Air pollution condition is affected by meteorological elements and emissions from air pollution sources. In order to evaluate the effect of air pollution control measures, we need to separate the contributions from air pollution sources. In this study, the KZ filter was used to decompose the time series of three pollutants including O₃, PM_2.5 and PM₁₀ as well as time series of several meteorological factors into long-term, short-term and seasonal components. Data of air pollutants were collected from six air quality monitoring sites in Tianjin during 2015~2017. The contribution of each component to the total variance of the original air quality data was calculated. Stepwise regression was used to establish the relationship between air pollutants (O₃, PM_2.5 and PM₁₀) and meteorological variables for each time scale. Our results showed that seasonal component contributed most to the total variance, followed by short-term component. Temperature and relative humidity were the major factors affecting seasonal and short-term changes of O₃. Temperature was positively correlated with short-term component. Relative humidity was negatively correlated with the seasonal component of O₃; Wind speed, air pressure and precipitation were negatively correlated with particle concentrations at short-term and seasonal time scales. Relative humidity was positively correlated with them. Temperature was positively correlated with short-term component of particles, and was negatively correlated with seasonal component. Long-term concentration of PM₁₀ showed a downward trend after removing the effects of meteorological factors. The concentration of PM_2.5 increased in early 2017, and for the rest of the study time periods, its concentration showed a downward trend. The long-term concentration of O₃ was increased during the studied years. We can conclude that the effect of particulate pollution control measures was significant. However, the pollution of O₃ was aggravated.