Method of manufacturing an object and system for assisting in the design of manufacturing an object

Abstract

The present application relates to methods of manufacturing objects and systems that assist in the design of manufacturing objects. A system that assists in the design of manufacturing objects includes a processor and a memory configured to store instructions. The processor is configured to receive first data representing a design of an object to be manufactured and second data representing a machine learning model. The processor is configured to execute the instructions to generate third data using the first data and the second data. The third data indicates at least one of a modification to the design of the object and a process condition for producing the object. The processor is configured to send the design of the object, the process condition, or both, to a manufacturing tool to enable production of the object. The machine learning model represents production data and is based at least in part on one or more of: object characteristics, process parameters, environmental factors, and quality data.

Description

Technical Field

[0001] The field of this disclosure generally relates to the design and manufacture of objects. Background Technology

[0002] Designers of objects to be manufactured can use metrics and tools that focus on performance rather than production and cost. Typically, designers may employ metrics such as cost per unit weight, but cannot quantify design features in terms of the quality or cost outcomes for the production system. Existing solutions provide designers with static and limited feedback on the results of design choices. The use of paper-based design manuals and guidelines is often not mandatory (if such manuals and guidelines exist). While designers may seek manufacturing feedback in some cases, this feedback is usually obtained through a post-design signing-out cycle, or more often through informal communication with manufacturing engineers who cannot truly understand the consequences of all design decisions related to every aspect of the production system. Therefore, many designs are not ideal in terms of manufacturability and cost.

[0003] A production system typically only emerges after the design process has progressed to a certain stage. If project schedules permit, the design can sometimes be incrementally adjusted. Often, the result is a production outcome that is unsatisfactory in terms of both quality and cost. Furthermore, because the updating and publication of specifications and quality documents are relatively slow, such documents usually do not reflect the latest technology. Summary of the Invention

[0004] In this example, a system is provided to assist in the design of an object for manufacturing. The system includes a processor and a memory configured to store instructions. The processor is configured to receive first data representing a design of an object to be manufactured and second data representing a machine learning model. The processor is configured to execute the instructions to generate third data using the first and second data. The third data indicates at least one of modifications to the design of the object and process conditions for producing the object. The processor is configured to send the design of the object, the process conditions, or both, to a manufacturing tool to enable the production of the object. The machine learning model represents production data and is based at least in part on one or more of the following: object characteristics, process parameters, environmental factors, and quality data.

[0005] In another example, a method for manufacturing an object is provided, the method comprising the steps of: using a processor to perform the following steps: receiving first data representing a design of an object to be manufactured and acquiring second data representing a machine learning model. The machine learning model represents production data and is based at least in part on one or more of the following: object features, process parameters, environmental factors, and quality data. The method further comprises: using the processor to perform the following steps: generating third data using the first data and the second data, wherein the third data indicates at least one of modifications to the design of the object and process conditions for manufacturing the object, and sending the design of the object, the process conditions, or both, to a manufacturing tool to enable the manufacture of the object.

[0006] In another example, a computer-readable medium is provided storing instructions executable by a processor to enable the production of an object by performing steps including: receiving first data representing a design of an object to be manufactured and acquiring second data representing a machine learning model. The machine learning model represents the production data and is based at least in part on one or more of the following: object features, process parameters, environmental factors, and quality data. These steps further include: generating third data using the first and second data, wherein the third data indicates at least one of modifications to the design of the object and process conditions for producing the object; and sending the design of the object, the process conditions, or both, to a manufacturing tool to enable the production of the object.

[0007] The features, functions, and advantages described herein can be implemented independently in various examples or combined in other examples. Further details can be found in the following description and figures. Attached Figure Description

[0008] Figure 1 This is a diagram illustrating an example of a system configured to assist in the design of objects based on a machine learning model.

[0009] Figure 2 It is based on the example. Figure 1 A diagram illustrating the specific implementation of the design.

[0010] Figure 3 It is based on the example. Figure 1 A diagram of a specific implementation of an object produced by the system.

[0011] Figure 4 Based on the example Figure 1 A diagram of a specific implementation of the production process of a machine learning model.

[0012] Figure 5 This is an example that can be derived from... Figure 1 The flowchart is a diagram illustrating an example of a system that assists in the design of objects.

[0013] Figure 6 It is a block diagram of a computing environment, based on an example, including aspects of a computing device configured to support computer-implemented methods and computer-executable program instructions (or code). Detailed Implementation

[0014] The aspects disclosed herein present systems and methods for designing objects using machine learning models. Design data for an object to be manufactured can be analyzed based on the machine learning model, and modifications can be suggested or automatically applied to the design to reduce production costs, lower expected defect rates, improve one or more other factors associated with manufacturing the object, or any combination of these aspects. The machine learning model can be updated based on real-time or near real-time production data, based on recent production events (e.g., in a globally distributed production environment).

[0015] The technical effects and examples of the disclosed systems and methods can provide direct and indirect results for design decisions regarding part costs and production system costs in real time, including analysis of design characteristic criteria for manufacturing processes (e.g., geometry, inventory, etc.) and assemblies (e.g., shape, fit, and tolerances, etc.). Real-time data, such as data from operation and disassembly laboratories, can be used, and machine learning techniques can be applied to provide designers with feedback on the latest performance of processes related to quality, cost, and production integration decisions. In parallel, machine learning can be used to frequently update specification and quality systems to reflect the latest techniques used by designers and for quality analysis. According to some examples, the system uses such updated data and specifications implemented within the design tools used by designers to create designs to provide direct analysis of the design. In this way, designers are allowed and encouraged to consider cost and manufacturing outcomes in addition to product performance (e.g., weight and function) when designing parts for manufacture. Therefore, by using the disclosed systems and methods, the impact on manufacturability and cost resulting from a lack of feedback from actual production operations and the delays associated with updating specification and quality documentation during routine design processes can be reduced or eliminated.

