Automated board design system, automated board design method, and automated board design program

Abstract

To provide an automated board design system configured to enable automated design of a board thereby significantly reducing the time and effort of a designer.SOLUTION: An automated board design system 1A which automatically designs a board includes: a leaning unit 15 which converts patterns of thinking of a board designer into an algorithm to learn the patterns of thinking of the board designer; a design application management unit 16 which stores and manages design applications 181,182, ...18k including the thinking pattern algorithm; a selection unit 11 which acquires design specification data, as various kinds of data necessary for board design, transmitted from a customer terminal 4 that a user uses, and selects the design applications 181,182, ...18k necessary for designing the board using the design specification data based on the acquired design specification data; and an automated designing unit 12 which automatically designs the board based on a design request and outputs design data as a result of the designing.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to an automatic substrate design system, an automatic substrate design method, and an automatic substrate design program for automatically designing the arrangement and wiring of electronic components on a printed circuit board, for example.

Background Art

[0002] In recent years, automatic design of printed circuit boards has been considered. As one of the design elements, a design support device that takes into account the shape and dimensions of components has been proposed. This design support device includes, for example, the technology described in Patent Document 1. This design support device calculates the height difference between a first electronic component and a second electronic component from first component information including the height information of the first electronic component mounted in a substrate with built-in electronic components and second component information including the height information of the second electronic component mounted in all substrates with built-in electromechanical components, and compares the difference with a preset reference value to determine the necessity of height adjustment members to be arranged on the first electronic component and / or the second electronic component.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] When arranging a plurality of electronic components with different heights in a substrate with built-in components, the design support device for a substrate with built-in electronic components described in Patent Document 1 above can determine the necessity of height adjustment and the detailed design of height adjustment members in a short time while ensuring the freedom of electronic component selection.

[0005] However, although the invention described in Patent Document 1 is effective when arranging a plurality of electronic components with different heights in a component-built substrate as described above, it is desired to automatically design not only the electronic components but also the wiring of the printed circuit board.

[0006] The present invention has been made in view of such problems, and an object thereof is to provide an automatic substrate design system, an automatic substrate design method, and an automatic substrate design program that enable automatic substrate design and can significantly reduce the labor of designers.

Means for Solving the Problems

[0007] In order to achieve such a problem, the invention according to claim 1 is an automatic substrate design system that automatically designs a substrate based on a customer's design requirements, and includes a learning means for algorithmizing the thinking pattern of a substrate designer and learning the thinking pattern of the substrate designer, and a design module management means for storing and managing a design module including the algorithmized thinking pattern, and a selection means for acquiring design specification data as various data necessary for designing the substrate transmitted from a customer terminal used by the customer, and selecting the design module necessary for designing the substrate according to the acquired design specification data based on the design specification data, and an automatic design means for automatically designing the substrate based on the design requirements by the design module selected by the selection means and outputting design data as a design result.

[0008] Further, the invention according to claim 2 is characterized in that, in addition to the configuration according to claim 1, it includes an evaluation means for evaluating whether or not the substrate on which the automatic design by the automatic design means has been performed has been designed along the thinking pattern, and an adjustment means for adjusting the design module by correcting the algorithm of the design module depending on the result of the evaluation by the evaluation means.

[0009] Further, the invention according to claim 3, in addition to the configuration according to claim 1, is characterized in that one of the design modules is formed based on the thinking pattern of one of the substrate designers, and a storage means for storing a plurality of the design modules formed based on the respective thinking patterns of the plurality of substrate designers is provided, and the automatic design means outputs a plurality of different design results using the plurality of design modules for the design requirements of the substrate based on one of the design specification data.

[0010] Further, the invention according to claim 4, in addition to the configuration according to any one of claims 1 to 3, is characterized in that the automatic design means compares model data, which is a model for which a specific substrate designer requests a design, with the existing design data, obtains difference data, which is the difference between the model data and the design data, the learning means learns the difference included in the difference data as the thinking pattern, and the automatic design means performs automatic design of the substrate using the thinking pattern based on the difference data.

[0011] Further, the invention according to claim 5 is an automatic substrate design method in an automatic substrate design system that performs automatic design of a substrate based on a customer's design requirements, and includes a learning procedure for algorithmizing the thinking pattern of a substrate designer and causing the learning means to learn the thinking pattern of the substrate designer, a design module management procedure for storing and managing in a design module management means a design module including the algorithmized thinking pattern, a selection procedure for obtaining design specification data, which is various data necessary for designing the substrate, transmitted from a customer terminal used by the customer, and selecting the design module necessary for designing the substrate according to the obtained design specification data, and an automatic design means that, in an automatic design procedure, performs automatic design of the substrate according to the design requirements using the design module selected by the selection procedure and outputs design data as a design result.

[0012] Also, the invention according to claim 6 is a program for causing a computer to execute an automatic substrate design method in an automatic substrate design system that automatically designs a substrate based on a customer's design requirements, the program including a learning procedure for algorithmizing the thinking pattern of a substrate designer and causing the learning means to learn the thinking pattern of the substrate designer, a design module management procedure for storing and managing in design module management means a design module including the algorithmized thinking pattern, a selection procedure for acquiring design specification data as various data necessary for designing a substrate transmitted from a customer terminal used by the customer and selecting, based on the acquired design specification data, the design module necessary for designing the substrate with the design data, and an automatic design means for performing automatic design of the substrate based on the design requirements by the design module selected by the selection procedure in the automatic design procedure and outputting design data as a design result.

