Systems and methods for detecting tool defects and classifying tool defects

Abstract

The present application relates to systems and methods for detecting tool defects and classifying tool defects. The system includes a control unit configured to receive one or more force signals from one or more force sensors coupled to one or both of a tool and an end effector. The one or more force signals are indicative of one or more forces exerted relative to the tool or end effector when the tool is operating on one or more components. The control unit is also configured to classify a state of an operating portion of the tool based on the one or more forces. The control unit can also be configured to predict a remaining life of the operating portion based on the one or more forces.

Description

Technical Field

[0001] The examples disclosed herein generally relate to systems and methods for detecting and classifying defects in tools, such as drill bits. Background Technology

[0002] During the manufacturing process, various components can be joined together. Different components can be joined together and manipulated using tools. As an example, during the manufacturing process of an aircraft, the outer skin of the wing is fixed to the spars, ribs, etc.

[0003] Defects in tools, such as drill bits, can cause significant delays and increased costs in manufacturing processes. For example, a defective tool, such as a drill bit with multiple chipped edges, can create anomalous holes in carbon fiber panels. The panels may then be unusable, leading to increased costs and time as these panels are reworked or scrapped, and new drilling processes are performed with replacement panels. As a concrete example, during the robotic-assisted assembly of wings for commercial aircraft, defective drill bits can result in anomalous holes (e.g., large burrs, undesirable surface roughness, hole misalignment, elliptical holes, flutter, etc.) and increased costs due to tool breakage and rework.

[0004] Due to the possibility of tool defects, hole quality needs to be inspected. For example, holes are manually inspected every fifty linear inches to determine the height of burrs relative to the drilled hole. Typically, current inspections are based on conventional qualification tests (and / or manual observation and experience) to define tool life thresholds, but do not utilize any processing monitoring. As can be understood, such periodic inspections can impact the manufacturing process. In particular, automated manufacturing is interrupted every 10 to 15 minutes for each inspection to ensure tooling remains effective and of quality. Summary of the Invention

[0005] There is a need for an efficient, effective, and accurate system and method for detecting the quality of tools (e.g., drill bits) during manufacturing processes. In view of this need, certain examples of this disclosure provide a system comprising a control unit configured to receive one or more force signals from one or more force sensors coupled to one or both of the tool and an end effector. The one or more force signals indicate one or more forces applied relative to the tool or end effector when the tool operates on one or more parts. The control unit is also configured to classify the state of the operating portion of the tool based on the one or more forces. In at least one example, the control unit is also configured to predict the remaining lifespan of the operating portion based on the one or more forces.

[0006] In at least one example, the system includes one or both of a tool and an end effector. In at least one example, the tool is a drill bit.

[0007] In at least one example, the control unit is also configured to train a machine learning model based on force data stored in memory. The control unit can also be configured to classify the state of the operating parts of the tool using the machine learning model. The control unit can be further configured to refine the machine learning model based on one or more force signals received from one or more force sensors.

[0008] In at least one example, the control unit is an artificial intelligence or machine learning system. In at least one example, the control unit includes an artificial neural network.

[0009] The system may also include a user interface with a display. The control unit is also configured to display information about the status of the operating components on the display.

[0010] Certain examples of this disclosure provide a method comprising the steps of: receiving, by a control unit, one or more force signals from one or more force sensors coupled to one or both of a tool and an end effector, wherein the one or more force signals indicate one or more forces applied relative to the tool or end effector when the tool operates on one or more components; and classifying, by the control unit, the state of an operating portion of the tool based on the one or more forces. In at least one example, the method further includes, by the control unit, predicting the remaining lifespan of the operating portion based on the one or more forces. Attached Figure Description

[0011] Figure 1 A simplified block diagram of a system according to an example of this disclosure is shown.

[0012] Figure 2 A graph of force data according to an example of this disclosure is shown.

[0013] Figure 3 A flowchart of an example method according to this disclosure is shown.

[0014] Figure 4 A schematic block diagram of a control unit according to an example of this disclosure is shown.

[0015] Figure 5 A perspective bottom view of an end effector according to an example of this disclosure is shown.

[0016] Figure 6 It shows Figure 4 An internal perspective view of the end effector.

[0017] Figure 7 A perspective front view of an aircraft according to an example of this disclosure is shown. Detailed Implementation

[0018] The foregoing summary of the invention and the following detailed description of certain examples will be better understood when read in conjunction with the accompanying drawings. As used herein, an element or step described in the singular and preceded by the word "a" or "an" should be understood to not necessarily exclude multiple elements or steps. Furthermore, the reference to "an example" is not intended to be construed as excluding the existence of additional examples also incorporated into the described features. Moreover, unless expressly stated to the contrary, examples that "comprise" or "have" an element or multiple elements having a particular condition may include additional elements that do not have that condition.

