KR101984760B1 - Self-designing modeling system and method using artificial intelligence - Google Patents

Abstract

The embodiments of the present invention may include loading at least one design design tool stored in a database and driving the design design tool, collecting design design data generated during the design process when the designer designs the object using the driven design tool, The collected design design data is machine-learned to accumulate learning design data, and a self-design model capable of self-design based on the learning design data is generated.
Accordingly, embodiments of the present invention replicate the capability of a designer with skillful design know-how, create a virtual designer who learned it, and the average level designers can also use the data created by the virtual designer It is possible to derive the design design quickly and optimally.

Description

[0001] SELF-DESIGNING MODELING SYSTEM AND METHOD USING ARTIFICIAL INTELLIGENCE [0002]

Various embodiments of the present invention are directed to techniques that can improve the accuracy and time of product design without the skilled artisan.

In recent years, with the development of computer technology and software technology, product design design technology is rapidly advancing. For example, a number of design tools have been developed to develop the product, enabling even more precise product design design and rapid development of design development.

As a result, design designers can easily design product design by easily handling the design tool. However, when the design tool is used, many efforts, experiences, and know-how (referred to as "proficiency" Product design can be completed.

Despite this, regardless of whether the proficiency is high or low, existing designers still use the design tool to reduce the quality of the product design when designing the product, and it takes a lot of time and money There is a problem in that it is increased.

Title: METHOD AND SYSTEM FOR SELF-LEARNING AND SELF-IMPROVING A SEMICONDUCTOR MANUFACTURING TOOL United States Patent Application 20080105670 Kind Code: A1 A method and system for self-learning and self-improvement of a semiconductor manufacturing tool.

Various embodiments of the present invention are directed to a self-designing modeling system and method for enhancing the completion rate of a design by reflecting machine learning of artificial intelligence into a designer's design.

According to an aspect of the present invention, there is provided an information processing apparatus including an input / output interface; A database for storing data or instructions; A tool driver for loading and driving at least one design design tool stored in the database; And an interceptor unit for collecting design design data generated during a design process through the input / output interface when the designer designs the object using the at least one driven design design tool; A self-designing device that includes a machine learning device that accumulates learning design data by machine learning of the collected design design data, and generates a self-design model capable of self-design based on the accumulated learning design data Modeling system.

In one embodiment, the design design data includes at least one of design input data input by the designer through the input / output interface, design output data output, design history data accumulated the design input data and the design output data .

In one embodiment, the machine learning apparatus includes a self-design modeling unit that improves the performance of the self-design model while the virtual designer continues to apply the design design data and the accumulated learning design data to the at least one design design tool .

In one embodiment, the self-design modeling unit enhances the performance of the self-design model by maximizing compensation of the learning design data by interaction between the design design data and the learning design data. And the like.

In one embodiment, the self-design modeling unit includes a map learning modeling unit that designates the learning design data as reference points and determines whether the design data matches the reference points to improve the performance of the self-design model can do.

In one embodiment, the self-design modeling unit classifies the learning design data and determines whether the design data exists within the range of the clustered learning design data to improve the performance of the self-design model. .

According to another aspect of the present invention, there is provided a method of designing a designing apparatus, the method comprising: loading at least one design design tool stored in a database in a designing apparatus; Collecting design design data generated in a design process when the designer designs the object using at least one design design tool driven in the design design apparatus; And storing the learning design data by machine learning the collected design design data in a machine learning apparatus and generating a self-design model capable of self-design based on the accumulated learning design data A self-design modeling method is provided.

In another embodiment, the design design data may include at least one of design input data input by a designer through an input / output interface, design output data outputted, and design history data accumulating the design input data and the design output data have.

In another embodiment, the self-design modeling method further comprises the steps of: in a machine learning apparatus, a virtual designer continuously applying the design design data and the accumulated learning design data to the at least one design design tool to improve the performance of the self- .

In another embodiment, the step of enhancing performance may be performed by interacting with the design design data and the learning design data to enhance the compensation of the learning design data to maximize the performance of the self-design model . ≪ / RTI >

In another embodiment, the step of improving performance includes the step of designating the learning design data as reference points and determining whether the design data corresponds to the reference points to improve the performance of the self-design model can do.

In another embodiment, the step of improving performance comprises grouping the learning design data and improving the performance of the self-design model by determining whether the design design data is present within the clustered learning design data can do.

As described above, various embodiments of the present invention can provide the following advantageous benefits by introducing artificial intelligence technology related to machine learning in design design.

First, the present invention replicates the ability of designers with proficient skill including design know-how as data, creates a virtual designer who learned it, and the average level designers can utilize the data created by the virtual designer And optimum design design can be derived.