Effect of the Invention

[0013] According to the inventions described in claim 1 and claim 5, by performing automatic design of a substrate using a design module including an algorithmized thinking pattern of a substrate designer and outputting design data as a design result, it becomes possible to automatically perform the design of the substrate, and the labor of the designer can be significantly reduced.

[0014] Also, according to the invention described in claim 2, by evaluating whether the design of the substrate for which automatic design has been performed is in accordance with the thinking pattern and adjusting the design module by correcting the algorithm of the design module depending on the result of the evaluation, the state of the substrate designed as a result of the automatic design is evaluated, and if the result of the evaluation is good, the result is reflected in the algorithm for future automatic design, so that a substrate of better quality can be automatically designed. Thereby, it is possible to automatically perform the design of the substrate while improving the quality of the design result.

[0015] Moreover, according to the invention described in claim 3, the automatic design means outputs a plurality of different design results by using a plurality of design modules each formed based on the thinking pattern of a single substrate designer for the design requirements of a substrate based on a single design specification data. Thus, for a single design requirement, a plurality of substrate designs can be performed using the thinking patterns of a plurality of substrate designers, making it possible to significantly improve the design quality of the substrate.

[0016] Furthermore, according to the invention described in claim 4, by comparing exemplary data that serves as a model for which a specific substrate designer requests a design with existing design data, obtaining difference data between the exemplary data and the design data, learning the differences included in the difference data as thinking patterns, and performing automatic design of the substrate using the thinking patterns based on the difference data, the differences between the existing design modules and the new design data can be obtained as difference data and reflected in subsequent automatic designs. Thereby, with a small amount of data, new thinking patterns can be acquired and the variations of automatic design can be expanded.

[0017] Moreover, according to the invention described in claim 6, the automatic substrate design method in the automatic substrate design system according to the present invention can be realized on various hardware.

Brief Description of the Drawings

[0018]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Embodiments for Carrying Out the Invention

[0019] Hereinafter, each embodiment of the present invention will be described.

[0020] [Embodiment 1 of the Invention] Embodiment 1 of this invention is shown in FIGS. 1-6.

[0021] [Basic Configuration] FIG. 1 is a system configuration diagram of the automatic board design system according to Embodiment 1 of the present invention. As shown in FIG. 1, the automatic board design system 1A of this Embodiment 1 includes a management server 2 and n (n≧1) designer terminals 3 1 , 3 2 , ··· 3 n and m (m≧1) customer terminals 4 1 , 4 2 , …, 4 m which are provided so as to be mutually communicable via a network 5.

[0022] [Management Server] The management server 2 is a server computer system provided in the design commission company 6 and has the function of a network server. Note that this design commission company 6 is a specific company that undertakes board design work, a specific for-profit organization, a non-profit organization, or the like.

[0023] The management server 2 executes commands transmitted from the designer terminals 3 1 , 3 2 , ··· 3 n and each customer terminal 4 1 , 4 2 , …, 4 m and processes various data such as design data, and the designer terminals 31 , 3 2 , ··· 3 n and the customer terminal 4 1 , 4 2 , …, 4 m and perform various data transmissions and receptions with them. Further, the management server 2 has the function of a database server, and various data such as design data and programs are recorded.

[0024] FIG. 2 is a diagram schematically showing the hardware configuration of the management server 2. As shown in the figure, the management server 2 includes a CPU 201, a RAM 202, a ROM 203 as a storage means, and an input / output device (I / O) 204.

[0025] The CPU (Central Processing Unit) 201 is a central arithmetic processing unit. The CPU 201 includes one or more cores. The RAM (Random Access Memory) 202 is a volatile memory that temporarily stores programs executed by the CPU 201, data processed by the CPU 201, and the like. The ROM (Read Only Memory) 203 is a storage medium that stores various data and programs. The ROM 203 may be a magnetic storage medium such as a magnetic disk, or a storage medium such as an EPROM or an EPPROM. The input / output device 204 is a communication interface, and performs various processes on signals and data for transmitting and receiving signals and data between the inside and outside of the management server 2.

[0026] Note that the management server 2 may be constructed on a cloud computer system (not shown).

[0027] Note that the hardware configuration described in FIG. 2 is also partially or entirely used in each designer terminal 3 1 , 3 2 , ···, 3 n and each customer terminal 4 1 , 4 2 , …, 4 m as well.

[0028] [Designer terminal, Designer] Designer terminal 3 1 , 3 2 , ··· 3 n are communication terminals used by a plurality of, for example, n (n ≥ 1) designers 7 who provide services for substrate design 1 , 7 2 , ··· 7 n .

