To effectively forecast the dynamic disasters caused by coal and rock failure,the mechanism of charges gen- eration and the monitoring principle of charge induced by loading on coal with deformation and failure are introduced. The monitoring experiment of charge in the course of uniaxial compression coal is carried out. The initial time of effec- tive charge generation is studied by the method based on hypothesis principle. The commonality characteristics of time series relationship with stress and induction charge in the process of evolution of coal failure are obtained. The time- frequency characteristics of charge signals in the early stage of loading and the late stage of failure are studied,and a low-pass filter based on the FIR method is designed to obtain the effective charge signals. Based on the time-frequency characteristics of induction charge in the process of coal failure evolution,the evaluation indexes of charge signals for instability failure are constructed. An intrinsically safe charge monitor is developed. The test of charge monitoring dy- namic appearance in the underground is developed. Also the effectiveness of monitoring results is verified. In conclu- sion,the results show that the method with the lilliefors test can identify the effective charge signals produced in the in- itial stage of coal loading. The charge signals of compaction stage are basically white noise. In the elastic stage,a small amount of low-amplitude effective charge signals which are difficult to identify and mixed with white noise will be pro- duced,and in the elastic-plastic transition stage,a charge signal with easy identification,relatively increased amplitude and poor continuity will be produced. In the plastic stage and failure stage,the effective charge signals increase with the increase of stress drop times and the increase of stress drop amplitude. The fluctuation amplitude of charge signals in the two stages is more intense and the continuity of high-amplitude charge signals is stronger. Through time-frequen- cy transformation,the main frequency of white noise of charge signals is not less than 50 Hz,the main frequency of charge signals produced by coal fracture is not more than 15 Hz,and the FIR low-pass filter designed can effectively reduce the noise signals and retain the high amplitude and fluctuation signals. Three evaluation indexes of charge accu- mulation,average charge intensity and charge variation coefficient are established,which can reflect the fracture state of coal body. Mean charge intensity can also reflect the stress level of coal body. The variation coefficient can more ac- curately identify the precursor points of coal body entering the accelerated failure stage. Under the influence of mining, the coefficient of variation of charge and the average charge intensity of each measuring point in the field are obviously larger than those without mining influence,and the coefficient of variation of charge decreases gradually with the dis- tance from the working face,and the average charge intensity reaches the maximum value in the range of the peak in- fluence of the abutment pressure at the advanced working face. It shows that the charge monitoring method can obtain the situation of dynamic failure of roadways. It can provide a new monitoring mean for the early warning of coal mine dynamic disasters. Key words:inducti