​​corresponding to the parameter codes after the optimal area search, and arrange them in ascending order of the loss function values ​​to obtain the sorted parameter codes; For the sorted n Parameter encoding, obtain the spiral search angle and spiral search radius: in, Indicates n The parameter encodes the corresponding spiral search angle, represents pi, represents the first spiral trajectory control coefficient, Indicates n The parameter encodes the corresponding spiral search radius, n =1,2,…,M, where M represents the total number of parameter codes; represents the second spiral trajectory control coefficient, represents the second random number between (0,1), Represents the third random number between (0,1); According to the spiral search angle and the spiral search radius, the first cooperation coefficient and the second cooperation coefficient are obtained as follows: in, Indicates n The first cooperation coefficient corresponding to the parameter encoding, Indicates k The parameter encodes the corresponding spiral search radius, Indicates k The parameter codes correspond to the spiral search angle, || represents the absolute value, sin represents the sine function, and max represents the maximum value function. Indicates n The second cooperation coefficient corresponding to the parameter encoding, cos represents the cosine function; According to the first cooperation coefficient and the second cooperation coefficient, the parameter coding after the optimal area search is searched, and the parameter coding after the full parameter coding cooperation search is obtained as follows: in, Indicates t The first n The parameter encoding after the optimal region search, Indicates the parameter encoding after the full parameter encoding cooperative search , Indicates the sorted n +1 parameter encoding, Indicates t Parameter center encoding during the learning process.

6. The AI ​​intelligent procurement and optimization method for aviation materials based on big data according to claim 5 is characterized in that: Performing an adaptive full-space search on the parameter coding after the cooperative search of the full-parameter coding to obtain the parameter coding after the adaptive full-space search includes: According to the loss function value of the parameter encoding after the cooperative search of all parameter encodings, the position importance corresponding to each parameter encoding is obtained as follows: in, Indicates t The first i The loss function value corresponding to the parameter encoding after the cooperative search of all parameter encodings, Indicates t The first i The importance of the position of the parameter coding corresponding to the full parameter coding cooperative search, i =1,2,…,M, where M represents the total number of parameter codes; According to the position importance corresponding to the parameter encoding, the adaptive attenuation factor is obtained as: in, Indicates t The first i The adaptive attenuation factor corresponding to the parameter coding after the cooperative search of the full parameter coding, Indicates t -1st learning process i The importance of the position of the parameter coding corresponding to the full parameter coding cooperative search, represents pi, and sin represents the sine function; According to the adaptive attenuation factor, the parameter coding after the cooperative search of the full parameter coding is searched, and the parameter coding after the adaptive full space search is obtained as follows: in, Indicates t The first i The parameter encoding after the cooperative search of all parameter encodings, Indicates the parameter encoding after adaptive full space search , Indicates t +1 optimal cumulative learning code in the learning process, Indicates t The optimal cumulative learning code in the learning process, represents the expansion factor, represents the fourth random number uniformly distributed between (0,1), represents the first optimal learning factor, represents the second optimal learning factor, represents the fifth random number between (0,1), represents the sixth random number between (0,1), Indicates t The optimal parameter encoding in the learning process.

7. The AI ​​intelligent procurement and optimization method for aviation materials based on big data according to claim 6 is characterized in that: Perform global mutation search on the parameter encoding after the adaptive full space search to obtain the parameter encoding after the global mutation search, including: Get the normal distribution parameters as: in, represents the first normal distribution parameter, u represents the second normal distribution parameter, It means that the normal distribution parameters follow the mean of 0 and the variance is The normal distribution of represents the first normal distribution factor, and =1; It means that the second normal distribution parameter follows the mean of 0 and the variance is The normal distribution of represents the second normal distribution factor, represents the gamma function, represents a random parameter factor between (0,1), represents pi, and sin represents the sine function; According to the normal distribution parameters, the parameter encoding after the adaptive full space search is searched, and the parameter encoding after the global variation search is obtained as follows: in, Indicates t The first j Parameter encoding after an adaptive full-space search, Indicates the parameter encoding after global mutation search , j =1,2,…,M, where M represents the total number of parameter codes; represents the seventh random number between (0,1), represents the symbolic function, represents the search step length, Representation and parameter encoding Different random parameter encodings.

8. The AI ​​intelligent procurement and optimization method for aviation materials based on big data according to claim 7 is characterized in that: The learning end condition is set as: when the total learning times is greater than or equal to the preset maximum learning times, the learning end condition is met.

9. The AI ​​intelligent procurement and optimization method for aviation materials based on big data according to claim 1 is characterized in that: Automatically optimize the procurement strategy based on the demand for aviation materials and the current inventory of aviation materials, and automatically generate purchase orders based on the procurement strategy, including: Obtain the current inventory of aviation materials stored in advance or entered by staff; Based on the demand for aviation materials, determine the total demand for aviation materials in multiple future cycles, and judge whether the total demand for aviation materials is less than the current inventory of aviation materials. If so, it is determined that the inventory of aviation materials is sufficient and no purchase is required. Otherwise, the total demand for aviation materials is subtracted from the current inventory of aviation materials to obtain the purchase quantity of aviation materials, and the purchase quantity of aviation materials is used as the purchase strategy; According to the procurement strategy, the aviation material procurement volume is allocated to suppliers according to preset rules, and a purchase order is automatically generated.

10. The AI ​​intelligent procurement and optimization method for aviation materials based on big data according to claim 1 is characterized in that: After the purchase order is automatically generated according to the purchase strategy, it also includes: sending the purchase order to a preset approval person, and after the approval person revise and approves it, transmitting the purchase order to the supplier to complete the intelligent procurement.

Images