Industrial-grade multi-station horizontal zone refining purification full-automatic control system and method

Abstract

The application discloses an industrial multi-station horizontal zone purification full-automatic control system and method, and belongs to the technical field of semiconductor material purification automation. The system comprises a multi-station visual measurement subsystem, an intelligent collaborative temperature control subsystem, a heating unit movement control subsystem, a multi-station collaborative scheduling module and a central controller. The visual measurement subsystem adopts a double-side camera array to collect the width of each station melting zone; the intelligent collaborative temperature control subsystem calculates the target power based on multi-heat source coupling compensation and feedforward feedback compound control; the collaborative scheduling module maintains an independent process state machine for each station, realizes time multiplexing scheduling of visual acquisition, time sequence synchronization of measurement-temperature control-movement and abnormal hierarchical processing. The application realizes the full automation of the industrial multi-station horizontal zone purification equipment for the first time, completes full-station coverage under limited hardware resources, achieves a single closed loop cycle of seconds, supports unattended continuous production, and has a perfect safety protection mechanism.

Description

Technical Field

[0001] This invention belongs to the field of semiconductor material purification automation technology, specifically relating to an industrial-grade multi-station horizontal zone melting purification fully automatic control system and method, particularly a fully automatic control system and method integrating visual measurement, intelligent temperature control and multi-station collaborative scheduling for industrial-grade horizontal zone melting purification equipment with multiple quartz tubes and multiple heating units. Background Technology

[0002] Horizontal zone melting purification is a mature technology for the physical purification of semiconductor materials using the principle of segregation. In this process, the material to be purified is loaded into a carrier boat, which is placed horizontally within a quartz tube. A heating unit moves slowly along the axial direction of the quartz tube, forming a localized melting zone (i.e., a "melting zone") on the material ingot, and this melting zone moves from one end of the material to the other. After multiple melting movements, impurities are concentrated at both ends of the material, while the middle portion is purified.

[0003] To improve purification efficiency and production capacity, industrial-grade horizontal zone melting equipment typically employs a parallel structure with multiple tubes and multiple stations. Taking a typical 4-tube system as an example, the equipment has four quartz tubes arranged in parallel, with three independently controlled heating units along its axial direction for a total of 12 heating units (i.e., 12 stations) operating simultaneously. The material inside each quartz tube needs to undergo multiple zone melting processes (usually 5 to 10 times), and the entire process cycle can take several days.

[0004] In terms of automation technology, existing technologies have attempted to improve single-stage zone melting purification. For example, CN118166421A discloses a method for determining the length of the molten zone using image recognition; CN102392294A discloses a preparation process integrating multiple purification methods. However, these technical solutions are all aimed at small-scale laboratory single-tube devices or floating zone single crystal growth, only solving the problem of a single stage and not involving the full-process automation of multi-station industrial-grade horizontal zone melting equipment. US4080172A discloses an automatic control method and device for floating zone refining, which uses two sensor arrays to locate the image of the melt / solid interface, measure the diameter, length, and "bag-like" or overhang angle of the molten zone, and processes the output of the sensor arrays into signals to operate the zone melting refining controller, thereby realizing the automation of the floating zone refining and single crystal semiconductor rod growth process. However, this technical solution still has the following shortcomings: First, it targets the floating zone single crystal growth process, where the material rod is vertically suspended and no carrier boat is required. This is fundamentally different from the horizontal zone refining purification process targeted by this invention in terms of equipment structure (vertical vs. horizontal), process purpose (single crystal growth vs. material purification), and controlled object (crystal diameter vs. melt zone width). Second, it is only a single-station solution and does not involve an industrial-grade multi-station architecture with multiple quartz tubes and multiple heating units operating in parallel. It also does not solve the problems of visual acquisition resource competition and process cycle coordination caused by multi-station parallelism. Third, its controlled object is only the melt zone parameters of a single material rod. It does not involve the movement control of the heating unit along the axial direction of the quartz tube and the management of the number of zone refining cycles. Furthermore, it does not integrate visual measurement, intelligent temperature control, and movement control into a complete multi-stage closed-loop system.

[0005] Specifically, simply combining existing single-stage technical solutions cannot directly achieve fully automated operation of industrial-grade equipment, and the following technical obstacles remain:

[0006] Hardware resource conflicts in multi-station visual acquisition: Configuring independent cameras for each of the 12 workstations is costly, space-constrained, and disrupts the equipment's temperature field. How can all workstations be covered with limited hardware?

[0007] Timing coordination between measurement, temperature control, and movement control: Visual measurement requires synchronization between image capture and heating unit position; temperature control decisions require timely response; and the heating unit's movement speed needs to match the zone melting process. How can these various stages be coordinated?

[0008] Multi-station parallel scheduling and exception handling: 12 workstations operate simultaneously, with different workstations potentially at different stages of the process. How to manage the process status of each workstation? How to safely handle exceptions?

[0009] In existing technologies, the operation of the aforementioned industrial-grade horizontal zone melting equipment mainly relies on manual operation. The specific process is as follows: operators continuously observe the melting zone conditions at each station through an observation window under high-temperature conditions → judge the suitability of the melting zone width based on experience → manually adjust the temperature setpoints of each heating unit → control the movement of the heating units to complete the number of zone melting cycles. This manual operation mode has the following prominent problems:

[0010] 1) High labor intensity and cannot be unattended: 12 workstations require continuous monitoring by operators, and a single batch can last for tens of hours, making it impossible to achieve 24 / 7 automated production.

[0011] 2) Difficulty in ensuring process consistency: The width of the melting zone is subject to subjective error due to manual visual inspection. The judgment standards of different operators and the same operator at different times are inconsistent, resulting in large differences in purification effects between batches and workstations.

[0012] 3) Response lag and low control precision: The closed-loop cycle of manual observation → judgment → temperature adjustment is usually several minutes. During this period, the width of the melting zone may have drifted significantly, affecting purification efficiency and quality.

