Electrochemical Method And Hardware Development For Rapid Detection Of Water Quality In Freshwater Aquaculture

In this study,nanomolecules were used to construct electrochemical sensors for detection of Aphanomyces and Cu²⁺.Based on this,a real-time intelligent water quality analysis and detection system was developed by using water temperature,pH value,dissolved oxygen,conductivity,ammonia nitrogen,Aphanomyces and Cu²⁺as detection parameters.This system is able to calculate BOD₅ at real time and predict BOD₅ of the following 3 days.Field test proved that this system can be used for real-time water quality detection and analysis of freshwater aquaculture.The study provides a technical guarantee for timely improving water quality of freshwater aquaculture and ensuring safety of aquatic products.The specific research contents are as follows:1.An unlabeled immunity biosensor was developed through covalently conjugating and immobilizing Aphanomyces invadans antibody（3gJC9）on graphene nanogold and cysteamine monolayer nanocomposites,and conditions for detection of A.invadans pathogen were oprimized.The optimized detection parameters were as follows:the modified electrode was soaked in 0.20μg/mL antibody solution for 90minutes,and the interaction time of the immune reaction was 10 min.The linear correlation range for detection of A.invadans concentration in this method was 0.2 to4μg/mL with a detection limit of 309 ng/mL.Through spiking test in actual water samples,the recovery rate was between 0.94 and 1.1,which meets the detection requirements of water quality targeting Aphanomyces in freshwater aquaculture.2.An Ag-Pd bimetallic nanoparticle composite material was synthesized and an electrochemical probe for detecting Cu²⁺in aquaculture water was constructed using Ag-Pd bimetallic nanoparticles.The feasibility of using Differential Pulse Adsorptive Stripping Voltammetry（DPAdSV）to detect Cu²⁺in aquaculture water was studied and the detection conditions were optimized.The results showed that,when the Ag-Pd/GCECu²⁺sensor was fabricated on a glassy carbon electrode（GCE,Φ=3.0mm）coated with a DMF dispersion with an Ag-Pd content of 1.0 mg/mL,at enrichment potential of-1.5 V,at enrichment time of 15 min,at buffer pH of 9.0,and using the DPAdSV method for scanning in the range of-0.3⁰.3V potential（step potential:4mV;pulse amplitude:0.05V）,its linear range and detection limit in the detection of Cu²⁺standard solution was 1.5³5 and 0.3μmol/L respectively.Compared with the reported Cu²⁺detection electrode,it had higher detection limit and stability.The Ag-Pd/GCE sensor for detection of Cu²⁺in water had strong anti-interference ability against As²⁺,Zn²⁺,Fe²⁺and Pb²⁺in aquaculture water.The detection of Cu²⁺concentration in water of freshwater aquaculture using external standard method had an RSD within 6%compared to ICP-OES detection method.3.Using open source hardware（Arduinoas）as tool,the hardware and software of mobile water quality inspection platform were designed and developed,and an intelligent water quality analysis and detection platform using 7 parameters including waterquality,pHvalue,dissolvedoxygen,conductance,ammonianitrogen,Aphanomyces and Cu²⁺was established.This platform was composed of the sensor node,Bluetooth node,aggregation node,local storage center,and network services.It transmitted the data collected by the local detection platform to the backend server through network,which laid a hardware and software foundation for later data analysis and diagnosis.After correcting the temperature of the sensor,in terms of the newly detected water temperature,pH value,dissolved oxygen,conductance and ammonia nitrogen,detection data of the platform had the relative error of less than±0.5%,±0.2%,±1%,±0.5%and±0.5%respectively and the accuracy of less than 1%,1%,1.5%,0.3%and 1.5%respectively,compared with the existing international standard detection methods.These test results meet the testing requirements for aquaculture water quality.4.The 7 parameter intelligent water quality monitor was used to continuously collect the parameters such as T,pH,DO,NH₃-N,K and Cu²⁺etc.in aquaculture water.The covariance analysis and principal component analysis of water quality parameters were carried out to discuss the correlation between parameters andtheir effects on water quality of aquaculture,and the intrinsic reasons for the change of parameters were analyzed from the perspective of biomass interaction.The 5parameters significantly related to BOD₅,including T,pH,DO,NH₃-N and Cu²⁺,were served as dependent variables to compare the feasibility of soft sensingbetween the multivariatelinearregressionmethod（MLR）andLevenberg-Marquardt Antipropagation Neural Network（LM-ANN）to BOD₅.The results showed that:the non-linear prediction model established by LM-ANN modeling can realize the soft sensing of BOD₅ effectively.When parameters such as T,pH,DO,NH₃-N and Cu²⁺were selected as input layer parameters,the number of hidden layer units was 8,and the output layer parameter was 1（BOD₅）,the prediction accuracy of LM-ANN model was the highest,and the correlation coefficient between the fitting value and the measured value could reach 0.964.The model can be used as a soft sensing method for BOD₅ estimation of aquaculture water.5.The feasibility of using time series method to alarm five parameters such as pH,DO,NH₃-N,Cu²⁺and K was discussed.Through timing diagrams and autocorrelation tests,it was found that thetime series of these five parameters were stationary and could be alarmed in advance with time series.The Matlab data processing software was used to build the NAR neural network.Through learning the historical water quality data and adjusting the number of hidden layers as well as delay orders of the neuron,the model parameters were trained and optimized.The results showed that:when the number of hidden layers and delay orders in the pH alarming model was at its best level of 12 and 5 respectively,the RSME of the predicted value was 4.85×10^-66 and the correlation coefficient was 0.995;when the number of hidden layers and delay orders in the DO alarming model was at its best level of 12 and 4 respectively,the RSME of the predicted value was 9.83×10^-66 and the correlation coefficient was 0.993;when the number of hidden layers and delay orders in the NH₃-N alarming model was at its best level of 8 and 6 respectively,the RSME of the predicted value was 7.91×10^-55 and the correlation coefficient was 0.998;when the number of hidden layers and delay orders in the K alarming model was at its best level of 13 and 7 respectively,the RSME of the predicted value was 8.55×10^-44 and the correlation coefficient was 0.999;when the number of hidden layers and delay orders in the Cu²⁺alarming model was at its best level of 10 and 4 respectively,the RSME of the predicted value was 9.43×10^-55 and the correlation coefficient was 0.998.By comparing the water quality parameters predicted with the trained NAR neural network model within 5 days,the results showed that the predictive data and detection data within the next 3 days has fairly high correlation and accuracy.The parameters such as pH,DO,NH₃-N,Cu²⁺and K within the next 3 days were used to fit BOD₅,and the results showed that fitting BOD₅ with the predicting value of the 2nd day has fairly high accuracy.Therefore,these 5 parameters could be used for early alarming of BOD₅.