Design of coal mine gas emission prediction system based on ARM

With the development of microelectronics and computer technology, embedded technology has gained a broad space for development. Especially since the 1990s, the development and popularization of embedded technology has become more eye-catching, and it has become the development direction of modern industrial control, communication and consumer products. In the coal mining industry, gas is one of the important factors that endanger the safe production of mines. Most of the current coal mine gas prediction systems send the physical quantities that affect gas outflow, such as concentration, humidity, wind speed, etc., to the central management system on the well for analysis and prediction. The prediction information is difficult to respond to the underground workers and systems in time, resulting in the inability to respond in the first time. The embedded gas outflow prediction system is easy to install in different mining areas underground, process and analyze the sensor data of the current mining area, and predict the gas outflow information of the mining area. The mining areas can not only communicate with each other, but also interact with the host computer in real time.
1 Overall design
The gas sensor converts the measured physical quantity gas outflow into an electrical signal, which is sampled by A/D conversion and converted into a digital signal, which is processed in the ARM processor. The S3C2440 core board based on the ARM920T core has a built-in STN/CSTN/TFT LCD controller and supports various LCDs with a resolution of less than 1 024×768, which are used to display prediction information, detection quantity, detection time, etc. The built-in 4-wire resistive touch screen controller is used for user interaction with the system, and system parameters can also be set through the keyboard. The 100 Mbit/s Ethernet controller is used for two-way information transmission with the host PC. The system software development is designed based on QT in the Fedora Linux environment. QT is an open source, cross-platform C++ graphical user interface application framework developed by Trolltech in Norway. It provides application developers with all the functions needed to build state-of-the-art graphical user interfaces. QT is completely object-oriented, easy to expand, and allows true component programming. The system divides each mining face of the mine into different gas outburst sources based on the source division method [1, 2], establishes a gas source division prediction model, and obtains and displays the prediction data after conversion. The system structure is shown in Figure 1.

2 Hardware Design
2.1 Chip and Memory Design
The system uses Samsung's S3C2440 embedded processor based on ARM 9 core. S3C2440 is widely used in PDA, mobile communication, router, industrial control and other fields. The chip integrates the following modules: 16 KB instruction cache, 16 KB data cache, MMU, external memory controller, LCD controller, NAND Flash controller, 4-channel PWM timer and 1 internal timer, 168-pin general GPIO, real-time clock, 8-channel 10-bit AD and touch screen interface, standard 20-pin JTAG debugging interface, etc. The memory uses standard 64 MB Nand-Flash for data storage and 64 MB SDRAM for program operation.
2.2 A/D sampling, display and interface design
The A/D conversion unit uses MAX1297AEEG to realize 12-bit parallel analog-to-digital conversion, which is directly connected to the I/O line of the core board, as shown in Figure 2. Since S3C2440 has its own LCD controller, the design of LCD controller is omitted. The display screen uses NEC's 3.5-inch piezoelectric touch LCD with a resolution of 240 × 320. The Ethernet interface uses the TC3097F-5 chip.

3 Software Design
3.1 BootLoader Porting
BootLoader is a small program that runs before the operating system kernel runs. Most BootLoaders are divided into two parts: Stage 1 and Stage 2. Stage 1 mainly contains the code for initializing the hardware architecture that depends on the CPU, and is usually implemented in assembly language. The tasks of this stage are: (1) Initializing basic hardware devices; (2) Preparing RAM space for the second stage; (3) Setting up the stack and jumping to the program entry point of the second stage. Stage 2 is usually completed in C language to implement more complex functions and make the program more readable and portable. The tasks of this stage are: (1) Initializing the hardware devices to be used in this stage and detecting the system memory mapping; (2) Reading the kernel image and the root file system image from Flash to RAM; (3) Setting the startup parameters for the kernel and calling the kernel. The system uses the open source BootLoader developed by MIZI Company of South Korea, namely vivi, and makes necessary cuts to vivi and ports it to the system. [page]

3.2 Linux kernel porting[5]
Since the system includes data analysis, communication between systems, and interaction with the host computer, it is considered to add an operating system to better manage and allocate resources. The system uses the latest Linux 2.6.14 kernel. The kernel porting is relatively complex, mainly including the modification of the Makefile file (such as setting the cross-compilation path, flash partition settings, etc.) and the configuration of kernel compilation items (make menuconfig): (1) Adding Yaffs2 file system support; (2) Porting of CS8900 network card driver; (3) LCD driver porting; (4) USB driver porting, etc.
3.3 Making the Yaffs2 file system
The file system is mainly made by using the busybox tool to make a minimum file system. After compiling and installing busybox-1.7.tar, a subdirectory _install will be generated in the busybox directory, and many Linux tools and commands will be integrated and compressed in the /bin directory. In addition, the dynamic shared libraries libQtCore.so.4, libQtGui.so.4, and libQtNetWork.so.4 that the QT program depends on must be added, and environment variables must be set.
3.4 Software Design Based on QT[6]
The application program of the system mainly includes the following modules:
(1) Core algorithm module. The mathematical model is established according to the source separation method, as shown in Figure 3.

The main gas emission sources include gas emission from the mining coal seam (including surrounding rock), gas emission from the adjacent coal seam, gas emission from the coal wall of the tunnel, gas emission from coal falling during tunneling, gas emission from the goaf of the mined area, and gas emission from the production area. Gas emission from
the mining face q1 = gas emission from the mining coal seam + gas emission
from the adjacent coal seam Gas emission from the tunneling face q2 = gas emission from the coal wall of the tunnel + gas emission from coal falling during tunneling

Where q4 is the gas emission in the goaf of the mined area.
(2) Communication module. The communication of the system includes communication with the host PC and communication with other subsystems. The communication interface adopts a 100 Mb/s Ethernet interface, and the communication protocol adopts a lightweight UDP protocol, which is suitable for network data transmission of short messages, a large number of clients, no special requirements for data security, and high requirements for response speed. QT provides a QUdpSocket class for writing UDP programs. An important function provided by the QUdpSocket class is broadcasting, which is just suitable for the system to send broadcast datagrams to the adjacent coal seam system in the form of broadcasting, so as to obtain the gas emission information of the adjacent coal seam.
(3) Information display GUI module. This module is used to interact with the operator and adopts a touch mode, which is more suitable for operation in a narrow space. QT's GUI class provides programmers with a wealth of operation controls, which can easily design a system with simple operation and user-friendly interface. The system display mainly includes the gas emission of the mining face, the gas emission of the excavation face, the current mining face wind speed, humidity, gas emission forecast information and detection time.
In view of the difference in gas emission in different depths and mining areas in the mine and the mutual influence of gas emission in adjacent mining areas, this paper designs a distributed gas prediction system based on ARM. In view of its special application environment, the hardware platform design of the system adopts S3C2440 with high reliability, wide application and mature technology as the core board; the software adopts the highly compatible Linux+QT design method to ensure the stability and reliability of the system. According to the historical gas emission data, in a mine with a coal seam thickness of 4.96m, a daily output of 3,000 t, a tunnel length of 1,000 m, a tunnel cross-section of 5 m2, an average gas content of 18.80 m3/t, and a height of 90 m from the ground, the system predicted value is 45.28 m3/min, and the actual value is 50.06 m3/min.