[0013] 4) Difficulty in coordinating multiple workstations: The heating units at 12 workstations move along the axial direction of the quartz tube, and the moving speed needs to be coordinated with the number of zone melting cycles. Manual management of the process cycle time of multiple workstations is prone to errors, such as heating units moving ahead or behind, or missing zone melting cycles.

[0014] In summary, there is currently no fully automated control system for industrial-grade multi-station horizontal zone melting purification equipment. How to integrate visual measurement, intelligent temperature control, and multi-station scheduling into a complete, collaborative, and reliable fully automated system is a pressing technical problem that needs to be solved in this field. Summary of the Invention

[0015] The technical problem this invention aims to solve is to address the lack of a fully automated control system in existing industrial-grade multi-station horizontal zone melting purification equipment. This invention provides an industrial-grade fully automated control system and method for multi-station horizontal zone melting purification. This system integrates visual measurement, intelligent temperature control, heating unit movement control, and multi-station collaborative scheduling into a complete closed-loop automated process. Under limited hardware resources, it achieves full-coverage monitoring and unattended operation of multi-tube, multi-station equipment, and possesses comprehensive process status management and anomaly handling capabilities.

[0016] The technical solution of this invention is as follows:

[0017] An industrial-grade, multi-station, fully automated control system for horizontal zone melting and purification is applied to a material horizontal zone melting and purification equipment with M quartz tubes and N heating units, where M ≥ 1 and N ≥ 1. Each quartz tube has one or more heating units arranged along its axial direction. Each heating unit and the material section heated within its corresponding quartz tube constitute an independent station. The number of stations is equal to the total number of heating units N. Each heating unit can move independently along the axial direction of its corresponding quartz tube. The system includes:

[0018] A multi-station visual measurement subsystem is used to acquire real-time melt zone width data for each heating unit's corresponding station. The visual measurement subsystem includes an industrial camera array symmetrically arranged on both sides of the quartz tube, and an image processing unit. The image processing unit executes a melt zone width measurement method based on bilateral vision: adaptive exposure optimization, deep learning-based coarse boundary localization, and traditional image processing-based fine boundary extraction are performed on the melt zone boundary images acquired synchronously by both cameras. The complete melt zone width is obtained by fitting the bilateral boundary line positions with the fixed geometric parameters of the heating unit. The visual measurement subsystem acquires melt zone boundary images for each station in a time-sharing or parallel manner according to a preset scheduling strategy, calculates and outputs the real-time melt zone width value for each station.

[0019] The intelligent collaborative temperature control subsystem is communicatively connected to the vision measurement subsystem. It receives real-time melting zone width values ​​from each workstation and calculates the target heating power for each heating unit based on the deviation between the melting zone width and the target width. The temperature control subsystem executes an intelligent collaborative temperature control method oriented towards multi-heat source coupling. "Multi-heat source coupling" refers to each heating unit acting as an independent heat source, whose heat influences the temperature field of adjacent workstations through radiation, conduction, etc., resulting in mutual interference. When calculating the target power: the thermal coupling compensation is calculated based on the relative positional relationship between each heating unit and the current power. The melting rate and solidification rate characteristics of each heating unit are extracted from historical operating data. Based on the current melting zone width, the thermal coupling compensation, and the aforementioned characteristics, the trend of melting zone width change is predicted. A feedforward and feedback composite control structure is used to calculate the target heating power.

[0020] Heating unit movement control subsystem: Communicatively connected to the drive motor of each heating unit, it controls the moving speed, direction, and stroke of each heating unit along the axial direction of the quartz tube. The movement control subsystem generates movement control commands for each heating unit based on preset zone melting process parameters (including the number of zone melting cycles, moving speed, and single movement stroke).

[0021] Multi-station collaborative scheduling module: As the core scheduling unit of the system, it is connected to the visual measurement subsystem, the intelligent collaborative temperature control subsystem, and the heating unit movement control subsystem. The multi-station collaborative scheduling module includes:

[0022] 1) Workstation Status Management Unit: Used to track the current process stage and zone melting progress of each workstation, manage the process state machine of each workstation, independently track the current process stage and zone melting progress of each workstation, and ensure that each workstation executes accurately according to the preset number of zone melting times and movement direction;

[0023] 2) Visual acquisition scheduling unit: used to dynamically allocate and arbitrate shooting tasks based on the real-time location of each heating unit and camera resources; generate time-division multiplexing acquisition scheduling instructions for the visual measurement subsystem; dynamically allocate shooting tasks to available cameras based on the real-time location of each heating unit and limited camera resources; and arbitrate and sort tasks according to workstation priority and shooting urgency when resource conflicts occur.

[0024] Timing Coordination Unit: Used to synchronize the execution rhythm of measurement, temperature control, and movement, ensuring the periodicity and synchronicity of closed-loop control; coordinating the timing relationship between measurement, temperature control, and movement to ensure the periodicity and synchronicity of the closed-loop process of "image acquisition → width measurement → temperature control decision → power execution → movement execution", so that the single closed-loop cycle reaches the second level;

[0025] 3) Monitor abnormal events in the process in real time and implement a graded protection strategy—classify abnormalities into three levels: minor, moderate, and severe, and take different response measures such as logging, triggering automatic recovery, and emergency shutdown alarm.

[0026] Central Controller: Serving as the system's hardware carrier and execution environment, it includes a processor, memory, communication interface, and human-machine interface. The central controller deploys the aforementioned subsystem modules, manages system operation in a unified manner, and provides functions such as process parameter setting, real-time status monitoring, historical data recording, and human-machine interaction.

[0027] This invention also provides an industrial-grade, multi-station, horizontal zone melting and purification fully automated control method, applied to the aforementioned system, comprising the following steps:

[0028] Step S1: System Initialization

[0029] After the equipment is powered on, it performs a system self-test, including checking the communication status of each camera, the status of each heating unit drive motor, and the status of each temperature sensor and actuator. After the self-test passes, preset process parameters are loaded, including the target melting zone width for each quartz tube. Heating unit moving speed V, total number of zone melting cycles Single zone melting process Then, reset each heating unit to its initial position (the first end of the quartz tube) and reset the camera to its initial shooting position.

