Abstract: Abstract: Wireless sensor network has been widely used in various types of industries, such as water quality detection. Due to all kinds of device faults and transmission faults, there are some outliers during water quality monitoring. In a large monitoring area, the difference between monitoring parameter values exists due to uneven distribution. The monitoring data in a single location is unsuitable to represent the real situation of the whole monitoring area. The data fusion method is used to fuse data in multiple locations. Traditional methods in water quality data fusion have problems of low accuracy and efficiency for limited to poor generalization and complex calculations. In response to requirements of high-precision for sensor data in water quality detection, a novel data fusion method based on a new support function IDTW-SF (Improved Dynamic Time Warping Distance Optimized Support Function) was proposed in this study. Based on the importance of dissolved oxygen in various water quality parameters, it was used as an example to study this research in this study. The purpose of data fusion was for correcting outliers to obtain high-quality data. Firstly, the consistency detection of sensor data improved the quality of the fusion data. With high computing complexity, the traditional Gaussian support function was a defective method in data fusion. The dissolved oxygen content was used for example to study the new data fusion method. An improved dynamic time warping distance IDTW (Improved Dynamic Time Warping Distance) was used to optimize a new support function SF (Support Function), thus calculated the support degree value between water quality time series data. Unlike the Gaussian support function, the SF function obtained mutual support degree of sensors without the exponent calculation. The weighted algorithm was used to complete the data fusion process. Based on Grey correlation analysis, the IDTW-SF combined the dynamic time warping distance. time segment strategy and Mahalanobis distance together. DTW algorithm was applied to replace the Euclidean distance and compute the distance between time series. Time series segmentation strategy was utilized to reduce the computation dimension of the DTW algorithm. To prove the validity of this fusion algorithm, various experiments were carried out on a water quality monitoring platform of aquaculture pond from May 24th, 2017 to May 29th, 2017 in the Freshwater Fisheries Research Center, Jiangsu Province. Different distance measures were applied to optimized the SF function and construct new support functions, such as Cos-SF (Cosine angle optimized SF) and DTW-SF. Meanwhile, existed support functions were realized in this study, such as the G function (Gaussian function). These functions were all used as a comparison. The experimental results showed that the mean absolute error value of the proposed fusion method was 0.279 2%. Compared with DTW-SF, Cos-SF, and SF, the mean absolute error value was reduced by 6.308 7%, 54.214 5%, and 16.306 9%, respectively. And the fusion effect of dissolved oxygen was improved obviously. The run time of the IDTW-SF was just 0.021 7 s. The fusion accuracy of the IDTW-SF support degree function was higher than the counterpart existed functions, and its efficiency was also high. Meanwhile, the residual value of IDTW-SF also had a distinct advantage over the other contrast functions. It was obvious that the combination of the IDTW algorithm and the SF method was reasonable and effective. To summarize, the proposed fusion algorithm in this study fully obtained the correlation of monitoring data and overcame shortcomings of traditional function, thus improved the accuracy and efficiency of fusion results. High fusion accuracy could meet the needs of water quality monitoring in the real world.