JP2020129267A - data structure - Google Patents

Abstract

To easily extract data based on a relation between members and a relation of constraint or the like in a data structure used by a design support device supporting design of a vehicle.SOLUTION: A data structure comprises: a first graph 31 including blocks indicating parameters regarding a structure of members of an existing vehicle or constraints on the structure; a second graph 33 including elements indicating a goal or a strategy set during development; and a knowledge graph 35 indicating a relation between the first graph and the second graph. The knowledge graph includes: nodes g11-g42 and s11-s41 corresponding to each of the blocks and the elements; an edge e1 connecting nodes corresponding to two blocks related to each other; an edge e2 connecting nodes corresponding to two elements related to each other; and at least one edge e3 connecting nodes corresponding to each of the blocks and the elements related to the blocks.SELECTED DRAWING: Figure 3

Description

The present disclosure relates to a data structure, particularly a data structure used in a design support device that supports vehicle design.

In a system including a large number of elements, such as a vehicle, each element such as a frame (member), an engine, a wheel, etc. is driven by being correlated with each other and cooperating with each other. Therefore, if the design of one of the elements is changed, the other elements are affected, and it is not easy to change the design of the system. Therefore, it is a design support system for designing a vehicle, and automatically changes the specifications of other related elements based on the input specifications input along with the design change of one element. What has been developed has been developed (for example, Patent Document 1).

JP, 2009-37509, A

Since the members forming the vehicle are correlated with each other, parameters such as the size and weight of the members forming the vehicle are determined by the relation between the members and the restrictions on the members. Therefore, it is a heavy burden for the user to extract necessary data from a large number of data at the time of designing.

In order to reduce the burden on the user when designing, it is conceivable that the design support device has a function of accepting a question (query) from the user and notifying the answer. It is preferable that the answer notified by the design support device is determined based on a relationship between members in a wide range and a relationship such as restrictions on members. Therefore, the data used in the design support device is required to have a structure that allows the data necessary for notification of the reply to be easily extracted based on these relationships.

In view of the above background, the present invention, in a data structure used in a design support device that supports a vehicle design by accepting a query from a user and notifying an answer to the query, includes relationships between members and constraints. The problem is to be able to easily extract the data corresponding to the answer based on the relationship.

In order to solve the above-described problems, an aspect of the present invention is to provide a design support device that supports a user's vehicle design by receiving a query from a user in a text format and notifying an answer to the query in a text format. A first graph (31), which is the data structure (X) of (1), including blocks (G11 to G42) associated with parameters related to the structure of existing members constituting the vehicle or constraints on the structure. , A relationship between a second graph (33) including elements (S11 to S41) associated with a goal set during the development of the vehicle or a strategy for the goal, and the first graph and the second graph And a knowledge graph (35) for indicating the nodes, the knowledge graph including nodes (g11 to g42, s11 to s41) corresponding to the blocks and the elements, and the nodes corresponding to the two blocks related to each other. An edge (e1) that connects the nodes, an edge (e2) that connects the nodes corresponding to the two elements related to each other, and an edge (e2) that connects the nodes corresponding to the block and the elements related to the block, respectively. And an edge (e3), the design support device performs a natural language process on the query to extract the intention of the question, and a conversation engine (9); and the question engine extracted by the conversation engine. A knowledge engine (11) that extracts the node corresponding to the query as a start node based on the intention, and the node corresponding to the answer from the nodes connected to the start node via the edge. And an expert frame (13) for performing extraction processing for extracting as an answer node, and the knowledge graph is used for the extraction processing by the expert frame.

According to this configuration, the data structure includes a knowledge graph, and the knowledge graph describes relationships between members and relationships between members and constraints on the members by edges. Thereby, the design support device can easily extract the data corresponding to the answer based on the relationship and the constraint between the members by referring to the knowledge graph.

Further, since the knowledge graph includes the nodes and edges corresponding to the second graph, the design support device extracts the data corresponding to the answer based on the wider relationship including the relationship between the goal and the strategy. be able to.

In the above aspect, each of the parameter, the constraint, the goal, and the strategy may be classified into a plurality of abstract levels (21, 23, 25, 27).

With this configuration, the design support apparatus can notify the abstract level answer suitable for the user.

In the above aspect, the abstraction level includes a first hierarchy (21) including any one of requirements of how the vehicle should function, physics describing the nature of the vehicle, and a design process. , A second hierarchy (23) including the parameters of the members included in the existing vehicle and the constraints required for the members, and a third hierarchy (3) including data obtained by computer simulation. 25) and a fourth hierarchy (27) containing data obtained by experiments with real objects, wherein in the extraction process, the expert frame is in the node connected to the start node via the edge. Therefore, it is preferable that the node associated with the constraint belonging to the first layer or the second layer can be extracted as the answer node.

With this configuration, the design support apparatus can output an answer based on the data included in the first layer or the second layer. Thereby, the design support device can output the answer based on the constraint included in the first layer or the second layer. As a result, the design support apparatus can notify the user of useful information not only in the result of the simulation and the result of the experiment using the actual product but also in the design including the design process and function.

In the above aspect, the expert frame corresponds to the constraint most relevant to the intent of the question extracted by the conversation engine from the nodes connected to the start node via the edge. A node may be extracted as the answer node.

According to this configuration, since the answer can be acquired by tracing the edge, the process of acquiring the answer is simplified.

In the above aspect, the design support device may include a user interface (7) capable of receiving the query from the user by voice and outputting the answer by voice.

According to this configuration, the user can input a query by voice and receive an answer by voice, so that the convenience of the design support apparatus is improved.

In the above aspect, the knowledge graph is provided with a weight for each of the edges, the weight indicating the strength of the relationship between the two nodes connected by the edge, and the question in which the weight is extracted by the conversation engine. Is preferably used for determining the strength of the relationship between the intention and the constraint.

