A video analysis-based algorithm resulting in power reduction in vacuum arc remelting

Abstract

The control system includes a vision system including an imaging device and a VAR monitoring system configured to determine a power adjustment stage of the VAR process based on images from the vision system and process parameters. The VAR monitoring system includes a vision analysis module configured to analyze images from the vision system to detect a melt marker based on a remelt image process model, and a prediction module configured to predict an operating characteristic of the VAR process related to a power adjustment relative to the melt marker location and the remelt prediction model. The VAR monitoring system is configured to initiate a power adjustment stage in response to the melt marker satisfying a predetermined melt marker condition, the operating characteristic of the VAR process satisfying a predetermined operating condition, or a combination thereof.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to and the benefit of U.S. Patent Application No. 63 / 180,961, filed April 28, 2021, the disclosure of which is incorporated herein by reference. [Technical Field]

[0002] The present disclosure relates to a system and method for monitoring a vacuum arc remelting (VAR) process in the production of metal ingots. [Background technology]

[0003] The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.

[0004] The vacuum arc remelting (VAR) process is commonly used to process high-performance titanium, zirconium, nickel-based alloys, and steels, among other alloys. Generally, VAR systems use an electrode that is slowly melted by passing an electric current through it, which then arcs to the molten metal contained in a crucible. The applied melting current is varied during the process to achieve the desired molten metal pool shape and ingot quality.

[0005] VAR systems typically follow a power control routine to perform the VAR process. Examples of power control routines include ramping power to a specified power output, supplying a specified power output to the electrodes for a given period of time, and gradually reducing the power supplied to the electrodes after the given period of time. However, reducing power too quickly or too slowly can result in quality issues. Summary of the Invention

[0006] This section provides a general overview of the disclosure and is not a comprehensive disclosure of the entire scope or every feature.

[0007] In one aspect of the present disclosure, a control system for a vacuum arc remelting (VAR) process is provided. The control system includes a vision system including an imaging device. The imaging device is configured to capture one or more images of an electrode in a vacuum chamber of the VAR process. The control system also includes a VAR monitoring system configured to determine a power adjustment stage for the VAR process based on the one or more images from the vision system and process parameters. The VAR monitoring system includes a vision analysis module configured to analyze the one or more images from the vision system to detect melt markers based on a remelt image process model, and a prediction module configured to predict operating characteristics of the VAR process associated with the power adjustment stage based on the process parameters and the remelt prediction model. The VAR monitoring system is configured to initiate a power adjustment stage in response to the melt markers satisfying a predetermined melt marker condition, the operating characteristics of the VAR process satisfying a predetermined operating condition, or a combination thereof.

[0008] In some forms, the fused marker comprises a pin, a slot, or a combination thereof.

[0009] In one form, the process parameters include a heat number, an electrode weight, a crucible identification, a furnace number, a furnace current, a furnace voltage, a furnace ram position, or a combination thereof.

[0010] In yet another aspect, for each image, the visual analysis module is configured to select a portion of the respective image to be analyzed by the refused image process model.

[0011] In yet another aspect, the vision system and VAR monitoring system initiate operation in response to the furnace ram position being in a predetermined position, the power input to the furnace of the VAR process being greater than a power setpoint, or a combination thereof.

[0012] In one aspect, the remelt prediction model is configured to predict a furnace ram position of a VAR process relative to a melt marker position.

[0013] In some embodiments, the control system further includes a primary VAR controller configured to control power to a furnace of the VAR process, and the VAR monitoring system is further configured to initiate a power adjustment phase by notifying the primary VAR controller to initiate a power ramp-down.

[0014] In one aspect of the present disclosure, a method for a vacuum arc remelting (VAR) process is provided, the method including acquiring one or more images of an electrode in a vacuum chamber, analyzing the one or more images to detect melt markers in the one or more images based on a remelt image process model, determining a power adjustment step for the VAR process based on the one or more images and process parameters, predicting operating characteristics of the VAR process based on the process parameters and the remelt prediction model, and initiating a power adjustment step in response to the melt markers satisfying a predetermined melt marker condition, the operating characteristics of the VAR process satisfying a predetermined operating condition, or a combination thereof.

[0015] In some forms, the fused marker comprises a pin, a slot, or a combination thereof.

[0016] In one form, the process parameters include a heat number, an electrode weight, a crucible identification, a furnace number, a furnace current, a furnace voltage, a furnace ram position, or a combination thereof.

[0017] In another aspect, the method further includes selecting a portion of each image to be analyzed by the remelting image process model.

[0018] In yet another aspect, the method further includes operating in response to the furnace ram position being at a predetermined position, the power input to the furnace of the VAR process being greater than a power setpoint, or a combination thereof.

[0019] In yet another aspect, the method further includes predicting a furnace ram position for the VAR process relative to the melt marker position.