[0029] In this Embodiment 1, the designers 7 as "substrate designers" 1 , 7 2 , ··· 7 n are members, for example, employees of the design commission company 6. The designers 7 1 , 7 2 , …, 7 n are preferably veteran designers proficient in design work, but designers other than veteran designers may also be used. The designer terminal 3 1 , 3 2 , ··· 3 n is provided in the design commission company 6. Note that the designer terminal 3 1 , 3 2 , ··· 3 n may be located outside the premises of the design commission company 6. The designer terminal 3 1 , 3 2 , ··· 3 n is a communication terminal capable of transmitting and receiving data to and from the management server 2 via the network 5 and performing data input and output, such as a personal computer, tablet, smartphone, etc. The designer terminal 3 1 , 3 2 , …, 3 n is provided with an operation unit (not shown) for inputting data, such as a mouse and keyboard, and a display unit (not shown), such as an LCD (Liquid Crystal Display) for displaying data in characters or images.

[0030] The designers 7 1 , 7 2 , …, 7 nBy limiting it to only the members of a specific group, which are the members of the design commissioning company 6, it has the effect of enhancing the profitability of the management server 2 and the effect of cultivating the artificial intelligence of the management server 2 with high quality by limiting the information sent to the management server 2 to only the information necessary for the substrate design. However, if the above effects are achieved, the designer 7 1 ,7 2 ,…,7 n may include persons other than the employees of the design commissioning company 6, that is, persons other than the members of the specific group.

[0031] Note that for the sake of simplicity of explanation, unless there is a particular need for distinction, the designer terminal 3 1 ,3 2 ,…,3 n is described as the designer terminal 3, and the designer 7 1 ,7 2 ,…,7 n is described as the designer 7.

[0032] [Customer terminal, customer] Customer terminal 4 1 ,4 2 ,…,4 m are communication terminals used by m customers (m≧1) 8 1 ,8 2 ,···8 m who request the substrate design service from the design commissioning company 6. The customer terminal 4 1 ,4 2 ,…,4 m is a communication terminal that can transmit and receive data with the management server 2 via the network 5 and can input and output data, similar to the designer terminal 3.

[0033] Customer 8 1 ,8 2 ,…,8 m are mainly individuals who do not belong to the design commissioning company 6, or employees of companies or organizations other than the design commissioning company 6. However, at least some of the customers 8 1 ,8 2 ,…,8 m may be members of the design commissioning company 6.

[0034] Note that for the sake of simplicity of explanation, unless there is a particular need for distinction, the customer terminals 4 1 , 4 2 , …, 4 m are referred to as customer terminal 4, and the customers 8 1 , 8 2 , …, 8 m are referred to as customer 8.

[0035] [Network] The network 5 is a data communication means that connects devices of various types, such as network devices, servers, PCs (Personal Computers), smartphones, mobile phones, etc., so that they can communicate with each other regardless of the type of constituent device. Specifically, for example, the Internet can be considered, but it is not limited to this. That is, the network 5 can be any network form, whether open or closed, and can be any network area, whether LAN (Local Area Network) or WAN (Wide Area Network).

[0036] [Design Platform] As shown in FIG. 1, a design platform 10 is implemented in the management server 2. This design platform 10 is a functional means realized by the program stored in the ROM 203 of the management server 2 being executed by the CPU 201.

[0037] The design platform 10 has the function of artificial intelligence (AI) and performs intelligent activities such as inference and problem-solving for the automatic design of circuit boards. The design platform 10 has a neural network configuration and realizes some or all of the functions required for the automatic design of circuit boards by machine learning.

[0038] As shown in FIG. 2, the design platform 10 includes, as functional means, a selection unit 11 as "selection means", an automatic design unit 12 as "automatic design means", an evaluation unit 13 as "evaluation means", an adjustment unit 14 as "adjustment means", a learning unit 15 as "learning means", a design application management unit 16 as "design module management means", and a replication unit 17.

[0039] The selection unit 11 selects, from among k (k≧1) design applications 18 used for substrate design 1 , 18 2 , ··· 18 k those suitable for each substrate design. This design application 18 1 , 18 2 , ··· 18 k will be described later.

[0040] The automatic design unit 12 performs automatic design of the circuit board using the design applications 18 1 , 18 2 , ··· 18 k selected by the selection unit 11.

[0041] The evaluation unit 13 evaluates the configuration, form, quality, etc. of the circuit board designed by the automatic design unit 12 according to a predetermined evaluation procedure.

[0042] The adjustment unit 14 makes adjustments such as correcting the component placement and wiring state of the designed circuit board based on the evaluation results evaluated by the evaluation unit 13.

[0043] The learning unit 15 has an image recognition function, learns from the circuit boards designed by actual designers using the results of image recognition, and generates design applications 18 1 , 18 2 , ··· 18 k . What the learning unit 15 learns based on image recognition is, for example, the component placement, wiring state, and modification mode on the circuit board. In addition, the learning unit 15 learns the evaluation results by the evaluation unit 13.

[0044] The design application management unit 16 manages the generated design applications 18 1 , 182 , ··· 18 k or the modified design application 18 1 , 18 2 , ··· 18 k records and manages them.

[0045] The duplication unit 17 is configured to duplicate the individuality of each designer 7 1 , 7 2 , …, 7 n by the function of AI, and is a configuration of j (j ≧ 1) digital clones 19 that duplicate the individuality (especially the individuality of component arrangement and wiring in design, etc.). 1 , 19 2 , ··· 19 j has. The duplication unit 17 duplicates the design data input from each designer 7 1 , 7 2 , …, 7 n from the designer terminal 3 1 , 3 2 , …, 3 n in large quantities by the corresponding digital clone 19 1 , 19 2 , ··· 19 j and stores the duplicated design data together with the design data before duplication in a storage means such as ROM203.