[0019] As described herein, examples of this disclosure provide systems and methods for measuring and recording physical borehole characteristics at high sampling rates during tool operation, such as during borehole processing. This data is used to train neural networks to detect, predict, and classify tool defects based on input data from sensors. The systems and methods described herein enable efficient and effective manufacturing, and increase sustainability through improved quality control, reduced confined space operation, and fully automated operation without the need for manual quality checks.

[0020] In at least one example, the systems and methods described herein automatically identify quality problems and cutting tool defects in the manufacturing process by using in-process measurements. This automated in-process defect detection reduces rework, extends drill bit life, and reduces manufacturing process time. In at least one example, the control unit analyzes in-process measurements to infer tool defects and determines the tool's health status by using signals from sensors to extract specific features representing the tool's health status compared to a new tool. In at least one example, machine learning regression and classification algorithms are trained based on input data from sensors (e.g., drill bit tip chipping and burr height) and the ability to extrapolate from minor defects to major defects before tool breakage.

[0021] The control unit receives and analyzes signals from sensors to extract specific features representing the tool's health relative to a perfect, fresh new tool. Furthermore, the control unit uses machine learning regression and classification algorithms, trained to extrapolate detection capabilities to new applications (e.g., new cutting tools, new hole sizes, different stacking, etc.).

[0022] Figure 1A simplified block diagram of a system 100 according to an example of this disclosure is shown. System 100 includes a tool 102 having an operating portion 104. As an example, tool 102 is a drill bit, and the operating portion 104 is a drill tip coupled to a spindle. As another example, tool 102 is a saw, and the operating portion 104 is a cutting blade. As another example, tool 102 is a stamping device, and the operating portion 104 is a stamping press. As yet another example, tool 102 is a laser forming device (such as a laser cutting device), and the operating portion 104 is a laser beam.

[0023] Tool 102 is operatively coupled to end effector 106 having a nose 108 (e.g., a pressure foot). As an example, the operating portion 104 of tool 102 is configured to fit into and pass through the nose 108 to operate on one or more parts 110. Alternatively, system 100 may not include end effector 106. Instead, tool 102 may be configured to operate on part 110 without end effector 106.

[0024] Tool 102 is configured to operate on one or more components 110. For example, two components 110 may be aligned with each other. Components 110 may be secured together by fasteners. Each component 110 may include one or more alignment holes 112, such as guide holes. The alignment holes 112 are used to provide positions for tool 102 to form expansion holes 116. For example, the operating portion 104 of tool 102 is configured to axially align with the central longitudinal axis 114 of each alignment hole 112. After alignment, tool 102 is operated such that the operating portion 104 engages component 110 about the central longitudinal axis 114 to cut into the material of component 110 surrounding the alignment holes 112 to form expansion holes 116, which are configured to receive fasteners that can be used to secure component 110 to another structure (e.g., another component 110).

[0025] Alternatively, component 110 may not include alignment hole 112. Alternatively, tool 102 is configured to operate to form holes (e.g., expansion holes 116) directly in component 110 without using alignment holes.

[0026] Each component 110 may be a panel, block, wall, sheet, bracket, connector, etc. In at least one example, the first component 110 may be the skin of the wing of an aircraft being manufactured, and the second component may be a shear connector to which the skin will be fixed.

[0027] To detect the condition (such as health status, quality, etc.) of the operating portion 104 of tool 102, one or more force sensors 120 are coupled to one or both of tool 102 and / or end effector 106. For example, force sensors 120 are used during operation of tool 102 to determine the health status of the operating portion 104. Examples of health status include: (a) completely intact and without defects, (b) one defect, such as chipping, (c) multiple defects, such as two or more chipping, etc.

[0028] In at least one example, force sensor 120 is coupled to either tool 102 or end effector 106. As another example, a first force sensor 120 is coupled to tool 102, and a second force sensor 120 is coupled to end effector 106. Force sensors 120 are configured to detect forces applied during operation of tool 102. Examples of force sensors 120 include load sensors, pneumatic load sensors, capacitive load sensors, strain gauges, hydraulic load sensors, transducers, etc. As another example, force sensor 120 may be coupled to a mandrel attached to a drill bit. In this example, force sensor 120 detects drilling processing data (such as torque applied to the part being drilled).

[0029] Control unit 122 communicates with force sensor 120, for example, via one or more wired or wireless connections. Control unit 122 receives force signal 130 from force sensor 120 indicating a detected force. In at least one example, control unit 122 may also communicate with tool 102 and be configured to control the operation of tool 102. Alternatively, control unit 122 may not be configured to control the operation of tool 102.