[0020] In one form, initiating the power regulation phase further includes initiating a power ramp-down routine.

[0021] In some embodiments, the predetermined operating condition includes determining whether a predicted furnace ram position of the electrode corresponds to a current furnace ram position.

[0022] In yet another aspect, the remelt image process model further includes performing one or more image processing routines to detect the melt markers using a deep convolutional neural network.

[0023] In yet another form, the remelt prediction model further includes using a regression routine to predict the operating characteristics based on one or more previously obtained process parameters associated with one or more previous VAR processes.

[0024] In one aspect of the present disclosure, a system is provided, including a processor and a non-transitory computer-readable medium containing instructions executable by the processor, the instructions including acquiring one or more images of an electrode in a vacuum chamber, analyzing the one or more images to detect melt markers in the one or more images based on a remelt image process model, determining a power adjustment step for a VAR process based on the one or more images and process parameters, predicting operating characteristics of the VAR process based on the process parameters and the remelt prediction model, and initiating a power adjustment step in response to the melt markers satisfying a predetermined melt marker condition, the operating characteristics of the VAR process satisfying a predetermined operating condition, or a combination thereof.

[0025] In one form, the remelt image process model further includes instructions for executing one or more image processing routines to detect the melt marker using a deep convolutional neural network.

[0026] In another form, the remelt prediction model further includes instructions for predicting operating characteristics based on one or more previously obtained process parameters associated with one or more previous VAR processes using a regression routine.

[0027] Further areas of applicability will become apparent from the description provided hereinafter. It should be understood that this description and specific examples are intended for purposes of illustration only and are not intended to limit the scope of the present disclosure. [Brief explanation of the drawings]

[0028] In order that the present disclosure may be properly understood, various aspects of the disclosure will now be described, by way of example, with reference to the accompanying drawings.

[0029] [Figure 1] FIG. 1 is a schematic diagram illustrating a vacuum arc remelting (VAR) system in accordance with the teachings of the present disclosure.

[0030] [Figure 2A] FIG. 2A is a top view of a slot in an electrode in accordance with the teachings of the present disclosure.

[0031] [Figure 2B] FIG. 2B is a top view of one or more pins of an electrode in accordance with the teachings of the present disclosure.

[0032] [Figure 3] FIG. 3 is a functional block diagram of a VAR system according to the teachings of the present disclosure.

[0033] [Figure 4] FIG. 4 is an example of a selected portion of an image identified by a VAR system according to the teachings of this disclosure.

[0034] [Figure 5] FIG. 5 is a flow chart of an example control routine in accordance with the teachings of the present disclosure.

[0035] The drawings described herein are for illustration purposes only and are not intended to limit the scope of the present disclosure in any way. DETAILED DESCRIPTION OF THE INVENTION

[0036] The following description is merely exemplary in nature and is not intended to limit the present disclosure, its applicability, or uses. It should be understood that throughout the drawings, corresponding reference numerals indicate like or corresponding parts and features.

[0037] The present disclosure provides a control system for a vacuum arc remelting (VAR) process that includes a VAR monitoring system that uses two detection schemes to identify when to initiate a power adjustment (i.e., power reduction) phase of the VAR process. Specifically, for the first detection scheme, the VAR monitoring system analyzes images from a vision system using a remelt image process model to determine whether a melt marker meets a predetermined melt marker condition. For the second detection scheme, the VAR monitoring system predicts the operating characteristics of the VAR process based on process parameters and a remelt prediction model. The VAR monitoring system initiates the power adjustment phase in response to the melt marker meeting the predetermined melt marker condition, the operating characteristics of the VAR process meeting the predetermined operating characteristics, or a combination thereof. Once detected, the VAR monitoring system instructs a primary VAR controller to initiate a power adjustment phase that gradually reduces power to the electrodes. Thus, a control system with the disclosed VAR monitoring system accurately transitions to the power reduction phase without relying on manual detection and / or operation.

[0038] One advantage of the present disclosure is that it automatically controls the power adjustment stage of the VAR process based on remelt image data and process parameters from a vision system. By automatically controlling the power adjustment stage of the VAR process to the electrode, the quality of the ingot melt of the VAR process is improved. Additionally, the VAR process described herein prevents or reduces material waste (e.g., electrode / molten ingot waste) while improving ingot productivity, furnace capacity utilization, and the cost-effectiveness of producing a consistently high-quality ingot melt.

[0039] 1 , VAR system 10 includes furnace 100, primary VAR controller 130, VAR monitoring system 200, and vision system 300. Furnace 100 heats electrode 102, which in one form heats housing 104 and crucible 106 that define vacuum chamber 108. Electrode 102 may be formed from a variety of conductive materials suitable for VAR processes, such as titanium, zirconium, nickel-based alloys, and steel, among other alloys. Crucible 106 may be formed from a variety of materials suitable for VAR processes, such as copper.