[0046] Note that among the above functional means, the configurations of the selection unit 11, the evaluation unit 13, and the adjustment unit 14 may be configured as functional means that do not use artificial intelligence, or may be performed by the user of the automatic circuit board design system 1A, for example, the customer 8 who uses the customer terminal 4.

[0047] Note that for the sake of simplicity of explanation, unless otherwise particularly distinguished, in this specification, the design application 18 1 , 18 2 , ··· 18 k is described as the design application 18. Also, unless otherwise particularly distinguished, the digital clone 19 1 , 19 2 , ··· 19 j 108 m is described as the digital clone 19.

[0048] [Procedures before operation start] FIG. 3 is a diagram schematically showing an overview of processing before and after the start of operation of the automatic substrate design system 1A according to this embodiment.

[0049] Step S1 in FIG. 3 schematically shows the processing procedure before the start of operation of the automatic substrate design system 1A according to this embodiment.

[0050] Before this automatic substrate design system 1A is operated, it is necessary to perform machine learning on the learning unit 15 (learning procedure). The machine learning here is mainly reinforcement learning. Based on the goodness or badness of the actions (processing) of the learning unit 15, an educator (for example, an administrator of the management server 2, etc. Hereinafter simply referred to as "educator") repeatedly gives or does not give rewards, and causes the learning unit 15 to acquire a method of obtaining the maximum reward. The reinforcement learning here is mainly imitation learning. Imitation learning is performed, for example, by the learning unit 15 repeatedly extracting feature amounts from the read exemplary data 21 and imitating these feature amounts. Note that, for this machine learning, learning other than reinforcement learning (supervised learning or unsupervised learning) may be used.

[0051] The machine learning of the learning unit 15 is performed using h (h≥1) pieces of exemplary data 21 of image data of existing circuit boards 1 , 21 2 , ··· 21 h and so on. The exemplary data 21 1 , 21 2 , ··· 21 h is generated from design data transmitted from the designer terminal 3 or design data replicated by the digital clone 19.

[0052] The educator causes the learning unit 15 to load at least one of the exemplary data 21 1 , 21 2 , ··· 21 h . The learning unit 15 reads the exemplary data 21 1 , 21 2 , ··· 21 hPerform image recognition for at least any one of them. Hereinafter, for the sake of simplicity of explanation, unless there is a particular need for distinction, exemplary data 21 in this specification 1 , 21 2 , ··· 21 h is described as exemplary data 21.

[0053] The learning unit 15 generates a design application 18 based on the result of image recognition, and the automatic design unit 12 outputs the design result of the circuit board using the generated design application 18. The educator gives or does not give a reward to the learning unit 15 according to the quality of the design result in the board design performed using the design application 18.

[0054] By repeating such processing, the learning of the learning unit 15 is performed.

[0055] The exemplary data 21 is one or both of the following (Data 1) and (Data 2). In the pre-operation learning, mainly (Data 1) is used as the exemplary data 21. (Data 1) Image data of a circuit board actually designed by an actual designer (Data 2) Image data of a circuit board designed by the automatic design unit 12 of the management server 2 in the past In this reinforcement learning, the learning unit 15 acquires information on various feature amounts from the result of image recognition of the exemplary data 21. The sources from which the feature amounts are acquired are mainly image information such as the types of components on the circuit board, the arrangement of components, the arrangement of wirings, and the overlapping conditions. The learning unit 15 generates a design application 18 having an algorithm for circuit design using the acquired feature amount information. The automatic design unit 12 of the management server 2 performs circuit design using this design application 18 and outputs the design result. The educator gives or does not give a reward to the learning unit 15 for the design result output by the automatic design unit 12. By repeating such processing and repeating the modification of the algorithm included in the design application 18, the learning unit 15 completes the design application 18.

[0056] In this case, it is desirable for the educator to give a higher reward when the design result of the automatic design unit 12 is closer to the exemplary data 21 that has been loaded. By giving such a reward, the learning unit 15 can acquire a circuit design close to that of an actual designer and reflect the acquisition result in the design application 18.

[0057] In this embodiment, the learning unit 15 acquires the thinking pattern of the circuit design of an actual designer from the exemplary data 21. This thinking pattern refers to the characteristics of the components and wirings appearing on the circuit boards designed by each designer. Specifically, it is the arrangement state and proximity / distance state between components on the circuit board, between components and wirings, and between wirings that each designer can select, and the selection states can be different for each individual designer.

[0058] Here, an actual designer designs a circuit board by thinking about how to make the circuit operate quickly without malfunction, and arranging components and wirings based on the result of the thinking. For example, arranging components with a large heat generation amount and heat-sensitive components apart, arranging components that generate electromagnetic waves apart, arranging a plurality of wirings together on the substrate end side, etc. Also, for example, the position of vias for installing components and wirings on the circuit board, the length of the wirings, the priority of the arrangement of components and wirings, etc.