[0030] Step S2: Process Task Generation and Allocation

[0031] Based on the equipment configuration (M quartz tubes, N heating units) and the applied process parameters, an independent process task sequence is generated for each station. The process task sequence for each station includes: the number of zone melting cycles, the direction of movement for each zone melting cycle (forward / reverse), the target melting zone width, and the current process stage identifier.

[0032] Step S3: Start the zone melting process

[0033] The central controller sends a start command to each heating unit, and the heating units begin to heat up to the preset initial temperature. When the temperature of the heating unit reaches near the melting point of the material, the heating unit is activated to move, and the molten zone begins to form.

[0034] Step S4: Multi-station visual acquisition scheduling

[0035] The multi-station collaborative scheduling module generates time-division multiplexing acquisition scheduling instructions for the visual measurement subsystem based on the real-time position information of each heating unit. The scheduling strategy includes:

[0036] 1) Time-sharing multiplexing scheduling: When the number of cameras is less than the number of workstations, one camera covers multiple workstations in a time-sharing manner. The field of view coverage of each camera along the quartz tube axis is divided into several shooting windows. Based on the current position of the heating unit of each workstation, it is determined which shooting window it falls into, and the workstation is dynamically added to or removed from the current acquisition queue.

[0037] 2) Synchronous Trigger Control: For workstations entering the acquisition queue, the corresponding camera is controlled to trigger image acquisition when the heating unit moves to the preset shooting reference position, and the left and right cameras are exposed synchronously.

[0038] 3) Resource conflict arbitration: When multiple workstations request the same camera resource at the same time, arbitration is carried out according to the workstation priority and the urgency of shooting to ensure that the image acquisition of key workstations is not missed.

[0039] Step S5: Real-time measurement of melt zone width

[0040] The vision measurement subsystem performs molten zone width measurement based on bilateral vision on the acquired images and outputs the real-time molten zone width value for each station. The measurement includes: adaptive exposure optimization of the molten zone boundary images acquired synchronously by bilateral cameras, coarse boundary localization based on deep learning, and fine boundary extraction based on traditional image processing. The complete width of the molten zone is obtained by fitting the bilateral boundary line positions with the fixed geometric parameters of the heating unit.

[0041] Step S6: Intelligent Collaborative Temperature Control Decision

[0042] The intelligent collaborative temperature control subsystem receives the melting zone width values ​​of each station output in step S5, executes the intelligent collaborative temperature control algorithm, and calculates the target power of each heating unit. The temperature control algorithm includes: calculating the thermal coupling compensation amount based on the relative positional relationship between each heating unit and the current power; extracting the melting rate characteristics and solidification rate characteristics of each heating unit from historical operating data; predicting the melting zone width change trend based on the current melting zone width, the thermal coupling compensation amount, and the aforementioned characteristics; and calculating the target heating power using a feedforward and feedback composite control structure.

[0043] Step S7: Heating unit movement control

[0044] The heating unit movement control subsystem controls each heating unit to move along the quartz tube axis at a preset speed V according to the process task sequence. This occurs when a heating unit completes a single zone melting stroke. Then, determine whether the preset number of melting zones has been reached:

[0045] a) If not reached, switch the movement direction (reverse), increment the melting count by 1, and continue to execute steps S4 to S7;

[0046] b) If the condition has been met, the workstation will enter the process completion stage.

[0047] Step S8: Process Status Monitoring and Anomaly Handling

[0048] Throughout the system's operation, the following status indicators are monitored in real time: whether the width of the melting zone at each workstation exceeds the preset safety range; whether each camera is functioning properly; whether the temperature of each heating unit exceeds the limit; and whether the movement position of each heating unit is abnormal. When an abnormality is detected, a tiered handling strategy is executed:

[0049] 1) Level 1 anomaly (minor anomaly, such as a momentary slight deviation in the width of the melt zone): Only logs are recorded, and the process continues to operate;

[0050] 2) Level 2 anomaly (medium anomaly, such as packet loss in camera communication or continuous deviation in melt zone width): Triggers an automatic adjustment strategy to attempt recovery and simultaneously issues a prompt to the operator; if the automatic recovery fails for a preset number of consecutive times, it is upgraded to a Level 3 anomaly;

[0051] 3) Level 3 abnormality (serious abnormality, such as heating unit overheating, camera completely disconnected, or movement exceeding safety limits): Immediately stop the operation of the relevant workstation or the entire machine, issue an audible and visual alarm, and wait for manual handling.

[0052] Step S9: Process Completion Judgment

[0053] The multi-station collaborative scheduling module monitors the process progress of each station in real time. When a station completes the preset number of zone melting operations, the station is marked as "completed". When all stations are marked as "completed", step S10 is executed; otherwise, steps S4 to S9 are executed repeatedly.

[0054] Step S10: Process completion and shutdown

[0055] The system controls each heating unit to gradually reduce its power according to a preset cooling curve until heating stops completely. Once the quartz tube has cooled to a safe temperature, the system enters standby mode, awaiting the next batch of operations.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] 1. For the first time, fully automated operation of industrial-grade multi-station horizontal zone melting and purification equipment has been achieved.

[0058] This invention does not simply piece together independent subsystems such as visual measurement, intelligent temperature control, and motion control. Instead, it organically integrates these processes into a closed-loop system through a multi-station collaborative scheduling module. This scheduling module takes the process state machine of each station as its core and uniformly manages the time-division multiplexing scheduling of visual acquisition, the timing triggering of temperature control decisions, and the process coordination of motion control. This enables the originally independent processes to work together to form a second-level rapid closed loop of "measurement → decision → execution → re-measurement", covering the entire process from process start-up to end, and truly realizing unattended automated production.

[0059] 2. Achieve full workstation coverage with limited hardware resources.

