Design of control system for pressure vessel weld inspection robot

The weld is the weakest part of the pressure vessel structure. Long-term operation under low or high temperature, high pressure, and corrosive conditions will easily cause weld fatigue cracking and defects such as pores, which will hinder the normal operation of the equipment. Therefore, it is necessary to regularly inspect and maintain the pressure vessel. At present, the weld inspection of pressure vessels is mostly carried out by manual magnetic particle inspection equipment [1]. The labor intensity is high and the efficiency is low. There is an urgent need for inspection robots to replace manual work [2-4]. This paper studies and designs the control system of the pressure vessel weld inspection robot and conducts relevant experimental tests.

1 Overall Design

The control system of the inspection robot is designed as a two-level master-slave structure. The control system structure is shown in Figure 1. The master system is the remote control end, which mainly completes the display of the original weld image collected by the robot and the inspection image, the identification and extraction of the weld, the deviation information processing of the weld tracking, and the issuance of motion control instructions. The slave system is the main control end. The inspection robot carries a camera to collect image information and sends it to the host computer through wireless transmission. The host computer uses the local area network formed by the wireless bridge and the network port to serial port module to transmit the robot motion control signal and other peripheral control signals to the main controller of the main end. The main controller receives the command of the host computer and controls the corresponding port to send data to control the robot wheel motor and the stepper motor used to lift the magnetic powder inspection equipment and other external devices. After processing the command of the host computer, the main controller needs to send feedback information to the host computer to feedback the current running status of the inspection robot.

2 Hardware Design

A stable and reliable hardware system is the guarantee for the detection and crawling robot to work. The hardware system design adopts a modular approach and unifies the design of similar functions, which can not only reduce the coupling between the units within the hardware system, but also reduce the difficulty of hardware system design [5]. The overall design of the hardware system includes controller, communication, motion control, peripheral device control and power supply module.

2.1 Controller Module

The remote controller uses the Guoyun GWBYT-VER:A2 industrial control computer, which is equipped with an Intel Celeron Bay Trail 1900 processor, 4G DDR3L memory, and has interfaces such as Mipcie, USB, and HDMI. According to the actual needs of the weld inspection robot, this industrial computer meets the requirements. The robot body uses the STM32F103C8T6 microcontroller as the main controller [9], which has a 32-bit Cortex-M3 core, 64 K bytes of flash memory, 3 communication serial ports, 4 timers, and 1 controller area network (CAN) interface.

2.2 Communication Module

The main controller of the robot body is connected to the switch through a serial port to network port module, the switch carried by the robot body is connected to the wireless gateway through a network cable, the wireless gateway and the remote control end communicate wirelessly through a wireless bridge, and the four devices including the weld image acquisition camera, magnetic particle image acquisition camera, serial port to network port module and wireless gateway are connected to the switch network port, and the wireless transmission of image data and control information is realized by using the WIFI local area network.

2.3 Motion Control Module

The moving parts of the wall-climbing robot are the movement of the moving wheels of the robot body and the lifting and lowering movement of the magnetic particle detection equipment. The magnetic particle detection equipment will only be lifted and lowered when the robot crosses the weld obstacle to prevent it from scratching the weld during the robot's movement. The lifting and lowering of the magnetic particle detection equipment is completed by two 42 stepper motors driving the lead screw to rotate. Therefore, this paper uses a micro-integrated driver PMD006P6 to control the stepper motor, which can be directly installed on the back cover of the 42 stepper motor. The output control current is adjustable in the range of 1A to 6A. For the robot's moving wheel, this paper uses a high-performance magnetic field oriented control driver, a high-power density brushless motor and a lightweight reducer. The motor model RMD-X6 S2 provides power output for the robot's movement and provides a CAN communication interface to the outside. Therefore, the CAN transceiver TJA1050 is used as a connection bridge between the microcontroller and the CAN bus.