[0030] In at least one example, the control unit 122 also communicates with the user interface 124, for example, via one or more wired or wireless connections, the user interface 124 including a display 126. The user interface 124 may be part of a computer workstation, a handheld device (such as a smartphone or tablet), etc. The display 126 may be an electronic monitor, a digital display, etc. As described herein, the control unit 122 may be configured to display information about the status of the tool 102 (such as the status of the operating portion 104) on the display 126.

[0031] The control unit 122 also communicates with the memory 128, for example, via one or more wired or wireless connections. The memory 128 may be separate from and distinct from the control unit 122. Optionally, the memory 128 may be part of the control unit 122. The memory 128 stores data about the operating characteristics of the component 110, tool 102, etc. For example, the memory 128 stores data about the material of the component 110. As another example, the memory 128 stores data about the forces applied to the operating parts of the tool at various stages of its health state. As another example, the memory 128 stores force data, which may be predetermined. Examples of force data include force vectors, force values, patterns, models, trends, etc.

[0032] Each material can be characterized by a specific operating force (e.g., drilling, cutting, etc.), which is used to calculate the operating force in the model stored in memory. The specific operating force, as a material property, can be recovered from the actual measured operating force, thrust, motor torque, etc.

[0033] In at least one example, the nose 108 abuts against the surface 111 of the component 110. The tool 102 may be coupled to the end effector 106 such that the operating portion 104 extends through the nose 108 and engages the component 110. The tool 102 provides tooling operations associated with the component 110. Examples of tooling operations include drilling, cutting, punching, laser forming, etc.

[0034] When tool 102 operates on component 110, force sensor 120 detects the forces applied by tool 102 and / or on component 110. For example, force sensor 120 may detect the thrust applied to component 110 by tool 102. As another example, force sensor 120 may detect the torque applied by tool 102, the bending moment of tool 102, and / or the radial displacement of tool 102.

[0035] In at least one example, memory 128 stores predetermined force data (such as force vectors, force values, patterns, force models, force trends, etc.) associated with tool 102 operating on component 110. For example, tool 102 may be a drill bit, and operating portion 104 may be a drill tip configured to enter a hole (e.g., align hole 112) and / or form a hole (e.g., expand hole 116).

[0036] In at least one example, the control unit 122 is configured to operate via machine learning to classify the state of the operating portion 104 of the tool 102. For example, data on one or more forces associated with the operating portion 104 at various stages of health is initially collected. Specifically, a first set of force data associated with a defect-free (e.g., new, intact) operating portion 104 is stored in memory 128. Furthermore, a second set of force data associated with a first defect value (e.g., a chipped edge formed on the operating portion 104) is also stored in memory 128. Then, a third set of force data associated with a second defect value (e.g., two chipped edges formed on the operating portion 104) is also stored in memory 128, and so on. That is, memory 128 stores force data for the range of health states of the operating portion 104. In at least one example, the range of health states is from a defect-free operating portion 104 to a completely defective operating portion 104 that is inoperable on the component.

[0037] During operation of tool 102, one or more force sensors 120 detect one or more forces applied to operating portion 104 and / or to component 110 by operating portion 104. Control unit 122 receives force signals 130 in real time indicating the forces during operation of tool 102. Control unit 122 then compares the forces (such as those detected by force sensors 120 and received via force signals 130) with data stored in memory 128 to determine the state of operating portion 104. In at least one example, if the force detected by force sensors 120 matches the dataset in memory 128, control unit 122 determines the state of operating portion 104.

[0038] As an example, if (one or more) forces match a first set of force data associated with the defect-free operating section 104, the control unit 122 determines that the operating section 104 currently operating on (one or more) components 110 is defect-free. The control unit 122 then outputs a status signal 132 to the user interface 124. The status signal 132 indicates the current real-time status of the operating section 104. The status of the operating section 104, including the status signal 132, is then displayed on the display 126. In this way, the control unit 122 operates the display 126 to show the current status of the operating section 104.

[0039] As another example, if the force (one or more) matches a second set of force data associated with a first defect value (e.g., a chipped edge formed on the operating portion 104), the control unit 122 determines that the operating portion 104 currently operating on the (one or more) components 110 has at least one defect (such as a chipped edge). The control unit 122 then outputs a status signal 132 associated with the current state to the user interface 124.

[0040] As another example, if (one or more) forces match a third set of force data associated with a second defect quantity (e.g., two chipped edges formed on operating portion 104), then control unit 122 determines that operating portion 104 currently operating on (one or more) components 110 has two or more defects (such as two chipped edges). Control unit 122 then outputs a status signal 132 associated with the current state to user interface 124.

