Abstract: For a long time, spontaneous coal combustion fires in the air mining area, coal pile, coal silo and other spaces occur from time to time, and it is more difficult to realize the rapid quantitative identification of the high temperature point of the spontaneous coal combustion disaster due to factors such as the complex environment of the working area and the bottleneck of the fire source detection technology. The acoustic temperature measurement technology has the characteristics of high measurement accuracy, wide temperature measurement range, large measurement space, etc., which can realize the accurate detection of the hidden fire source location in the air mining area and other hidden fire sources and is very promising for development. At present, this technology is still in the stage of basic and laboratory research in the field of temperature measurement of loose coals, and a lot of research still needs to be carried out. Based on the superior performance of pseudo-random sequences, they were introduced as acoustic source signals, and an experimental system for acoustic temperature measurement of loose coals was constructed according to the principle of acoustic temperature measurement and the principle of pseudo-random sequence acoustic source signal generation. The main part of the system includes an acoustic test system, a programmed warming system, a soundproofing system and a coal sample box, and the accuracy of the system was verified by combining experimental tests and simulation and modelling methods. Using a combination of theoretical analysis, Matlab simulation and experimental testing, we carried out a study on the distortion characteristics of the pseudo-random sequence acoustic source signal, determined the optimal processing method for this signal, and successfully applied it to the temperature measurement of loose coals. The results show that pseudo-random sequences can be used as sound source signals for the acoustic temperature measurement of loose coal bodies in the space of air-mining areas, coal piles, and coal silos, etc. However, it is necessary to apply the quadratic correlation PHAT–β algorithm to the frequency interval (1 000−3 000 Hz) of the pseudo-random sequences sound source signals to process them, so as to make the bandwidth narrower and to concentrate the energy. The pseudo-random sequence of sound source signals is used to measure the flying time of acoustic waves in a loose coal body, and it is found that the error of the acoustic flying time measurement results is less than 5% at different distances, and it is verified by comparison. When the pseudo-random sequence is used as the acoustic source signal for temperature measurement of loose coals, the average absolute error between the inverted temperature and the temperature of coal samples with different grain sizes is 2.051 ℃, with an average error rate of 5.293%, which is capable of inverting the coal temperature in a more accurate and reliable way.