FCA-based construction method and system for aircraft engine blade body

Abstract

The present invention proposes a method and system for constructing an aircraft engine blade ontology based on FCA, including: constructing a formal background with a thesaurus and a text set in the field of aircraft engine blade manufacturing as data sources; converting the formal background into a concept lattice; extracting concepts, attributes and instances from the concept lattice to organize them into an aircraft engine blade ontology. The present invention combines the two through the FCA method, thereby taking into account both efficiency and accuracy. In order to further improve the integrity and accuracy of the ontology, the ontology can be constructed through multiple FCA iterations. Adding an ontology quality assessment link in the entire construction link can effectively improve the integrity and accuracy of the ontology. In this link, it is necessary to conduct a quality assessment on the constructed ontology, evaluate the constructed ontology, and modify the ontology that does not meet the requirements before conducting the next round of iterations. Better results can be achieved after multiple iterations.

Description

Technical Field

[0001] The invention belongs to the field of aero-engine blade manufacturing, and in particular relates to a method for constructing an aero-engine blade body based on FCA. Background Art

[0002] With the continuous development of aviation industry, the manufacturing of aircraft engine blades has received more and more attention. There is a large amount of knowledge data in the manufacturing of aircraft engine blades. As time goes by, new knowledge data will be generated in the future. In addition, the technical level and engineering experience of experts in the field, the prior knowledge, the processing case knowledge accumulated by the workshop over the years, and the structured processing data collected in real time are scattered in various application systems of aircraft blade processing. These heterogeneous data cannot be effectively integrated and shared, forming the so-called "data island" state, making it impossible for manufacturing companies to deeply use the existing data and knowledge to provide guidance for subsequent processing and manufacturing, process optimization, etc., so as to improve production efficiency. Therefore, how to effectively manage the knowledge of aircraft engine blades and their processing has always been a problem that needs to be solved.

[0003] Building a knowledge base is an effective way to solve this problem. In the construction of knowledge bases for manufacturing, there are mainly three methods: relational, XML and ontology. Although the relational method is intuitive, it is difficult to reason about dynamic knowledge with relational data and the reasoning efficiency is low; the XML-based method still has the above problems, and the XML expression method is poor in visualization, and the knowledge representation is relatively obscure; the ontology-based representation method has good expression visualization effect and complete reasoning mechanism, and gradually becomes the mainstream method. At present, there are two major types of ontology construction methods: top-down and bottom-up. Top-down ontology construction mainly relies on manual work. For enterprises in the field of aero-engine blade processing, the latest research on blade processing mostly exists in the form of unstructured or semi-structured documents, which have the characteristics of numerous concepts, complex processes, and huge amounts of knowledge and data. Therefore, relying on manual construction of ontology is inefficient, has a long cycle, and is costly; while bottom-up ontology construction is to extract information from a large amount of data through relevant technologies, and then further aggregate the information to form an ontology. This ontology construction process does not require human intervention and is highly efficient, but there is a problem that the constructed ontology is not accurate enough. Summary of the invention

[0004] The purpose of the present invention is to overcome the problem of the prior art that efficiency and accuracy cannot be taken into account at the same time, and proposes a method for constructing an aircraft engine blade body based on FCA (Formal Concept Analysis).

[0005] In view of the shortcomings of the prior art, the present invention proposes a method for constructing an aero-engine blade body based on FCA, which includes:

[0006] Step 1: construct a formal background with a thesaurus and a text set in the field of aircraft engine blade manufacturing as data sources. The formal background is a triple K=(G, M, R), where G is a set of objects, M is a set of attributes, and R represents a binary relationship between G and M.

[0007] Step 2: According to the attribute set M and the instance set G in the formal background, the formal background is converted into a binary relationship matrix between instances and attributes; all instance sets in the binary relationship matrix are traversed to obtain the object MG with the largest attribute set; according to the attribute set M and the object MG, all combinations whose attribute sets are less than or equal to the object MG are obtained, and combined with the instance set G, all formal concepts C in the formal background are obtained;

[0008] Step 3, traverse all the formal concepts C to establish a hierarchical relationship between concepts, and obtain a partial order set by adding the top-level formal concept and the bottom-level formal concept to the hierarchical relationship. According to the position of the concepts in the partial order set, adjust the concepts in the partial order set to obtain a concept lattice, and extract the concepts, attributes and instances in the concept lattice to organize them into the ontology of aircraft engine blades.

