KR-102964043-B1 - OPERATION SERVER FOR EVALUATING ENGINE VALUE USING A VECTOR ANALYSIS METHOD AND OPERATION METHOD THEREOF

KR102964043B1KR 102964043 B1KR102964043 B1KR 102964043B1KR-102964043-B1

Abstract

An operating server and a method of operation thereof are disclosed for evaluating engine value using a vector analysis method. The method of operation of the present invention includes the step of classifying into a plurality of engine groups among a plurality of steps for evaluating the value of an engine, the method of operation of the present invention includes the step of providing high-dimensional feature vectors corresponding to each of the plurality of engines as input to a pre-set clustering algorithm, and the step of classifying the plurality of engines into a plurality of clusters by comparing the distance between the pre-set distance threshold value and the high-dimensional feature vectors to generate a plurality of engine groups.

Inventors

남교훈
김봉선

Assignees

브이엠아이씨주식회사

Dates

Publication Date: 20260512
Application Date: 20260414

Claims (1)

As an operation method performed on an operating server that evaluates engine value using a vector analysis method, A step of collecting engine information related to multiple engines and obtaining attribute-specific data for each engine based on the attributes of the data included in the collected engine information; A step of generating source information for generating feature vectors of each engine by processing the data for each attribute in at least one of encoding, normalization, and embedding according to a pre-set preprocessing method corresponding to the attributes of the data for each attribute; A step of generating a feature vector of each engine based on the source information above, and inputting the feature vector into an artificial neural network model trained to output a high-dimensional feature vector for a plurality of engines to generate a high-dimensional feature vector of each engine that reflects the attribute-specific data of each engine; A step of applying the high-dimensional feature vector of each of the above engines to a clustering algorithm to classify the plurality of engines into a plurality of engine groups based on similarity between engines; A step of calculating a distance value between a high-dimensional feature vector between a plurality of engines within a specific engine group containing the engine to be evaluated and the engine to be evaluated among the plurality of engine groups above; A step of calculating a relative value score of the engine to be evaluated based on the distance value and reference value information representing price-related information for a plurality of reference engines representing engines within the specific engine group; and The method includes the step of calculating an estimated value score of the engine to be evaluated using a weight calculated based on the relative value score and the average value score of the specific engine group; wherein The step of obtaining data by the above attributes is, The data included in the above engine information is classified into one of a plurality of pre-set attributes, namely categorical attributes, numeric attributes, and unstructured attributes, and data for each attribute is generated, The above categorical attributes include manufacturer, engine model, QEC inclusion status, and operating environment classification information; the above numerical attributes include LLP remaining life, time elapsed since maintenance, cumulative cycle count since maintenance, and metal ion concentration in engine oil; and the above atypical attributes include maintenance reports, BSI images, and diagnostic log files. The step of generating the above source information is, A step of performing preprocessing to convert the data of the above-mentioned categorical attributes into discrete vectors using a pre-set encoding method to correspond a unique dimension to each category within a word set and convert them into discrete vectors; A step of performing preprocessing to normalize the data of the above numerical attributes into Z-scores based on the mean and standard deviation; A step of performing preprocessing to convert text data included in the above unstructured attributes into fixed-length vectors by applying a natural language processing-based sentence embedding model and image data into fixed-length vectors by applying a convolutional neural network-based image embedding model; and The method further includes the step of generating source information for generating the feature vector using the attribute-specific data on which the above preprocessing has been performed; The step of generating the above high-dimensional feature vector is, A step of generating an engine-specific feature vector, which is generated by combining the values of the source information for each engine, as input data for the artificial neural network model; A step of providing the above input data to the above artificial neural network model that is pre-trained to output an embedding vector that preserves the distance corresponding to the similarity between similar engines; and The method further includes the step of defining the embedding vector output for the input data in the artificial neural network model as a high-dimensional feature vector reflecting the characteristics of the attributes of each engine. The step of classifying into the above plurality of engine groups is, A step of providing high-dimensional feature vectors corresponding to each of the plurality of engines as inputs to a preset clustering algorithm; and The method further includes the step of generating the plurality of engine groups by classifying the plurality of engines into a plurality of clusters through a comparison of the distance between a preset distance threshold and the distance between the high-dimensional feature vectors. The step of calculating the above distance value is, A step of calculating a distance value between a high-dimensional feature vector of an engine to be evaluated and a high-dimensional feature vector of each reference engine that is an engine within a specific engine group including the engine to be evaluated; and The step of calculating the average of the distance values calculated above and determining the calculated average as the distance value for the engine subject to evaluation; further comprising Operation method of an operating server evaluating engine value.