[0059] The arrangement, proximity / distance state, etc. of components and wirings on the circuit board based on such design concepts and thinking reflect the design concepts and thinking of actual designers, and significant individual differences appear depending on the learning depth and experience of the designers. Also, such design concepts and thinking of actual designers can be replaced with an algorithm for automatically arranging components and wirings on the circuit board.

[0060] Therefore, having the learning unit 15 learn the positions of the components and wirings of the exemplary data 21 based on image recognition and reflecting the learning result in the algorithm of the design application 18 is synonymous with reflecting the design concepts and thinking of an actual designer when designing a substrate as a thinking pattern in the design application 18.

[0061] Therefore, through such reinforcement learning of the learning unit 15, the algorithms of the learning unit 15 and the design application 18 can acquire the design concepts and thinking patterns of the designers of the actual circuit board as thinking patterns, and cause the management server 2 to perform the design of the circuit board reflected as the thinking pattern imitating the design concepts and thinking of the actual designer.

[0062] The design application 18 including the algorithmized thinking pattern is stored in the design application management unit 16 and managed by the design application management unit 16 (design module management procedure).

[0063] In this reinforcement learning, it is desirable that the exemplary data 21 used for the generation or modification of one design application 18 is repeatedly learned using a plurality of circuit board data classified based on a predetermined criterion. This is because by performing learning with the classified exemplary data 21, it becomes possible to generate a design application 18 in which the design concepts and thinking for each classification target are reflected as thinking patterns.

[0064] Examples of the predetermined criterion for classifying the design application 18 include those shown in the following (Criterion 1)-(Criterion 5).

[0065] (Criterion 1) A specific designer involved in the design of the substrate. For example, the images of one or more circuit boards designed by a specific designer A in the past are used as the exemplary data 21. Similarly, the images of one or more circuit boards designed by designers B, C,... in the past are used as the exemplary data 21. Thereby, the learning unit 15 can generate a design application 18 in which the design concepts and thinking of each circuit board designer are reflected.

[0066] (Standard 2) Designers involved in the design of the substrate who belong to a specific group. For example, take the images of one or more circuit boards designed in the past by the designers (either one or multiple) of each company such as Company D, Company E, Company F, etc. as the exemplary data 21. Thereby, the learning unit 15 generates a design application 18 based on the exemplary data in which the design concepts and thoughts of the design for each group are accumulated. In this Embodiment 1, the specific group mainly corresponds to the design commissioning company 6.

[0067] (Standard 3) An unspecified number of designers involved in the design of the substrate. For example, take the images of one or more circuit boards designed in the past by unspecified designers (either one or multiple) who responded to a recruitment via the Internet as the exemplary data 21. This (Standard 3) is mainly used in [Embodiment 2 of the Invention] described later.

[0068] (Standard 4) Exemplary data 21 with different numbers of learning times. For example, in (Standard 1), the result of learning by the exemplary data 21 of the designs made by Designer A five times in the past is the first generation, and then the result of learning by the exemplary data 21 of the designs made by Designer A five more times (a total of ten times) is the second generation, and then the result of learning by the exemplary data 21 of the designs made by Designer A five more times (a total of fifteen times) is the third generation, and so on. Different design applications 18 are generated based on the exemplary data of different generations. (Standards 2)-(3) also generate design applications 18 using exemplary data 21 with different numbers of learning times in the same way.

[0069] (Standard 5) Exemplary data 21 for each type of circuit board. For example, in (Standard 1), the single-sided circuit board, double-sided circuit board, and multi-layer circuit board designed by Designer A in the past are used as three different types of exemplary data 21 according to their types. (Standards 2)-(4) also use exemplary data 21 with different numbers of learning times in the same way.

[0070] Note that the above (Criterion 1)-(Criterion 5) are examples of classification criteria, and the design application 18 may be generated using the exemplary classification data 21 obtained by appropriately combining the above (Criterion 1)-(Criterion 5). Alternatively, the design application 18 may be generated using the exemplary classification data 21 other than the above (Criterion 1)-(Criterion 5).

[0071] FIG. 4 schematically shows an example of a design module generated by the automatic substrate design system 1A of this embodiment. The figure shows k (k≧1) different design applications 18 based on a plurality of exemplary data 21 classified according to different criteria, such as the above (Criterion 1)-(Criterion 5). 1 ,18 2 ,···18 k is shown in a state where they are generated. For example, in FIG. 4, the design application 18 1 is generated by performing reinforcement learning 10 times using the design data of a specific designer A as the exemplary data 21, and the design application 18 2 is generated by performing reinforcement learning 20 times using the design data of designer A as the exemplary data 21, and the design application 18 3 is generated by performing reinforcement learning 30 times. The design application 18 4 is generated by performing reinforcement learning 25 times using the design data of a specific designer B as the exemplary data 21, and the design application 18 5 is generated by performing reinforcement learning 35 times. The design application 18 6 is generated by performing reinforcement learning 20 times using the design data of a specific designer C as the exemplary data 21.

[0072] FIG. 5 schematically shows another example of a design module generated by the automatic substrate design system 1A of this embodiment. In FIG. 5, the design application management unit 16 classifies and manages a plurality of, for example, 15 design applications 18 1 ,18 2 ,···18 15 into a plurality of, for example, 3 categories 20 for each type of substrate to be designed. Specifically, the design application management unit 16 includes a first category 22a for double-sided substrate design (design applications 18 1 ,18 7 etc.), a second category 22b for multilayer substrate design (design applications 182 , 18 8 , etc.), and a third category 22c for flexible substrate design (design application 18 3 , 18 9 , etc.). By classifying and managing in this way, the design application management unit 16 can easily and appropriately automatically design one or more substrate designs according to the type of substrate to be designed, using the design applications 18 corresponding to the types.