According to this configuration, by using the weight, it is possible to easily acquire the constraint deeply related to the intention of the question extracted by the conversation engine.

According to the above configuration, in the data structure used in the design support device that supports the design of the vehicle by receiving the query from the user and notifying the answer to the query, the relation between the members and the relation such as the constraint are Based on this, the data corresponding to the answer can be easily extracted.

Configuration diagram of a design support device using a data structure according to the present invention (A) Explanatory drawing of the 1st graph, and (B) Explanatory drawing of the 2nd graph Illustration of Knowledge Graph Diagram showing an example of the method selection model The figure which shows an example of a thesaurus and an intention understanding model Sequence diagram for explaining the processing performed in the design support device

An embodiment of a data structure according to the present invention will be described below with reference to the drawings.

The data structure X is used in the design support device 1 that supports the design of the vehicle. Design support here means data about existing products (existing vehicles) and their parts, goals and strategies created in the design process of existing products, by accepting questions from users involved in design. The answer is to be derived based on the data and the like and to notify the user.

The design support device 1 is composed of a computer including a central processing unit (CPU), a memory, and a hard disk 2. As shown in FIG. 1, the design support device 1 includes a data storage unit 3, an information block 5, a user interface 7, a knowledge engine 11, and an expert frame 13.

The data storage unit 3 classifies and saves the data of the existing product structure and the design including the goal and process at the time of development of the product into a plurality of different abstract levels in the hard disk 2. More specifically, the data storage unit 3 divides the data into four hierarchies in descending order of abstraction level and stores the data. The data 21 of the first layer defines the concept, and includes parameters of the system (vehicle in this embodiment) to be designed, requirements of how the system should function, physical laws describing them, and equations. , And/or data relating to the design process. The data 23 of the second layer includes the parameters of the members constituting the system designed based on the requirements and the physical laws described in the information of the first layer, and the data regarding the functions required for the members. The data 25 of the third layer includes data obtained by computer simulation. The fourth level of data 27 represents the physical space of the actual material, test sample, and includes data obtained from real-world experiments, drawings, photographs, reports and other physical evidence.

The information block 5 stores a graph defining a data relationship in a memory. The graph stored in the information block 5 is the first graph 31 associated with the data regarding the structure of the existing product in the data storage unit 3, and the data regarding the goal and process at the time of development of the product in the data storage unit 3. 2 and a knowledge graph 35 that defines the relationship between the first graph 31 and the second graph 33.

In other words, the data held in the information block 5 of the design support device 1 includes the first graph 31, the second graph 33, and the knowledge graph 35 that defines the relationship between the two graphs 31 and 33. It has a data structure X. Each of the first graph 31, the second graph 33, and the knowledge graph 35 is a database having a graph structure including vertices and directed lines (arrows) connecting the vertices based on the relation, and is a so-called graph. It is a database.

The first graph 31 is a graph database showing a system structure of an existing product described by SysML as shown in FIG. 2A, and includes a plurality of blocks as vertices. Each block includes a table of contents, a summary, a file name, etc. of the data stored in the data storage unit 3, and is associated with the data stored in the data storage unit 3. The first graph 31 includes, as blocks, a General Constructor (hereinafter, general block 37) and a Constrain block (hereinafter, constraint block 39). The general block 37 is associated with data including parameters (load capacity, size, weight, etc.) related to the entire existing product (that is, past design target) or data including parameters related to members constituting the past design target. It is a block. The constraint block 39 is a block associated with data regarding constraints on the entire existing product (overall design target in the past) or data regarding constraints regarding members forming the existing product (past design target).

FIG. 2A shows a part of the first graph 31 according to this embodiment. The first graph 31 includes blocks S11, S21 to S27, S31, and S41 (hereinafter referred to as S11 to S41). S11 is a general block 37 associated with data relating to parameters such as the size, weight, and load capacity of the entire vehicle. S21 is a general block 37 associated with data relating to parameters such as dimensions and weights of members constituting the skeleton of the vehicle. S22 is a general block 37 associated with data relating to parameters such as engine room dimensions. S23 is a general block 37 associated with data relating to parameters such as dimensions of side members constituting the member, and S24 is a general block 37 associated with data relating to parameters such as dimensions of cross members constituting the member, S25 is a general block 37 associated with data relating to parameters such as dimensions of side members constituting the engine room. S23 and S25 may be configured to be associated with the same data. S26 is a constraint block 39 associated with data in which the functions generally required for members including side members and cross members are described. S27 is a constraint block 39 associated with the data in which the functions required of the side members forming the engine room are described. S31 is a general block 37 associated with the load bearing simulation data for the side member alone, and S41 is a general block 37 associated with the load bearing test data for the side member.

The general blocks 37 are connected to each other by solid arrows based on the inclusion relation of corresponding members. However, the direction of the arrow is set to face from the general block 37 corresponding to the included member to the general block 37 corresponding to the included member. More specifically, S21 and S22 are connected by a solid arrow pointing to S11. This means that the vehicle (S11) includes the member (S21) and the engine room (S22). S23 and S24 are connected by a solid arrow pointing to S21. This means that the member (S21) includes the side member (S23) and the cross member (S24). S25 is connected by a solid arrow pointing to S22. This means that the engine room (S22) includes the side member (S25). S31 is connected by a solid arrow pointing to S23. This means that the load-bearing simulation data (S31) is included as the information regarding the side member (S23). S41 is connected to S23. This means that the data (S41) of the load bearing test (experiment) for the side member is included in the information on the side member (S23).