Second, the present invention improves to a self-design model that can be self-designed utilizing a virtual designer trained based on design design data generated in designing, so that a human designer can design a self- Can be used as a self-design model that can be self-designed to reach the optimum design quickly and accurately.

Third, the present invention can reduce the time and cost required to derive a new design design, and can capitalize on the capabilities of advanced designers.

The effects of the present invention are not limited to those described above, and include various other effects that can be clearly understood by those skilled in the art to which the present invention belongs.

BRIEF DESCRIPTION OF THE DRAWINGS The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of the specification, serve to explain the principles of the invention. It is to be understood, however, that the technical features of the present invention are not limited to the specific drawings, and the features disclosed in the drawings may be combined with each other to constitute a new invention.
1 is a block diagram schematically showing a self-design modeling system according to an embodiment of the present invention.
FIG. 2 is a block diagram showing the structure of a design designing apparatus and a machine learning apparatus according to an embodiment of the present invention in more detail.
3 is a block diagram showing the structure of a self-designing modeling unit according to an embodiment of the present invention.
4 is a flowchart illustrating a self-design modeling method according to another embodiment of the present invention.
FIG. 5 is a flowchart showing a machine learning algorithm applied to the self-design modeling method of FIG.

In the following description and the claims, the terms "comprises", "comprising" and the like mean that the constituent elements are provided as one of the various constituent elements unless otherwise specifically stated, Is not to be excluded.

In addition, the suffix " part " for the constituent elements used in the following description and the claims is not limited to the meaning or role itself, May mean a unit or block that performs a specific function by distinguishing hardware or software or functions performed by combining hardware and software.

It should be understood that the self-design modeling system described in this specification refers to a system that reflects the artificial intelligence design design technique unlike the conventional design system.

1 is a block diagram schematically showing a self-design modeling system according to an embodiment of the present invention.

Referring to FIG. 1, a self-design modeling system according to an embodiment of the present invention includes a design design apparatus 100, a machine learning apparatus 200, and a database 300.

The design design apparatus 100 can generate design design data that is repeatedly processed in the design design tool when a design designer carries out a design design for an object on the design design tool. The design design data mentioned may be structured design or modified edited data of the object. The editing data may include correction, deletion and change results generated in the process of designing the object.

The objects mentioned may be any product or device structure that requires design design, including commodities, computers, etc., as well as large structures such as resource plants, nuclear power plants, chemical plants, and ships.

The machine learning apparatus 200 is a system that learns the data processed in the design apparatus 100, for example, repetitively processed design design data through a trained virtual designer, thereby generating a self-design model a self-design model can be created.

The database 300 stores programmed design tool data, stores the processed data in the design design apparatus 100 and the machine learning apparatus 200 described above, and may further store instructions related to the data.

The referred database 300 is a concept including a computer-readable recording medium. It includes not only a consultation database but also a database in a broad sense including a data record based on a file system, It can be included in the category of the database referred to in the present invention as long as it can retrieve and extract data.

Meanwhile, the design design apparatus 100, the machine learning apparatus 200, and the database 300 may be connected to each other by wire communication or wireless communication, or may be connected to each other by a wired communication network or a wireless communication network.

For example, a serial communication, an infrared communication (not shown) may be performed between the designing design apparatus 100 and the machine learning apparatus 200, between the design design apparatus 100 and the database 300, and between the machine learning apparatus 200 and the database 300 And can be connected to a cable-type wired communication capable of bidirectional communication.

Or, the design design apparatus 100 and the machine learning apparatus 200 may be connected to each other through a wired communication network or a wireless communication network including a distributed network. For example, when a plurality of computing devices are configured as one machine learning apparatus 200 through a distributed network, one machine learning apparatus 200 can be connected to the design design apparatus 100 on a one-to-one basis.

Hereinafter, the design design apparatus 100 and the machine learning apparatus 200 will be described in more detail.

FIG. 2 is a block diagram showing the structure of a design designing apparatus and a machine learning apparatus according to an embodiment of the present invention in more detail.

2, the designing apparatus 100 according to the present embodiment may include an input / output interface 110, a tool driving unit 120, an interceptor unit 130, a control unit 140, and a communication interface 150 And the machine learning apparatus 200 may include a communication interface 210 and a self design modeling unit 220.

First, in the designing design apparatus 100, the input / output interface 110 can interface data signals transmitted / received between the input device 111 and the control unit 140, and can interface with the interceptor unit 130 / And interfacing data signals transmitted / received between the interceptor unit 130 / controller 140 and the machine learning apparatus 200 between the interceptor unit 130 /

The input device 111 may include a unidirectional remote control, a bi-directional remote control, a mouse, and a touch screen, and the output device 112 may include a display device and a printer, etc. for viewing or printing the processing results of the data.