[0073] [Procedures after operation start] Step S2 in FIG. 3 schematically shows the processing procedures after the operation start of the automatic substrate design system 1A of this Embodiment 1. Also, FIG. 6 schematically shows the procedures during substrate design of the automatic substrate design system 1A of this Embodiment 1.

[0074] As shown in FIG. 6, the automatic substrate design system 1A of this embodiment is used in the automatic design procedures for component placement and wiring of the substrate (step S20) after the procedure (step S10) in which the customer 8 inputs design specification data (data such as the data of the outer shape of the substrate, circuit diagram, parts list, netlist data, etc.) using the customer terminal 4 when designing the substrate. As shown in FIG. 6, when the automatic design of component placement and wiring is performed in step S20 and the design data 24 is generated, further post-processing design is performed (step S30), and the output of component coordinates is performed (step S40).

[0075] When the design application 18 is generated in step S1 shown in FIG. 3 and the automatic substrate design system 1A starts operation, the customer 8 using the customer terminal 4 requests the design of a circuit board using the automatic substrate design system 1A from the design receiving company 6. Specifically, the customer 8 requests the design of a circuit board from the management server 2 from the customer terminal 4 (see step S10 in FIG. 6).

[0076] Customer 8 transmits information necessary for the design of the circuit board as design specification data 23 from the customer terminal 4 to the management server 2 (see step S10 in FIG. 7). The information necessary for the design of the circuit board in the design specification data 23 may be, for example, data on the outer shape and dimensions of the circuit board (DXF format), data on the circuit diagram (Kicad format), a parts list of parts to be installed in the circuit (Excel format), a net list of the data of the circuit board (txt format), etc., as shown in FIG. 3 (see step S10 in FIG. 6). The data transmitted from the customer terminal 4 is read into the automatic design unit 12 of the management server 2.

[0077] When the design specification data is read, the automatic design unit 12 performs the design of the circuit board (see step S2 in FIG. 3, steps S20 in FIGS. 3 and 6, automatic design procedure).

[0078] The automatic design unit 12 selects and calls a design application 18 necessary for automatic design (selection procedure), and performs the design of the circuit board using the algorithm of this design application 18.

[0079] As shown in step S20 of FIG. 3, the automatic design unit 12 arranges parts on the circuit board by the algorithm of the design application 18 (step S21), and also performs wiring on the circuit board (step S22). By performing steps S21 and S22 based on the algorithm of the design application 18, the automatic design unit 12 performs the design of the circuit board reflecting the design concept and thinking of the designer of the reference data 21 used at the time of generating the design application 18 (automatic design procedure), and as a result of the design, generates the design data 24 shown in FIG. 6.

[0080] As shown in FIG. 3, when the arrangement of parts (step S21) and wiring (step S22) on the circuit board are completed, the evaluation unit 13 evaluates the circuit board automatically designed by the automatic design unit 12 (step S23).

[0081] The evaluation by the evaluation unit 13 in step S23 is performed, for example, by comparing the design data 24 of the circuit board automatically designed by the automatic design unit 12 with the feature amounts extracted from the model data 21. By this comparison, portions where the automatically designed circuit board matches the model data 21, portions where there are differences, and the state of the differences in the portions where there are differences are detected. Also, in step S23, the evaluation unit 13 evaluates the match or difference between the circuit board and the model data 21.

[0082] As an evaluation mode of the evaluation unit 13, for example, among the circuit boards automatically designed by the automatic design unit 12, portions that match the component arrangement and wiring status of the model data 21 are highly evaluated, and portions that do not match are poorly evaluated. Also, for example, among the circuit boards automatically designed by the automatic design unit 12, for portions that do not match the component arrangement and wiring status of the model data 21, portions where the efficiency has increased compared to the model data 21 (for example, portions where the wiring has become shorter, portions where the number of components can be reduced, portions where the electromagnetic wave influence on specific components has decreased, etc.) are highly evaluated. The evaluation unit 13 evaluates the circuit board by various other evaluation methods.

[0083] The evaluation by the evaluation unit 13 in step S23 is desirably performed as the same processing as reinforcement learning using rewards. That is, the evaluation unit 13 gives a high reward to the highly evaluated portions of the circuit board and gives a low reward or no reward to the poorly evaluated portions.

[0084] In step S23, the learning unit 15 learns the evaluation by the evaluation unit 13. For example, the learning unit 15 learns the portions of the circuit board that received high rewards and the portions that received low rewards or no rewards among the circuit boards automatically designed by the automatic design unit 12.

[0085] In step S23, the learning unit 15 reflects the results of machine learning on the algorithm of the design application 18 by means of a genetic algorithm (GA).

