Abstract: Aiming at the problem that the dust mass measurement of the existing dust monitoring devices in the process of farmland dust monitoring was susceptible to the influence of ambient temperature and humidity, resulting in zero drift and a large measurement error, this study designed a dust monitoring device based on temperature and humidity compensation. The device included temperature and humidity sensors, mass sensor, data acquisition system, etc., which could realize cross-regional, remote, real-time online measurement of dust mass. The zero drift and temperature and humidity data of the four dust monitoring devices collected by the temperature and humidity correlation test showed that the temperature and humidity had a significant correlation with the zero drift of the dust monitoring device (the mean values of |p| were 0.59 and 0.62, respectively),and the correlation between zero drift of the dust monitoring device and absolute humidity was more stable than relative humidity performance, a linear expression of absolute humidity and zero drift of the dust monitoring device was provided, the humidity coefficient of the expression was positively correlated with temperature. When the temperature changed from 40 ℃ to 60 ℃, the single-factor test of the temperature and the output signal of dust monitoring device showed that the zero drift was 65.7 times that of the sensitivity change, zero drift was a key factor affecting the measurement accuracy of dust monitoring device. A compensation model of the zero-point drift dust monitoring device of BP network based on temperature and humidity was established, the experimental results showed that the data fluctuation range of the proposed model was 8.8% lower than that of the basing on temperature-humidity second-order polynomial compensation model, and 60.3% lower than that of the basing on temperature compensation model of BP network. A field measurement experiment was conducted for the dust monitoring device based on temperature and humidity compensation of this paper. The results showed the device in this paper reduces the final measurement error by 57.5% and the fluctuation of measurement results by 34.2% compared to the uncompensated device, effectively reduced the measurement error of dust mass. The research results could provide technical support for the research on online real-time monitoring of farmland dust.