[0041] As described above, by comparing one or more forces detected by one or more force sensors 120 with force data stored in memory 128, control unit 122 is able to classify the state of operating portion 104 in real time during actual operation of tool 102. As can be understood, detected forces may not match specific force data stored in memory 128. Therefore, control unit 122 uses machine learning (which is trained on force data stored in memory 128) to determine states that may not be specifically stored in memory 128. For example, force signal 130 received from force sensor 120 may fall between one or more magnitudes of a first set of force data and a second set of force data. Control unit 122 analyzes the force data within force signal 130 with respect to the force data stored in memory 128 to determine the current health status between the level of the stored force data and the rate of change of the detected forces over time (such as during operation of tool 102). Based on the detected forces and the rate of change over time (e.g., trend), control unit 122 predicts how long operating portion 104 can continue to operate before reaching a specific state stored in memory 128. The control unit 122 then outputs a signal (e.g., status signal 132) to the user interface 124, which then displays a prediction on the display 126 regarding the remaining operational lifespan (until one or more health status levels are reached, such as single defect, multiple defects, inoperability, etc.).

[0042] In at least one example, control unit 122 trains a machine learning model based on a dataset stored in memory 128. Upon receiving new force sensor data from one or more force sensors 120 and classifying the health status of operating part 104 based on such received data, control unit 122 further refines the health status classification. For example, the newly received data may represent health statuses different from those previously stored as training data in memory 128, and is then used as another dataset for future training of the machine learning model. Therefore, the training and learning of the machine learning model is an iterative process with increasing precision and accuracy over time. Control unit 122 uses the machine learning model to classify the health status of operating part 104 and predict the operational integrity of operating part 104.

[0043] As described herein, control unit 122 is configured to monitor the operation of tool 102 in real time to determine the current real-time state of operating portion 104 and predict the remaining service life of operating portion 104. Therefore, system 100 does not require interruption of the manufacturing process (i.e., the operation of tool 102 on part 110) to investigate operating portion 104 to determine its state. In this way, the system and method described herein allow for increased efficiency and less waste in the manufacturing process (e.g., by determining the real-time state, operating portion 104 can be easily replaced before becoming defective, thereby eliminating, minimizing, or otherwise reducing the likelihood of defective parts).

[0044] As described herein, system 100 includes a control unit 122 that receives one or more force signals 130 from one or more force sensors 120 coupled to one or both of tool 102 or end effector 106. The one or more force signals 130 indicate one or more forces applied relative to tool 102 or end effector 106 when the tool operates on one or more parts 110. Control unit 122 is also configured to classify the state of operating portion 104 of tool 102 based on one or more forces. In at least one example, control unit 122 is also configured to predict the remaining lifespan of operating portion 104 based on one or more forces. In at least one example, control unit 122 is also configured to train a machine learning model based on force data stored in memory 128. Control unit 122 is also configured to classify the state of operating portion 104 of tool 102 using the machine learning model.

[0045] Figure 2 A graph of force data according to an example of this disclosure is shown. Reference Figure 1 and Figure 2For example, the control unit 122 may display force data on the display 126. Alternatively, the control unit 122 may not display force data on the display 126. Multiple force sensors 120 are configured to detect one or more forces during operation of the tool 102 on multiple components 110. For example, force sensors 120 may detect the thrust of the operating portion 104 into the component 110, the torque of the operating portion 104, and the bending moment of the operating portion 104. In at least one example, memory 128 stores: a first set of force data 202a, 202b, and 202c, which are associated with the thrust, torque, and bending moment of a defect-free (e.g., new, intact) operating portion 104, respectively; a second set of force data 204a, 204b, and 204c, which are associated with the thrust, torque, and bending moment of an operating portion 104 having one defect (such as a chipped edge), respectively; and a third set of force data 206a, 206b, and 206c, which are associated with the thrust, torque, and bending moment of an operating portion having multiple defects (such as two chipped edges), respectively. For example, a chipped edge may form on the corner of the drill bit tip.

[0046] It has been found that the bending moment of the operating portion 104 (e.g., the drill tip) provides a reliable indicator of the health of the operating portion. As an example, regarding a defect-free, intact operating portion 104, there is no peak bending moment when the operating portion 104 enters the component 110. Conversely, when a defect is present (such as a chipped corner of the operating portion 104), it generally does not negatively affect the quality of the hole formed within the component 110. However, when the operating portion 104 enters the component 110, the defect can be detected as a peak bending moment. This defect leads to an increase in force at a second corner of the operating portion 104, and the operating portion 104 will wear down with repeated use, forming holes of reduced quality (such as more than 5, 10, 15, or more drill holes). Furthermore, when two defects are present (such as chipped corners), such multiple defects significantly affect the hole quality, such as by increasing the height of the burr relative to the formed hole.