[0086] For example, in step S23, when the parts placement and wiring situation, which are highly rewarded in the design application 18, match the reference data, the learning unit 15 corrects the parts placement and wiring situation of the design application 18 to approximate the reference data. Also, for example, in step S23, when the parts placement and wiring situation, which are highly rewarded in the design application 18, do not match the reference data 21, the learning unit 15 corrects the parts placement and wiring situation of the design application 18 to approximate the designed circuit.

[0087] When the design of the circuit board is completed according to the procedures of steps S21 to S23, the design data 24 of the circuit board is transmitted from the management server 2 to the customer terminal 4 of the customer 8 that requested the design. As shown in FIG. 6, after the design data 24 of the circuit board is generated, the automatic design unit 12 performs the design of the post - processing process (step S30) according to the instructions of the customer 8 or automatic processing, and further, the automatic design unit 12 outputs the coordinate information of the parts on the circuit board (step S40). Thereby, the design process of the circuit board is completed.

[0088] [Operational Effects] As described above, in this Embodiment 1, in the automatic design unit 12, the automatic design of the substrate is performed using the design application 18 including the algorithmized thinking pattern of the substrate designer, and the design result is output as the design data 24. Thus, it becomes possible to automatically perform the design of the substrate, and the labor of the designer can be significantly reduced.

[0089] In this Embodiment 1, in the evaluation unit 13, it is evaluated whether the substrate on which the automatic design has been performed is designed in accordance with the thinking pattern. In the adjustment unit 14, the design application 18 is adjusted by modifying the algorithm of the design application 18 depending on the evaluation result of the evaluation unit 13. Then, the state of the substrate designed as a result of the automatic design is evaluated. When the evaluation result is good, the result is reflected in the algorithm of future automatic designs, and a substrate with better quality can be automatically designed.

[0090] In this embodiment, the automatic design unit 12 can perform a plurality of printed circuit board designs for one design requirement by using a plurality of different design results output by using a plurality of design applications 18 each formed based on the thinking pattern of one printed circuit board designer, for the design requirements of the printed circuit board based on one design data.

[0091] [Embodiment 2 of the Invention] Embodiment 2 of this invention is shown in FIGS. 7 and 8.

[0092] FIG. 7 is a diagram showing the overall configuration according to Embodiment 2 of the automatic printed circuit board design system according to the present invention. In this Embodiment 2, the same or corresponding parts as those in Embodiment 1 are denoted by the same reference numerals and will not be described again.

[0093] The automatic printed circuit board design system 1B of this Embodiment 2 includes, as terminals, the designer terminals 3 1 , 3 2 , ··· 3 n , the customer terminals 4 1 , 4 2 , …, 4 m shown in FIG. 1, and in addition, g (g ≧ 1) other company complementary terminals 25 1 , 25 2 , …, 25 g which are each provided so as to be communicable with other components via the network 5.

[0094] The other company complementary terminals 25 1 , 25 2 , …, 25 g are information communication terminals similar to the designer terminals 3 and the customer terminals 4, and are used by a plurality of other designers 26 1 , 26 2 , …, 26 g for transmitting and receiving data of the designed circuit boards and the like in the design of the circuit boards.

[0095] The other designers 26 1 , 26 2 , …, 26 gjare mainly members of groups other than the group to which Designer 7 belongs, such as employees of companies other than the company to which Designer 7 belongs, members of groups other than for-profit or non-profit organizations to which Designer 7 belongs, and so on. However, Other Designer 26 1 ,26 2 ,…,26 g may be an unspecified individual who does not belong to a specific group. Also, Designer 7 may be an unspecified individual who does not belong to a specific group. Other Designer 26 1 ,26 2 ,…,26 g is preferably a veteran designer who is proficient in design work, but may also be a designer other than a veteran designer.

[0096] In addition, for the sake of simplicity of explanation below, unless there is a particular need for distinction, Other Company Complementary Terminal 25 1 ,25 2 ,…,25 g is described as Other Company Complementary Terminal 25, and Other Designer 26 1 ,26 2 ,…,26 g is described as Other Designer 26.

[0097] FIG. 8 is a diagram schematically showing a partial overview of the processing procedure of the second embodiment. In the learning unit 15 of the design platform 10, reference data 27 as design data 24 of a circuit board designed by Other Designer 26, which is input by Other Designer 26 to Other Company Complementary Terminal 25, is supplied. This reference data 27 is exemplary data 21 conforming to the standards required by a company other than the design commissioning company 6, for example, Company D (D is the company name).

[0098] As shown in FIG. 8, the learning unit 15 compares the reference data 27 with the design data 24 designed by the design application 18 based on the exemplary data 21 already recorded. The learning unit 15 extracts difference data 28, which is the difference between the reference data 27 and the design data 24. The learning unit 15 learns this difference data 28 and records it as the design application 18 according to the exemplary data 21 of Company D 1 . The reference data 27 is compared with the exemplary data 21 already recorded 21 1 ​By learning and recording it as the difference data 28 from, a large learning result can be obtained with a small amount of data.

[0099] As shown in FIG. 8, the learning unit 15 repeatedly learns the difference data 28 many times. Then, the learning unit 15 acquires the thinking patterns and design senses of the circuit designs of other designers 26, especially skilled other designers 26. Thereby, the management server 2 can realize a circuit design in which the thinking patterns and design senses of other designers 26 are reflected in the subsequent circuit design.