[0009] The FCA-based aero-engine blade body construction method includes:

[0010] Step 4: Determine whether the current ontology meets the expected goal. If so, save the current ontology as the final result. Otherwise, take the current ontology as the old ontology and execute steps 1 to 3 again to obtain a new ontology. Then merge the new ontology with the old ontology as the final result.

[0011] In the FCA-based aero-engine blade body construction method, the expected goal includes the number of bodies and / or the comprehensiveness of the description of body attributes.

[0012] In the FCA-based aero-engine blade ontology construction method, the process of obtaining the new ontology includes updating the subject word list according to the new knowledge to be added to the old ontology; and editing the old ontology to remove erroneous or redundant content in the old ontology.

[0013] The present invention also proposes a method for constructing an aero-engine blade body based on FCA, which includes:

[0014] Step 1: construct a formal background with a thesaurus and a text set in the field of aircraft engine blade manufacturing as data sources. The formal background is a triple K=(G, M, R), where G is a set of objects, M is a set of attributes, and R represents a binary relationship between G and M.

[0015] Step 2: According to the attribute set M and the instance set G in the formal background, the formal background is converted into a binary relationship matrix between instances and attributes; all instance sets in the binary relationship matrix are traversed to obtain the object MG with the largest attribute set; according to the attribute set M and the object MG, all combinations whose attribute sets are less than or equal to the object MG are obtained, and combined with the instance set G, all formal concepts C in the formal background are obtained;

[0016] Step 3, traverse all the formal concepts C to establish a hierarchical relationship between concepts, and obtain a partial order set by adding the top-level formal concept and the bottom-level formal concept to the hierarchical relationship. According to the position of the concepts in the partial order set, adjust the concepts in the partial order set to obtain a concept lattice, and extract the concepts, attributes and instances in the concept lattice to organize them into the ontology of aircraft engine blades.

[0017] The FCA-based aero-engine blade body construction method includes:

[0018] Step 4: Determine whether the current ontology meets the expected goal. If so, save the current ontology as the final result. Otherwise, take the current ontology as the old ontology and execute steps 1 to 3 again to obtain a new ontology. Then merge the new ontology with the old ontology as the final result.

[0019] In the FCA-based aero-engine blade body construction method, the expected goal includes the number of bodies and / or the comprehensiveness of the description of body attributes.

[0020] In the FCA-based aero-engine blade ontology construction method, the process of obtaining the new ontology includes updating the subject word list according to the new knowledge to be added to the old ontology; and editing the old ontology to remove erroneous or redundant content in the old ontology.

[0021] The present invention also proposes a storage medium for storing a program for executing any one of the above-mentioned methods for constructing an aircraft engine blade body based on FCA.

[0022] The present invention also proposes a client, which is used for any of the above-mentioned FCA-based aircraft engine blade body construction systems.

[0023] It can be seen from the above scheme that the advantages of the present invention are:

[0024] The FCA-based aero-engine blade body construction method is a semi-automatic body construction method that can take into account both the efficiency and accuracy of body construction. At the same time, the body quality assessment link is added to the traditional FCA construction process, which further improves the accuracy of body construction. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is the overall flow chart of the present invention;

[0026] Figure 2 A flow chart for the process of converting the formal context into a concept lattice and visualizing the concept lattice;

[0027] Figure 3 This is the concept lattice corresponding to the formal background in Table 1. DETAILED DESCRIPTION

[0028] When the inventor was conducting research on knowledge management in the field of aero-engine blade manufacturing, he found that the knowledge representation method based on XML and relational database had defects such as obscure knowledge representation, low reasoning efficiency and poor scalability, which were determined by its own characteristics. After research, the inventor found that the knowledge representation method based on ontology can be used to represent knowledge, thereby solving the above defects; the two existing methods, top-down and bottom-up, have some problems. The top-down method is inefficient because it is completely constructed manually; the bottom-up method does not use existing prior knowledge, resulting in a low accuracy rate in constructing the ontology. The present invention combines the two through the FCA method, thereby taking into account both efficiency and accuracy. In order to further improve the integrity and accuracy of the ontology, the ontology can be constructed through multiple FCA iterations. Adding an ontology quality assessment link in the entire construction link can effectively improve the integrity and accuracy of the ontology. In this link, it is necessary to conduct a quality assessment on the constructed ontology, evaluate the constructed ontology, and modify the ontology that does not meet the requirements before the next round of iteration. After multiple iterations, better results can be achieved.