[0047] Figure 3 A flowchart illustrating an example method according to this disclosure is shown. References Figure 1 and Figure 3 At 300, a force dataset relating to different levels of health status of the operating portion 104 of tool 102 is stored in memory 128. At 302, control unit 122 uses the force dataset to train a machine learning model. For example, control unit 122 trains the machine learning model to determine different levels of health status based on the stored data, and to interpolate health levels between the stored datasets by determining factors such as rate of change and trend.

[0048] At 304, the tool 102 is operated relative to one or more components 110. For example, the tool 102 is operated to form a hole in the component 110 using the operating portion 104.

[0049] At 306, during actual real-time operation of tool 102, the control unit receives force data output by force sensor 120 via force signal 130. Force sensor 120 detects one or more forces associated with operating portion 104 of tool 102 (e.g., forces applied by and / or entering operating portion 104, forces applied by operating portion 104 to component 110, forces applied by component 110 to operating portion 104, etc.).

[0050] At 308, the control unit 122 uses a machine learning model to classify the current health status of the operating part based on the forces detected by the force sensor 120 during the actual real-time operation of the tool 102. Then, at 302, the method can utilize such information to further refine the machine learning model.

[0051] At 310, the control unit 122 then predicts the remaining lifespan of the operating section 104 based on the current state of the operating section 104 and the changes in force detected over time during the operation of the tool 102. Optionally, the method may exclude 310.

[0052] Figure 4 A schematic block diagram of a control unit 122 according to an example of the present disclosure is shown. In at least one example, the control unit 122 includes at least one processor 400 in communication with a memory 402. The memory 402 stores instructions 404, received data 406, and generated data 408. Figure 4 The control unit 122 shown is merely exemplary and not limiting.

[0053] As used herein, the terms “control unit,” “central processing unit,” “CPU,” “computer,” etc., can include any processor-based or microprocessor-based system, including systems using microcontrollers, reduced instruction set computers (RISC), application-specific integrated circuits (ASICs), logic circuits, and any other circuitry or processors, including systems capable of performing the functions described herein, including hardware, software, or combinations thereof. These are merely exemplary and are therefore not intended to limit the definition and / or meaning of these terms in any way. For example, control unit 122 may be or include one or more processors configured to control operations, as described herein.

[0054] The control unit 122 is configured to execute a set of instructions stored in one or more data storage units or elements (such as one or more memories) to process data. For example, the control unit 122 may include or be coupled to one or more memories. The data storage units may also store data or other information as desired or required. The data storage units may be in the form of physical memory elements within an information source or processor.

[0055] The instruction set may include various commands that instruct the control unit 122, acting as a processor, to perform various examples of methods and specific operations related to the subjects described herein. The instruction set may be in the form of a software program. Software can take various forms, such as system software or application software. Furthermore, software may be a collection of individual programs, a subset of programs within a larger program, or a part of a program. Software may also include modular programming in the form of object-oriented programming. The processor's processing of input data may be in response to user commands, the results of previous processing, or a request from another processor.

[0056] The diagrams in this document may illustrate one or more control or processing units, such as control unit 122. It should be understood that a processing or control unit may represent a circuit, a portion thereof, which may be implemented as hardware (e.g., software stored on a tangible and non-transitory computer-readable storage medium, such as a computer hard disk drive, ROM, RAM, etc.) having associated instructions for performing the operations described herein. The hardware may include state machine circuits hardwired to perform the functions described herein. Optionally, the hardware may include electronic circuitry comprising and / or connected to one or more logic-based devices, such as microprocessors, processors, controllers, etc. Optionally, control unit 122 may represent processing circuitry, such as one or more field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), microprocessors, etc. The circuits in the various examples may be configured to execute one or more algorithms to perform the functions described herein. One or more algorithms may include aspects of the examples disclosed herein, whether or not explicitly identified in the flowcharts or methods.

[0057] As used herein, the terms “software” and “firmware” are interchangeable and include any computer program stored in data storage units (e.g., one or more memories) for execution by a computer, including RAM memory, ROM memory, EPROM memory, EEPROM memory, and non-volatile RAM (NVRAM) memory. The data storage unit types described above are merely exemplary and therefore do not limit the types of memory that can be used to store computer programs.