The constraint block 39 is connected to the general block 37 corresponding to the member to which the constraint is applied by an arrow. However, the arrow is configured to point from the constraint to the general block 37 corresponding to the member to which the constraint is applied. More specifically, S26 is connected by dashed arrows pointing to S23, S24, and S25. This means that the request described in S26 is a restriction on the side member (S23, S25) and the cross member (S24). Similarly, S27 is connected by a dashed arrow pointing to S25. This means that the request described in S27 is a restriction on the side member (S25) forming the engine room (S22). However, in FIG. 2A, the inclusion relation and the constraint relation are shown by arrows for convenience, but the present invention is not limited to this mode. For example, the arrow tip of the arrow in FIG. May be.

The first graph 31 includes blocks associated with the first layer data 21; blocks associated with the second layer data 23; blocks associated with the third layer data 25; And blocks associated with the data 27 of the fourth hierarchy. More specifically, as shown in FIG. 2A, S11 is a block associated with the first layer data 21 and S21 to S25 are blocks associated with the second layer data 23, respectively. S31 is a block associated with the third layer data 25, and S41 is a block associated with the fourth layer data 27.

The second graph 33 is a graph database showing a target structure at the time of development described by GSN as shown in FIG. 2B, and includes a plurality of elements as vertices. Each element is associated with the data stored by the data storage unit 3, and each element describes the table of contents or the summary of the associated data. Elements are classified into three types, goals, strategies, and evidence, depending on the type of associated data. Elements classified as goals are associated with past design targets, that is, data indicating goals set during the development of existing products. Elements classified as strategies are associated with data indicating strategies for achieving the goals set during the development of existing products. The elements classified as evidence are associated with data acquired by experiments, computer simulations, or the like in the past development process of the design target. In the following, for simplification, each element in the second graph 33 will be described by using its type name (goal, strategy, evidence) as necessary.

FIG. 2B shows a part of the second graph 33 of this embodiment. The second graph 33 includes elements G11, G21, G22, G31 to G33, and G41 to G43 (hereinafter referred to as G11 to G43). In FIG. 2B, goals are indicated by rectangles, strategies are indicated by parallelograms, and evidence is indicated by ellipses. The second graph 33 is a goal (G11) described as an improvement in vehicle safety performance, a strategy (G21) of improving the withstand load of the vehicle body for the goal (G11), and a side member design for the strategy (G21). A goal (G22), a strategy (G31) of simulation for the goal (G22) of designing the side member, a goal (G32) of load bearing calculation of the side member for the strategy (G31), and a load bearing calculation result for the goal (G32) And the evidence (G33). In the present embodiment, the second graph 33 further includes a strategy (G41) of load bearing test for the goal (G22) of designing the side member, a goal (G42) of executing an actual test for the strategy (G41), and a goal. The evidence (G43) that is the result of the actual test for (G42) is included.

As shown in FIG. 2B, the goal and the strategy for realizing the goal are connected to each other by a solid arrow pointing from the goal to the strategy. A strategy and a goal for realizing the strategy are connected to each other by a solid arrow pointing from the strategy to the goal. More specifically, G11 is connected to G21 by a solid arrow pointing to G21, G21 is connected to G22 by a solid arrow pointing to G22, and G22 is a solid line pointing to G31 and G41 by a solid arrow pointing to G31. Are respectively connected to G41. G31 is connected to G32 by a solid arrow pointing to G32, and G32 is connected to G33 by an arrow pointing to G33. G41 is connected to G42 by a solid arrow pointing to G42, and G42 is connected to G43 by an arrow pointing to G43. These arrows correspond to the design process.

Like the first graph 31, the second graph 33 includes elements associated with data corresponding to the first to fourth hierarchies. In the present embodiment, G11 is associated with the first layer data 21, G21 and G22 are associated with the second layer data 23, G31 to G33 are associated with the third layer, and G41 to G43 are associated with each other. It is associated with the data 27 of the fourth layer.

As shown in FIG. 3, the knowledge graph 35 has nodes (S11 to S41, G11 to G43) as vertices and arrows (hereinafter, edges) connecting the first graph 31 and the second graph 33. It was obtained by converting it into a graph database.

More specifically, the knowledge graph 35 is obtained as follows. First, the blocks S11 to S41 of the first graph 31 are converted into nodes s11 to s41, respectively, and are connected by an edge (first edge e1) indicating inclusion and constraint, as in FIG. 2A. Next, the elements G11 to G43 of the second graph 33 are converted into nodes g11 to g43, respectively, and the nodes g11 to g43 are connected by an edge (second edge e2) as in FIG. 2B. However, the nodes s11 to s41 corresponding to the blocks S11 to S41 of the first graph 31 are respectively connected by arrows in the opposite direction to the case of the first graph 31.

Further, the blocks having high commonality between the blocks S11 to S41 of the first graph 31 and the elements G11 to G43 of the second graph 33 are respectively assigned to the nodes corresponding to the blocks of the second graph 33 from the node Connection is made by an edge (third edge e3) toward the node corresponding to the element of the graph 31 of 1. In the present embodiment, a node g32 corresponding to the element G32 (load bearing calculation) of the second graph 33 and a node s31 corresponding to the block S31 (side member load bearing simulation data) of the first graph 31 are provided. , From the node g32 to the node s31. The commonality between blocks and elements may be determined by the commonality of words and images included in their contents. With this configuration, each node is connected by an edge based on the design development process and design considerations. As a result, the knowledge graph 35 holds the relationship of the data stored in the data storage unit 3.

Each of the first edge e1, the second edge e2, and the third edge e3 is given a weight that is a numerical value indicating the strength of the relationship between the two nodes connected to the edge. The weight is determined based on the past experience of an experienced design developer (expert, expert) and the questions asked by the user so far. For example, the weight may be determined by an expert who inputs a question into the design support apparatus 1 and performs supervised machine learning for evaluating an answer derived based on his own experience. In the present embodiment, the weight is set to increase as the relationship between the two nodes connected by the edge increases. As shown in FIG. 3, the weight is represented by a numerical value between 0 and 1.

