Research And Development Of Water Quality Information Monitoring System For Aquaculture Based On Unmanned Aerial Vehicle

To maintain a good aquaculture environment and ensure the quality of aquatic products,regular monitoring of the water quality of aquaculture is necessary.Traditional monitoring methods include empirical methods,underwater sensor monitoring,and sampling for inspection,but all have certain drawbacks.Firstly,empirical methods rely heavily on observers for visual observation and odor detection,which can only determine limited water quality parameters,and due to human factors,the accuracy cannot be effectively guaranteed.Underwater monitoring sensors are easily corroded if left underwater for a long time,affecting monitoring accuracy.If the sensors are replaced regularly,it increases labor and economic costs.Sampling for inspection is too expensive and has low timeliness.In recent years,with the development of unmanned remote sensing technology and remote sensing image processing technology,it is possible to solve the above problems and achieve low-cost,high-timeliness,and high-precision monitoring of water quality.This paper studies a ground-to-air coupled water quality information inversion method: using a drone equipped with a multi-spectral camera to capture multi-spectral images of aquaculture waters,after completing the pre-processing of multi-spectral images,extract the remote sensing image pixel brightness values(DN values)of the multi-spectral images.The obtained water quality sample parameters are analyzed for correlation with the extracted remote sensing image pixel brightness values,and sensitive bands or band combinations are identified.Genetic algorithms and genetic-ant colony hybrid algorithms are used to establish inversion models for three parameters: chemical oxygen demand(COD),water temperature(T),and acidity and alkalinity(p H).The obtained models are verified,analyzed,and compared to determine the inversion model based on the genetic-ant colony hybrid algorithm.The verification results show that the inversion model method can accurately and quickly invert water quality parameters in aquaculture waters and has a certain degree of reliability and feasibility.The main work of this paper is as follows:(1)Design an independent mobile application based on DJI drone and obtain multispectral images of aquaculture water surface.Developed a drone flight control app based on DJI Mobile SDK that supports DJI M300 RTK model.The app can plan drone flight routes,control drone flight,and capture images.Long Guangyu Chen MS600 Pro multispectral camera was mounted on the drone gimbal at a runway-type fish farming area in Huzhou,Zhejiang Province.Multispectral images were captured using the self-developed app,and 28 sets of water quality samples and water surface multispectral images were collected.The chemical oxygen demand,water temperature,and p H data of the collected water quality samples were detected and recorded using sensors for data analysis and modeling.(2)After completing the multispectral image processing work and conducting correlation analysis,the geometric correction,radiation correction,and vignetting correction were performed on the multispectral images.The remote sensing image element brightness value,that is,DN value,of each band image was extracted,and the DN value of each band or different band combinations was calculated for correlation with water quality parameters.The best band or band combination for each water quality parameter was determined based on the correlation analysis..(3)To improve the efficiency of water quality monitoring and reduce monitoring costs,genetic algorithms and ant colony-genetic hybrid algorithms were used to construct inversion models for water quality parameters,and accuracy tests were conducted.The conclusion was that the genetic-ant colony hybrid algorithm was better for modeling.