Abstract: In order to improve the timeliness of environmental monitoring in shallow sea and avoid the irrationality of subjective evaluation of water quality, the narrowband internet of things (NB-IoT) multi-node method is introduced to transmit back the environmental parameters in shallow sea, and the data fusion technique is used at the remote terminal to integrate them for scientific evaluation of water quality grade. In this paper, a two-stage parallel fusion method is adopted for environmental parameter fusion. Before fusion, the correlation function in fuzzy theory is used to eliminate the abnormal data caused by environmental noise and sensor itself; then the adaptive weighted fusion algorithm is used for the first-level fusion. Finally, the fuzzy comprehensive evaluation method is used to merge the first-level fusion results into the decision-making level to realize the water quality grade evaluation of shallow sea environment. It is verified by experiments that the above method can obtain water quality environmental parameters more timely and effectively, and the relative error of the first level fusion is less than 6.5%, which can directly show the water quality grade of the monitoring area in the form of probability, and improve the accuracy and reliability of water quality grade evaluation.