[0040] In one form, the furnace 100 can be configured to perform a VAR process to gradually melt the electrode 102 to form an ingot 110 including a pool of molten metal 112. During the VAR process, the primary VAR controller 130 is configured to lower the electrode ram 114 to vertically position the electrode 102 in close proximity to (e.g., adjacent to and / or near) the pool of molten metal 112 (e.g., an ingot melt). When the furnace ram position of the electrode ram 114 (e.g., the vertical position of one or both ends of the electrode ram 114 within the vacuum chamber 108) indicates that the electrode 102 will be in close proximity to the pool of molten metal 112, the primary VAR controller 130 is configured to supply power to the electrode 102 to generate an electric arc within the vacuum chamber 108. In some forms, the electric arc is configured to form a continuous melt between the electrode 102 and the pool of molten metal 112. To perform the functions described herein, the primary VAR controller 130, coupled to a power supply (not shown), may include driver circuits, actuators, switches, power converters, and / or other suitable electronic components to provide power to the electrode 102 and adjust the position of the electrode 102 within the crucible 106.

[0041] In one embodiment, the furnace 100 also includes one or more appropriate sensors and / or electrical hardware (not shown) for acquiring process parameters, such as, for example, physical and / or electrical properties of the electrode 102, as described in further detail below. In one embodiment, the furnace 100 includes a coolant guide 116, a coolant inlet 118, a coolant outlet 120, and a coolant chamber 122 defined by the outer wall of the vacuum chamber 108 and the inner wall of the housing 104. In one embodiment, a coolant (e.g., water) is supplied into the coolant chamber 122 via the coolant inlet 118 to reduce the temperature of the crucible 106. The coolant flows upward through the inner side 116A of the coolant guide 116, downward through the outer side 116B of the coolant guide, and exits the coolant chamber 122 via the coolant outlet 120. It should be understood that other systems for cooling the crucible 106 may be provided, such as, for example, a forced air system, and are not limited to the examples described herein.

[0042] In one form, the primary VAR controller 130 is configured to supply power to the electrodes 102 according to a power control routine. By way of example, the primary VAR controller 130 initially ramps power to a steady-state power (e.g., a predetermined steady-state current and / or voltage value) and then supplies the steady-state power to the electrodes 102. The primary VAR controller 130 is also configured to ramp down power from the steady-state power to zero in response to receiving a notification from the VAR monitoring system 200 to initiate a power adjustment phase, as described in more detail below.

[0043] The vision system 300 includes an imaging device 302 configured to capture images of the electrode 102 within the crucible 106. The imaging device 302 may include, but is not limited to, a two-dimensional camera and a three-dimensional camera. In one form, the imaging device 302 includes one or more serial digital interface (SDI) output connections. In one form, the SDI output connections provide a locking feature that allows a connecting interface cable to be locked in place, allowing image data to be transmitted over long distances (e.g., up to 300 feet). While the exemplary imaging device 302 may be positioned vertically, facing downward and toward the electrode 102B, those skilled in the art should understand that cameras can have a variety of positions and orientations and are not limited to the examples described and illustrated herein.

[0044] In one embodiment, the imaging device 302 is configured to capture an image of the end of the electrode 102 to capture an image of one or more melting markers. As used herein, a "melting marker" refers to a predetermined feature of the electrode 102 that is used to initiate a power adjustment phase. By way of example, with reference to FIGS. 2A-2B, the melting marker may include a slot 132 on the electrode 102A and / or a pin 134 on the electrode 102B. It should be understood that the melting marker may include other features of the electrode 102A, 102B and is not limited to the examples described herein. While examples of pins and slots are shown, one skilled in the art should understand that the pins and slots may have various shapes, geometries, and sizes and are not limited to the examples described and illustrated herein.

[0045] In one form, the predetermined melting marker condition includes an open or closed slot in the electrode, or an edge of the melting marker. In one example, the edge of the melting marker includes the absence of the pin 134 in the electrode.