Description

Operation server for evaluating engine value using a vector analysis method and method of operation therewith The present invention relates to an operating server and a method of operation for evaluating engine value using a vector analysis method, and more specifically, to an operating server and a method of operation for evaluating engine value by applying a vector analysis method for a similar evaluation of an aircraft engine in relation to aviation assets. With the growth of the aviation industry, the transaction volume of aircraft and engine assets is increasing rapidly, and the aircraft leasing market is expanding, particularly due to the growth of low-cost carriers (LCCs). In response to these market changes, the demand for quantitative valuation of high-value assets such as aircraft and engines is also surging. However, existing aircraft engine asset valuations have primarily been based on limited numerical information, such as aircraft age, cumulative flight hours, and maintenance cycles. Consequently, there have been limitations in comprehensively reflecting the impact of various factors on asset value, such as the actual operational status of the engine, maintenance history, and failure history. Furthermore, aircraft engines account for a significant portion of the aircraft's total value and are active assets in the second-hand market, making accurate technical analysis and valuation essential. In particular, the individual values of multiple engines installed in a single aircraft can vary significantly depending on their operational history and maintenance levels; valuation methods that fail to reflect these differences risk inaccurate investment decisions and potential financial losses. In reality, technical reports on engines are written in non-standardized formats and exist in various forms, such as PDFs, Excel spreadsheets, and images, making it difficult for non-experts to interpret the content and quantify asset value. Due to the unstructured and heterogeneous nature of such data, there are numerous instances where financial institutions and investors make investment decisions regarding aviation assets without expertise or reliable analysis, leading to information asymmetry within the aviation finance market. Specifically, existing valuation methods are limited to comparative analysis between similar assets and predominantly rely on single variables, failing to consider diverse operational and maintenance conditions. Consequently, there is a need to develop intelligent valuation methods capable of reflecting the actual condition of aircraft engines and enabling quantitative comparisons between similar assets. FIG. 1 is a schematic diagram showing an environment in which an operation method of an operating server performing engine value evaluation based on engine similarity evaluation according to one embodiment is performed. FIG. 2 is a diagram illustrating an exemplary configuration of hardware for an operating server that performs engine value evaluation based on engine similarity evaluation according to one embodiment. FIG. 3 is a flowchart illustrating an operation method performed on an operating server that performs engine value evaluation based on engine similarity evaluation according to one embodiment. FIG. 4 is a flowchart illustrating a method for generating source information in an operating server that performs engine value evaluation based on engine similarity evaluation according to one embodiment. FIG. 5 is a diagram illustrating the process of generating source information in an operating server that performs engine value evaluation based on engine similarity evaluation according to one embodiment. FIG. 6 is a flowchart illustrating a method for generating high-dimensional feature vectors in an operating server that performs engine value evaluation based on engine similarity evaluation according to one embodiment. FIG. 7 is a flowchart illustrating a method for creating an engine group in an operating server that performs engine value evaluation based on engine similarity evaluation according to one embodiment. FIG. 8 is a flowchart illustrating a method for determining the distance value of an engine to be evaluated in an operating server that performs engine value evaluation based on engine similarity evaluation according to one embodiment. The present invention is susceptible to various modifications and may have various embodiments; specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the invention to specific embodiments, and it should be understood that the invention includes all modifications, equivalents, and substitutions that fall within the spirit and scope of the invention. Similar reference numerals have been used for similar components in the description of each drawing. Terms such as first, second, A, B, etc., may be used to describe various components, but said compo