2.4 Power Module

Since the robot body is equipped with many 12 V electrical devices, the current output capacity of the 12 V circuit should be as large as possible to meet the load requirements. Therefore, for the 12 V power supply, this article uses the BUCK topology circuit to implement it. Because the LM5116 operating frequency is programmable from 50 kHz to 1 MHz, it has the characteristics of emulating current ramp mode control, providing inherent line feedforward, loop compensation and undervoltage lockout, and is suitable for a wide voltage input buck system, so the LM5116 synchronous buck controller is selected to control the on and off of the switch tube. Based on the product manual of the LM5116 buck controller, the input voltage is 22 to 48 V, the output is 12 V, and the switching frequency is 525 kHz.

3 Control system software design

The software of the robot control system includes two parts: the main controller program of the wall-climbing robot and the remote control program. The two parts transmit data through the TCP protocol. The software system structure of the wall-climbing robot is shown in Figure 2. Since the main controller of the robot body STM32F103C8T6 is a single-core processor, it is responsible for few tasks, mainly control command data parsing, motor motion control, attitude sensor data acquisition and power control of external devices. The main controller of the remote end is a multi-core processor, and the software system is implemented using multi-threading, which is mainly responsible for functions such as visual interface display, working mode switching, real-time image acquisition, weld extraction, data distribution and processing, and TCP communication services.

3.1 Ontology software system design

In order to make full use of processor resources to realize the robot body motor movement, attitude sensor data acquisition, and external device power control tasks, the robot body main program is designed in an interrupt response manner. Four interrupts are set in the system: data analysis serial port interrupt, attitude sensor data acquisition interrupt, and 2 timer interrupts. After the system is started, the system is initialized first, including the configuration of the system clock, setting interrupt priority grouping, data analysis and attitude sensor data acquisition serial port baud rate, CAN bus working mode and baud rate, and initialization timer interrupt. After that, the program enters the main loop to perform task processing. If the robot body main controller receives the control command sent by the remote end, serial port 1 receives the interrupt response, parses the data and saves it. If the control command is not received or the command is in the wrong format, the time judgment of this situation will be made in the timer interrupt 3. If this is the case within 1 s, it is considered that the remote end does not need the robot to move or the communication data is abnormal. The stop motion flag is set and the shutdown command is sent to the motor in the main loop. If the main controller of the main body parses the motion task from the received instructions, it controls the speed and direction of the motors on both sides of the robot to perform weld tracking movement; if the analysis is to read the posture information, the yaw, pitch and roll angle information of the current robot is saved through the receiving interrupt of serial port 3; if the analysis is the lifting task of the magnetic particle detection equipment, the channel 1 of timer 1 is controlled to output a set of complementary PWM pulses, which are sent to the stepper motor driver to complete the lifting and lowering of the magnetic particle detection equipment; if the analysis is the external device control task, such as turning on the camera auxiliary lighting, the control port level of the robot main controller changes and acts on the lighting power control circuit.

3.2 Remote software system design

The remote control system software is designed by QT Creator, which has excellent cross-platform features and rich application programming interfaces. It uses multi-threaded processing and signal and slot mechanisms to implement the main thread of interface display, image acquisition, TCP communication service and other sub-thread programming in the remote software system.

1) The interface displays the main thread

When the main thread program of the interface is started, the object of the main event loop QApplication class and the main window MainWindow class will be created first; then the image acquisition, weld processing, data management and TCP service sub-threads will be initialized in the main window class and the connection between the signal in the main window class and the slot function in the sub-thread will be established through the connect function; then the buttons and TCP services in the interface will be initialized; finally, the main window class will be displayed. When the collected image, working status, TCP communication data and attitude sensor data are the latest status, these data will be updated in real time to the corresponding area in the window for display. If the image acquisition thread has no latest image data stream update, the no signal icon will be displayed in the main interface window. If other display information is not updated, the current status information of the sub-thread will be printed out.