[0046] VAR monitoring system 200 initiates a power adjustment step based on images from vision system 300 and / or process parameters of furnace 100. Specifically, VAR monitoring system 200 is configured to detect melt markers and perform image processing routines on the captured images to determine whether the melt markers meet predetermined melt marker conditions. In one form, a melt marker meets the predetermined melt marker condition if VAR system 200 detects that pin 134 and / or slot 132 have a predetermined geometric shape, area, orientation, size, and / or dimensions. In one example, the predetermined melt marker condition indicates the presence of a melt marker end or the absence of a pin. In another example, the predetermined melt marker condition indicates the presence of an open slot. VAR monitoring system 200 initiates a power adjustment step if the predetermined melt marker condition exists. VAR monitoring system 200 is also configured to predict operating characteristics of the VAR process based on the process parameters and initiates a power adjustment step in response to the operating characteristics meeting the predetermined operating conditions. Examples of process parameters include, but are not limited to, heat number, physical characteristics of the electrode 102 (e.g., electrode weight, electrode material, melt marker location, location of the bottom surface of the electrode, etc.), physical characteristics of the ingot melt (e.g., height of the ingot melt), identification number (e.g., identification number of the crucible 106, identification number of the furnace 100, or a combination thereof), electrical characteristics (e.g., voltage and / or current supplied to the furnace 100 via the electrode 102), arc area (e.g., gap between the bottom surface of the electrode and the ingot melt), and / or furnace ram position of the electrode 102. Additional details regarding predetermined operating conditions are described below.

[0047] 3 , a functional block diagram of a VAR system 10 having a furnace 100, a primary VAR controller 130, a VAR monitoring system 200, and a vision system 300 is provided. In one embodiment, the VAR system 10 further includes a process parameter controller 150, a control room 320, and a VAR history database 330. It should be readily understood that any one of the components of the VAR system 10 may be provided in the same location or distributed at different locations (e.g., via one or more edge computing devices) and communicatively coupled accordingly. In one embodiment, the furnace 100, the primary VAR controller 130, the process parameter controller 150, the VAR monitoring system 200, the vision system 300, the control room 320, and the VAR history database 330 are communicatively coupled using wired and / or wireless communication protocols (e.g., Bluetooth®-type protocols, cellular protocols, Wireless Fidelity (Wi-Fi)-type protocols, Near Field Communication (NFC) protocols, Ultra-Wideband (UWB) protocols, etc.).

[0048] In one form, the process parameter controller 150 obtains process parameters of the furnace 100 and selectively provides process parameters and / or control signals to the vision system 300, the VAR monitoring system 200, and / or the primary VAR controller 130. By way of example, the process parameter controller 150 provides control signals to the vision system 300 and the VAR monitoring system 200 when at least one of the following occurs: the furnace ram position of the electrode 102 is at a predetermined position; the power input to the furnace 100 via the electrode 102 is greater than a power setpoint; and / or the weight of the electrode 102 is at a predetermined weight limit. In one form, the control signals may instruct the vision system 300 and the VAR monitoring system 200 to initiate a vision analysis routine, an operating characteristic prediction routine, and / or a power adjustment step routine as described herein.

[0049] In one form, vision system 300 includes an imaging device 302 and an image processing module 304. Image processing module 304 is configured to segment image data acquired from imaging device 302 into one or more frames and convert the image data into a pixel format compatible with VAR monitoring system 200. Optionally, in some forms, image processing module 304 provides the segmented and converted image data (hereinafter referred to as "processed image data") to one or more display / computing devices located in control room 320 and / or VAR history database 330 for further processing.

[0050] In one embodiment, the VAR monitoring system 200 includes a visual analytics module 210 and a prediction module 220. In one embodiment, the visual analytics module 210 is configured to analyze processed image data from the vision system 300 and detect melting markers based on a remelted image process model 212. By way of example, the remelted image process model 212 is a deep convolutional neural network configured to perform various image processing routines to detect melting markers. Specifically, the remelted image process model 212 may include one or more convolutional layers defined by any suitable combination of parameters, including, but not limited to, kernel dimensions, number of kernels, stride values, padding values, input / output channels, bit depth, feature map width / length, and rectified linear unit (ReLU) activation layers. The remelted image process model 212 may also include residual layers, downsampling layers, flattening layers, and other convolutional neural network layers to analyze images from the vision system 300. It should also be understood that the refused image process model 212 may be trained according to various known training routines, and a description of the various training routines will be omitted for the sake of brevity. In one embodiment, the refused image process model 212 is configured to perform a semantics-based image processing routine on processed image data from the vision system 300. By way of example, the refused image process model 212 includes one or more reference images of the electrode 102 semantically labeled with fused markers (e.g., slots 132 and / or pins 134) during training of the refused image process model 212. In one embodiment, the one or more reference images of the electrode 102 may include one or more images having the same and / or different image qualities. The image quality may include one or more variables such as orientation, distortion, sharpness, vignetting, noise, brightness, dynamic range, color accuracy, flare, uniformity, lateral chromatic aberration, etc. In one embodiment, the reference image of the electrode 102 includes one or more orientations and / or one or more angles of the electrode 102.In one example, the reference image includes approximately 10,000 images, each image identifying a respective melt marker and a respective melt marker condition, such as the presence of a pin and whether each pin includes a pin end or is absent. In another example, the reference image includes approximately 10,000 images, each image identifying a respective melt marker and a melt marker condition, such as the presence of a slot and whether each slot is an open slot or a closed slot.