[0016] Furthermore, examples of this disclosure enable consideration of environmental factors of the manufacturing facility at the design, production, or both stages. For example, production quality may be affected by variations in factors such as ambient temperature and humidity in the factory floor. According to some examples, the system adjusts the design analysis of the object to be manufactured based on the actual or predicted environmental conditions of the manufacturing facility, and can generate design modifications, adjustments to the processing parameters to be used when manufacturing the object, or a combination of both, to improve production quality and reduce defects during production operations.

[0017] The accompanying drawings and the following description illustrate examples. It should be understood that, although not explicitly described or shown herein, those skilled in the art will be able to design various arrangements that embody the principles described herein and are included within the scope of the claims following this description. Moreover, any examples described herein are intended to aid in understanding the principles of this disclosure and should be construed as not limiting. As a result, this disclosure is not limited to the specific embodiments or examples described below, but is limited by the claims and their equivalents.

[0018] Specific examples are described herein with reference to the accompanying drawings. Throughout this description, common features are designated using common reference numerals. In some drawings, multiple instances of a particular type of feature are used. Although these instances are physically and / or logically distinct, each instance uses the same reference numerals, and different instances are distinguished by adding letters to the reference numerals. When features are referred to herein as groups or types (e.g., when no particular feature among these features is being referred to), the reference numerals do not include distinguishing letters. However, when a specific feature among multiple features of the same type is referred to herein, the reference numeral is used with distinguishing letters. For example, see... Figure 1 The diagram illustrates several designs and is associated with the designations 120A and 120B. When referring to a specific design among these designs (e.g., design 120A), the distinguishing letter "A" is used. However, when referring to any one of these designs or to them as a group, the designation 120 is not used with the distinguishing letter.

[0019] As used herein, various terms are used only for the purpose of describing specific examples and are not intended to be restrictive. For example, unless the context explicitly indicates otherwise, descriptions of the singular form are also intended to include the plural form. Furthermore, the terms “comprise,” “comprises,” and “comprising” are used interchangeably with “include,” “includes,” or “including.” Additionally, the term “wherein” is used interchangeably with the term “where.” As used herein, “exemplary” indicates an example, implementation, and / or aspect and should not be construed as restrictive or indicating a preferred or preferred implementation. As used herein, ordinal terms used to modify an element (e.g., structure, component, operation, etc.) (e.g., “first,” “second,” “third,” etc.) do not themselves indicate any priority or order of that element relative to another element, but merely distinguish that element from another element with the same name (but used for ordinal items). As used in this article, the term “set” refers to a grouping of one or more elements, and the term “multiple” refers to multiple elements.

[0020] As used herein, unless the context explicitly indicates otherwise, the terms “generate,” “calculate,” “use,” “select,” “access,” and “determine” are interchangeable. For example, “generate,” “calculate,” or “determine” a parameter (or signal) can refer to actively generating, calculating, or determining the parameter (or signal), or it can refer to using, selecting, or accessing the parameter (or signal) that has already been generated (e.g., generated by another component or device). As used herein, “coupled” can include “communicationally coupled,” “electrically coupled,” or “physically coupled,” and may (or optionally) include any combination of these. Two devices (or components) can be coupled directly or indirectly (e.g., communically coupled, electrically coupled, or physically coupled) via one or more other devices, components, wires, buses, networks (e.g., wired networks, wireless networks, or combinations thereof). As an illustrative, non-limiting example, two electrically coupled devices (or components) can be included in the same device or in different devices and can be connected via electronic devices, one or more connectors, or inductive coupling. In some examples, such as two devices (or components) connected communicatively in electrical communications, they can send and receive electrical signals (digital or analog signals) directly or indirectly (e.g., via one or more wires, buses, networks, etc.). As used herein, "direct connection" is used to describe two devices connected without intermediate components (e.g., communicatively, electrically, or physically).

[0021] Figure 1 An example of system 100 is depicted, which is configured to apply machine learning to update and improve manufacturing processes. System 100 includes: a design device 102, a manufacturing tool 104, a detector 106, and a machine learning device 108. System 100 is configured to apply a machine learning model 126 based on production data to modify or supplement the design of an object 110 to be manufactured. Figure 1 The example shown is object 110A, which serves as a component of another article to be manufactured, for example... Figure 3The components of the aircraft shown. Alternatively, as an illustrative and non-limiting example, object 110 may be a ship, automobile, or other vehicle or structure, or a component of a vehicle or structure. Alternatively or additionally, object 110 to be manufactured may be an article assembled from other manufactured components produced at globally distributed manufacturing locations and transported to an assembly location. Object 110 may be any type of physical component to be manufactured, for example, via composite processes (e.g., Automated Tape Laying (ATL) or Automated Fiber Placement (AFP) processes) or non-composite processes (e.g., additive manufacturing processes (e.g., three-dimensional (3D) printing) or subtractive manufacturing processes (e.g., machining)). Modified design and supplementary process information are used during the manufacture of object 110, and the resulting production data is used to update machine learning model 126. Modifying or extending the design of object 110 using machine learning model 126 enables the manufacture of object 110 with reduced production costs and fewer defects compared to conventional designs that are not based on machine learning feedback from the production system.

[0022] The design device 102, manufacturing tool 104, detector 106, and machine learning device 108 are interconnected via one or more networks to enable data communication. For example, the design device 102 may be connected to the manufacturing tool 104 via one or more wireless networks, one or more other wired networks, or any combination of both. Two or more of the design device 102, manufacturing tool 104, and machine learning device 108 may be located in the same location or geographically distributed among each other.

[0023] Design apparatus 102 includes a processor 112 coupled to a memory 114. The memory 114 includes a computer-readable medium storing instructions 116 executable by the processor 112. The instructions 116 are executable to initiate, perform, or control operations that assist in the design and / or manufacture of an object 110, executor of an object 110A (e.g., an aircraft component).

[0024] Processor 112 includes design module 118, which can be implemented at least partially by processor 112 executing instructions 116. Processor 112 can be implemented as a single processor or as multiple processors, for example, using a multi-core configuration, a multi-processor configuration, a distributed computing configuration, a cloud computing configuration, or any combination thereof. In some examples, one or more portions of design module 118 are implemented by processor 112 using dedicated hardware, firmware, or a combination of both.