The user interface 7 is a device for receiving a question (query) from the user, outputting it in a text format, and notifying the user of the answer. The text format here means data composed only of characters that the user can understand.

In the present embodiment, the user interface 7 includes the microphone 41 to which the voice from the user is input. When there is a voice input from the user to the microphone 41, the user interface 7 receives the voice as a query by converting the voice into a text format. After that, the user interface 7 outputs the query to the conversation engine 9 in a text format. The user interface 7 may further include a keyboard 43 and an input/output port 45 (USB port or LAN port) for storing drawings, photographs, numerical data such as simulation results in the hard disk 2.

The user interface 7 includes a speaker 47 that receives text data from the conversation engine 9 to the user and outputs it as voice. The user interface 7 converts the text format data from the conversation engine 9 into voice and notifies the user of the text format data by voice. In the present embodiment, the user interface 7 includes a monitor 49 for displaying drawings, experimental data, photographs, images and textual data.

The conversation engine 9 is composed of software executed by the CPU. The conversation engine 9 holds a dictionary 51 for parsing textual data and an intention understanding model 52. In this embodiment, the dictionary 51 and the intention understanding model 52 are stored in the memory.

The dictionary 51 includes a thesaurus dictionary 51a. The synonym dictionary 51a classifies words based on meanings, and holds a plurality of synonyms and synonyms corresponding to the semantics, as shown in FIG. A semantic element indicates a meaning associated with a word, and a synonym includes a plurality of words having the same meaning.

As shown in FIG. 5, the intent understanding model 52 associates a question intent 52a with a query corresponding to the question intent 52a, which is supposed to be issued by the user (hereinafter, assumed text 52b). It is a database. In the intention understanding model 52, for example, "I want to increase the load resistance of the side frame", "I want to increase the load resistance of the side frame", etc. as the assumption text 52b corresponding to the intention 52a of the question "Increase of load resistance". It is included. Further, the intention understanding model 52 may include the intention 52a of the question "specification construction" and the corresponding assumed text 52b "I want to build a specification that satisfies the performance". Further, the intention understanding model 52 may include the intention 52a of the question "factor analysis" and the corresponding assumed text 52b "I want to analyze the failure factor". As shown in FIG. 5, the intent identifier 52c, which is an ID, is given to each intent 52a of the question.

The conversation engine 9 executes natural language processing on the query input from the user interface 7, extracts a keyword from the query, and outputs the keyword to the knowledge engine 11. The keyword here is an index word for outputting an answer based on the data held in the data storage unit 3 and the information block 5. The keywords extracted by the conversation engine 9 are preferably set based on the table of contents and the summary described in the blocks of the first graph 31 and the elements of the second graph 33.

After that, the conversation engine 9 refers to the intention understanding model 52 to understand the intention of the question indicated by the query, and outputs it to the knowledge engine 11. More specifically, the conversation engine 9 searches the assumed text 52b of the intention understanding model 52 for any that match the query. If there is a query that corresponds, the intention 52a of the corresponding question is output to the knowledge engine 11, and if it does not exist, the synonym dictionary 51a is referenced and the words included in the query are replaced with synonyms. For example, when the query "I want to increase the load capacity of the side frame" is received, the conversation engine 9 refers to the synonym dictionary 51a and replaces "rise" with "increase" to change the load resistance of the side frame. I want to increase” (hereinafter, conversion query). Next, the conversation engine 9 searches the supposed text 52b of the intention understanding model 52 for the same thing as the conversion query. If there is the same thing, the intent 52a of the corresponding question is output to the knowledge engine 11, and if not, the synonym dictionary 51a is referenced again to replace the query, and the assumed text 52b of the intent understanding model 52 is searched. To do. As described above, by substituting words included in the query with synonyms, the user's question intention is appropriately output even when the query is ambiguous or even when the assumed text 52b does not match the query. can do.

Upon receiving an input from the expert frame 13 or the knowledge engine 11, the conversation engine 9 performs natural language processing on the input data to construct an answer in text format, and outputs it as text format data to the user interface 7. .. However, when a plurality of answers are constructed by the input from the expert frame 13, the conversation engine 9 constructs a question for selecting an answer and outputs it to the user interface 7 as text format data. Further, the conversation engine 9 causes the monitor 49 to display the drawing, the experiment data, the photograph, and the image stored in the data storage unit 3 as necessary.

The knowledge engine 11 is configured as software executed by the CPU. The knowledge engine 11 determines a tracing method of the knowledge graph 35 for obtaining an answer based on the output result of the conversation engine 9, and outputs it to the expert frame 13.

For example, when the intention of the question is “specification construction”, the knowledge engine 11 traces from the node associated with the data of the upper hierarchy to the node associated with the data of the lower hierarchy, If the intent is “factor analysis”, it is recommended to output a tracing method “going back from a failed goal to find an incorrect design method”.

In this embodiment, the knowledge engine 11 holds a method selection model 61. As shown in FIG. 4, the method selection model 61 is a database including a question intent 61a and a tracing method 61b corresponding to the question intent 61a. The method selection model 61 may include an intention identifier 61c corresponding to each intention of the question. In this embodiment, the method selection model 61 is stored in the memory.

The intent 61a of the question includes "increased load resistance". A constraint trace is associated with the “increase in withstand load” as a tracing method 61b. The constraint trace is a constraint in which only the arrow tip side of the arrow is connected by tracing (tracing) an edge from the start node in the knowledge graph 35, and the constraint is associated with the data belonging to the first hierarchy or the second hierarchy. This is a method of extracting a node corresponding to.

The question intention 61a includes "display of simulation result". The calculation result trace is associated with the "display of simulation result" as the tracing method 61b. The calculation result trace is a method of extracting a node corresponding to data belonging to the third layer by tracing an edge from a start node.