[0058] In at least one example, all or part of the systems and methods described herein may be, or otherwise include, an artificial intelligence (AI) or machine learning system capable of automatically performing the operations of the methods also described herein. For example, control unit 122 may be an AI or machine learning system. In at least one example, control unit 122, as an AI or machine learning system, may automatically classify the force data described herein, rather than (or in addition to) relying on predetermined force data, and / or automatically predict the remaining lifespan of the operating parts of the tool. These types of systems can be trained and / or self-trained from external information to iteratively improve accuracy by how data is analyzed to determine the health status of the operating parts of the tool and / or predict their remaining lifespan. Over time, these systems can be improved by determining shape, force data, etc., with increased accuracy and speed, thereby significantly reducing the likelihood of any potential errors. The AI ​​or machine learning systems described herein may include techniques enabled by adaptive predictive capabilities and exhibit at least some degree of autonomous learning to automate and / or enhance pattern detection (e.g., identifying irregularities or regularities in data), customization (e.g., generating or modifying rules to optimize record matching), etc. The system may be trained and retrained using feedback from one or more previous analyses of data received from force sensor 120. Based on this feedback, the system can be trained by adjusting one or more parameters, weights, rules, criteria, etc., used in the system analysis. This process can be performed using data instead of training data and can be repeated multiple times to repeatedly improve the determination of the state and remaining lifetime of the tool's operational parts. Training minimizes conflicts and disturbances by executing an iterative training algorithm, in which the system is retrained using an updated dataset and based on feedback examined before the system's most recent training. This provides a robust analytical model that can better determine the state of the tool's operational parts.

[0059] In one example, control unit 122 includes or represents an artificial neural network (ANN) that recognizes patterns in a visual representation of data and classifies patterns based on the content of the recognized patterns (e.g., assigning categories to the recognized patterns, such as category #1, category #2, etc.). Using specially trained ANNs in this way provides improvements over traditional methods of determining classifications, including more accurate real-time determination of the state of surgical parts. An ANN can be implemented in software, hardware, or a combination of both. The structure of an ANN can be a series of layers, where each layer includes one or more artificial neurons arranged in an array of one or more neurons. Each of these neurons can include or represent a register, a microprocessor, and at least one input. Each neuron can produce an output or activation based on an activation function that uses the output of the previous layer and a set of weights as input. Each neuron in the neuron array can be connected to another neuron in the same or another layer via one or more synaptic circuits. The synaptic circuits can include memory for storing synaptic weights. An example of such an ANN can be a deep neural network with an input layer, an output layer, and multiple fully connected hidden layers. In some examples, the ANN (e.g., control unit 122) can be implemented by an application-specific integrated circuit (ASIC) that is specifically tailored for the particular artificial intelligence applications described herein and offers superior computing power and reduced power consumption compared to conventional computers.

[0060] Training data can be generated by receiving continuous data at control unit 122 and discretizing the continuous data using control unit 122. Optionally, control unit 122 can be trained using a pre-trained model. The training data or pre-trained model can be remotely received by control unit 122 via one or more networks. The training data can be historical data, which the neural network can use to learn patterns in the received data to identify or detect the same (or similar) patterns in other data. The trained ANN monitors additional visual representations of the data to identify and classify patterns. If the trained ANN detects one or more patterns, the trained ANN can classify the patterns to generate classification data that can be output to the user and / or used to retrain the ANN.

[0061] The ANN in control unit 122 can continue to learn to improve pattern recognition in data visualization and to improve the classification of the recognized patterns. This continuous learning can occur, for example, by changing the output generated by one or more neurons in response to receiving the same input (e.g., neurons producing different outputs after the change), changing the activation function of one or more neurons, changing one or more weights, and / or changing one or more connections between neurons (or which neurons are connected to each other). Changing one or more of these factors can cause the ANN to produce different outputs than before the change (e.g., recognizing different patterns and / or selecting different classifications).

[0062] Figure 5 A perspective bottom view of an end effector 106 according to an example of this disclosure is shown. The nose 108 is configured to abut against the surface 111 of the member 110 (e.g., ...). Figure 1 (As shown). For example, the nose 108 is configured to clamp perpendicular to surface 111. The nose 108 includes an opening 109 leading to a channel, and a tool 102 (as shown). Figure 1 The operating part 104 (such as the drill tip) passes through the channel.

[0063] Figure 6 It shows Figure 5 Internal perspective view of the end effector 106. (Reference) Figure 1 and Figure 6 The force sensor 120, coupled to the end effector 106, may be a load sensor configured to detect forces, such as the clamping load 500 applied to the nose 108 by component 110. Force sensor 120, another force sensor, and / or a force sensor coupled to tool 102 detect forces (such as drill bit thrust 502) applied by tool 102. Control unit 122 can then determine the measured force 504 (such as the measured load) by subtracting the drill bit thrust 502 from the clamping load 500.

[0064] Figure 7 A perspective front view of an aircraft 600 according to an example of this disclosure is shown. The aircraft 600 includes a propulsion system 612, including, for example, an engine 614. Optionally, the propulsion system 612 may include more engines 614 than shown. The engines 614 are carried by the wings 616 of the aircraft 600. In other examples, the engines 614 may be carried by a fuselage 618 and / or a tail 620. The tail 620 may also support a horizontal stabilizer 622 and a vertical stabilizer 624. The fuselage 618 of the aircraft 600 defines an interior cabin 630, which includes a flight deck or cockpit, one or more work areas (e.g., galley, carry-on baggage area, etc.), one or more passenger areas (e.g., first class, business class, and instructor area), one or more lavatories, etc. Reference Figures 1 to 7The examples disclosed herein can be used during the manufacture of various parts of the aircraft 600. For example, the wing skin is a component that is attached to other internal parts (such as spars, ribs, etc.).