[0025] Processor 112 is configured to receive first data 122 representing a design 120A of an object 110A to be manufactured. For example, the design 120A may be provided by a designer operating design device 102. For illustration, the design 120A may be generated graphically by the designer via a graphical user interface (GUI), based on non-graphical data (e.g., a set of points representing surface data or a set of geometric parameters (e.g., component dimensions, orientation, and position), or a combination of both. In some examples, the first data 122 is graphically generated design data received from the designer via the GUI, non-graphical design data received from the designer (e.g., via a keyboard), or a combination of both. In some examples, the first data 122 may be retrieved from one or more stored files (e.g., computer-aided design (CAD) model files). In exemplary examples, the first data 122 includes a CAD model of design 120A or a portion of a CAD model of design 120A.

[0026] The processor 112 is also configured to receive second data 124 representing a machine learning model 126. As further described below, the machine learning model 126 represents production data and may be based at least in part on object features, process parameters, environmental factors, and quality data, as further described below.

[0027] The processor 112 is configured to automatically generate third data 128, based on the operation of the design module 118 and on the first data 122 and the second data 124, indicating at least one of a modification 130 to the design 120A of the object 110A and process conditions 132 for the production of the object 110A. In an example, the design module 118 is configured to analyze the first data 122 according to a machine learning model 126 to determine one or more of the following: costs associated with the production of the design 120A, one or more geometric features in the design 120A identified as associated with defects, an estimated defect rate for the manufacture of the object 110A based on the design 120A, one or more other factors associated with manufacturing the object 110A based on the design 120A, or any combination of these. Based on such factors, the design device 102 can, for example, suggest a modification 130 to the designer via prompts at a graphical user interface. The modification 130 may include adjustments to at least one of the object geometry and the ply angle, as referenced... Figure 2 Further description.

[0028] In some examples, design device 102 is configured to generate estimation data 190, which can be presented to designers or operators of design device 102 to provide feedback in the form of estimates associated with production factors. In a particular implementation, estimation data 190 indicates a supply chain distribution estimate 194, a footprint estimate 196, a cost estimate 198, or a combination of these based on design 120 and second data 124 (e.g., estimation data 190 is generated by applying machine learning model 126 to design 120). Supply chain distribution estimate 194 may include an estimate of how geographically distributed the elements (“supply chain”) for manufacturing object 110A are, which can provide an indication of the cost or time associated with retrieving said elements for manufacturing object 110A from geographically distant locations. Footprint estimate 196 may include an estimate of the size of a factory floor or what portion of available factory space or equipment is predicted to be used for manufacturing object 110A, which can provide an indication of the opportunity cost associated with manufacturing object 110A. As an illustrative, non-limiting example, cost estimate 198 may include an estimate of the total manufacturing cost of object 110A, and may include costs associated with parts, materials, and labor related to the manufacture of object 110A. Operators of design facility 102 may generate one or more updates to design 120A based on current manufacturing production data represented by machine learning model 126 to reduce one or more of the following: supply chain distribution estimate 194, footprint estimate 196, or cost estimate 198.

[0029] Processor 112 is configured to send design 120B, process conditions 132, or both of object 110A to manufacturing tool 104. Design 120B and process conditions 132 enable the production of object 110A at manufacturing tool 104. For example, if no modification to design 120A is indicated based on machine learning model 126, design 120B may not be modified relative to design 120A. Alternatively, design 120B may include a modified version of design 120A after modification 130 is introduced or after other changes are made by the designer. For example, in some examples, design device 102 generates one or more messages based on third data 128 for the designer to indicate proposed modification 130 to design 120A. Alternatively, in some examples, design device 102 automatically applies modification 130 to design 120A before sending design 120B to manufacturing tool 104. (See also...) Figure 2 Additional examples of how to operate the design module 118 are provided.

[0030] Manufacturing tool 104 is configured to apply manufacturing process 140 to produce object 110A. In an exemplary example, manufacturing process 140 includes an automated fiber placement (AFP) process 142. AFP process 142 can be performed based on one or more process parameters 178A, such as temperature 146, pressure 148, tension, or roller type 152. AFP process 142 is also affected by one or more environmental factors 180A, such as ambient temperature 162 or humidity 164. In some examples, process parameters 178A are determined based on process conditions 132 (e.g., temperature 134, tension 136, roller type 138, pressure 139, or any combination thereof) provided by design device 102, which are determined by design device 102 to help reduce defects, costs, or other criteria. One or more process parameters 178A used by manufacturing tool 104 may include default parameters or may be set by the operator of manufacturing tool 104.

[0031] In the exemplary example, process parameter 178A corresponds to the adjustable parameters used during AFP process 142, wherein temperature 146 indicates resin temperature, pressure 148 indicates pressure applied by the fiber placement end during fiber bundle placement, tension 150 indicates the amount of controlled tension held on the fiber bundle by the fiber placement end, and roller type 152 indicates the type of compaction roller used by the fiber placement end. In addition to process parameter 178A, environmental factors 180A may also affect the defect rate associated with AFP process 142. As a non-limiting example, ambient temperature 142 indicates room temperature near the AFP machine where object 110A is generated, and humidity 164 represents the humidity near the AFP machine.

[0032] In some examples, one or more process parameters 178A are determined based on process condition 132, and these process parameters can be updated for each production run of object 110A without changing design 120B. A first design / production feedback sub-loop can be used, in conjunction with production shop floor feedback or by designers or modifiers, to determine the process condition 132 for each production run of object 110A in real time or iteratively. For illustration, multiple production runs can be performed at manufacturing tool 104. Each production run can receive updated process conditions 132 from design device 102 without modifying design 120B. Due to the update of process condition 132, process parameters 178A can be changed directly between production runs. Alternatively, a second design / production feedback sub-loop can be used, where process parameters 178A can be changed during processing, for example, based on surrounding conditions (e.g., environmental factor 180), via machine learning model 126.