The question display 61a included in the method selection model 61 may include "display of experiment result". A trace method of extracting the node corresponding to the data belonging to the fourth layer by tracing the edge from the start node may be associated with the “display of the experimental result”.

In the present embodiment, the knowledge engine 11 outputs the start node, which is the start point of tracing, and the tracing method to the expert frame 13 based on the output result of the conversation engine 9. More specifically, first, one element of the second graph 33 is selected based on the keyword extracted by the conversation engine 9, and the node corresponding to the selected element is extracted as the start node. For example, the knowledge engine 11 may be configured to extract, as a start node, a node corresponding to an element that contains the most input keywords (for example, load bearing, side members, etc.).

After that, the knowledge engine 11 refers to the method selection model 61 and outputs the tracing method corresponding to the intention of the question input from the conversation engine 9. More specifically, when the intention of the question “increase in load bearing” is input, the knowledge engine 11 has the same intention as the question input from the conversation engine 9 from the question intent 61 a of the method selection model 61. Extract things. At this time, the knowledge engine 11 may extract the same intention identifier corresponding to the intention of the question input from the conversation engine 9 from the intention identifier 61c of the method selection model 61. After that, the knowledge engine 11 refers to the method selection model 61 to acquire the tracing method corresponding to the intent or the intent identifier of the extracted question, that is, the constraint trace. After that, the knowledge engine 11 outputs to the expert frame 13 an instruction to execute the constraint trace with the extracted node as the start node.

When the intention of the question “display simulation result” is input, the knowledge engine 11 refers to the method selection model 61 and acquires a corresponding calculation result trace as a tracing method. After that, the knowledge engine 11 outputs to the expert frame 13 an instruction to execute the calculation result trace with the extracted node as the start node.

In this embodiment, the knowledge engine 11 includes an inference engine 62 that makes an inference based on the intention of the question output by the conversation engine 9. The inference engine 62 includes the existing vehicle structure stored in the data storage unit 3, past simulation data, past test data, past failure analysis, that is, information on the third layer and information on the fourth layer. Inference is performed using a known algorithm with reference to the data, and an appropriate coping method is output. The inference engine 62 outputs data indicating a coping method to the conversation engine 9 as needed.

The expert frame 13 is configured as software executed by the CPU. The expert frame 13 traces the knowledge graph 35 stored in the information block 5 according to the tracing method determined by the knowledge engine 11, and performs an extraction process of extracting a node corresponding to the answer to the query from the user as an answer node.

However, when the expert frame 13 extracts a plurality of nodes that can be answer nodes (hereinafter, answer node candidates), it extracts the node most closely related to the keyword extracted by the conversation engine 9 from the answer node candidates as the answer node. To do. At this time, the expert frame 13 may use the weight for determining the strength of the relationship. In this embodiment, when a plurality of answer node candidates are extracted, the expert frame 13 extracts the shortest route from the start node to each node using the knowledge graph 35, and the average value of the weights of the nodes provided in the route. Is calculated, and the node most closely related to the start node is extracted as the answer node. After that, the expert frame 13 extracts the data associated with the answer node from the data storage unit 3 and outputs it to the conversation engine 9 as data corresponding to the answer to the query. After outputting the data corresponding to the answer to the conversation engine 9, the expert frame 13 increases the weight of the edge on the route of the shortest route.

At this time, the expert frame 13 may output the answer node to the inference engine 62 when the answer node is a node corresponding to the strategy. The inference engine 62 makes an inference according to the design method described in the strategy corresponding to the answer node, and takes an appropriate countermeasure (for example, in the past designs, the side frame is often made of steel having a thickness of 0.5 mm or more, Etc.) and outputs them to the conversation engine 9. Thereby, the conversation engine 9 notifies the user from the user interface 7 of the coping method derived by the inference engine 62.

Further, when the expert frame 13 extracts a plurality of nodes which can be the answer nodes, the expert frame 13 may output information of all the extracted nodes to the inference engine 62. The inference engine 62 may rank the nodes in a suitable order as an answer, and output the data associated with the nodes to the conversation engine 9 according to the rank. The inference engine 62 may perform the ordering by calculating the average value of the weights in the route from the start node to each node and arranging the nodes in order from the one having the largest average value of the weights. The inference engine 62 may output the average value of the calculated weights as the edge strength to the conversation engine 9 together with the data. At this time, the conversation engine 9 may notify the user of the data received from the inference engine 62 and the strength of the corresponding edge from the user interface 7.

In the present embodiment, the design support device 1 further includes an analysis platform 71 and a knowledge input system 73. The analysis platform 71 and the knowledge input system 73 are both configured as software executed by the CPU.

The analysis platform 71 provides an environment in which well-known algorithms necessary for performing inference by the inference engine 62 can be executed. In the present embodiment, the analysis platform 71 can execute algorithms such as a Bayesian model, a neural network, a decision tree, and image recognition. The analysis platform 71 may be further configured to provide an environment for executing an algorithm of natural language processing performed in the conversation engine 9.

The knowledge input system 73 receives the input from the user interface 7 and uses the algorithm of the natural language processing, and the data stored in the data storage unit 3 and the first graph 31 and the second graph 31 stored in the information block 5. The graph 33 and the knowledge graph 35 are updated. More specifically, when the user interface 7 receives the data converted into the text format data, the knowledge input system 73 first analyzes the relationship between the words included in the data in order to understand the sentence structure. Then, the knowledge input system 73 determines the abstraction level of the data based on the relation between the sentence structure and the words, and stores it in the data storage unit 3. Further, the knowledge input system 73 adds the block of the first graph 31 and the element of the second graph 33 based on the sentence structure and the relationship between words and associates them with the data stored in the data storage unit 3. Further, the knowledge input system 73 adds the nodes and edges of the knowledge graph 35 according to the added blocks of the first graph 31 and the added elements of the second graph 33. Further, the knowledge input system 73 saves numerical data such as drawing, photograph, and simulation data input via the input/output port 45 in the data storage unit 3 and the information block 5 as necessary. It is possible to add a block corresponding to the created first graph 31, an element of the second graph 33, and a node corresponding to the knowledge graph 35. As a result, the data stored in the data storage unit 3 and the graph stored in the information block 5 are changed according to the input from the user, so that the information can be accurately provided to the user.

