Modeling and identification for soft sensor systems based on the separation of multi-dynamic and static characteristics

Data-driven soft sensor is an effective solution to provide rapid and reliable estimations for key quality variables online. The secondary variables affect the primary variable in considerably different speed, and soft sensor systems exhibit multi-dynamic characteristics. Thus, the first contribution is improving the model in the previous study with multi-time-constant. The characteristics-separation-based model will be identified in substep way,and the stochastic Newton recursive(SNR) algorithm is adopted. Considering the dual-rate characteristics of soft sensor systems, the proposed model cannot be identified directly. Thus, two auxiliary models are first proposed to offer the intersample estimations at each update period, based on which the improved algorithm(DAM-SNR) is derived. These two auxiliary models function in switching mechanism which has been illustrated in detail. This algorithm serves for the identification of the proposed model together with the SNR algorithm, and the identification procedure is then presented. Finally, the laboratorial case confirms the effectiveness of the proposed soft sensor model and the algorithms.

Research on the calibration method of precipitation micro- physical characteristics sensor

In order to diminish the effect of the ambient light and CCD pixel non-uniformity to the Precipitation Micro-physical Characteristics Sensor,a modified calibration scheme was designed and calibration experiments in sunny,cloudy,night,different location of sample space were carried out. Firstly,the characteristics of particle images which affected by ambient light and different location of sample space were analyzed. Secondly,the relevance betw een particle image features and parameters of image processing were discussed. Finally,the parameter setting scheme were determined,the radium of median filtering algorithm is 3 pixels,the defocusing radius of point spread function( PSF) is 7 pixels,the radium of erosion is 3 pixels,and the binary threshold is obtained from the Area-thresh relationship. The results show that the new scheme could deal with the image calibration well,the average errors of equivolumetric diameter was 0. 041 mm with standard deviation of 0. 115 mm,and the average errors of the axis ratio was 0. 011 with standard deviation of 0. 085. The new scheme works well in the field observation too,the observed axis ratio is consistent with the empirical relationship that proposed by Beard. The relative error of accumulation precipitation is-3. 06% after calibration,w hich is improved 1. 94% low er than the initial one without calibration.

ANALYSIS OF SPECTRAL CHARACTERISTICS AMONG DIFFERENT SENSORS BY USE OF SIMULATED RS IMAGES

This research,by use of RS image_simulating method,simulated apparent reflectance images at sensor level and ground_reflectance images of SPOT_HRV,CBERS_CCD,Landsat_TM and NOAA14_AVHRR’s corresponding bands.These images were used to analyze sensor’s differences caused by spectral sensitivity and atmospheric impacts.The differences were analyzed on Normalized Difference Vegetation Index(NDVI).The results showed that the differences of sensors’ spectral characteristics cause changes of their NDVI and reflectance.When multiple sensors’ data are applied to digital analysis,the error should be taken into account.Atmospheric effect makes NDVI smaller,and atmospheric correction has the tendency of increasing NDVI values.The reflectance and their NDVIs of different sensors can be used to analyze the differences among sensor’s features.The spectral analysis method based on RS simulated images can provide a new way to design the spectral characteristics of new sensors.