[0033] Detector 106 is configured to inspect object 110A after manufacturing to determine defect data 166. For example, defect data 166 may indicate one or more observed defects, such as the location, number, and type of the observed defects. For illustrative purposes, as a non-limiting example, defects that may be observed for AFP process 142 may include creases, wrinkles, twisted cellulose fibers, creases, and meandering fiber bundles.

[0034] Machine learning device 108 is configured to generate and update machine learning model 126 based on received production data 168. The received production data 168 includes information such as object features 176, process parameters 178, environmental factors 180, and quality data 182. For example, object features 176 may include descriptions of curvature regions of object 110A, layer angles of object 110A, one or more other features, or combinations thereof. Process parameters 178 may include process parameters 178A used during the manufacture of object 110A, environmental factors 180 may include environmental factors 180A present during the manufacture of object 110A, and quality data 182 may represent defect data 166 generated during the inspection of object 110A (e.g., quality data 182 includes data indicating defects observed in the manufactured object).

[0035] In some examples, production data 168 includes globally distributed production data. For instance, in addition to production data from manufacturing tool 104, production data 168 may also include data provided by distributed manufacturing locations 170, 172. For illustration, manufacturing locations 170 and 172 may be located in geographical regions different from the location of manufacturing tool 104.

[0036] Machine learning device 108 is configured to train machine learning model 126 based on received production data 168. Training machine learning model 126 based on the latest production data 168 enables machine learning model 126 to represent the current state of production and manufacturing facilities of system 100. In some examples, machine learning model 126 is continuously updated as globally distributed production data 168 is received in a near-instantaneous or real-time manner.

[0037] In an illustrative example, the designer using design device 102 may be located in California when generating design 120A, while the manufacturing tool 104 used to manufacture object 110 may be located in a manufacturing plant in China. Although the designer may not know the ambient temperature and humidity in the manufacturing plant, or how these conditions might affect the production of object 110A (e.g., the defect rate during production runs in the manufacturing plant), machine learning model 126 may guide modification 130 to one or more values ​​of design 120A, process conditions 132-139, or both, based on the current environmental conditions in the manufacturing plant. For example, production data 168 may indicate that, based on the latest production run in the manufacturing plant under current environmental conditions, using a lower resin temperature (e.g., temperature 134) and a higher fiber placement end pressure (e.g., pressure 139) would result in a lower defect rate compared to using default values.

[0038] Therefore, system 100 includes a mechanism that uses current production data to analyze and modify design 120 to improve one or more production factors (e.g., required floor space, predicted defect rate, material costs, manufacturing costs, or one or more factors that are not normally readily available to designers). Manufacturing tool 104 is operated based on the output of design device 102 to generate object 110A, and production data corresponding to object 110A (e.g., object characteristics, process parameters, and quality measurements) is used to further update machine learning model 126.

[0039] although Figure 1 A design apparatus 102 is depicted within system 100, interoperating with manufacturing tool 104, detector 106, and machine learning device 108. However, it should be understood that each component of system 100 can operate independently of one or more, or all, of the other components of system 100. For example, in some examples, design apparatus 102 can operate as part of a pure design facility that receives data representing a machine learning model (e.g., second data 124) but is independent of any particular manufacturing or production facility. As another example, in some examples, manufacturing tool 104 can operate as part of a manufacturing or production facility independent of any particular design facility. Also as another example, in some examples, machine learning device 108 can be operated by a provider of artificial intelligence (AI) services, which is also independent of any particular design facility or manufacturing or production facility.

[0040] Although the design device 102, manufacturing tool 104, detector 106, and machine learning device 108 are described as separate components, in other examples, the described functions of two or more of the design device 102, manufacturing tool 104, detector 106, and machine learning device 108 may be performed by a single component. In some examples, each of the design device 102, manufacturing tool 104, detector 106, and machine learning device 108 may be represented in hardware, such as via an application-specific integrated circuit (ASIC) or a field-programmable gate array (FPGA), or the operations described with reference to these components may be performed by a processor that executes computer-readable instructions.

[0041] Although for the sake of clarity, Figure 1 Specific examples are illustrated, but such examples are not intended to be limiting. For example, although manufacturing process 140 is described as including AFP process 142, in other examples, in addition to or in lieu of AFP process 142, the manufacturing process may include one or more other processes, such as another composite process (e.g., Automated Tape Laying (ATL)) or non-composite processes (e.g., additive manufacturing processes (e.g., 3D printing), subtractive manufacturing processes (e.g., machining), one or more other manufacturing processes, or any combination of these processes). Because the third data 128, process parameters 178A, and environmental factors 180A are provided as examples for use with AFP process 142, in examples using manufacturing processes other than AFP process 142, the third data 128, process parameters 178A, and environmental factors 180A may be modified to include parameters and factors more relevant to that particular manufacturing process, while omitting parameters and factors irrelevant to that particular manufacturing process.

[0042] Figure 2 A graphical representation of a specific implementation of design 120 is illustrated. As shown, design 120 has an object geometry 202 resembling a planar rectangular sheet that has been bent into an approximate U-shape (sometimes also called a "saddle" shape). The curvature of this sheet is non-uniform, with two curvature regions 204A and 204B indicating areas where the curvature is higher than other areas of design 120. As shown, in addition to the curvature along the U-shape (e.g., along the length of the rectangular sheet), design 120 also exhibits curvature on one or more other dimensions (e.g., along the width of the rectangular sheet).

[0043] Design 120 also includes a lamellar angle 210. As shown, the lamellar angle 210 indicates the angular orientation of the fiber bundles forming the composite lamellars, for example in... Figure 1In the automated fiber placement process 142, in a multilayer sheet material, each sheet can have a different sheet angle, for example, by increasing the sheet angle of each sheet by 45 degrees relative to the sheet angle of the previous sheet to enhance the structural integrity of the object 110A. Sheet angle 210 can represent the initial sheet angle of design 120, based on which the sheet angle of each subsequent sheet is determined. Although in some examples, sheets are layered to cover the same area, in other examples, a single sheet covers a different area than other sheets and can form a patch or partial cover of the entire shape. Such a patch does not necessarily correspond to any sheet in the preceding or following sheet in terms of position, size, or angle.

