中文摘要：

为了探究湛江市空气NO2浓度的变化特征及时频特性,以达到有效防治空气污染,利用数理统计方法和时频分析技术的短时傅立叶变换、小波分解分析、平滑伪Wigner-Ville分布分析、重排谱图及Hough变换等方法进行了分析和研究。结果表明,湛江市空气NO2浓度较全国大多数城市偏低,逐日差异明显,变化辐度波动大,资料离散度高。分布形态属正偏（右偏）分布,偏斜程度较大,分布陡峭,为尖峰分布。湛江市空气NO2浓度的变化逐年呈明显减少趋势,年内跨度也明显减少,波动变小,同时NO2逐日资料的年分布形态偏斜程度明显减少,尖峰分布逐年变得更为平缓。NO2季节变化比较明显,年周期性清楚。NO2年内变化具有冬春季节值大,夏秋季节值小,变化相对平缓的年趋势变化特性;冬春季节振辐大,夏秋季节振辐小,小波分解变化信号的频率从低层至高层细节系数逐层降低的年振荡波动特性。湛江市空气NO2浓度的重排谱图与平滑伪Wigner-Ville分布趋势相近,但没有交叉项,时频聚集性好,可解释性强。平滑伪Wigner-Ville分布的Hough变换可发现湛江市空气NO2浓度年内的辐值变化峰值。

英文摘要：

In order to control air pollution effectively, variations characteristics of NO2 in Zhanjiang were analyzed by means of Numerical Statistic method and time-frequency analysis,including short time discrete Fourier transform, wavelet composition, smoothing pseudorange Wigner-Ville distribution,recomposition spectrogram and Hough Transform method. The results showed that： the NO2 concentration in Zhanjiang was lower than that in most cities in China. It changed obviously day by day and wave motion of the Variations peak type was wide, which indicated a right asymmetric distribution shape with great decline. Distributional pattern classified forward bias （right avertence） distribution,decline level was comparatively large, distributional craggedness, this was leptokurtic distribution. The variance and annual spans of the NO2 appeared decreasing trends these years in Zhanjiang, thus the wave motion diminished. Simultaneously, the decline level of the distributional pattern of e NO2 every data was obviously reduced . The leptokurtic distribution was dropoff year by year. The seasonal variations of NO2 were comparatively obvious, the annual periodicity was distinct. Variations of annual NO2 was great in winter and spring, while small in the other two seasons. Variation Characteristics of annual trend was relatively dropoff ; The remarkable peak and vale of amplitude distribution mainly appeared in winter and spring, while small during summer and autumn, the frequency of variation signals showed the wave property of annual NO2 Oscillation with detail coefficient decreasing from low levels to high levels of Wavelet Decomposition. Distributed tendency of smoothing pseudorange Wigner-Ville distribution and recomposition spectrogram of NO2 was similar. But which haven＇t cross terms. Aggregation of Time-Frequency was good, Interpretability was powerful. By means of Hough transform of smoothing pseudorange Wigner-Ville distribution, it found the peak value of the peak Variations of annual NO2 in Zhanjiang.