[0051] Thus, the remelted image process model 212 may perform a semantic-based image processing routine based on the processed image data and compare the classified objects with the melting markers in the reference image to detect the melting markers in the processed image data. It should be understood that the remelted image process model 212 may perform other image processing routines (e.g., difference-based image processing routines) and is not limited to the examples described herein. In one form, the visual analysis module 210 is configured to select portions of each image to be analyzed by the remelted image process model 212. In one form, the visual analysis module 210 may be configured to filter the processed image data such that the remelted image process model 212 analyzes pixels related to contours / edges of the electrode 102 and / or brightness / color transitions in the processed image data. As an example, as shown in FIG. 4 , the visual analysis module 210 filters / removes portion 360 of image 350 and selects section 370 to be analyzed based on portion 370 having both a nominal pixel contour 372 representing the edge of electrode 102 and one or more protruding pixel areas 374 representing one or more potential pins 134 of electrode 102.

[0052] In another embodiment, the remelted image process model is configured to use a bounding box regression model to select portions of each image to be analyzed by the remelted image process model 212. In one example, the visual analytics module 210 may be configured to localize objects to select portions of each image to be analyzed by the remelted image process model 212. The bounding regression model is trained using a loss function, such as mean squared error or mean absolute error, for a reference image and a bounding box of the melted markers in each of the reference images. The bounding box includes at least two bounding box coordinates, such as an upper-left (x, y) coordinate and a lower-right (x, y) coordinate. The visual analytics module uses the bounding box regression model to predict the bounding box coordinates to define the bounding box. The predicted bounding box corresponds to the selected portion of each image to be analyzed by the remelted image process model 212. In some variations, the visual analytics module 212 is configured to provide data regarding the location of the bounding box relative to the image and provide the data to the vision system 300, which may then provide a digital representation of the bounding box on an image of the VAR process displayed by a monitor in the control room 320. Thus, a technician visualizing the VAR process is informed of the portions of the electrode 102 that are expected to have melt markers.

[0053] In one form, the visual analytics module 210 is configured to generate an image-associated confidence score indicating whether the processed image data from the vision system 300 includes a melted marker. By way of example, the convolutional neural network may output a prediction distribution (e.g., a delta-based prediction interval, a Bayesian-based prediction interval, or a mean-variance estimation-based interval) indicating the probability that the melted marker was correctly detected and / or whether the detected melted marker satisfies a predetermined melted marker condition according to the remelt image process model 212. If the prediction distribution value is greater than a threshold, the visual analytics module 210 determines that the melted marker satisfies the predetermined melted marker condition. Additionally, if the melted marker satisfies the predetermined melted marker condition, the VAR monitoring system 200 broadcasts a notification to the primary VAR controller 130 to initiate a power adjustment phase to ramp down the power from steady-state power to a low power setpoint.

[0054] In one aspect, the prediction module 220 is configured to predict operating characteristics of the VAR process based on the process parameters and the remelt prediction model 222. In one aspect, the operating characteristics of the VAR process include, but are not limited to, furnace ram position. In one aspect, the process parameters used to predict the operating characteristics of the VAR process may include, but are not limited to, a melt marker position, an electrode weight, a position of the lower surface of the electrode, an arc area, an ingot melt height, etc. In one aspect, the prediction module 220 predicts the furnace ram position of the VAR process relative to the melt marker position. In one aspect, the remelt prediction model 222 predicts the operating characteristics of the VAR process by inputting the process parameters into a multivariate regression-based equation / model.

[0055] As an example, the remelt prediction model 222 is configured to predict the furnace ram position of the VAR process as an operating characteristic based on the process parameters. Specifically, the remelt prediction model 222 executes a recursive routine to obtain process parameters associated with a previous VAR process, such as the furnace ram position during a previous power adjustment step in which power was ramped down to zero, and calculate a predicted furnace ram position of the electrode 102. In one embodiment, each previous power adjustment step is ramped down to zero in association with a melt marker position and / or a melt marker condition being satisfied. Thus, the remelt prediction model 222 determines / updates coefficients used to predict the furnace ram position at various times in addition to predicting the furnace ram position. It should be understood that the remelt prediction model 222 may predict other operating characteristics based on the process parameters and is not limited to the examples provided herein. In one embodiment, the VAR monitoring system 200 is further configured to determine whether the predicted operating characteristic of the VAR process satisfies a predetermined operating condition. The predetermined operating condition is met when the operating characteristic corresponds to a numeric or qualitative value indicating, for example, that the operating characteristic is above (or below) a predetermined threshold, within a predetermined numerical tolerance, and / or corresponds to a predetermined qualitative value (e.g., the current furnace ram position is at the predetermined threshold). As an example, VAR monitoring system 200 determines whether a predicted furnace ram position corresponds to the current furnace ram position. In one form, VAR monitoring system 200 may broadcast a notification to primary VAR controller 130 to initiate a power adjustment phase to ramp down power based on the predicted operating characteristic of the VAR process. As an example, VAR monitoring system 200 may broadcast a command to initiate a power adjustment phase if the predicted furnace ram position corresponds to the current furnace ram position.

