Virtual water quality monitoring method based on improved depth limit learning machine

Abstract

The invention discloses a virtual water quality monitoring method based on an improved depth limit learning machine. Aiming at solving the problems of high construction cost, complicated equipment maintenance and the like of monitoring stations in water quality monitoring, through historical data of the stations related to virtual monitoring positions, the improved depth limit learning machine isproposed for realizing water quality prediction of virtual positions. According to the virtual water quality monitoring method, in order to extract the robustness characteristic with the more invariance property in water quality data collected by a water quality monitoring network, a new deep limit learning machine model is developed, the new deep limit learning machine model introduces local denoising criteria and shrinkage penalty terms into a self-encoder based on the limit learning machine; then on the basis, water quality parameter actual values of the virtual positions are predicted through a weighed extreme value learning machine, and thus water quality monitoring of unknown positions is achieved; and real-time water quality information of the unknown stations can be predicted better in real time, and the good prediction accuracy is achieved.

Description

Technical field [0001] The invention relates to a virtual water quality monitoring method, in particular to a virtual water quality monitoring method based on an improved deep extreme learning machine. Background technique [0002] Water quality monitoring should provide as much information as possible on the current state of the water body on a spatial scale, and highlight areas that may require new management measures, or determine whether current management practices are adequate. Therefore, the greater the number of monitoring points in the entire water body, the higher the probability that they will accurately represent their current state. However, according to market research, in China, the construction cost of a small automatic monitoring station with five water quality parameters (permanganate index, ammonia nitrogen, total phosphorus, water temperature, pH) is as high as 4 million yuan, not even including the later ones. Equipment maintenance and human resources. Ther...

AI Technical Summary

Problems solved by technology

However, the current neural network models used to predict the spatial distribution of water quality are often shallow models, which are easy to fall into local optimum, and have insufficient ability to propose high-level features of water quality spatial data, which is obviously insufficient compared with deep models.

Images