[0060] Through the time-sharing multiplexing scheduling strategy in the multi-workstation collaborative scheduling module, one camera can cover multiple workstations in shifts. Taking a 4-pipe 12-workstation device as an example, only 4 cameras are needed to complete the visual monitoring of all workstations, which significantly reduces hardware costs compared to the "one camera per workstation" solution. At the same time, it avoids the damage to the equipment's temperature field caused by installing a large number of cameras.

[0061] 3. Precise coordination of multi-station process cycles.

[0062] The system maintains an independent process state machine for each workstation, uniformly managing the number of zone melting cycles, movement direction, and process stages. This ensures that each workstation executes precisely according to preset process parameters, eliminating problems such as omissions and incorrect directions that are prone to occur in manual management, and significantly improving process consistency between batches.

[0063] 4. Closed-loop control offers rapid response and high precision.

[0064] The system tightly integrates bilateral vision-based melt zone width measurement with intelligent collaborative temperature control for multi-heat source coupling, forming a rapid closed loop of "measurement → decision → execution → re-measurement". The single closed-loop cycle reaches the second level, far superior to the minute-level response of manual operation. The melt zone width control accuracy meets the process requirements of industrial-grade horizontal zone melting purification.

[0065] 5. Comprehensive anomaly monitoring and security protection.

[0066] The system features a tiered anomaly handling mechanism, capable of real-time monitoring of key indicators such as melting zone width, equipment status, and communication status, and automatically taking corresponding measures when anomalies occur. This three-tiered anomaly handling strategy ensures safe equipment operation and prevents material scrapping or equipment damage caused by undetected anomalies.

[0067] 6. Good scalability and versatility.

[0068] The system architecture of this invention is independent of the specific number of workstations; it can be adapted to industrial-grade horizontal zone melting equipment of different specifications, such as 2-tube, 4-tube, and 6-tube systems, by modifying configuration parameters. The visual measurement and temperature control algorithms are modularly designed, allowing for individual upgrades or replacements, facilitating system maintenance and functional expansion. Attached Figure Description

[0069] Figure 1 This is an overall architecture diagram of the industrial-grade multi-station horizontal zone melting purification fully automatic control system provided in an embodiment of the present invention.

[0070] Figure 2 This is a schematic diagram of the station layout and system coverage of the 4-tube 12-station horizontal zone melting equipment in an embodiment of the present invention. In the figure: Q1~Q4—the 1st to 4th quartz tubes; H1~H3—the 1st to 3rd heating units corresponding to each quartz tube; C1_L, C2_L—the 1st and 2nd cameras on the left; C1_R, C2_R—the 1st and 2nd cameras on the right.

[0071] Figure 3 The above is an overall flowchart of the fully automatic control method provided in the embodiments of the present invention.

[0072] Figure 4 This is a schematic diagram of time-sharing multiplexing scheduling for multi-station visual acquisition.

[0073] Figure 5 This is the timing diagram for closed-loop control.

[0074] Figure 6 This is a schematic diagram of the process state machine at the workstation.

[0075] Figure 7 This is a flowchart of multi-workstation collaborative scheduling.

[0076] Figure 8This is a flowchart for handling anomalies in a hierarchical manner. Detailed Implementation

[0077] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments described herein are only for explaining the present invention and are not intended to limit the present invention.

[0078] Example 1: Fully automatic control system applied to a 4-tube, 12-station industrial-grade horizontal zone melting equipment

[0079] This embodiment takes a typical industrial-grade horizontal zone melting purification device with 4 quartz tubes and 3 heating units arranged along its axial direction in each quartz tube as an example. That is, the number of quartz tubes M=4, the total number of heating units N=12 (3 heating units are arranged along the axial direction in each quartz tube), and the number of workstations is equal to the total number of heating units, that is, 12 workstations Q1-H1, Q1-H2, Q1-H3, Q2-H1...Q4-H3, to explain the specific implementation of the present invention in detail.

[0080] I. System Overall Architecture

[0081] Reference Figure 1 This diagram illustrates the communication connections between the central controller, the multi-station vision measurement subsystem, the intelligent collaborative temperature control subsystem, the heating unit movement control subsystem, and the multi-station collaborative scheduling module in a modular block diagram format. The central controller, as the hardware carrier, is located at the center of the architecture. Each subsystem is connected to the central controller via a data bus. The multi-station collaborative scheduling module is connected to the vision measurement, temperature control, and movement control subsystems to achieve timing coordination. The fully automatic control system in this embodiment includes the following components:

[0082] 1. Multi-station vision measurement subsystem

[0083] Hardware configuration: A multi-functional visual monitoring device is used, including four industrial CMOS cameras (two on each side), mounted on a guide rail and movable along the quartz tube axis. The cameras are connected to an industrial control computer via Gigabit Ethernet.

[0084] The industrial CMOS camera has a resolution of 1920×1200 pixels, a frame rate of no less than 30fps, and is equipped with a 12mm fixed-focus lens with a field of view of 30° horizontally and 20° vertically. The camera is mounted on both sides of the quartz tube, at a distance of 200~300mm from the outer wall of the tube, ensuring that the field of view covers the area where the melting zone is located. The repeatability of the guide rail moving mechanism is no less than ±0.1mm to ensure the accuracy of camera switching between different shooting positions.

[0085] Software Function: It implements a molten zone width measurement method based on dual-view vision, performs adaptive exposure optimization, deep learning coarse localization, and traditional image processing fine extraction on the molten zone boundary images acquired synchronously by dual cameras. It obtains the complete width of the molten zone by fitting the position of the dual-view boundary line with the fixed geometric parameters of the heating unit, and outputs the molten zone width value of each station in real time.

[0086] 2. Intelligent Collaborative Temperature Control Subsystem

[0087] The control software module, running on the industrial computer, exchanges data with the vision measurement subsystem via shared memory or message queues. It implements an intelligent collaborative temperature control method for multi-heat source coupling. When calculating the target power, it calculates the thermal coupling compensation based on the relative position of each heating unit and the current power. It extracts melting / solidification rate characteristics from historical operating data to predict changes in the melt zone width and employs a feedforward and feedback composite control structure to calculate the target power for each heating unit. Power commands are sent to the thyristor power regulators of each heating unit via the Modbus TCP protocol.