[0056] 5, a flow chart illustrating an example routine 500 for implementing a VAR process is shown. At 504, the process parameter controller 150 obtains process parameters from the furnace 100. At 508, the process parameter controller 150 determines whether the process parameters indicate that the VAR process is active. If the process parameters indicate that the VAR process is active, the routine 500 proceeds to 512. Otherwise, the routine 500 returns to 504 until the VAR process is active.

[0057] At 512, the process parameter controller 150 activates the VAR monitoring system 200 and the vision system 300. At 520, the vision system 300 processes the image data acquired by the imaging device 302. At 524, the vision system 300 provides the processed image data to the VAR monitoring system 200, and the process parameter controller 150 provides the process parameters to the VAR monitoring system 200.

[0058] At 528, the visual analytics module 210 analyzes the processed image data to detect melt markers, as described above. At 532, the prediction module 220 predicts the operating characteristics of the VAR process based on the process parameters, as described above. At 536, the VAR monitoring system 200 determines whether the visual analytics module 210 determines whether the melt markers meet predetermined melt marker conditions or whether the predicted operating characteristics meet predetermined operating conditions (e.g., the furnace ram position corresponds to the current furnace ram position). At 536, if the melt markers do not meet the predetermined melt marker conditions and the operating characteristics are not met, the routine 500 proceeds to 520. At 536, if the melt markers meet the predetermined melt marker conditions, the operating conditions are met, or a combination thereof, the routine 500 proceeds to 540, where the VAR monitoring system 200 notifies the primary VAR controller 130 to ramp down (reduce) the power from the steady-state value to a low power setpoint. For example, if the melt marker meets the predetermined melt marker condition and the operating characteristics are met, the VAR monitoring system 200 notifies the primary VAR controller 130 to ramp down the power from the steady state value to a low power set point.

[0059] Routine 500 is merely one example of a monitoring routine for a control system of the present disclosure, and other routines may be utilized. For example, in one variation, instead of issuing a notification to the primary VAR controller 130, the VAR monitoring system 200 may issue a notification to a technician in the control room 320, causing the technician to initiate a power adjustment step.

[0060] Unless otherwise expressly stated herein, all numerical values ​​expressing mechanical / thermal properties, compositional proportions, dimensions and / or tolerances, or other characteristics are understood to be modified by the term "about" or "approximately" for purposes of describing the scope of this disclosure. This modification is desirable for various reasons, including industrial practices, material, manufacturing, and assembly tolerances, and test performance.

[0061] As used herein, the phrase at least one of A, B, and C should be construed to mean a non-exclusive logical OR (A OR B OR C), and not to mean "at least one of A, at least one of B, and at least one of C."

[0062] In this application, the terms "controller" and / or "module" may refer to, be part of, or include an application specific integrated circuit (ASIC), a digital, analog, or mixed analog / digital discrete circuit, a digital, analog, or mixed analog / digital integrated circuit, a combinational logic circuit, a field programmable gate array (FPGA), a processor circuit (shared, dedicated, or group) that executes code, a memory circuit (shared, dedicated, or group) that stores code executed by the processor circuit, other suitable hardware components that provide the described functionality (e.g., an op-amp circuit integrator as part of a heat flux data module), or a combination of some or all of the above, such as, for example, a system on a chip.

[0063] The term memory is a subset of the term computer-readable medium. As the term computer-readable medium, as used herein, does not include transient electrical or electromagnetic signals propagating through a medium (e.g., on a carrier wave, etc.), and therefore the term computer-readable medium may be considered to be tangible and non-transitory. Non-limiting examples of non-transitory tangible computer-readable media are non-volatile memory circuits (e.g., flash memory circuits, erasable programmable read-only memory circuits, or mask read-only circuits), volatile memory circuits (e.g., static random access memory circuits or dynamic random access memory circuits), magnetic storage media (e.g., analog or digital magnetic tape or hard disk drives), and optical storage media (e.g., CDs, DVDs, or Blu-ray discs).

[0064] The apparatus and methods described in this application may be implemented in part or entirely by a special-purpose computer created by configuring a general-purpose computer to perform one or more specific functions embodied in a computer program. The functional blocks, flowchart components, and other elements described above serve as software specifications and may be converted into a computer program by the routine work of an engineer or programmer.