[0029] The present invention includes the following key technical points:

[0030] Key point 1: Adopting ontology-based knowledge representation model in the field of aero-engine blade manufacturing. The knowledge in the field of aero-engine blade manufacturing is numerous and complex, and will increase over time. Adopting ontology-based knowledge representation model can not only intuitively display the connection between knowledge, but also be easy to expand.

[0031] Key point 2: Use the FCA method to realize the ontology construction in the field of aero-engine blade manufacturing. On the one hand, the mathematical model of FCA can ensure that the ontology constructed by FCA has a high accuracy rate; on the other hand, the algorithm of FCA ontology construction is easy to implement by computer programming, which ensures that the method has a fast speed.

[0032] Key point 3: Integrate the ontology quality assessment link into the entire construction process. In order to further improve the accuracy of the ontology and the integrity of the knowledge base, an ontology quality assessment link is added in an iterative process.

[0033] To make the above features and effects of the present invention more clearly and understandably described, specific embodiments are given below and will be described in detail in conjunction with the accompanying drawings of the specification as follows.

[0034] The present invention will be further described in detail below with reference to the accompanying drawings.

[0035] A method for constructing an aero-engine blade body based on FCA, as Figure 1 shown, includes the following steps:

[0036] Step 1: Construct a formal context with the thesaurus and text set in the field of aero-engine blade manufacturing as data sources;

[0037] Step 2: Convert the formal context into a concept lattice;

[0038] Step 3: Extract concepts, attributes, and instances from the concept lattice and organize them into an aero-engine blade body;

[0039] Step 4: After domain experts evaluate the quality of the ontology, repeat Steps 1-3. Since knowledge is constantly updated, if new knowledge needs to be added to the original ontology, the thesaurus also needs to be updated; if there is no new ontology to add for a period of time, experts edit the existing ontology according to their experience and knowledge to remove some incorrect or redundant content, thereby improving the accuracy of the ontology.

[0040] 1. Construct a formal context

[0041] Constructing the context of formal concepts, that is, the formal context, is the premise of formal concept analysis. The data sources are of two types: one is the classic and complete thesaurus in this field, and the other is the text document set in this field. The thesaurus is easy to determine the structural relationship between concepts, but the semantics and attributes in the thesaurus are not rich enough; there are a large number of concepts and concept relationships in the text document set, but it is not easy to extract. Therefore, it is necessary to fuse them.

[0042] For the data in the text document set, natural language processing needs to be performed to extract the corresponding concepts and relationships. The specific process is as follows: (1) Use a Chinese lexical analysis system to perform word segmentation on the aero-engine blade text document set to obtain a dataset with part-of-speech tags; (2) Filter stop words such as "de", "di", "de" from the obtained dataset; (3) Extract core nouns and verbs according to the characteristics of Chinese grammar; (4) Take the extracted core nouns as formal objects and the corresponding objects of these nouns as the formal attributes of the objects, so as to obtain a formal context based on this text document set.

[0043] After obtaining the formal context corresponding to the text document set, the formal context obtained from the thesaurus data can be fused through relevant fusion algorithms to obtain a more complete formal context.

[0044] 2. Convert the formal context into a conceptual context

[0045] For the sake of convenience, the following is an example of the core definition of FCA.

[0046] Definition 1. A formal context is a triple K = (G, M, R), where G is a set of objects, M is a set of attributes, and R represents a binary relation between G and M. To represent an object g and an attribute m in a relation R, we can express it as gRm or (g, m)∈R.

[0047] Definition 2. Given an object set G, for an object subset The set of attributes common to all objects in A is given by For a subset of attributes The set of objects with all the properties in B is given by express.

[0048] Definition 3. A formal concept (A, B) in a formal background (G, M, R) where If A'=B and B'=A are satisfied, then A and B are respectively called the extension (Extent) and connotation (Intent) of the formal concept (A, B).