[0088] 3. Heating Unit Movement Control Subsystem

[0089] The hardware includes: stepper motor drivers, limit switches, and position encoders for each heating unit. The drivers are connected to the industrial computer via an RS485 bus.

[0090] Software functions: Generate movement control commands (speed, direction, start / stop) based on process tasks, monitor the real-time position of each heating unit, and handle limit trigger events.

[0091] 4. Multi-station collaborative scheduling module

[0092] Reference Figure 5 As the core scheduling logic of the system, it runs as an independent thread on the industrial control computer. The main data structures include:

[0093] 1) Workstation Status Table: Records the current process stage, number of melting zones, melting zone width, target width, and movement direction of 12 workstations;

[0094] 2) Camera Resource Table: Records the current location, field of view coverage, and current workstation list of 4 cameras;

[0095] 3) Scheduling queue: Manages the data collection tasks to be executed.

[0096] 5. Central controller (industrial computer)

[0097] The industrial control computer is equipped with a processor, memory, solid-state drive, gigabit Ethernet card, RS485 communication card and GPU accelerator card (for deep learning inference) and runs fully automatic control software.

[0098] 6. Human-computer interaction interface

[0099] The 23-inch industrial touch screen provides functions such as parameter setting, status monitoring, image preview, and historical query.

[0100] II. Equipment workstation layout

[0101] Reference Figure 2 , Figure 2 This diagram illustrates the layout of the 4-tube, 12-station horizontal zone melting equipment in this embodiment, along with the camera coverage relationship. The diagram, viewed from a top-down perspective, shows the relative positions of four horizontally parallel quartz tubes Q1-Q4, three heating units H1-H3 arranged along the axial direction of each tube, and four cameras C1_L, C2_L, C1_R, and C2_R symmetrically arranged on both sides of the quartz tubes. Dashed boxes indicate the field of view coverage of each camera (each camera covers the same side boundary of two heating units on each of the two quartz tubes), and arrows indicate the movable direction of the cameras along the guide rails. Each heating unit and the material segment heated within its corresponding quartz tube constitute an independent station. In other words, each pair of "heating unit + the material segment within the quartz tube heated by that heating unit at the current moment" constitutes one station, and the total number of stations equals the total number of heating units.

[0102] In this embodiment, four quartz tubes Q1 to Q4 are arranged horizontally in parallel. Each quartz tube has three heating units (H1 to H3) arranged axially, for a total of 12 workstations. The workstations are numbered as follows: Q1-H1, Q1-H2, Q1-H3, Q2-H1, and so on.

[0103] Four cameras, C1_L, C2_L, C1_R, and C2_R, are symmetrically arranged on both sides of the quartz tubes. Each camera covers the same-side boundary of two heating units on each of the two quartz tubes. The cameras can be moved along guide rails to extend coverage to a third heating unit.

[0104] III. Fully Automatic Control Method Flow

[0105] Reference Figure 3 , Figure 3 A flowchart illustrating the overall process of the fully automatic control method of the present invention is shown. This flowchart, in the form of step boxes and decision boxes, illustrates the complete process from system initialization (S1) to process shutdown (S10). Steps S4 to S9 constitute a closed-loop control loop, representing the periodic iterative process of visual data acquisition, temperature control decision-making, motion control, and status monitoring. The fully automatic control method of this embodiment includes the following steps:

[0106] Step S1: System Initialization

[0107] After the system is powered on, the industrial computer automatically starts the fully automatic control software. The software first executes a self-test program:

[0108] a) Send a Ping command to check the network connectivity of the four cameras;

[0109] b) Read the communication status of the power regulator of each heating unit;

[0110] c) Read the position status and limit signals of the drive motors of each heating unit.

[0111] After the self-test passes, preset process parameters are loaded from the configuration file, including the target melting zone width, heating unit moving speed, total number of zone melting cycles, and single zone melting stroke, etc., and loaded according to preset values. Then, each heating unit is controlled to return to its initial position (the head end of the quartz tube), and the camera returns to its initial shooting position.

[0112] Step S2: Process Task Generation and Allocation

[0113] The collaborative scheduling module creates process task objects for each of the 12 workstations based on equipment configuration and process parameters. Each task object contains the following attributes:

[0114]

[0115] Step S3: Start the zone melting process

[0116] The industrial control computer sends heating commands to each heating unit via Modbus TCP, and the power gradually increases to the preset initial value. When the temperature of the heating unit reaches near the melting point of the material, movement control is activated: each heating unit moves along the quartz tube at a speed V. As the heating units move, a molten zone begins to form on the material ingot.

[0117] Step S4: Multi-station visual acquisition scheduling

[0118] Reference Figure 4 , Figure 4 This diagram illustrates the time-division multiplexing scheduling of multi-station visual acquisition according to the present invention. The diagram shows, in a two-dimensional format along the time and position axes, the process of a camera switching between three shooting positions P1, P2, and P3 along the guide rail, and the time sequence relationship of each station's heating unit being added to the acquisition queue when it moves into its corresponding field of view. Since the number of cameras (4) in this embodiment is less than the number of stations (12), a time-division multiplexing scheduling strategy is required.

[0119] Taking the left-side camera C1_L as an example, its movement along the guide rail is divided into 3 shooting positions:

[0120] a) Position P1: The field of view covers the left boundaries of H1 and H2 of quartz tubes Q1 and Q2;

[0121] b) Position P2: The field of view covers the left boundaries of H2 and H3 of quartz tubes Q1 and Q2;

[0122] c) Position P3: The field of view covers H3 and the end area of ​​quartz tubes Q1 and Q2.

[0123] The scheduling logic is as follows:

[0124] 1) The collaborative scheduling module polls the real-time location of each heating unit at a preset frequency (obtained through the motion control subsystem).