[0044] In some examples, defects are statistically more likely to occur in regions of higher curvature, and the likelihood of defects is influenced by the lamellar angle 210. For example, compared to a design in which the fiber bundles cross the curvature region 204A at a 45-degree angle to the gradient of the object surface at the curvature region 204A (e.g., where the lamellar angle 210 is approximately 90 degrees), the fiber bundles are predicted to be more likely to bend in a design where the fiber bundles are placed across the curvature region 204A at a 45-degree angle to the gradient of the object surface at the curvature region 204A (e.g., as shown in the figure).

[0045] use Figure 2 Design 120 is used as an example, based on a specific implementation. Figure 1 The operation of the design device 102 includes a design module 118 that receives first data 122 and runs a classifier on the first data 122 representing design 120A to determine whether one or more modifications will reduce the estimated defect rate associated with manufactured object 110A. For example, second data 124 may include a classifier (e.g., a set of weights defining a trained neural network) that is based on a machine learning model 126 and configured to detect when the relationship between the sheet angle 210 and the curvature at curvature regions 204A and 204B indicates an estimated defect rate exceeding a threshold amount. For illustration, the design module 118 may perform a series of calculations including: scanning design 120A and generating feature data corresponding to design 120 (e.g., generating surface gradient data, edge data, data associated with other features (e.g., protrusions, holes, etc.), or any combination of these), feeding the feature data into the classifier, and selectively identifying modifications to design 120A based on the classifier's output.

[0046] For example, if the classifier's output indicates that the estimated defect rate is within acceptable limits (e.g., less than or equal to a threshold amount), design module 118 may not generate any modifications. Otherwise, if the classifier's output indicates that the estimated defect rate is outside acceptable limits (e.g., above a threshold amount), design module 118 may generate suggested modifications, such as by identifying potential sources of defects (e.g., "the curvature at this sheet angle is too large."), for designers to adapt by applying changes to design 120A.

[0047] In some examples, design module 118 generates proposed design changes as modification 130 (e.g., "Increasing the lamination angle to 60 degrees will increase manufacturing yield to an acceptable level."). In some examples, after a classifier detects that the estimated defect rate of design 120A is above a threshold, design module 118 automatically tests alternative designs to identify one or more modifications that will reduce the estimated defect rate below the threshold amount. For example, design module 118 may iteratively adjust the lamination angle 210 in 5-degree increments and generate an updated test design that matches the geometry 202 of design 120A but uses incremental values ​​for the lamination angle 210. Design module 118 may run the classifier for each test design in the test designs to identify one or more designs with an estimated defect rate below the threshold level and may determine the value of the lamination angle 210 for the identified designs.

[0048] In some examples, design module 118 can generate a test design by adjusting multiple parameters to identify combinations of changes to design 120A that result in a reduced estimated defect rate. For example, in addition to adjusting the ply angle 210 (or instead of adjusting the ply angle), design module 118 can also adjust (e.g., reduce) the curvature at a first curvature region 204A, the curvature at a second curvature region 204B, or both. Therefore, design module 118 can be configured to perform a search process (e.g., a steepest descent search) that involves adjusting the values ​​of multiple parameters to locate one or more sets of these parameter values ​​that result in an estimated defect rate less than a threshold amount.

[0049] In some examples, design module 118 notifies the user of design device 102 of identified layer angle values, curvature adjustments, other suggested parameter changes, or combinations thereof, as proposed modifications to design 120A (e.g., modification 130). In some examples, design module 118 prompts the user of design device 102 for the proposed modifications to design 120A and, in response to user approval of modification 130, automatically updates design 120A to generate design 120B. In some examples, design module 118 automatically adjusts design 120A to implement modification 130 and generates design 120B without first requesting and receiving user approval.

[0050] Although the estimated defect rate has been described with reference to other examples, it should be understood that in other examples, design module 118 may apply one or more other classifiers to detect other factors, in addition to or in place of the classifier that detects the estimated defect rate based on geometry and lamination angles. For example, second data 124 may include data based on machine learning model 126 representing a classifier for production space standards, supply chain distribution, floor space, cost, one or more other factors, or any combination thereof. In a manner similar to that described above with reference to the defect rate estimated based on geometry 202 and lamination angles 230, design module 118 may be configured to run various classifiers on design 120 and adjust design 120 to determine values ​​for design parameters, process conditions 132, or both, to meet various criteria. As a result, design 120B for manufacturing process 140 may be improved compared to initial design 120A in terms of defect rate, production cost, floor space, etc.

[0051] Figure 3 This is a diagram illustrating an example of an object 110B that can be manufactured based on the machine learning systems and techniques described in reference system 100. As shown, object 110B is an aircraft comprising an assembly of multiple components, such as object 110A (e.g., a panel), door 110C, and window 110D (as a typical, non-limiting example). Each of these components can be produced by manufacturing tool 104 or at one or more production facilities 170-172 and transported to a central facility for assembly. Although object 110B is depicted as an aircraft, in other examples, object 110B may be another type of vehicle (e.g., a spacecraft, seaplane, or land vehicle), structure (e.g., an aircraft pylon, office building, or bridge), or another type of object.

[0052] Figure 4 Based on Figure 1A diagram illustrating a specific implementation of a manufacturing process 400 using a machine learning model. Manufacturing process 400 includes a design phase 402, a manufacturing system phase 404, and a specification and design manual (“DM”) phase 406. Each of phases 402-406 is linked to a successive machine learning phase 408. In one specific implementation, design phase 402 includes a design device 102, manufacturing system phase 404 includes manufacturing tooling 104 and manufacturing locations 170, 172, and machine learning phase 408 includes a machine learning device 108.

[0053] Design phase 402 can be configured as "Design X," where X represents one or more design objectives, such as manufacturing, assembly process capability / control, cost, etc. Design phase 402 can implement "manufacturing license" via part definition, production system definition, and continuous learning specifications. Design phase 402 can be implemented as an extended project management software tool for design and continuous learning. This project management software tool can access actual manufacturing data and process information, link to the production system of production system phase 404 (e.g., reuse), format data for continuous learning in continuous machine learning phase 408, and provide relevant feedback to designers. Such a project management software tool could, for example, be available from the Galorath Corporation of El Segundo, California.

