The accident rate at Ukrainian coal mines is high. One of the reasons for such accident rate is a high methane concentration level in the underground air. This situation often provokes blasts with a large number of victims. For the control of methane concentration methane level, mine air automatic control systems. The control systems used now, have become out of date and have exhausted their potential at many coal mines. In order to improve safety of work in coal mines, it is necessary to replace the existing systems. One of the short-term low-coast solutions for the problem is an upgrade of existing systems.

The end result of this work should be an algorithm for digital processing of discrete signal levels, underground air control algorithm and a microprocessor structure which can be used as the basis for an implementation of the methane concentration control system.

Conversion of an analog system into a digital system requires many changes not only in the hardware part.

In order to prove the applicability of DSP in this system, it is necessary to conduct a number of researches:

• analysis of existing algorithms and development of the specialized algorithm;
• determination of an optimum dimension for the algorithm;
• study of the data representation format for calculation;
• study of the input data format;
• study of algorithm reaction to interferences in an electrical circuit;
• determination of a dynamic range;
• development of an algorithm for the analysis of underground air;
• a grounded selection of a DSP processor;
• development of a microprocessor structure.

Methane concentration control is carried out using the hardware/software complex KAGI. Signals from the methane concentration sensors (MCS, ДМ) arrive to the KAGI through a signaling device (SD, АС).

The SD performs signal transformation for the signals arriving from three MCS into a single signal which is transmitted to the surface information receiver unit.

The structure of the control system

If methane concentration does not exceed the permissible level, the SD generate a continuous sinusoidal waveform signal from each connected MCS. These signals are combined and transmitted across a two-wire line to the surface. The signal frequencies of the MCS are respectively equal to 14, 20 and 26 kHz.

Сombined signals and transmitted

When methane concentration detected by one of the sensors exceeds the permissible level then the current sensor pulses are transmitted with the frequency 3 – 5 Hz. If a connection with MCS is lost, then SD transmits video pulses too. The absence of the carrier frequency signal the loss of connection between SD and surface or the absence of the power supply voltage.

Video pulses

The SD is generated a telephone call signal with frequency 1 kHz too.

KAGI’s signal processing system is used for SD, which based on analog method of data processing.

Analog filters are complexity devices and they require tuning during producing process and exploitation.

Lacks of analog schemes:

• susceptibility by temperature influencing;
• presence of admittances;
• component deterioration.

These lacks reason price rising of fabrication cost and service of analog system.

The applicability of digital signal processing system allows solutions of this task. Research of this work is one of the solutions for signal processing of SD, which is based on Fast Fourier Transform (FFT).

Optimum dimension’s determination of algorithms for realization wit DSP reason to make a number of researches. They were made in MATLAB design system.

Accretion of FFT size changes resolving ability, higher calculable expenses, bigger size of memory, in addition probability of processor register’s overflow rises.

The resolving ability of FFT algorithm can be not so large, because sinusoidal frequencies of SD signal are wide-spaced for 6kHz. The research proved that an optimum dimension for algorithm is 128.

FFT-128

Applicability of DSP processors with floating point eliminates defects of overflow, but its cost is higher then fixed point processors. The research proved that using of 16-bit fixed point processors is enough for task.

Analysis of quantization effects were also researched in this work. Quantization of input signal by level doesn’t influence essentially of signal spectrum. Therefore, 8-bit ADC can be used in system.

Quant FFT

Research of algorithms showed is good reaction on the noise effects (represent on the animation).

Signal with noise and result of FFT

The experiments showed results:

• development of the real FFT algorithm;
• data representation format for calculation is [16 10];
• input data format is 8-bit;
• interferences in electrical circuit are satisfying;
• the FFT algorithm can be designed with 16-bit processors.

1. Рабинер Л., Гоулд Б. Теория и применение цифровой обработки сигналов/ Пер. с англ.; Под ред. Ю.А. Александрова.- М.: Мир, 1978.- 836с.
2. Сергиенко А. Б. Цифровая обработка сигналов: Учебник для вузов. 2-е изд. – СПб.: Питер, 2006. – 751 с.: ил.
3. В. Дьяконов MATLAB 6: Учебный курс. СПб.: Питер, 2001. – 592 с.: ил.
4. Научно-техническое руководство по цифровой обработке сигналов. 2-е изд. Стивен В. Смит. 1997-1999, Калифорнийское техническое издательство Сан-Диего, Калифорния. Перевод выполнен фирмой Автэкс Санкт-Петербург. Авторы перевода: Покровский В. Н., Силантьев В. И., 2001.
5. Калабеков Б. А. Микропроцессоры и их применение в системах передачи и обработки сигналов. Учеб. пособие для вузов, 1988, 368 с.