[0065] Figure 7 An example of aircraft 600 is shown. It should be understood that the size, shape, and construction of aircraft 600 may differ from... Figure 7 The differences are shown in the illustration. Optionally, the examples of this disclosure can be used with a variety of other means of transportation. For example, instead of an aircraft, the means of transportation can be a land-based vehicle, such as a car, bus, train carriage, etc. As another example, the means of transportation can be a boat. As another example, the means of transportation can be a spacecraft. Optionally, the examples of this disclosure can be used with fixed structures, such as residential or commercial buildings.

[0066] Furthermore, this disclosure includes examples pursuant to the following terms:

[0067] Clause 1. A system comprising:

[0068] Control unit, the control unit being configured to:

[0069] One or more force signals are received from one or more force sensors coupled to one or both of the tool and the end effector, wherein the one or more force signals indicate one or more forces applied relative to the tool or the end effector when the tool operates on one or more components, and

[0070] The state of the operating parts of the tool is classified based on one or more forces.

[0071] Clause 2. The system according to Clause 1, wherein the control unit is further configured to predict the remaining lifespan of the operating part based on the one or more forces.

[0072] Clause 3. The system according to Clause 1 or 2 further includes one or both of the tool or the end effector.

[0073] Clause 4. The system according to any one of Clauses 1 to 3, wherein the tool is a drill bit.

[0074] Clause 5. The system according to any one of Clauses 1 to 4, wherein the control unit is further configured to train a machine learning model based on force data stored in memory.

[0075] Clause 6. The system according to Clause 5, wherein the control unit is further configured to classify the state of the operational portion of the tool by using the machine learning model.

[0076] Clause 7. The system according to Clause 5 or 6, wherein the control unit is further configured to refine the machine learning model based on the one or more force signals received from the one or more force sensors.

[0077] Clause 8. The system according to any one of Clauses 1 to 7, wherein the control unit is an artificial intelligence or machine learning system.

[0078] Clause 9. The system according to any one of Clauses 1 to 8, wherein the control unit comprises an artificial neural network.

[0079] Clause 10. The system according to any one of Clauses 1 to 9 further includes a user interface having a display, wherein the control unit is also configured to display information about the status of the operating portion on the display.

[0080] Clause 11. A method comprising the steps of:

[0081] The control unit receives one or more force signals from one or more force sensors coupled to one or both of the tool and the end effector, wherein the one or more force signals indicate one or more forces applied relative to the tool or the end effector when the tool operates on one or more components; and

[0082] The control unit classifies the state of the operating parts of the tool based on the one or more forces.

[0083] Clause 12. The method according to Clause 11, further comprising the step of: the control unit predicting the remaining lifespan of the operating part based on the one or more forces.

[0084] Clause 13. The method described in accordance with Clause 11 or 12, wherein the tool is a drill bit.

[0085] Clause 14. The method according to any one of Clauses 11 to 13, the method further comprising the step of: training a machine learning model by the control unit based on force data stored in a memory.

[0086] Clause 15. The method according to Clause 14, wherein the classification step includes: using the machine learning model.

[0087] Clause 16. The method according to Clause 14 or 15, the method further comprising the step of refining the machine learning model by the control unit based on the one or more force signals received from the one or more force sensors.

[0088] Clause 17. The method according to any one of Clauses 11 to 16, wherein the control unit is an artificial intelligence or machine learning system.

[0089] Clause 18. The method according to any one of Clauses 11 to 17, wherein the control unit comprises an artificial neural network.

[0090] Clause 19. The method according to any one of Clauses 11 to 18, further comprising the step of: the control unit displaying information about the status of the operating portion on a display of the user interface.

[0091] Clause 20. A system comprising:

[0092] The tool has an operating part;

[0093] One or more force sensors are coupled to one or both of the tool and the end effector, wherein the one or more force sensors are configured to output one or more force signals indicating one or more forces applied relative to the tool or the end effector when the tool operates on one or more components;

[0094] A user interface, the user interface having a display; and

[0095] Control unit, the control unit being configured to:

[0096] Machine learning models are trained based on force data stored in memory.

[0097] Receive one or more force signals from one or more force sensors.

[0098] By using the machine learning model to classify the state of the operational portion of the tool based on one or more forces,

[0099] The remaining lifespan of the operating part is predicted based on one or more of the forces mentioned above.