[0054] Production system phase 404 can be modeled as a physical layout and value stream (costs and processes), including the preferred supply chain. The value stream and layout can be continuously updated based on design decisions, and relevant feedback can be provided to designers at design phase 402. As a non-limiting example, designer feedback may include information such as "20 other designs use prepreg, using thermoplastics requires additional equipment - value [X] and footprint," for example. Figure 1 The combination of estimated data 190 and modification 130. As an illustrative, non-limiting example, real-time evaluation of design decisions may include factors such as space requirements, supply chain, distribution, footprint, or data costs of top-line products or subassemblies.

[0055] As an illustrative, non-limiting example, the specification and DM phase 406 may include updates and revisions to the specifications, DM, D6, and quality control documents. For example, specifications that are conventionally implemented as only infrequently (e.g., annually or less) updated can be continuously updated through production learning and targeted testing. To illustrate, as production technologies and equipment are updated, immediate or near-immediate updates to the specifications can be published via updated data provided to and processed by the continuous machine learning phase 408. The updated specifications can be used, for example, via... Figure 1 The second data 124 provides one or more updated classifiers to guide and constrain the design of the new object 110 to be manufactured, said updated classifiers being used to classify the design 120A as conforming to or not conforming to specifications.

[0056] Manufacturing process 400 can provide a process control loop to link variations in the process to the quality (or defects) of parts manufactured according to those variations. Manufacturing process 400 can provide a basis for accelerated design qualification, data from offline machine learning experiments (e.g., accelerated curing, high-entropy alloy changes), and relevant real-time feedback to designers, and can also provide predictive capabilities for designers.

[0057] By interfaceing continuous machine learning phase 408 with updates from production system phase 404 and applying the updated machine learning data to specification and DM phase 406 and design phase 402, production process 400 provides an architecture and working mechanism that enables continuous learning for the design of manufacturing, assembly, process capabilities, detailed part costs, and production system costs.

[0058] Figure 5 An example of a method 500 for manufacturing an object is depicted, including the use of a processor to execute the steps. In a particular implementation, method 500 is... Figure 1 The design device 102 is used to perform this.

[0059] The method 500 includes the following steps: at 502, receiving first data representing a design 120 of an object 110 to be manufactured. For illustration, representing... Figure 1 The first data 122 of the design 120A of object 110A is received at the processor 112 of the design device 102.

[0060] The method 500 includes the following steps: at 504, acquiring second data representing a machine learning model. For illustration, the second data 124 representing the machine learning model 126 is received at the processor 112 of the design device 102. In some examples, globally distributed production data (e.g., production data 168 including data from distributed manufacturing locations 170, 172) is used to update the machine learning model.

[0061] The machine learning model represents production data and is based at least in part on one or more of the following: object characteristics, process parameters, environmental factors, and quality data. For illustration, machine learning model 126 represents production data 168 and may be based at least in part on object characteristics 176, process parameters 178, environmental factors 180, and quality data 182. In the illustrative example, the production object includes an automated fiber placement (AFP) process, such as AFP process 142. Figure 1 As shown, process parameters 178A may include at least one of temperature 146, pressure 148, tension 150, and roll type 152; environmental factors 180A may include at least one of ambient temperature 162 and humidity 164; and quality data 182 may include defect data 166 indicating defects observed in the produced object 110. Object features 176 may include curvature regions, such as… Figure 2 The curvature region is 204.

[0062] The method 500 includes the following steps: at 506, generating third data using first data and second data. At 506, the third data indicates at least one of a modification to the design of the object and the process conditions for producing the object. For example, processor 112 automatically generates third data 128, which indicates at least one of a modification 130 to the design 120A of object 110A and the process conditions 132 for producing object 110A. In some examples, the modification includes: adjusting at least one of the object geometry and lamination angles (e.g., adjusting the object geometry 202 (e.g., by reducing the curvature in curvature region 204)) and adjusting... Figure 2 The lamellar angle is 220°. Figure 1 In its implementation, the process conditions may include: temperature 134, tension 136, roller type 138, pressure force 139, or any combination thereof.

[0063] In some examples, the method 500 includes the following steps: at 508, generating data indicative of supply chain distribution estimates, land area estimates, or cost estimates based on design and second data. For illustration, processor 112 may generate estimation data 190 indicative of supply chain distribution estimates 194, land area estimates 196, or cost estimates 198.

[0064] The method 500 includes the following steps: at 510, sending the design, process conditions, or both of an object to a manufacturing tool to enable the production of the object. For example, processor 112 sends the design 120B, process conditions 132, or both of an object 110A to a manufacturing tool 104 to enable the production of object 110A.

[0065] By using machine learning models based on production data to generate design modifications or process conditions for object design, production factors associated with manufacturing objects can be improved. For example, production factors such as required floor space, predicted defect rates, material costs, manufacturing costs, or one or more factors not normally readily available to designers can be evaluated, and designs can be modified in light of these factors. Updating the machine learning model based on production data (including production data associated with manufacturing object 110) completes a feedback loop or cycle between object design, object manufacturing, and the machine learning model, enabling enhanced accuracy and speed of adaptive design to address changing conditions at the production facility.

[0066] Figure 6 This is a block diagram of a computing environment 600 comprising a computing device 610 configured to support computer-implemented methods and computer-executable program instructions (or code) according to this disclosure. For example, the computing device 610 or a portion thereof is configured to initiate, execute, or control references. Figures 1 to 5 Instructions describing one or more operations.

[0067] The computing device 610 includes one or more processors 620. The processors 620 are configured to communicate with system memory 630, one or more storage devices 640, one or more input / output interfaces 650, one or more communication interfaces 660, or any combination thereof. System memory 630 includes volatile memory devices (e.g., random access memory (RAM) devices), non-volatile memory devices (e.g., read-only memory (ROM) devices, programmable read-only memory, and flash memory), or both. System memory 630 stores operating system 632, which may include a basic input / output system for booting computing device 610 and a complete operating system enabling computing device 610 to interact with users, other programs, and other devices. System memory 630 stores system (program) data 636, such as first data 122, second data 124, third data 128, other data, or combinations thereof.