Next, the operation of the design support device 1 according to this embodiment will be described. As shown in FIG. 6, when the user inputs a voice "I want to increase the load resistance of the side frame" into the microphone 41, the user interface 7 uses the input voice as a text-based query (for example, "Saidou Fure"). -I want to give you a lot of information") and output it to the conversation engine 9.

The conversation engine 9 performs natural language processing on the query converted by the user interface 7 using the dictionary 51, and converts the query into a sentence "I want to increase the load resistance of the side frame". Next, the conversation engine 9 extracts four keywords “side frame”, “load bearing”, “raise”, and “tai” from the query.

Further, the conversation engine 9 refers to the intention understanding model 52 to extract the intention of the question corresponding to the query. More specifically, the conversation engine 9 refers to the intent understanding model 52, extracts an assumed text including the same as the query, and acquires the intent “increased load bearing” of the corresponding question. Then, the conversation engine 9 uses the four keywords (“side frame”, “load bearing”, “raise”, and “tai”) corresponding to the query and the intention of the question (“increase load bearing”) as a knowledge engine. Output to 11.

The knowledge engine 11 extracts the node corresponding to the element containing the most keywords output from the conversation engine 9 from the second graph 33 as the start node. In the example of FIG. 2B, the knowledge engine 11 extracts the node g32 including the “side member” and the “withstand load” as the start node. Then, the knowledge engine 11 refers to the method selection model 61, and acquires a constraint trace corresponding to the intention “increase in load bearing” of the question input from the conversation engine 9 as a tracing method. Next, the knowledge engine 11 outputs to the expert frame 13 an instruction to execute the constraint trace in which the start node is the node g32.

When the execution instruction of the constraint trace with the start node as g32 is input, the expert frame 13 traces the knowledge graph 35 stored in the information block 5 from the node g32 according to the constraint tracing method, as shown in FIG. That is, the expert frame 13 follows the edge from the start node g32 in the knowledge graph 35, only the arrowheads of the arrows are connected, and the node s26 corresponding to the constraint associated with the data belonging to the first hierarchy or the second hierarchy, And s27 are selected as answer node candidates.

Next, the expert frame 13 uses the weight to extract, from the answer node candidates s26 and s27, the node corresponding to the constraint most closely related to the keyword extracted by the conversation engine 9 as an answer node. More specifically, the expert frame 13 extracts the shortest route R1 (see the dashed arrow in FIG. 3) from the start node g32 to the node s26. After that, the expert frame 13 calculates the average value of weights (0.8+0.4+0.3)/3=0.50 by taking the sum of the weights of the nodes described in the route R1 and dividing by the number of nodes. .. Next, the expert frame 13 extracts the shortest route R2 from the start node g32 to the node s27 (see the double-dotted line arrow in FIG. 3), and similarly to the route R1, the average value of the weights of the route R2 (0. 8+0.4+0.3+0.3+0.8)/5=0.52 is calculated. The expert frame 13 extracts s27 having a large average weight value from the answer node candidates s26 and S27 as an answer node, and associates with the answer node s27 from the data in the data storage unit 3 The data showing the constraint of “the largest energy absorption” and “deformed so as not to affect the other vehicle” is extracted and output to the conversation engine 9. As described above, by using the weight, it is possible to easily extract and acquire the constraint most closely related to the keyword extracted by the conversation engine 9.

After outputting the data associated with the answer node s27 to the conversation engine 9, the expert frame 13 increases the weight of each edge located on the route of the shortest route R1 by 0.01.

As shown in FIG. 6, the knowledge engine 11 further performs inference by the inference engine 62 using a well-known algorithm based on the keyword output from the conversation engine 9 to perform an appropriate coping method (for example, in the side frame). The data indicating the increase of the cross-sectional area, etc.) is output to the conversation engine 9.

When the coping method is input from the inference engine 62, the conversation engine 9 uses the coping method input from the inference engine 62 as data in an appropriate text format (for example, "How about increasing the cross-sectional area of the side frame"). And output to the user interface 7. The user interface 7 generates a voice from the speaker 47 according to the text format data indicating the coping method input from the conversation engine 9 to notify the coping method to the user.

Further, the conversation engine 9 gives the respective data to the user in response to the two inputs of the knowledge engine 11 "the largest energy absorption in the vehicle body at the time of a front collision" and "deform so as not to affect the opponent vehicle". The two corresponding answers, more specifically, "Please try to absorb the largest amount of energy in the vehicle body at the time of a frontal collision," and "Please transform it so that it does not affect the other vehicle." Build one answer. After that, construct a question for the user to identify one of the multiple answers (for example, "Does the side member have the largest energy absorption in the vehicle body at the time of front collision?") 7 is output from the speaker 47 as voice. Thereafter, the conversation engine 9 receives the user's answer to the question from the user interface 7 (for example, “yes”), identifies an appropriate answer from the plurality of answers, and outputs it to the user interface 7. The user interface 7 notifies the user of the answer input from the conversation engine 9 by causing the speaker 47 to generate, for example, a voice "Please transform so as not to affect the other vehicle".