[0049] Definition 4. If (A1, B1) and (A2, B2) are two formal concepts of a formal background, then (equivalent to ), then (A1, B1) is called the subconcept of (A2, B2), and (A2, B2) is called the superconcept of (A1, B1), denoted by (A1, B1) ≤ (A2, B2). The relation ≤ is the order between formal concepts. β(G, M, R) represents the set of all (G, M, R) formal concepts ordered in this way, which is called the concept lattice of the formal background (G, M, R).

[0050] This step is to transform the obtained formal background into a concept lattice and visualize the concept lattice - generate the corresponding Hasse diagram. The input of the algorithm is the formal background, and the output is the concept lattice and the Hasse diagram. Figure 2 The steps are as follows:

[0051] Step 1: Generate all formal concepts. First, obtain the attribute set M and instance (object) set G in the formal background, and convert the formal background into a binary relationship between instance and attribute; then traverse all instance sets to obtain the object with the largest attribute set, denoted as MG; then, based on the attribute set M and the largest attribute set MG, obtain all combinations whose attribute sets are less than or equal to the largest attribute set. Finally, from these combinations and instance sets, obtain all formal concept sets C according to definition ⒉.

[0052] Table 1 A formal background

[0053]

[0054] It can be obtained that the largest object set G is {1,2,3,4,5}, and the largest attribute set M is {a,b,c,d}. By traversing M, we can generate subsets of M, such as {a}, {b}, {c}, {d}...{a,b,c},{a,b,c,d}. Similarly, by traversing the object set G, we can also get its subsets: {1},{2},...{1,2,3,4},{1,2,3,4,5}.

[0055] For the attribute set {a}, the set of objects containing attribute a is {1,2,4,5}. By definition 2, we know that {a}'={1,2,4,5}. At the same time, for the object set {1,2,4,5}, its common attribute (the intersection of the attributes of each object) is {a}. By definition 2, we know that {1,2,4,5}'={a}. Therefore, {a}'={1,2,4,5} and {1,2,4,5}'={a}. By definition 3, we know that ({1,2,4,5},{a}) is a formal concept, {1,2,4,5} is the extension, and {a} is the connotation.

[0056] For the attribute set {d}, the set of objects containing attribute d is {1,4}. By definition 2, we know that {d}'={1,4}. For the object set {1,4}, the common attributes it contains are {a,b,d}. By definition 2, we know that {1,4}'={a,b,d}. Because {d}'={1,4} but {1,4}'={a,b,d}≠{d}, by definition 3, we know that ({1,4},{d}) is not a formal concept.

[0057] According to Definition 2, we can get the formal concept set C' of the formal background in Table 1, which has 6 formal concepts, namely:

[0058] #2:({1,2,4,5},{a}) #3:({1,3,4},{b}) #4:({2,3},{c})

[0059] #5:({2},{a,c}) #6:({3},{b,c}) #7:({1,4},{a,b,d})

[0060] #2: This is just a notation for convenience, indicating that this is the second formal concept.

[0061] Step 2: Establish a direct predecessor-successor relationship between nodes, that is, a partial order relationship on the concept lattice.

[0062] First, traverse the formal concept set C and establish the hierarchical relationship between concepts. Since the constructed concept lattice is complete, it is necessary to add the top-level formal concept (Top) and the bottom-level formal concept (Bottom), and obtain a partially ordered set (H, ≤) according to the hierarchical relationship of all concepts. Specifically:

[0063] First, since the constructed concept lattice is complete, it is necessary to add the top-level formal concept (Top) and the bottom-level formal concept (Bottom) to C' to obtain C. Taking the formal background corresponding to Table 1 as an example, its bottom-level concept has the most attributes, and this formal concept is #8: ({}, {a, b, c, d}); its top-level concept has the most objects (instances), and its formal concept is #1: ({1, 2, 3, 4, 5}, {}). Therefore, the formal concept set C consists of 8 formal concepts in total, namely:

[0064] #1:({1,2,3,4,5},{}) #2:({1,2,4,5},{a}) #3:({1,3,4},{b}) #4:({2,3},{c})

[0065] #5:({2},{a,c}) #6:({3},{b,c}) #7:({1,4},{a,b,d}) #8:({},{a,b,c,d})

[0066] By traversing the formal concept set C and establishing the hierarchical relationship between concepts according to the partial order relationship in Definition 4, the corresponding concept lattice can be obtained.