[0100] The machine learning model is refined based on the force signals received from the one or more force sensors, and

[0101] The display shows information about the status and remaining lifespan of the operating part.

[0102] As described herein, examples of this disclosure provide efficient, effective, and accurate systems and methods for detecting the quality of tools (such as drill bits) during manufacturing processes.

[0103] While various spatial and directional terms (such as top, bottom, lower, middle, horizontal, horizontal, vertical, front, etc.) may be used to describe examples of this disclosure, it should be understood that these terms are used only relative to the orientation shown in the accompanying drawings. The orientation may be reversed, rotated, or otherwise changed such that an upper portion is a lower portion, or vice versa, a horizontal portion becomes a vertical portion, etc.

[0104] As used herein, structures, constraints, or elements “constructed” to perform a task or operation are structurally formed, constructed, or adapted in a manner corresponding to the task or operation. For clarity and to avoid ambiguity, objects that can only be modified to perform a task or operation are not “constructed” to perform the task or operation as used herein.

[0105] It should be understood that the above description is intended to be exemplary and not restrictive. For example, the above examples (and / or aspects thereof) may be used in combination with each other. Furthermore, many modifications may be made to adapt particular situations or materials to the teachings of the various examples of this disclosure without departing from its scope. While the dimensions and types of materials described herein are intended to define aspects of the various examples of this disclosure, these examples are by no means restrictive and are exemplary. Many other examples will be apparent to those skilled in the art upon review of the above description. Therefore, the scope of the various examples of this disclosure should be determined by reference to the appended claims and the full scope of their equivalents. In the appended claims and the detailed description herein, the terms “comprising” and “wherein” are used as simple English equivalents of the corresponding terms “comprising” and “wherein”. Furthermore, the terms “first,” “second,” and “third,” etc., are used merely as labels and are not intended to impose numerical requirements on their objects. Moreover, the limitations of the following claims are not written in the form of means plus function unless and until such a claim is limited by the phrase “means for” followed by a functional statement without further structure.

[0106] This written description uses examples to disclose various examples of this disclosure, including best practices, and also enables any person skilled in the art to practice the various examples of this disclosure, including making and using any device or system and performing any incorporated methods. The patentable scope of the various examples of this disclosure is defined by the claims, and may include other examples that would occur to a person skilled in the art. Such other examples are intended to fall within the scope of the claims if the example has structural elements that are not indistinguishable from the literal language of the claims, or if the example includes equivalent structural elements that are not substantially indistinguishable from the literal language of the claims.

Claims

1. A system for detecting defects in a tool and classifying the defects in the tool, the system comprising: Control unit, the control unit being configured to: One or more force signals are received from one or more force sensors coupled to one or both of the tool and the end effector, wherein the one or more force signals indicate one or more forces applied relative to the tool or the end effector when the tool operates on one or more components, and The state of the operating parts of the tool is classified based on one or more forces.

2. The system according to claim 1, wherein, The control unit is also configured to predict the remaining lifespan of the operating part based on the one or more forces.

3. The system of claim 1, further comprising one or both of the tool and the end effector.

4. The system according to claim 1, wherein, The tool in question is a drill bit.

5. The system according to claim 1, wherein, The control unit is also configured to train a machine learning model based on force data stored in memory.

6. The system according to claim 5, wherein, The control unit is also configured to classify the state of the operational portion of the tool by using the machine learning model.

7. The system according to claim 5, wherein, The control unit is also configured to refine the machine learning model based on the one or more force signals received from the one or more force sensors.

8. The system according to claim 1, wherein, The control unit is an artificial intelligence or machine learning system.

9. A method for detecting defects in a tool and classifying the defects in the tool, the method comprising the following steps: The control unit receives one or more force signals from one or more force sensors coupled to one or both of the tool and the end effector, wherein the one or more force signals indicate one or more forces applied relative to the tool or the end effector when the tool operates on one or more components; and The control unit classifies the state of the operating parts of the tool based on the one or more forces.

10. A system for detecting defects in a tool and classifying the defects in the tool, the system comprising: The tool has an operating part; One or more force sensors are coupled to one or both of the tool and the end effector, wherein the one or more force sensors are configured to output one or more force signals indicating one or more forces applied relative to the tool or the end effector when the tool operates on one or more components; A user interface, the user interface having a display; and Control unit, the control unit being configured to: Machine learning models are trained based on force data stored in memory. Receive one or more force signals from one or more force sensors. By using the machine learning model to classify the state of the operational portion of the tool based on one or more forces, The remaining lifespan of the operating part is predicted based on one or more of the forces mentioned above. The machine learning model is refined based on the force signals received from the one or more force sensors, and The display shows information about the status and remaining lifespan of the operating part.

Images