When the user utters "Please display the simulation result of the side frame" through the microphone 41, the conversation engine 9 selects three keywords "side frame", "simulation", and "result" from the query. Extract. Further, the conversation engine 9 refers to the intention understanding model 52 and acquires “display of simulation result” as the intention of the question corresponding to the query. The conversation engine 9 outputs three keywords (“side frame”, “simulation”, and “result”) corresponding to the query and the intention of the question (“display of simulation result”) to the knowledge engine 11.

The knowledge engine 11 selects g32 as the start node based on the keyword. Next, the knowledge engine 11 refers to the method selection model 61, and acquires the calculation result trace corresponding to the intention of the question “display of simulation result” input from the conversation engine 9 as a tracing method. After that, the knowledge engine 11 instructs the expert frame 13 to execute the calculation result trace with g32 as the start node. As a result, the expert frame 13 extracts the node g33 associated with the data 25 of the third layer as the answer node, and the conversation engine 9 causes the monitor 49 of the user interface 7 to display the simulation result. After that, the expert frame 13 increases the weight of the edge connecting g32 and g33 by 0.01.

Next, the effect of the data structure X according to this embodiment will be described. In designing and developing a new vehicle, execution of simulations and experiments with real objects are indispensable. However, it often takes time and money to perform simulations and experiments using real objects, and the data about knowledge, experience, and failures so far are saved so that unnecessary experiments and simulations can be performed and experiments can be omitted. Is preferably provided.

Data such as knowledge, experience, and failures in existing product design/development processes up to now are often stored in two types: a first graph 31 and a second graph 33. The first graph 31 is mainly created after the development is completed, and corresponds to a specification or the like in which much knowledge about parameters such as dimensions of existing products is described. On the other hand, in the second graph 33, it is mainly created during product development, and many objectives in the design/development process and processes including failures are described. It is desirable that the design support device 1 constructs an answer to the user based on the first graph 31 and the second graph 33 of the existing product.

However, when the data associated with the first graph 31 and the second graph 33 is decomposed into word levels to form a graph database, relationships between blocks such as elements and constraints and elements such as goals and strategies are It is difficult to inform the user of the appropriate answer because the relationship is lost.

A data structure X including a first graph 31, a second graph 33, and a knowledge graph 35 stores relationships between blocks and elements. Therefore, the design support device 1 can notify the user of the answer by tracing (tracing) the knowledge graph 35 according to the relationship between blocks such as elements and constraints and the relationship between elements such as goals and strategies. As a result, it is possible to notify the user of a more appropriate answer based on the knowledge, experience, and failure data, and it is possible to omit unnecessary experiments and simulations and experiments.

Further, since the construction of the answer is performed by tracing the knowledge graph 35, the process for obtaining the answer is simpler than the case of using the algorithm such as inference.

The knowledge graph 35 is associated with data belonging to the four abstract levels of the first to fourth layers. As a result, the design support apparatus 1 can construct an answer based on a wide range of abstract level data, as compared with the case of tracing the knowledge graph 35 associated with one abstract level. This makes it possible to provide the user with more appropriate information.

Further, the knowledge graph 35 is configured based on the first graph 31 and the second graph 33. As a result, the design support device 1 makes an answer based on a wider range of data than in the case where the knowledge graph 35 is configured based on only one of the first graph 31 and the second graph 33. Can be output. This allows the user to be provided with more appropriate information.

Inexperienced design developers often refer to the second graph 33 described by GSN to proceed with the design. On the other hand, an expert often proceeds with the design by referring to the first graph 31 described by SysML and the second graph 33 described by GSN. By using the knowledge graph 35, the design support device 1 can output an answer based on a wider range of data, which is particularly useful for inexperienced design developers.

As the design support device 1, it is possible to use an expert system. In the expert system, an expert constructs a knowledge base in the form of "IF (if)... THEN (if...)", and inference is performed based on the data. Therefore, it is not suitable to implement the case where there is not only one answer, and when there are multiple points to be noted, it is difficult to configure to notify the user of multiple answers.

When a plurality of data are associated with the answer node, the design support device 1 makes a plurality of answers (“The side member has the largest energy absorption in the vehicle body at the time of a frontal collision.”) An answer "Please configure so that it does not affect the other vehicle" is provided) and a question that prompts the user to select an appropriate answer. Thereafter, the design support device 1 selects an answer according to the user's reply to the question. In this way, the design support device 1 can build a plurality of answers. Further, the design support device 1 can promptly notify the user of an appropriate answer by asking the user a question to select an answer from a plurality of answers.

When the intention of the query question is design change or impact evaluation, it is preferable to notify the user of an answer other than the simulation result or the experimental result by the actual product. In the design support device 1, when the intention of the query question is “increase in load bearing”, the answer is constructed based on the data indicating the constraint associated with the data 23 of the first layer or the second layer. It As a result, the user is notified of an answer other than the simulation result and the actual experiment result. As described above, in the design support device 1, since the data is classified into the plurality of abstract levels in the data storage unit 3, it is possible to construct an answer based on the data of the appropriate abstract level and notify the user of the answer. it can.

In the design support device 1, the user can input a query by voice and receive an answer by voice. In this way, since the interaction between the design support device 1 and the user is performed based on the voice, the user does not need to input the query using the keyboard 43.

Further, the weight given to the edge by the expert frame 13 is updated by the voice exchange between the design support apparatus 1 and the user, and the knowledge input system 73 stores the data stored in the data storage unit 3, the first data. The graph 31, the second graph 33, and the knowledge graph 35 are updated. By updating the weight given to the edge, each time the design support apparatus 1 is used, the weight between the nodes located at the route traced by the expert frame 13 becomes large. As a result, the relationship between the nodes is changed, and the expert frame 13 can more efficiently extract the data of high importance to the user. Furthermore, the weight, the data stored in the data storage unit 3, the first graph 31, the second graph 33, and the knowledge graph 35 are updated by not only the input from the keyboard 43 but also the interaction with the voice. It is possible to save the trouble of inputting data using the keyboard 43. As a result, the design support device 1 becomes easier for the user to use, and data from the user can be stored in the data storage unit 3 more quickly and efficiently.