[0125] 2) For camera C1_L, check whether the heating unit within its coverage area is in the "to be photographed" state (i.e., the melting zone has been formed but the current melting zone has not been completed).

[0126] 3) If a heating unit enters the field of view of the current camera position, a data acquisition task is generated and added to the data acquisition queue.

[0127] 4) When the acquisition queue is not empty, the camera is triggered to perform image acquisition when the heating unit moves to the preset shooting reference position. The left and right cameras are triggered synchronously.

[0128] 5) When the heating unit moves out of the current camera's field of view, if the workstation still needs to be monitored, the scheduling module determines whether the camera needs to be moved to the next shooting position. If so, a guide rail movement command is sent.

[0129] The triggering time of visual data acquisition is synchronized with the control cycle of the intelligent collaborative temperature control subsystem. (Refer to...) Figure 5 , Figure 5 The closed-loop control timing diagram of this invention is shown. This diagram, in waveform form, illustrates the sequence and synchronization relationship of the global synchronization pulse, image acquisition execution, temperature control calculation execution, power distribution execution, and motion control execution within a control cycle on the time axis. The horizontal axis represents time, and the vertical axis represents the execution status of each module (high level indicates execution in progress). Specifically, the multi-station collaborative scheduling module generates a global synchronization pulse with a fixed period of 1 second. This pulse simultaneously triggers the image acquisition and temperature control subsystems of each camera in the current acquisition queue to read the latest melt zone width value measured in the previous cycle from the station status table and perform temperature control calculation. The acquisition task and the temperature control calculation task are executed in parallel thread mode on the industrial control computer. After completing the image measurement, the acquisition thread writes the result to shared memory, and the temperature control thread reads the latest data from shared memory at the beginning of each cycle. The calculation method for the shooting urgency U is as follows: ,in, The time interval between workstation i and the last successful image acquisition. The maximum allowable acquisition interval is represented by Ri; Ri is a coefficient reflecting the urgency of the process stage, with R in the initial stage being... i The value is higher than the steady-state value; Si This is a normalized state deviation index calculated based on the molten zone width deviation; the larger the deviation, the closer Si is to 1. These are weighting coefficients. During scheduling... Camera resources are sorted from largest to smallest and allocated preferentially.

[0130] Reference Figure 7 , Figure 7 The flowchart illustrating the multi-station collaborative scheduling process of this invention is shown. The flowchart, presented in the form of step boxes, illustrates the complete scheduling loop of the scheduling module, from polling the status of each station, generating acquisition tasks, arbitrating resource conflicts, to triggering image acquisition. The arbitration step prioritizes multiple stations based on the urgency (U) of the shooting. Through this scheduling strategy, four cameras can achieve time-sharing full coverage of 12 stations, ensuring that image acquisition at critical process stages is not missed.

[0131] Step S5: Real-time measurement of melt zone width

[0132] The vision measurement subsystem performs a bilateral vision-based measurement of the melt zone width for each acquired image and outputs the melt zone width value for the corresponding workstation. The measurement results are updated to the workstation status table via shared memory.

[0133] The deep learning-based coarse boundary localization employs the U-Net semantic segmentation network. The input is a 1024×768 pixel grayscale image simultaneously acquired by dual cameras, and the output is a pixel-level segmentation mask for the molten zone boundary region. This network uses ResNet-34 as the encoder backbone and is pre-trained on a dataset of 2000 zone fusion images labeled with molten zone boundaries. During training, a cross-entropy loss function and an Adam optimizer are used. The traditional image processing fine extraction performs Canny edge detection (low threshold 30, high threshold 80) and sub-pixel curve fitting (using quadratic curve fitting interpolation with 7 control points) within the coarse localization region (the region 50 pixels outside the boundary segmentation mask). After the bilateral boundary line fitting is completed, the calibration parameters of the left and right cameras and the fixed geometric dimensions of the heating unit (outer diameter of the heating coil) are considered. ), calculate the full width of the melt zone: ,in and These are the position coordinates of the melting zone boundary lines identified by the left and right cameras in the spatial coordinate system.

[0134] Step S6: Intelligent Collaborative Temperature Control Decision

[0135] The intelligent collaborative temperature control subsystem operates at a preset control frequency. In each control cycle, the latest melt zone width value for each workstation is read from the workstation status table, and the intelligent collaborative temperature control algorithm is executed.

[0136] 1) Calculate the thermal coupling compensation for each heating unit (considering power variations between adjacent workstations);

[0137] 2) Extract the historical melting / solidification rate characteristics of each workstation;

[0138] Target power is calculated using a feedforward + feedback composite control. Calculate the thermal coupling compensation amount as follows:

[0139] ,in: This is the thermal coupling compensation amount for the i-th heating unit. The current power of the j-th adjacent heating unit The coupling coefficient is pre-calibrated experimentally. Melting and solidification rate characteristics of each heating unit are extracted from historical operating data. Based on the current melt zone width, thermal coupling compensation, and the aforementioned characteristics, the trend of melt zone width variation is predicted. A feedforward and feedback composite control structure is used to calculate the target heating power. The control law of the composite control structure is: ,in Let be the target heating power of the i-th heating unit at time t. The output is the feedforward control term corresponding to the i-th heating unit. For the deviation of the molten zone width, Let be the target melting zone width at time t. The actual width of the molten zone at time t. , , For control parameters, This represents the cumulative amount of error (integral term), used to eliminate steady-state error; The rate of change of error (differential term) is used to predict the trend of error change and suppress oscillations, and τ is the integral variable. The PID parameters are tuned using the Ziegler-Nichols critical proportional gain method during the equipment commissioning phase, and the feedback control period is set to 1 second.

[0140] In this invention, "coupling" refers to the mutual thermal influence between adjacent heating units through thermal radiation, thermal conduction, etc. When the power of a heating unit changes, its heat will be conducted to adjacent workstations, causing an unexpected change in the melting zone width of the adjacent workstations. The "thermal coupling compensation amount" is the correction value applied to the target power of the heating unit to offset the above-mentioned mutual thermal influence.