[0065] Because the description of the present disclosure is merely exemplary in nature, variations that do not depart from the essence of the disclosure are intended to be within the scope of the disclosure. Such variations are not to be regarded as a departure from the spirit and scope of the disclosure. 。 The inventions described in the original claims of this application are set forth below. [1] A control system for a vacuum arc remelting (VAR) process, comprising: a vision system including an imaging device configured to capture one or more images of an electrode within a vacuum chamber of the VAR process; a VAR monitoring system configured to determine a power adjustment stage of the VAR process based on the one or more images from the vision system and process parameters; The VAR monitoring system includes: a visual analysis module configured to analyze the one or more images from the vision system to detect a melt marker based on a remelt image process model; a prediction module configured to predict an operating characteristic of the VAR process associated with the power adjustment stage based on the process parameters and a remelt prediction model; Including, the VAR monitoring system is configured to initiate the power adjustment step in response to the melt marker satisfying a predetermined melt marker condition, the operating characteristics of the VAR process satisfying a predetermined operating condition, or a combination thereof. Control system. [2] The melt marker includes a pin, a slot, or a combination thereof; [1] The control system according to [1]. [3] The process parameters include a heat number, an electrode weight, a crucible identification, a furnace number, a furnace current, a furnace voltage, a furnace ram position, or a combination thereof; [1] The control system according to [1]. [4] For each image, the visual analysis module is configured to select a portion of the each image to be analyzed by the refused image process model. [1] The control system according to [1]. [5] the vision system and the VAR monitoring system initiate operation in response to the furnace ram position being in a predetermined position, the power input to the furnace of the VAR process being greater than a power setpoint, or a combination thereof; [1] The control system according to [1]. [6] the remelt prediction model is configured to predict a furnace ram position of the VAR process relative to a melt marker position. [1] The control system according to [1]. [7] further comprising a primary VAR controller configured to control power to a furnace of the VAR process, wherein the VAR monitoring system is further configured to initiate the power adjustment phase by notifying the primary VAR controller to initiate a power ramp-down. [1] The control system according to [1]. [8] A method for controlling a vacuum arc remelting (VAR) process, comprising: acquiring one or more images of the electrode within the vacuum chamber; analyzing the one or more images to detect melt markers in the one or more images based on a remelt image process model; determining a power adjustment stage of a VAR process based on the one or more images and process parameters; predicting an operating characteristic of the VAR process based on the process parameters and a remelt prediction model; initiating the power adjusting step in response to the melting marker satisfying a predetermined melting marker condition, the operating characteristics of the VAR process satisfying a predetermined operating condition, or a combination thereof. method. [9] The melt marker includes a pin, a slot, or a combination thereof; [8] The method described in [8].

[10] The process parameters include a heat number, an electrode weight, a crucible identification, a furnace number, a furnace current, a furnace voltage, a furnace ram position, or a combination thereof; [8] The method described in [8].

[11] further comprising selecting a portion of each image to be analyzed by the remelted image process model. [8] The method described in [8].

[12] further comprising operating in response to a furnace ram position being in a predetermined position, a power input to the furnace of the VAR process being greater than a power setpoint, or a combination thereof. [8] The method described in [8].

[13] further comprising predicting a furnace ram position of the VAR process relative to a melt marker position. [8] The method described in [8].

[14] Initiating the power adjustment step further includes initiating a power ramp-down routine. [8] The method described in [8].

[15] the predetermined operating condition includes determining whether a predicted furnace ram position of the electrode corresponds to a current furnace ram position; [8] The method described in [8].

[16] The remelt image process model further comprises performing one or more image processing routines to detect the melt marker using a deep convolutional neural network. [8] The method described in [8].

[17] The remelt prediction model further comprises using a regression routine to predict the operating characteristics based on one or more previously obtained process parameters associated with one or more previous VAR processes. [8] The method described in [8].

[18] A system for controlling a vacuum arc remelting (VAR) process, comprising: a processor; a non-transitory computer-readable medium containing instructions executable by the processor; The instruction: acquiring one or more images of the electrode within the vacuum chamber; analyzing the one or more images to detect melt markers in the one or more images based on a remelt image process model; determining a power adjustment stage for the VAR process based on the one or more images and process parameters; predicting operating characteristics of the VAR process based on the process parameters and a remelt prediction model; initiating the power adjustment step in response to the melting marker satisfying a predetermined melting marker condition, the operating characteristics of the VAR process satisfying a predetermined operating condition, or a combination thereof. system.

[19] The remelt image process model further comprises instructions for executing one or more image processing routines to detect the melt marker using a deep convolutional neural network. The system described in

[18] .

[20] the remelt prediction model further comprises instructions for predicting the operating characteristic based on one or more previously obtained process parameters associated with one or more previous VAR processes using a regression routine. The system described in

[18] .