Weights are determined based on expert evaluation. Thereby, the user can be notified of the answer based on the evaluation of the expert. If a plurality of nodes are extracted by the expert frame 13 and a plurality of answers are notified together with the strength of the edge, the user can understand the expert's evaluation of the answer by the strength of the edge.

Although the description of the specific embodiment is finished above, the present invention is not limited to the above embodiment and can be widely modified and implemented. In the above embodiment, the design support device 1 outputs the answer by tracing the knowledge graph 35 corresponding to one existing product, but the present invention is not limited to this mode. The design support device 1 may be configured to hold a plurality of knowledge graphs 35, extract an appropriate knowledge graph 35 from the plurality of knowledge graphs 35, and output an answer by tracing the extracted knowledge graph 35. ..

In the above embodiment, the first layer in the first graph 31 includes only the parameters of the system to be designed, but the present invention is not limited to this mode. The first layer of the first graph 31 may include a block corresponding to a physical law describing a property of a system to be designed or a design development process.

In the above embodiment, the first graph 31 is described by SysML and the second graph 33 is described by GSN, but the invention is not limited to this aspect. The first graph 31 and the second graph 33 may each be a graph database, and may be, for example, a graph database described by UML or a graph database obtained by a GQM (Goal Question Metric) method. However, the first graph 31 is preferably a graph database in which parameters and constraints are described, and the second graph 33 is preferably a graph database in which goals and strategies are described. Further, in the above embodiment, the second graph 33 includes the elements of goals, strategies, and evidences, but may further include contexts.

In the above embodiment, the dictionary 51 is configured to be included in the software executed by the CPU, but the present invention is not limited to this mode, and the data storage unit 3 holds the dictionary 51 in the hard disk 2 and the conversation engine 9 May be configured to appropriately refer to the dictionary 51 held in the data storage unit 3.

In the above embodiment, an example in which a plurality of nodes are extracted by the expert frame 13 and a plurality of answers are notified to the user together with the edge strength is shown, but the embodiment is not limited to this. For example, instead of notifying the user of the edge strength itself, the edge strength may be converted into a numerical value that monotonically increases in the edge strength or a ratio determined to be useful by an expert and notified to the user.

In the above embodiment, the design support device 1 is configured by one computer, but the present invention is not limited to this mode. The design support device 1 may be configured by cooperating a plurality of computers connected by a network such as the Internet. The design support device 1 may be configured by a so-called cloud computing system including a plurality of servers on the network.

In the above embodiment, the design support device 1 is used for vehicle design support, but the present invention is not limited to this mode. The design support device 1 may be applied to, for example, design support for transportation equipment or machines such as ships and aircrafts.

1: Design support device 3: Data storage unit 5: Information block 7: User interface 9: Conversation engine 11: Knowledge engine 13: Expert frame 21: First layer data 23: Second layer data 25: Third Layer data 27: fourth layer data 31: first graph 33: second graph 35: knowledge graph 37: general block 39: constraint block 41: microphone 43: keyboard 45: input/output port 47: speaker 49: Monitor 51: Dictionary 61: Inference engine 71: Analysis platform 73: Knowledge input system S11 to S41: Blocks G11 to G42: Elements s11 to s41, g11 to g42: Nodes e1, e2, e3: Edge X: Data structure

Claims (6)

A data structure of a design support device that supports a user's vehicle design by receiving a query from a user in a text format and notifying an answer to the query in a text format,
A first graph including a block associated with a parameter relating to a structure of an existing member constituting the vehicle or a constraint on the structure;
A second graph including elements associated with goals set during the development of the vehicle or strategies for reaching the goals;
A knowledge graph showing the relationship between the first graph and the second graph,
The knowledge graph connects the nodes corresponding to the blocks and the elements, the edges connecting the nodes corresponding to the two blocks related to each other, and the nodes corresponding to the two elements related to each other. And at least one edge connecting the nodes corresponding to the block and each of the elements related to the block,
The design support device performs a natural language process on the query to extract a question intent, and a node corresponding to the query based on the question intent extracted by the conversation engine. A knowledge engine for extracting as a start node, and an expert frame for performing an extraction process for extracting the node corresponding to the answer as an answer node from the nodes connected to the start node via the edge,
A data structure, wherein the knowledge graph is used in the extraction process by the expert frame. The data structure according to claim 1, wherein each of the parameter, the constraint, the goal, and the strategy is classified into a plurality of abstract levels. The abstraction level is included in an existing vehicle, with a first hierarchy that includes any one of requirements for how the vehicle should function, physics describing the nature of the vehicle, and the design process. Includes a second hierarchy including the parameters of the member and the constraints required for the member, a third hierarchy including data obtained by computer simulation, and data obtained by actual experiment Including the fourth hierarchy,
In the extraction processing, the expert frame is associated with the constraint belonging to the first layer or the second layer as the answer node from the nodes connected to the start node via the edge. The data structure according to claim 2, wherein the node can be extracted. The expert frame includes the answer node that corresponds to the constraint that is most closely related to the intent of the question extracted by the conversation engine from the nodes that are connected to the start node through the edge. The data structure according to any one of claims 1 to 3, wherein the data structure is extracted as. The data structure according to claim 4, wherein the design support device includes a user interface capable of receiving the query from the user by voice and outputting the answer by voice. In the knowledge graph, a weight indicating the strength of the relationship between the two nodes connected by the edge is given to each of the edges,
The data structure according to claim 4 or 5, wherein the weight is used to determine the strength of the relationship between the intent of the question extracted by the conversation engine and the constraint.

Images