[0068] System memory 630 includes one or more applications 634 (e.g., instruction sets) executable by processor 620. As an example, the one or more applications 634 include those executable by processor 620 to initiate, control, or perform references. Figures 1 to 6 Instructions describing one or more operations. For example, the one or more applications 634 include instructions executable by processor 620 to initiate, control, or perform one or more operations described by reference design module 118, design device 102, manufacturing tool 104, machine learning device 108, or a combination thereof.

[0069] In one particular implementation, system memory 630 includes a non-transitory computer-readable medium storing instructions that, when executed by processor 620, cause processor 620 to initiate, execute, or control operations to assist in the design of an object. The operations include: receiving, via the processor, first data representing the design of an object to be manufactured; obtaining, via the processor, second data representing a machine learning model that represents production data and is at least partially based on object features, process parameters, environmental factors, and quality data; and automatically generating, via the processor, third data based on the first and second data, indicating at least one of modifications to the object's design and process conditions for producing the object, and sending the object's design, process conditions, or both, to a manufacturing tool to enable the production of the object.

[0070] The one or more storage devices 640 include non-volatile storage devices, such as disks, optical disks, or flash memory devices. In a particular example, storage device 640 includes both removable and non-removable storage devices. Storage device 640 is configured to store an operating system, an image of the operating system, applications (e.g., one or more of applications 634), and program data (e.g., program data 636). In a particular aspect, system memory 630, storage device 640, or both include tangible computer-readable media. In a particular aspect, one or more of storage devices 640 are external to computing device 610.

[0071] The one or more input / output interfaces 650 enable the computing device 610 to communicate with one or more input / output devices 670 to facilitate user interaction. For example, the one or more input / output interfaces 650 may include a display interface, an input interface, or both. The processor 620 is configured to communicate with the device or controller 680 via the one or more communication interfaces 660. For example, the one or more communication interfaces 660 may include a network interface. The device or controller 680 may, for example, include a manufacturing tool 104, a machine learning device 108, one or more other manufacturing processes, or any combination thereof.

[0072] In conjunction with the described system and method, an apparatus for assisting in the design of an object is disclosed. This apparatus includes means for automatically generating third data based on first data representing the design of the object to be manufactured and second data representing a machine learning model. The third data indicates at least one of modifications to the object's design and process conditions for producing the object, wherein the machine learning model represents production data and is at least partially based on object characteristics, process parameters, environmental factors, and quality data. In some examples, the means for automatically generating the third data corresponds to design device 102, processor 112, computing device 610, processor 620, one or more other devices, or a combination thereof. In exemplary implementations, the means for automatically generating the third data may be as described with reference to... Figure 2 To perform the operation, please refer to the above instructions.

[0073] The equipment also includes means for sending the design, process conditions, or both of an object to a manufacturing tool to enable the production of the object. For example, the means for sending may correspond to... Figure 6 The one or more communication interfaces 660 mentioned above are configured to send data to one or more other devices of the manufacturing tool, or a combination of both.

[0074] In some examples, a non-transitory computer-readable medium storage instruction, when executed by a processor, causes the processor to initiate, execute, or control operations for performing some or all of the functions described above. For example, the instruction may be executable to implement... Figures 1 to 6 One or more operations or methods. In some examples, Figures 1 to 6 One or more of the operations or methods may be implemented by one or more processors executing instructions (e.g., one or more central processing units (CPUs), one or more graphics processing units (GPUs), one or more digital signal processors (DSPs)), by dedicated hardware circuitry, or any combination of both.

[0075] The examples described herein are intended to provide a general understanding of the structure of various examples. These examples are not intended as a complete description of all components and features of devices and systems utilizing the structures or methods described herein. Many other examples will be apparent to those skilled in the art upon review of this disclosure. Other examples can be utilized and derived from this disclosure to allow for structural and logical substitutions and changes without departing from the scope of this disclosure. For example, method operations may be performed in a different order than those shown in the figures, or one or more method operations may be omitted. Therefore, this disclosure and the accompanying drawings are to be considered illustrative rather than restrictive.

[0076] Furthermore, although specific examples have been illustrated and described herein, it should be understood that any subsequent arrangement designed to achieve the same or similar results may replace the specific examples shown. This disclosure is intended to cover any and all subsequent modifications or variations of the various examples. Combinations of the foregoing examples, as well as other examples not specifically described herein, will be apparent to those skilled in the art upon review of this specification.

[0077] Furthermore, this disclosure includes examples pursuant to the following provisions:

[0078] Clause 1. A method (500) for manufacturing an object (100), said method comprising the following steps:

[0079] Perform the following steps using the processor (112):

[0080] Receive (502) first data (122) representing the design (120) of the object to be manufactured;

[0081] Acquire (504) second data (124) representing a machine learning model (126), the machine learning model representing production data (168) and based at least in part on one or more of the following: object features (176), process parameters (178), environmental factors (180), and quality data (182);

[0082] Using the first data and the second data, a third data (128) is generated (506), wherein the third data indicates at least one of a modification (130) to the design of the object and the process conditions (132) for producing the object; and

[0083] The design of the object, the process conditions, or both are sent (510) to the manufacturing tool (104) to enable the production of the object.

[0084] Clause 2. The method according to Clause 1, wherein the machine learning model is updated using globally distributed production data.

[0085] Clause 3. The method according to any one of Clauses 1 to 12, the method further comprising the step of: using the processor to perform the following steps: generating (508) data indicating a supply chain distribution estimate (194), a land area estimate (196), or a cost estimate (198) based on the design and the second data.

[0086] Clause 4. The method according to any one of Clauses 1 to 3, wherein the production of said object includes an automated fiber placement (AFP) process (142).

[0087] Clause 5. The method according to any one of Clauses 1 to 4, wherein the modification includes adjusting at least one of the object geometry (202) and the layer angle (210).