[0100] In this second embodiment, by causing the learning unit 15 to learn the reference data 27 of other designers 26 (as the difference data 28), the learning unit 15 acquires the thinking patterns of the circuit designs of other designers 26. In addition to the exemplary data 21 generated based on the design data of the designers 7 of the design commission company 6, by causing the learning unit 15 to learn the reference data 27 of the circuit boards designed by other designers 26, the design ideas and learning results of more diverse designers can be reflected in the algorithm of the design application 18. Then, it is possible to realize various substrate designs that reflect the design ideas and thoughts of more real designers 7 and other designers 26 as thinking patterns. Thereby, it becomes possible to further improve the design quality in the automatic design of the substrate.

[0101] [Other Embodiments of the Invention] Although Embodiments 1 and 2 of the present invention have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, replacements, changes, and combinations can be made without departing from the gist of the invention. These embodiments and their modifications are included in the scope and gist of the invention, and are also included in the invention described in the claims and its equivalent scope.

[0102] In Embodiments 1 and 2 of the present invention, the learning unit 15 reflects the results of machine learning in the algorithm of the design application 18 by means of a genetic algorithm (GA). However, the learning unit 15 may be configured not to use the genetic algorithm. On the other hand, at least a part of the functional means other than the learning unit 15 of the design platform 10 may be configured to repeat learning using the genetic algorithm and perform processing while reflecting the learning results.

Explanation of Signs

[0103] 1A, 1B Automatic substrate design system 3, 3 1 , 3 2 , ··· 3 n Designer (substrate designer) 11 Selection unit (selection means) 12 Automatic design unit (automatic design means) 13 Evaluation unit (evaluation means) 14 Adjustment unit (adjustment means) 15 Learning unit (learning means) 16 Design application management unit (design module management means) 18, 18 1 , 18 2 , ··· 18 k Design module 21, 21 1 , 21 2 , ··· 21 h Model data 23 Design specification data 24 Design data 26, 26 1 , 26 2 , ··· 26 g Other designers (substrate designers) 27 Reference data (model data) 28 Difference data

Claims

1. An automatic substrate design system for automatically designing a substrate based on a customer's design requirements, comprising: learning means for algorithmizing the thinking pattern of a substrate designer and learning the thinking pattern of the substrate designer; design module management means for storing and managing a design module including the algorithmized thinking pattern; selection means for acquiring design specification data as various data necessary for designing a substrate transmitted from a customer terminal used by the customer, and selecting the design module necessary for designing the substrate according to the acquired design specification data based on the design specification data; automatic design means for automatically designing the substrate based on the design requirements by using the design module selected by the selection means and outputting design data as a design result An automatic substrate design system, characterized in that it is provided.

2. evaluation means for evaluating whether or not the substrate designed by the automatic design means is designed along the thinking pattern; The automatic substrate design system according to claim 1, further comprising adjustment means for adjusting the design module by correcting the algorithm of the design module depending on the result of the evaluation by the evaluation means.

3. One of the design modules is formed based on the thinking pattern of one of the substrate designers, storage means for storing a plurality of design modules formed based on the respective thinking patterns of a plurality of the substrate designers, The automatic design means outputs a plurality of different design results by using a plurality of the design modules for the design requirements of the substrate based on one of the design specification data The automatic substrate design system according to claim 1, characterized in that.

4. The automatic design means compares model data, which is a model for which a specific substrate designer requests a design, with existing design data, and acquires difference data, which is the difference between the model data and the design data, The learning means learns the difference included in the difference data as the thinking pattern, The automatic substrate design system according to any one of claims 1 to 3, characterized in that the automatic design means automatically designs the substrate by using the thinking pattern based on the difference data.

5. An automatic substrate design method in an automatic substrate design system for automatically designing a substrate based on a customer's design requirements, comprising: A learning procedure for algorithmizing the thinking pattern of a substrate designer and causing the learning means to learn the thinking pattern of the substrate designer, A design module management procedure for storing and managing in design module management means a design module including the algorithmized thinking pattern, A selection procedure for acquiring design specification data as various data necessary for the design of a substrate transmitted from a customer terminal used by the customer, and selecting the design module necessary for the design of the substrate according to the acquired design specification data based on the design specification data, An automatic design procedure for automatically designing the substrate based on the design requirements by the design module selected by the selection procedure in the automatic design procedure, and outputting design data as a design result, An automatic substrate design method characterized by comprising the above.

6. A program for causing a computer to execute an automatic substrate design method in an automatic substrate design system for automatically designing a substrate based on a customer's design requirements, A learning procedure for algorithmizing the thinking pattern of a substrate designer and causing the learning means to learn the thinking pattern of the substrate designer, A design module management procedure for storing and managing in design module management means a design module including the algorithmized thinking pattern, A selection procedure for acquiring design specification data as various data necessary for the design of a substrate transmitted from a customer terminal used by the customer, and selecting the design module necessary for the design of the substrate according to the acquired design specification data based on the design data, An automatic substrate design program characterized by causing the computer to execute a selection procedure for selecting the design module necessary for the design of the substrate according to the acquired design specification data based on the design specification data, and an automatic design procedure for automatically designing the substrate based on the design requirements by the design module selected by the selection procedure in the automatic design procedure, and outputting design data as a design result.

Images