[0067] Take formal concepts #4:({2,3},{c}) and #6:({3},{b,c}) as examples. The object set of #6 is A1={3}, and the object set of #4 is A2={2,3}. Obviously, According to Definition 4, #4 is the superconcept of #6 (or #6 is the subconcept of #4), and there is no formal concept #0 such that #6≤#0 and #0≤#4 hold, so #6≤#4 exists.

[0068] Therefore, according to Definition 4, we can obtain the following partial order relation ≤ (taking the formal background of Table 1 as an example): #8≤#7, #8≤#5, #8≤#6, #7≤#2, #7≤#3, #5≤#2, #5≤#4, #6≤#3, #6≤#4, #2≤#1, #3≤#1, #4≤#1.

[0069] The formal concept set C and the relation ≤ on C constitute the concept lattice, denoted as (H, ≤).

[0070] Step 3: Visualize the concept lattice and generate the corresponding Hasse diagram.

[0071] In order to intuitively show the relationship between formal concepts, the concept lattice can be represented in the form of a Hasse diagram. A node in the Hasse diagram corresponds to a formal concept in the concept lattice, and the edge of the Hasse diagram represents the existence of a partial order relationship between formal concepts. If there is #7≤#2, then there will be an edge connecting #7 and #2. Figure 3 Shown is the concept lattice corresponding to the formal background in Table 1.

[0072] Since the partially ordered set (H, ≤) has been obtained, the corresponding Hasse diagram can be easily obtained through some visualization tools. On the Hasse diagram, it is easy to obtain the hierarchical relationship between concepts, so as to explore the implicit associations and relationships between concepts, and the obtained concepts can be edited according to needs. There are two main editing methods: one is to modify the concepts on the concept grid directly based on experience, including adding / deleting some attributes for a certain concept, such as adding the "blade material" attribute to "blade"; the other is given by the design program prompt. The main principle is: when two concepts fall in the same position, according to the theory of the concept grid, the two concepts should be the same concept. For example, if the two concepts of "CNC machine tool" and "digital control machine tool" are in the same position of the concept grid, they should be the same concept. This judgment is easily given by the relevant concept grid construction and visualization tools. This only needs to be judged according to the actual situation. If it is indeed the same concept, it is merged; otherwise, other content is added to distinguish them from each other and become different concepts.

[0073] 3. Extract concepts, attributes, and instances from the concept lattice and organize them into the aircraft engine blade ontology.

[0074] Applying the concept lattice to the construction of the ontology model is essentially a mapping process. The concept lattice generated by the formal background is used as the initial ontology, and relevant content (attributes, instances) is extracted from the concept lattice to expand the initial ontology. The basic idea of ​​the algorithm is as follows:

[0075] (1) Directly remove the bottom-level concepts, map the nodes in the Hasse diagram corresponding to the concept lattice to the concepts in the ontology, and assign a name to each concept node. This name should cover the extension and connotation of the concept.

[0076] (2) The hierarchical relationship of the concept lattice corresponds to the hierarchical relationship between the concepts of the ontology, and the ontology trunk is obtained;

[0077] (3) The connotation of the concept is mapped to the attribute of the ontology concept, the extension is mapped to the instance of the ontology, and the corresponding attribute value is mapped to the attribute value of the instance in the ontology;

[0078] (4) Add attributes, axioms, instances, etc. to the ontology backbone to expand the initial ontology.

[0079] 4. Repeat, iterate, and improve the ontology

[0080] The obtained ontology may still be incomplete or inaccurate. In order to obtain a better ontology, after a round of "formal background construction-concept lattice-domain ontology", domain experts need to evaluate the constructed ontology. If the accuracy, completeness and other indicators of the ontology reach the expected goals, you can stop. The completeness includes the number of ontologies and the comprehensiveness of the ontology's attribute description; otherwise, continue to repeat the above steps. In this repetitive and iterative process, the completeness and accuracy of the ontology will increase until the expected effect is achieved. In this method of constructing the ontology of aircraft engine blades based on FCA, relevant personnel only participate in the concept and ontology evaluation, and most of the ontologies and their relationships are automatically completed by FCA.