In the design design apparatus 100, the tool driving unit 120 can operate a design design tool when the power is turned on. For example, when the tool driving unit 120 is turned on, the tool driving unit 120 can receive at least one design design tool stored in the database 300 and activate the loaded design design tool. have.

The design design tool mentioned is stored in the database 300 and the loaded design design tool can be stored in a volatile or nonvolatile memory 121 or the like and can be programmed software that helps to design any object Can point to. For example, a design tool is a software tool that assists in synthesis, analysis, modeling, etc. in software design and may take the form of at least one of a simulator, an analysis aid, a design expression processor, a documentation or a generator.

In the design design apparatus 100, the interceptor unit 130 can collect design design data generated in the design process of any object when the designer designs the object using any design design tool driven by the designer.

The mentioned design design data may include design input data and processed design output data that the designer inputs through the input interface.

For example, when a design designer inputs a design input data required to complete an object through an input interface while using a design design tool driven by the design designer, the interceptor 130 can recognize and collect input design input data , And when the design input data is applied to the design design tool and operated, the design output data output as the operation processing result can be further collected.

When the design input data and the design output data described above are stored in the memory or the database 300, the interceptor 130 stores the design input data and the design output data stored in the memory 300 or the database 300, May be further collected from memory or database 400.

The design history data mentioned above may be a result of accumulating design input data and design output data generated in a design process repeatedly performed by a design designer, or may be a result of being modified based on design input data and design output data.

In the design design apparatus 100, the control unit 140 can control the flow of data transmitted and received between the input / output interface 110, the tool driving unit 120, the interceptor unit 130, and the memory / database 300 . The control unit 140 may include at least one processor (microprocessor), and may include the interceptor unit 130 described above.

In the design design apparatus 100, the communication interface 150 converts the data to fit the at least one of a wired communication, a wireless communication, a wired communication network, and a wireless communication network and transmits the converted data to another device.

For example, the communication interface 150 may convert the design design data processed by the interceptor 130 to a wired communication or a wired communication network and transmit the data to the machine learning device 200 or the database 300 to be described later have.

On the other hand, in the machine learning apparatus 200, the communication interface 210 converts the result processed by the self-design modeling unit 210 to be described later into at least one of wired communication, wireless communication, wired communication network, To the machine learning device 200 or the database 300.

In the machine learning apparatus 200, the self-design modeling unit 220 can receive design design data from the interceptor unit 130 of the design apparatus 100 via wire communication or wireless communication, And learning design data is accumulated, and a self-design model capable of self-design can be created based on the learning design data.

As more of the mentioned learning design data accumulate, the performance of the self-design model can be further improved. For example, the self-design modeling unit 220 can accumulate the learning design data from the design design data through a machine learning algorithm in which the virtual designer applies the collected design design data to the rules applied by at least one design design tool And the performance of the self-design model can be improved while continuing to apply the accumulated learning design data to at least one design design tool.

In this way, the self-design model can optimize the design of the target object, thereby improving the design design data designed by the design designer, thereby shortening the time consumed by the design designer and improving the completeness of the object design.

Hereinafter, the self-design modeling unit 220 applying three types of machine learning techniques will be described in more detail.

3 is a block diagram showing the structure of a self-designing modeling unit according to an embodiment of the present invention.

Referring to FIG. 3, the self-designing modeling unit 220 according to the embodiment of the present invention enhances learning design data and applies the reinforcement learning, the guidance learning, and the non-learning learning algorithm to at least one designing tool A self-design model can be created.

 To this end, the self-design modeling unit 220 may include an enhanced learning modeling unit 221, a map learning modeling unit 222, and a non-mapping learning modeling unit 223. The self-design modeling unit 220 may be a controller including at least one processor.

Herein, the reinforcement learning modeling unit 221 receives the design design data from the design design apparatus 100, maximizes the compensation of the learning design data by interaction between the received design design data and the accumulated learning design data based on the received design design data , The performance of the self-design model can be improved by applying the compensated learning design data to at least one design design tool.

The map learning modeling unit 222 can design the learning design data as reference points and determine whether or not the design design data matches the reference points to improve the performance of the self-design model.

For example, when the design design data matches the reference points of the learning design data, the map learning modeling unit 222 regards the accumulated learning design data as having high accuracy and transmits the learning design data having high accuracy to at least one design design The performance of the self-design model can be improved. If the design design data does not match the reference points of the learning design data, it is regarded that the accumulated learning design data is less accurate and the corresponding learning design data is stored in at least one It is not applied to design design tools.

