Research On Portable Microscopic System For Field Detection Of Microalgae

With the development of society,a large amount of wastewater is discharged at rivers and lakes,which induces the eutrophication of water easily,and results in the explosive proliferation of microalgae leading to ecological disasters such as red tide and water bloom.Due to the high frequency of algal blooms,detecting the species and density of microalgae as well as developing a prediction model are critical in order to detect the explosive growth of microalgae,prevent economic losses,and avoid ecological calamities.At present,microalgae in natural waters are usually detected by taking water samples back to the laboratory or using portable instruments on site.Taking water samples back to the laboratory for testing takes a long time,and microalgae are prone to morphological changes after leaving the original growth environment,which will affect the test results.The in-site detection method is to use portable instruments to detect chlorophyll,carotenoids and other substances in microalgae for indirect detection.It is only able to detect the overall algae content in the water body,but cannot distinguish specific algae that are easy to erupt,and lacks selectivity.In this paper,a portable microscopic system for in-site detection of microalgae is developed.Through acquiring images of microalgae in water samples in-site,it can intelligently identify algae species and count algae density to provide data support for algal bloom prediction model.The research contents are as follows:(1)A smartphone-based portable microalgae microscopic device is studied.Using the principle of single-lens magnification and combining with a mobile phone camera,the highresolution microalgae image with a resolution of up to 2 μm is obtained.The device eliminates the standard eyepiece structure,replaces the optical amplification element with a spherical lens with a variable curvature,and combines the mobile phone camera to achieve magnification of 50-2000 times.(2)Research on microalgae recognition system based on deep learning algorithm is carried out.Four kinds of algae images were labeled manually,and a training data set of the microalgae recognition model was constructed.The yolov5 deep learning network is used to train the recognition model,and the optimal model obtained by training has a precision of 0.85,a recall of 0.9,and a mAP0.5 of 0.92.(3)Developing the microalgae detection mobile phone app,and realizing the operation of deep learning model on mobile phone.The modules of microscopic imaging,data storage,microalgae recognition and data analysis are integrated to improve the portability of the system.The microalgae recognition model is deployed on the mobile terminal.After obtaining the microalgae microscopic image on site,the microalgae species and quantity can be recognized by calling the deep learning model through the app,and the density of microalgae is calculated.(4)Carrying out microalgae detection experiments to test the feasibility and accuracy of the system.Taking Crotalaria as an example,the deep learning model for microalgae recognition is trained with an accuracy rate of 94%.The mathematical model of microalgae density calculation was constructed and the microalgae density gradient experiment was carried out.Under four different orders of microalgae density,the relative errors of the results measured by the classic microscope counting method and the results measured by this system are 3.03 percent,2.64 percent,4.08 percent,and 0.84 percent,respectively.The gradient close to the density of microalgae in natural water was selected for detection.The correlation coefficient R between the number of microalgae in the field of vision of the developed portable device and the actual density is 0.9895.In the 100 days water bloom simulation experiment,the detection results of the system accord with the change trend of microalgae density in the development of natural water bloom.