Claims

1. 1. A control system for a vacuum arc remelting (VAR) process, comprising: a vision system including an imaging device configured to capture one or more images of an electrode within a vacuum chamber of the VAR process; a VAR monitoring system configured to determine a power adjustment stage of the VAR process based on the one or more images from the vision system; The VAR monitoring system comprises: a visual analysis module configured by a processor to analyze the one or more images from the vision system to detect pins, slots, or combinations thereof of a melt marker based on a remelt image process model; the VAR monitoring system is configured to initiate the power adjustment step by a VAR controller in response to the pin, the slot, or the combination thereof of the melting marker satisfying a predetermined melting marker condition. Control system.

2. The VAR monitoring system is configured to determine the power adjustment stage further based on a process parameter; The process parameters include a heat number, an electrode weight, a crucible identification, a furnace number, a furnace current, a furnace voltage, a furnace ram position, or a combination thereof; The control system of claim 1 .

3. and for each image, the visual analysis module is configured to select a portion of the respective image to be analyzed by the remelting image process model. The control system of claim 1 .

4. the vision system and the VAR monitoring system initiate operation in response to the furnace ram position being at a predetermined position, the power input to the furnace of the VAR process being greater than a power setpoint, or a combination thereof; The control system of claim 1 .

5. The method further comprises a prediction module configured by a processor to predict a furnace ram position of the VAR process associated with the power adjustment stage based on process parameters and a remelt prediction model; the VAR monitoring system is configured to initiate the power adjustment step in response to the furnace ram position of the VAR process satisfying a predetermined operating condition; the remelt prediction model is configured to predict the furnace ram position of the VAR process relative to a melt marker position. The control system of claim 1 .

6. a primary VAR controller configured to control power to a furnace of the VAR process, the VAR monitoring system further configured to initiate the power adjustment phase by notifying the primary VAR controller to initiate a power ramp-down. The control system of claim 1 .

7. 1. A method for controlling a vacuum arc remelting (VAR) process, comprising: acquiring one or more images of the electrode within the vacuum chamber; analyzing the one or more images to detect pins, slots, or combinations thereof of melt markers in the one or more images based on a remelt image process model; determining a power adjustment stage of a VAR process based on the one or more images; initiating the power adjusting step by a VAR controller in response to the pin, the slot, or the combination thereof of the melting marker satisfying a predetermined melting marker condition. method.

8. The method of claim 7, wherein determining the power adjustment stage of the VAR process is further based on a process parameter; The process parameters include a heat number, an electrode weight, a crucible identification, a furnace number, a furnace current, a furnace voltage, a furnace ram position, or a combination thereof; The method of claim 7.

9. selecting a portion of each image to be analyzed by the remelting image process model. The method of claim 7.

10. and operating in response to a furnace ram position being in a predetermined position, a power input to the furnace of the VAR process being greater than a power setpoint, or a combination thereof. The method of claim 7.

11. and predicting a furnace ram position of the VAR process relative to a melt marker position. The method of claim 7.

12. Initiating the power adjustment phase further includes initiating a power ramp-down routine. The method of claim 7.

13. The method of claim 12, further comprising predicting a furnace ram position of the VAR process based on process parameters and a remelt prediction model; initiating the power adjustment step is in response to the furnace ram position of the VAR process satisfying a predetermined operating condition; the predetermined operating condition includes determining whether a predicted furnace ram position of the electrode corresponds to a current furnace ram position; The method of claim 7.

14. the remelt image process model further comprises performing one or more image processing routines to detect the pins, the slots, or the combination thereof of the melt marker using a deep convolutional neural network. The method of claim 7.

15. the remelt prediction model further comprises using a regression routine to predict the operating conditions based on one or more previously obtained process parameters associated with one or more previous VAR processes. The method of claim 13.

16. 1. A system for controlling a vacuum arc remelting (VAR) process, comprising: a processor; a non-transitory computer-readable medium containing instructions executable by the processor; The instruction: acquiring one or more images of the electrode within the vacuum chamber; analyzing the one or more images to detect melt markers in the one or more images based on a remelt image process model including a deep convolutional neural network; determining a power adjustment step for the VAR process based on the one or more image and process parameters; predicting operating characteristics of the VAR process based on the process parameters and a remelt prediction model; initiating the power adjustment step by a VAR controller in response to the melting marker satisfying a predetermined melting marker condition, the operating characteristics of the VAR process satisfying a predetermined operating condition, or a combination thereof. system.

17. the remelt prediction model further comprising instructions for predicting the operating characteristic based on one or more previously obtained process parameters associated with one or more previous VAR processes using a regression routine.

17. The system of claim 16.

18. The remelting image process model includes a deep convolutional neural network. The control system of claim 1 .

19. The remelting image process model includes a deep convolutional neural network. The method of claim 7.

Images