The non-instruction learning modeling unit 223 can group the learning design data and determine whether design design data exists within the range of the clustered learning design data, thereby improving the performance of the self-design model.

For example, the non-map learning modeling unit 223 proves that the design design data is sufficiently reflected in the learning design data if the design design data created by the design designer falls within the range of the clustered learning design data, The performance of the self-design model can be improved by applying the design data to at least one design design tool. If the design design data is not within the range of the clustered learning design data, the learning design data is insufficient. The learning design data is not applied to at least one design design tool.

As described above, in this embodiment, the performance of the self-design model is improved by using any of the reinforcement learning, the guidance learning, and the non-affinity learning algorithms to derive an optimal design plan and shorten the design time.

FIG. 4 is a flowchart illustrating a self-designing modeling method according to another embodiment of the present invention, and FIG. 5 is a flowchart illustrating a machine learning algorithm applied to the self-designing modeling method of FIG.

Referring to FIG. 4, the self-designing modeling method according to the embodiment of the present invention includes steps 410 through 440 for design design applying artificial intelligence machine learning algorithm through the design design apparatus 100 and the machine learning apparatus 200 . ≪ / RTI >

First, in step 410, when the design design apparatus 100 is powered on, at least one design design tool stored in the database 300 may be loaded, and a loaded design design tool may be activated and activated have.

The design design tools mentioned may be stored in volatile memory or non-volatile memory (not shown) and / or database 300, and may refer to programmed software that aids in the design of any object. For example, a design tool is a software tool that assists in synthesis, analysis, modeling, etc. in software design and may take the form of at least one of a simulator, an analysis aid, a design expression processor, a documentation or a generator.

In step 420, the design design apparatus 100 can collect design design data generated in the design process of any object when the design designer designs the object using any design design tool driven. The mentioned design design data may include design input data and processed design output data that the designer inputs through the input interface.

For example, the designing design apparatus 100 can recognize and collect design input data inputted by the design designer, and when the deeply drawn design input data is applied to the design design tool and operated, More design output data can be collected.

In step 420, the design design apparatus 100 accumulates the design input data and the design output data stored in the memory or the database 300, when the design input data and the design output data described above are stored in the memory or the database 300 The design history data can be further collected from the memory or database 400. [

The design history data mentioned above may be a result of accumulating design input data and design output data generated in a design process repeatedly performed by a design designer, or may be a result of being modified based on design input data and design output data.

In step 430, the machine learning apparatus 200 may accumulate the learning design data by mechanically learning the design design data received from the design design apparatus 100, and based on these, a self-design model Can be generated.

As more of the mentioned learning design data accumulate, the performance of the self-design model can be further improved.

That is, in step 440, the machine learning apparatus 200 accumulates the learning design data from the design design data through the machine learning algorithm that applies the collected design design data to the rules applied by at least one design design tool And the performance of the self-design model can be improved while continuing to apply the accumulated learning design data to at least one design design tool.

Here, step 440 may include steps 441 through 443 as a concrete example of the machine learning algorithm as shown in FIG.

In step 441, the machine learning apparatus 200 receives design design data from the design design apparatus 100, and maximizes the compensation of the learning design data by interaction between the received design design data and the accumulated learning design data And apply the compensated learning design data to at least one design design tool to improve the performance of the self-design model.

In step 442, the machine learning apparatus 200 can design the learning design data as reference points, and determine whether or not the design design data matches the reference points, thereby improving the performance of the self-design model.

For example, when the design design data matches the reference points of the learning design data, the machine learning apparatus 200 regards the accumulated learning design data as having high accuracy and transmits the learning design data with high accuracy to at least one design design tool The performance of the self-design model can be improved. If the design design data does not fit the reference points of the learning design data, it is regarded that the accumulated learning design data is less accurate, It is not applied to a single design tool.

In step 443, the machine learning apparatus 200 may group the learning design data and determine whether the design design data exists within the clustered learning design data to improve the performance of the self-design model.

For example, the machine learning apparatus 200 proves that the design design data is sufficiently reflected in the learning design data when the design design data created by the design designer is within the range of the clustered learning design data, The performance of the self-design model can be improved by applying the design data to at least one design design tool. If the design design data is not within the range of the clustered learning design data, the learning design data is insufficient, Data is not applied to at least one design design tool.

The self-design modeling method described above can be implemented in the form of program instructions that can be executed through various computer components and recorded on a computer-readable medium.

The computer readable medium may be any medium accessible by the processor. Such media can include both volatile and nonvolatile media, removable and non-removable media, storage media, and computer storage media.

