Research On Signal Processing And Interference Compensation Of Water Quality Monitoring Parameters Based On Wide Spectrum

Environmental pollution as a companion to the industrial development stage get widely attention in society.The process of water pollution is slower and easier to be ignored,its harm to human health is greater and the recovery process is slower than air pollution.The traditional reagent method has large area,long detection time,need to replace reagents regularly and is easy to cause secondary pollution.At present,this method replaced by spectral analysis gradually.Spectral analysis not only has the characteristics of no reagents,fast detection,analysis of various water quality parameters and does not affect the water quality environment of the water source,so it is the mainstream trend of future water quality analysis.Based on the principle of broad-spectrum water quality detection,the paper designs a wide-spectrum water quality multi-parameter monitor.The monitor has functions such as automatic water supply,drainage,multi-stage water filtration,data analysis,and data remote.In order to overcome the noise and temperature interference factors in the process of spectral water quality detection,the three methods of waveform similarity,signal-to-noise ratio and root mean square error are used to compare and analyze the denoising effects of various filtering algorithms.The recursive least squares(RLS)adaptive filtering algorithm has the best noise reduction effect.To reduce the temperature and measurement interference,a turbidity compensation model and a temperature compensation model are established.In order to overcome the overlapping interference phenomenon of the absorbance of each parameter in the actual water body,the derivative spectroscopy method of spectral analysis is used to model the absorbance derivative of the characteristic wavelength at different concentrations of turbidity,total nitrogen,CODcr and TOC.Aiming at the nonlinear problem of the absorbance and concentration of the characteristic wavelength of total phosphorus,the method of particle swarm optimization based on least squares-support vector machine is using to model and achieve good results.Using a self-designed wide-spectrum water quality multi-parameter monitor a lot of experiment.It is verified by experiments that the equipment can measure the turbidity,total nitrogen,CODcr and TOC parameters in real time and accurately.The repeatability error of turbidity is 2.1%,the zero drift is 1.2%,the linearity error is 1.5%,and the actual water sample comparison test error is 6.4%.The repeatability error of total nitrogen is 3%,the zero drift is 1.8%,the range drift is 2%,the linearity error is 2%,and the actual water sample comparison test error is 5.6%.The repeatability error of total phosphorus is 3.2%,the zero drift is 1.9%,the range drift is 2.1%,the linearity error is 2.5%,and the actual water sample comparison test error is 6.3%.The COD reproducibility is 3.7%,the zero drift is 2.2 mg/L,the range drift is 2.8%,and the actual water sample comparison test error is 5.9%.The experimental results show that the wide-spectral detection system can accurately measure the parameters at the same time,which has good practical application value.