[0141] The thermal coupling compensation is calculated as follows: For the i-th heating unit, consider the coupling effect of its adjacent j-th heating unit (j=i-1 or i+1) on it, and the coupling compensation is... = ,in The current power of the j-th heating unit. These are elements in the coupling coefficient matrix that have been experimentally calibrated beforehand.

[0142] The calculation results are sent to the power regulator of the corresponding heating unit via ModbusTCP for execution.

[0143] Step S7: Heating Unit Movement Control

[0144] The heating unit movement control subsystem drives each heating unit to move at a constant speed V. When a heating unit completes a single stroke... (When moving from one end of the quartz tube to the other):

[0145] a) Increase the number of zone melting cycles at this workstation by 1;

[0146] b) Determine if the number of zone melting cycles has been reached. If the target is not reached, switch the movement direction (reverse) and continue moving; if the target is reached, mark the station as "completed", the heating unit stops moving, and enters the cooling program.

[0147] The typical value of a single closed-loop cycle is ≤3s, of which image acquisition and transmission takes about 200ms, melt zone width measurement and calculation takes about 100ms, temperature control decision calculation takes about 100ms, power command issuance takes about 5ms, heating unit movement control response takes about 500ms, and the remainder is scheduling overhead and communication margin. The system generates a global synchronization pulse with a fixed period of 1s, waiting for the next round of synchronization trigger acquisition.

[0148] Step S8: Process Status Monitoring and Anomaly Handling

[0149] Reference Figure 8 , Figure 8 A flowchart illustrating the anomaly classification handling process of this invention is provided. This flowchart uses decision boxes and action boxes to show the complete processing flow from status monitoring and anomaly level determination (Level 1 / Level 2 / Level 3) to response measures for each level. Level 2 anomalies include an automatic recovery loop, escalating to Level 3 after consecutive failures. Throughout the system's operation, the status monitoring module continuously checks the following indicators:

[0150] 1) Melting zone width: Check whether it exceeds the preset safety range. According to the degree of deviation, it is divided into three levels: slight deviation, continuous deviation, and severe deviation.

[0151] 2) Camera communication: Check communication delay and connection status, divided into two levels: packet loss warning and disconnection alarm;

[0152] 3) Heating unit temperature: Check if it exceeds the safe operating limit;

[0153] 4) Movement position: Check whether it is within the allowable travel range, divided into two levels: position deviation warning and limit trigger alarm;

[0154] 5) Anomaly classification and handling strategy:

[0155] a) Level 1 anomaly (minor): Only logs are recorded; the process continues to operate.

[0156] b) Level 2 Anomaly (Medium): Triggers an automatic adjustment strategy to attempt recovery, while alerting the operator; if automatic recovery fails consecutively to the preset number of times, it will be upgraded to Level 3;

[0157] c) Level 3 Abnormality (Severe): Immediately stop the operation of the relevant workstation or the entire machine, issue an audible and visual alarm, and wait for manual intervention.

[0158] The automatic recovery strategy for Level 2 anomalies includes: if the camera communication packet loss is the cause, triggering the ModbusTCP reconnection mechanism with a 5-second retry interval, retrying 3 times consecutively; if the melting zone width continues to deviate, adjusting the current heating power in steps down or up within a safe range (±5% of the current power), with a step size of 1%, adjusting every 60 seconds until the deviation returns to normal, while simultaneously checking image quality and adjusting exposure parameters; if the heating unit temperature exceeds the limit, reducing the heating power. The above strategies can be executed a maximum of 3 times; after 3 consecutive failures, the anomaly is upgraded to Level 3.

[0159] Step S9: Process Completion Judgment

[0160] The collaborative scheduling module iterates through the workstation status table in each control cycle and counts the number of workstations in the "completed" state. When all 12 workstations are marked as "completed", proceed to step S10; otherwise, continue looping through steps S4 to S9.

[0161] Reference Figure 6 , Figure 6 A schematic diagram of the process state machine for each workstation of the present invention is shown. The diagram illustrates the transition conditions (such as the establishment of a stable melting zone, completion of a single melting operation, and completion of a total number of melting operations) between the initial stage, intermediate steady-state stage, and final stage, using state boxes and transition arrows. The process state machine transition logic for each workstation is as follows:

[0162] a) Initial stage: From startup to the establishment of a stable melting zone

[0163] b) Intermediate steady-state stage: Normal zone melting operation

[0164] c) Final stage: The final zone melting is completed, and cooling begins.

[0165] Step S10: Process completion and shutdown

[0166] After all workstations are completed, the system enters the process completion process:

[0167] 1) Each heating unit gradually reduces its power according to the preset cooling curve;

[0168] 2) After the quartz tube has cooled to a safe temperature, stop all camera data acquisition;

[0169] 3) Generate batch reports, including: melt zone width curves for each station, abnormal event records, process duration, etc.

[0170] 4) The system enters standby mode, waiting for the operator to take out the purified material and load the next batch.

[0171] IV. System Validation Results

[0172] The fully automated control system described in this embodiment has been practically verified on a 4-tube, 12-station industrial-grade horizontal zone melting purification equipment, and has been running continuously for multiple batches.

[0173] 1) Automation coverage: Except for loading / unloading which requires manual operation, the entire process is automated.

[0174] 2) Melt zone width control accuracy: The fluctuation range of the melt zone width for all batches at each station meets the process requirements.

[0175] 3) Batch consistency: The purity of the purified material remains highly consistent across different batches.

[0176] 4) Effectiveness of anomaly handling: In simulated anomaly tests (artificially induced camera network outages, overheating, and other faults), the system responded correctly according to the preset strategy, and no safety incidents occurred.