The storage medium may be any type of storage medium such as RAM, flash memory, ROM, EPROM, electrically erasable read only memory (" EEPROM "), registers, hard disk, removable disk, compact disk read only memory Or any other type of storage medium.

Computer storage media includes removable and non-removable, nonvolatile, and nonvolatile storage media implemented in any method or technology for storing information such as computer readable instructions, data structures, program modules or other data, Volatile media.

Such computer storage media may be embodied as program instructions, such as RAM, ROM, EPROM, EEPROM, flash memory, other solid state memory technology, CDROMs, digital versatile disks (DVDs) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage, Lt; RTI ID = 0.0 > and / or < / RTI >

Examples of program instructions may include machine language code such as those produced by a compiler, as well as high-level language code that may be executed by a computer using an interpreter or the like.

While the present invention has been described in connection with what is presently considered to be the most practical and preferred embodiment, it is evident that many alternatives, modifications and variations will be apparent to those skilled in the art. .

Therefore, the embodiments described above are merely illustrative and are not intended to limit the scope of the present invention to the foregoing embodiments. It is to be understood that the flowcharts shown in the drawings are merely illustrative examples for achieving the most desirable results in the practice of the present invention, and that other steps may be added or some steps may be deleted.

The scope of the present invention will be defined by the appended claims, but all changes or modifications derived from the equivalents, as well as those directly derived from the claims, are also included in the scope of the present invention. .

100: design learning device 110: input / output interface
111: input device 112: output device
120: tool driving unit 130: interceptor unit
140: control unit 150: communication interface
200: machine learning device 210: communication interface
220: self-design modeling unit 221: reinforcement learning modeling unit
222: map learning modeling unit 223: non-map learning modeling unit

Claims (12)

Input / output interface;
A database for storing data or instructions;
A tool driver for loading and driving at least one design design tool stored in the database;
Wherein the designer design data for an object using the at least one driven design tool, design design data generated during a design process through the input / output interface, the object is a plant, An interceptor section for collecting data or structure change data;
A self-design model capable of self-design is generated by accumulating learning design data by machine learning the collected design design data, and applying the accumulated learning design data to the at least one design design tool Machine learning device
, ≪ / RTI &
Wherein the machine learning apparatus includes a self design modeling unit that improves the performance of the self design model by repeatedly applying the accumulated learning design data to the at least one design design tool by a virtual designer.
Self - Design Modeling System. The method according to claim 1,
The design design data includes:
And at least one of design input data inputted by the designer through the input / output interface, design output data outputted, design history data accumulating the design input data and the design output data. delete The method according to claim 1,
The self-design modeling unit,
A reinforcement learning modeling unit that improves the performance of the self-design model by improving the compensation of the learning design data by maximizing interaction between the design design data and the learning design data;
The modeling system comprising: The method according to claim 1,
The self-design modeling unit,
Designing modeling unit for designating the learning design data as reference points, determining whether the design data corresponds to the reference points, and improving the performance of the self-design model;
The modeling system comprising: The method according to claim 1,
The self-design modeling unit,
A non-instruction learning modeling unit for grouping the learning design data and determining whether the design design data exists within the clustered learning design data to improve the performance of the self-design model;
The modeling system comprising: Loading at least one design tool stored in a database in a design apparatus;
Design design data generated in a design process when the designer designs the object using at least one design design tool driven in the design design apparatus, the object is a plant, and the design design data includes structural design data Or structural change data;
Accumulating learning design data by mechanically learning the collected design design data in a machine learning apparatus and generating a self-design model capable of self-design based on the accumulated learning design data; And
Improving the performance of the self-design model by repeatedly applying the accumulated learning design data to the at least one design design tool by a virtual designer in a machine learning apparatus;
/ RTI > The method of claim < RTI ID = 8. The method of claim 7,
The design design data includes:
Wherein the design input data includes at least one of design input data input by a designer through an input / output interface, output design output data, and design history data accumulating the design input data and the design output data. delete 8. The method of claim 7,
Wherein improving the performance comprises:
Improving the performance of the self-design model by interacting with the design design data and the learning design data in a direction maximizing compensation of the learning design data;
/ RTI > The method of claim < RTI ID = 8. The method of claim 7,
Wherein improving the performance comprises:
Designing the learning design data as reference points and determining whether the design data corresponds to the reference points to improve the performance of the self-design model;
/ RTI > The method of claim < RTI ID = 8. The method of claim 7,
Wherein improving the performance comprises:
Clustering the learning design data and improving performance of the self-design model by determining whether the design design data exists within the clustered learning design data;
/ RTI > The method of claim < RTI ID =

Images