[0088] Clause 6. The method according to any one of Clauses 1 to 5, wherein the process parameters include at least one of temperature (146), pressure (148), tension (150), and roller type (152).

[0089] Clause 7. The method according to any one of Clauses 1 to 6, wherein the environmental factor includes at least one of ambient temperature (162) and humidity (164).

[0090] Clause 8. The method according to any one of Clauses 1 to 7, wherein the quality data includes data indicating observed defects in the manufactured object (166).

[0091] Clause 9. The method according to any one of Clauses 1 to 8, wherein the object feature includes a curvature region (204).

[0092] Clause 10. The method according to any one of Clauses 1 to 9, wherein the process conditions include temperature (134), tension (136), roller type (138), pressure (139), or any combination thereof.

[0093] Clause 11. A system (102) for assisting in the design of manufacturing an object (110), said system comprising:

[0094] The processor (114) is configured to store instructions (116); and the processor (112) is configured to receive first data (122) representing a design (120) of an object to be manufactured and second data (124) representing a machine learning model (126), execute the instructions to generate third data (128) using the first data and the second data, wherein the third data indicates at least one of a modification (130) to the design of the object and process conditions (132) for producing the object, and send the design of the object, the process conditions, or both to a manufacturing tool (104) to enable the production of the object, wherein the machine learning model represents production data (168) and is based at least in part on one or more of the following: object features (176), process parameters (178), environmental factors (180), and quality data (182).

[0095] Clause 12. The system according to Clause 11, wherein the production of the object includes an automated fiber placement (AFP) process (142).

[0096] Clause 13. The system according to any one of Clauses 11 to 12, wherein the modification includes adjusting at least one of the object geometry (202) and the layer angle (210).

[0097] Clause 14. The system according to any one of Clauses 11 to 13, wherein the process parameters include at least one of temperature (146), pressure (148), tension (150), and roller type (152).

[0098] Clause 15. The system according to any one of Clauses 11 to 14, wherein the environmental factors include at least one of ambient temperature (162) and humidity (164).

[0099] Clause 16. The system according to any one of Clauses 11 to 15, wherein the quality data includes data (166) indicating observed defects in the produced object, and wherein the object features include curvature regions (204).

[0100] Clause 17. The system according to any one of Clauses 11 to 16, wherein the object includes an aircraft (110B), components of an aircraft (110A, 110C, 110D), a vehicle or structure, components of a vehicle or structure, or any combination thereof.

[0101] Clause 18. The system according to any one of Clauses 11 to 17, wherein the process conditions include temperature (134), tension (136), roller type (138), pressure (139), or any combination thereof.

[0102] Clause 19. A computer-readable medium (114) storing instructions (116) executable by a processor (112) to enable the production of an object (110) by performing steps including:

[0103] Receive the first data representing the design of the object to be manufactured;

[0104] Acquire second data representing a machine learning model, the machine learning model representing production data and based at least in part on one or more of the following: object features, process parameters, environmental factors, and quality data;

[0105] Third data is generated using the first data and the second data, wherein the third data indicates at least one of modifications to the design of the object and the process conditions for producing the object; and

[0106] The design of the object, the process conditions, or both, are sent to the manufacturing tool to enable the production of the object.

[0107] Clause 20. The computer-readable medium according to Clause 19, wherein the modification includes adjusting at least one of the object geometry and the layer angle, wherein the process parameters include at least one of temperature, pressure, tension and roller type, and wherein the environmental factors include at least one of ambient temperature and humidity.

[0108] An abstract of this disclosure is provided without being construed as limiting the scope or meaning of the claims. Furthermore, in the foregoing detailed description, various features may be combined or described in a single implementation for the purpose of simplifying this disclosure. The foregoing examples are illustrative and not limiting of this disclosure. It should also be understood that many modifications and variations can be made in accordance with the principles of this disclosure. As reflected in the foregoing claims, the claimed subject matter may refer to less than all features of any of the disclosed examples. Therefore, the scope of this disclosure is defined by the foregoing claims and their equivalents.

Claims

1. A method for manufacturing an object, the method comprising: using a processor to perform the following steps: receiving first data representing a design of an object to be manufactured, the object having object characteristics including a curvature region and a ply angle; obtaining second data representing a machine learning model, the machine learning model representing production data and being based at least in part on the object characteristics and quality data, the quality data representing defect data generated when inspecting the object, wherein the second data includes a classifier that is a set of weights defining a trained neural network, the classifier being configured to detect when a relationship between the ply angle and a curvature at the curvature region indicates an estimated defect rate that exceeds a threshold amount, the detection being based on a gradient of a surface of the object at the curvature region and the ply angle; generating third data using the first data and the second data, wherein the third data indicates a modification to the design of the object, wherein the modification includes an iterative adjustment to the ply angle in the object to reduce the estimated defect rate below the threshold amount; and sending the modified design of the object to a manufacturing tool to enable production of the object, wherein the machine learning model is continuously updated as production data is received.

2. The method of claim 1, wherein, the machine learning model is updated using production data from a global distribution.

3. The method of claim 1 or 2, further comprising: using the processor to perform the following step: generating data indicating a supply chain distribution estimate, a footprint estimate, or a cost estimate based on the design and the second data.

4. The method of claim 1 or 2, wherein, the production of the object includes an automated fiber placement (AFP) process.

5. The method of claim 1 or 2, wherein, the modification includes adjusting at least one of an object geometry and a ply angle.

6. The method of claim 1 or 2, wherein, the second data is further based on process parameters, and the process parameters include at least one of a temperature, a pressure, a tension, and a roller type.

7. The method of claim 1, wherein, the second data is further based on environmental factors, and the environmental factors include at least one of an ambient temperature and a humidity.

8. The method of claim 1 or 2, wherein, the quality data includes data indicating defects observed in the produced object.

9. The method of claim 1 or 2, wherein, the object includes an aircraft, a component of an aircraft, a vehicle or structure, a component of a vehicle or structure, or any combination thereof.

10. A system to assist in the design of a manufactured object, the system comprising: a memory configured to store instructions; and a processor configured to perform the method of any of claims 1-9.

Images