[0177] V. Other Embodiments

[0178] The above embodiments use a 4-tube, 12-station equipment as an example, but the system architecture of the present invention has good scalability:

[0179] 1) For 2-tube equipment (each quartz tube has 3 heating units arranged along its axis, for a total of 6 workstations): only 2 cameras are needed (1 on the left and 1 on the right), and the scheduling logic is simplified accordingly;

[0180] 2) For 6-tube equipment (each quartz tube has 3 heating units arranged along its axis, for a total of 18 workstations): 6 cameras can be configured (3 on each side), and the workstation status table of the collaborative scheduling module is expanded to 18 items;

[0181] 3) If the number of heating units per tube in the equipment increases (e.g., 4), only the guide rail travel and the number of shooting positions need to be adjusted, while the core scheduling algorithm remains unchanged.

[0182] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An industrial-grade multi-station horizontal zone melting and purification fully automatic control system, applied to a material horizontal zone melting and purification equipment having several quartz tubes and several heating units, wherein each quartz tube has at least one heating unit arranged along its axial direction, and each heating unit and the material section heated within the corresponding quartz tube constitute an independent station, and each heating unit can move independently along the axial direction of the corresponding quartz tube; characterized in that, The system includes: A multi-station vision measurement subsystem is used to collect the melting zone width data of each station in real time. The intelligent collaborative temperature control subsystem is communicatively connected to the visual measurement subsystem and is used to calculate the target heating power of each heating unit based on the deviation between the width of the molten zone and the target width. The heating unit movement control subsystem is used to control the movement of each heating unit along the axial direction of the quartz tube; The multi-station collaborative scheduling module is connected to the vision measurement subsystem, the intelligent collaborative temperature control subsystem, and the heating unit movement control subsystem, respectively, to coordinate the action sequence of each subsystem and to execute a graded processing strategy for abnormal events detected in the process. The central controller, as the hardware carrier of the system, communicates with each subsystem and is used to deploy and manage the operation of each subsystem.

2. The system according to claim 1, characterized in that, The multi-station vision measurement subsystem includes: an industrial camera array symmetrically arranged on both sides of a quartz tube and a guide rail moving mechanism that drives at least some of the cameras in the industrial camera array to move along the axial direction of the quartz tube; and an image processing unit for performing molten zone width measurement based on bilateral vision: performing boundary recognition on the molten zone boundary images synchronously acquired by the bilateral cameras, and fitting the complete width of the molten zone by combining the position of the bilateral boundary lines with the fixed geometric parameters of the heating unit; the image processing unit uses a deep learning model for coarse boundary localization, and uses traditional edge detection and curve fitting for sub-pixel level fine extraction within the coarse localization area.

3. The system according to claim 1, characterized in that, The intelligent collaborative temperature control subsystem executes an intelligent collaborative temperature control method for multi-heat source coupling, including calculating the target power: calculating the thermal coupling compensation amount based on the relative positional relationship between each heating unit and the current power; extracting the melting rate characteristics and solidification rate characteristics of each heating unit from historical operating data; predicting the trend of melting zone width change based on the current melting zone width, thermal coupling compensation amount and the aforementioned characteristics; and calculating the target heating power using a feedforward plus feedback composite control structure.

4. The system according to claim 1, characterized in that, The multi-station collaborative scheduling module includes: Workstation status management unit: used to track the current process stage and zone melting progress of each workstation; Visual acquisition and scheduling unit: used to dynamically allocate and arbitrate shooting tasks based on the real-time location of each heating unit and camera resources; Timing coordination unit: used to synchronize the execution rhythm of measurement, temperature control, and movement, ensuring the periodicity and synchronization of closed-loop control.

5. The system according to claim 1, characterized in that, The central controller includes a processor, memory, and communication interface, as well as a human-machine interface; the central controller deploys the above-mentioned subsystem modules and provides process parameter setting, real-time status monitoring, historical data recording, and human-machine interaction functions.

6. The system according to claim 2, characterized in that, The horizontal zone melting purification equipment has 4 quartz tubes, each with 3 heating units arranged along its axial direction, for a total of 12 workstations; the industrial camera array has 4 cameras, which achieve full coverage of the 12 workstations through time-sharing multiplexing scheduling.

7. An industrial-grade, multi-station, horizontal zone melting and purification fully automated control method, applied to the system described in any one of claims 1 to 6, characterized in that, Includes the following steps: Step S1: Start the zone melting process, the heating unit heats up and begins to move along the quartz tube; Step S2: The multi-station collaborative scheduling module coordinates the timing of visual measurement, temperature control, and motion execution; Step S3: The vision measurement subsystem performs melt zone width measurement and outputs the real-time melt zone width value for each station; Step S4: The intelligent collaborative temperature control subsystem calculates the target heating power of each heating unit based on the melting zone width value and issues the command for execution; Step S5: The heating unit movement control subsystem controls each heating unit to move at a preset speed; Step S6: Determine whether all workstations have completed the preset zone melting task. If not, return to step S2 and repeat. If completed, execute the process end process.

8. The method according to claim 7, characterized in that, Step S2 involves coordinating the timing, including time-division multiplexing acquisition scheduling: dividing the field of view coverage of each camera along the quartz tube axis into several shooting windows; when a heating unit enters a shooting window, adding that station to the current acquisition queue; when multiple stations simultaneously request the same camera resource, arbitrating and sorting them according to station priority and shooting urgency; and adjusting the camera position through the guide rail moving mechanism to expand the effective coverage of a single camera.

9. The method according to claim 7, characterized in that, During the cyclic execution, the process status is also monitored in real time and a hierarchical processing strategy is executed, which includes: Level 1 anomaly: Only logs are recorded; the process continues to operate. Level 2 anomaly: Triggers an automatic adjustment strategy to attempt recovery, while alerting the operator; if the automatic recovery fails consecutively to the preset number, it is upgraded to a Level 3 anomaly. Level 3 anomaly: Immediately stop the operation of the relevant workstation or the entire machine and issue an audible and visual alarm.

10. The method according to claim 7, characterized in that, The preset speed and preset zone melting task are determined based on preset process parameters, which include: the target melting zone width corresponding to each quartz tube, the moving speed of the heating unit, the total number of zone melting operations, and the single zone melting stroke.

Images