[0081] The following is a system embodiment corresponding to the above method embodiment. This embodiment can be implemented in conjunction with the above embodiment. The relevant technical details mentioned in the above embodiment are still valid in this embodiment. In order to reduce repetition, they are not repeated here. Accordingly, the relevant technical details mentioned in this embodiment can also be applied in the above embodiment.

[0082] The present invention also proposes a method for constructing an aero-engine blade body based on FCA, which includes:

[0083] Step 1: construct a formal background with a thesaurus and a text set in the field of aircraft engine blade manufacturing as data sources. The formal background is a triple K=(G, M, R), where G is a set of objects, M is a set of attributes, and R represents a binary relationship between G and M.

[0084] Step 2: According to the attribute set M and the instance set G in the formal background, the formal background is converted into a binary relationship matrix between instances and attributes; all instance sets in the binary relationship matrix are traversed to obtain the object MG with the largest attribute set; according to the attribute set M and the object MG, all combinations whose attribute sets are less than or equal to the object MG are obtained, and combined with the instance set G, all formal concepts C in the formal background are obtained;

[0085] Step 3, traverse all the formal concepts C to establish a hierarchical relationship between concepts, and obtain a partial order set by adding the top-level formal concept and the bottom-level formal concept to the hierarchical relationship. According to the position of the concepts in the partial order set, adjust the concepts in the partial order set to obtain a concept lattice, and extract the concepts, attributes and instances in the concept lattice to organize them into the ontology of aircraft engine blades.

[0086] The FCA-based aero-engine blade body construction method includes:

[0087] Step 4: Determine whether the current ontology meets the expected goal. If so, save the current ontology as the final result. Otherwise, take the current ontology as the old ontology and execute steps 1 to 3 again to obtain a new ontology. Then merge the new ontology with the old ontology as the final result.

[0088] In the FCA-based aero-engine blade body construction method, the expected goal includes the number of bodies and / or the comprehensiveness of the description of body attributes.

[0089] In the FCA-based aero-engine blade ontology construction method, the process of obtaining the new ontology includes updating the subject word list according to the new knowledge to be added to the old ontology; and editing the old ontology to remove erroneous or redundant content in the old ontology.

[0090] The present invention also proposes a storage medium for storing a program for executing any one of the above-mentioned methods for constructing an aircraft engine blade body based on FCA.

[0091] The present invention also proposes a client, which is used for any of the above-mentioned FCA-based aircraft engine blade body construction systems.

Claims

1. A method for constructing an aircraft engine blade body based on FCA, characterized in that: include: Step 1: construct a formal background with a thesaurus and a text set in the field of aircraft engine blade manufacturing as data sources. The formal background is a triple K=(G, M, R), where G is a set of objects, M is a set of attributes, and R represents a binary relationship between G and M. Step 2: According to the attribute set M and the instance set G in the formal background, the formal background is converted into a binary relationship matrix between instances and attributes; all instance sets in the binary relationship matrix are traversed to obtain the object MG with the largest attribute set; according to the attribute set M and the object MG, all combinations whose attribute sets are less than or equal to the object MG are obtained, and combined with the instance set G, all formal concepts C in the formal background are obtained; Step 3, traverse all the formal concepts C to establish a hierarchical relationship between concepts, obtain a partial order set by adding the top-level formal concept and the bottom-level formal concept to the hierarchical relationship, and adjust the concepts in the partial order set according to the positions between the concepts in the partial order set to obtain a concept lattice, extract the concepts, attributes and instances in the concept lattice and organize them into the ontology of aircraft engine blades; Step 4: Determine whether the current ontology meets the expected goal. If so, save the current ontology as the final result. Otherwise, take the current ontology as the old ontology and execute steps 1 to 3 again to obtain a new ontology. Then merge the new ontology with the old ontology as the final result. Among them, the expected goal includes the number of ontologies and / or the comprehensiveness of the description of ontology attributes; The process of obtaining the new ontology includes updating the thesaurus according to the new knowledge to be added to the old ontology, and editing the old ontology to remove erroneous or redundant content in the old ontology.

2. A storage medium, characterized in that: A program for storing and executing the method for constructing an aircraft engine blade body based on FCA as described in claim 1.

3. A client, characterized in that: A method for constructing an aircraft engine blade body based on FCA as described in claim 1.

Images