The hottest new on-line ultraviolet spectrum water

2022-08-15
  • Detail

A new UV spectrum water quality analyzer

Abstract: This paper introduces a self-developed water quality analyzer that comprehensively applies UV spectrum measurement and intelligent computing technology, and describes in detail its measurement principle, hardware structure, software function and intelligent computing method. The actual measurement results show that the water quality analyzer is simple to operate and suitable for continuous water quality analysis and batch water sample detection

key words: water quality analysis; Ultraviolet spectrum; Spectrophotometry; Intelligent computing

Introduction

in view of the increasing demand for environmental protection and the increasingly serious water pollution, the state has stipulated several standards for the content of water pollutants. There are two kinds of widely used water quality parameters: one is to directly reflect the specific components of water, such as the concentration of metal ions; The other is called substitute parameters, such as cod, BOD, TOC, etc. Alternative parameters can simply and quickly reflect the physical, chemical and microbial characteristics of water

many scientific research institutions and professional manufacturers at home and abroad are studying how to determine these alternative water quality parameters. At present, the main technical principles of automatic analyzer include chemical titration, electrochemical measurement, visible and ultraviolet spectrophotometry [7]. The principle of the first method is to determine the content of water quality parameters through chemical titration. Its disadvantages are long measurement time, complex operation and maintenance, high operation cost, and secondary pollution. The principle of the second method is that while the organic matter in the water is oxidized on the surface of the working electrode, there will be current changes on the working electrode. When the potential of the working electrode is balanced, the change of current is linear with the water quality parameters in the water. Water quality parameters can be measured by calculating the change of current. Its main characteristics are fast measurement speed, simple instrument structure and no secondary pollution. The disadvantage is that the electrode current change produced by different electrochemical methods only has a linear relationship with one alternative parameter, but has a nonlinear relationship with other water quality parameters. Therefore, generally, the analyzer based on the principle of electrochemical method can only measure one water quality parameter [2]. The third method is a quantitative analysis method based on the absorption law, which uses the absorption degree of the molecules or ions of the measured substance to the characteristic electromagnetic radiation. Experiments show that UV absorbance can reflect the degree of organic pollution in water, especially for a large class of aromatic organic compounds and organic compounds with double bonds in water. Many data also show that UV absorbance has a certain correlation with some main water quality substitution parameters [4]. Therefore, it is of great theoretical and practical significance to obtain water quality parameters by analyzing UV absorbance. This paper introduces a new UV spectrum water quality analyzer. Based on the principle of UV spectrophotometry, the full band spectrum of water samples in the UV region is collected, and the characteristic spectrum in the full spectrum is obtained by analyzing the correlation of spectral data. Then, the algorithm of intelligent software is used to analyze the relationship between spectrum and water quality parameters, and the relevant prediction model is established. The experimental results show that the water quality analyzer has the characteristics of fast measurement speed, high accuracy, good tracking performance and simple operation. Compared with similar products, it has great advantages

1 system measurement principle

this instrument is a water quality analyzer based on UV spectrophotometry. Different from foreign analyzers with the same principle, the light source used in foreign instruments only produces a single wavelength spectrum, because UV UV254 is approximately linear with Cod, while the relationship between other wavelengths and water quality parameters is more complex. In order not to lose the useful information about water quality parameters provided by other wavelengths, the light source used by the instrument produces a spectrum of 250 ~ 470nm. The influence of different wavelengths on the change of water quality parameters is analyzed by intelligent software, and the relationship model between the two is established. And many foreign instruments will use some chemical reagents more or less in use. For example, several water quality analyzers of dkk-toa company in Japan belong to this type. The instrument designed by the author can protect it well, and it does not need any chemical reagent in use, so it can obtain various water quality parameters more quickly and directly. Compared with previous water quality analyzers, this system has the advantages of fast measurement speed, good repeatability, low cost and no secondary pollution

<7. Experimental speed of film tensile testing machine p>2 system hardware structure

the analyzer adopts modular design. In addition to facilitating the expansion of different measurement units in the future, the main purpose is to prevent mutual interference between modules and improve the stability of instrument operation

the instrument is divided into three parts: water circulation system, sampling system, water quality analysis and display system

water circulation system includes water inlet system and water outlet system. The water inlet system can continuously provide water samples to the analyzer, and the water inlet pump and solenoid valve will lead the sewage into the sampling system under the control of the built-in software of the analyzer. The effluent system will discharge the sewage from the sampling system and obtain the current sewage sample

the sampling system includes water quality physical and chemical parameters acquisition system and spectral data acquisition system. The physical and chemical parameters acquisition system of water quality mainly includes sensors and transmitters. The sensor obtains common water quality parameters including pH value, temperature, conductivity, etc. The transmitter sends it to the industrial computer. The spectrum data acquisition system is mainly composed of light source, optical fiber and spectrometer. The light source is used to provide a stable UV laser. The sewage is received by the spectrometer and sent to the industrial computer after treatment. Optical fiber is the transmission medium of optical signal

the water quality analysis and display system processes the data obtained by the sampling system and displays the final results. It mainly includes industrial computer, a/d card, relay and LCD touch screen for display. The industrial computer can complete most of the functions of the computer. In addition to the built-in intelligent analysis software, it also provides many expansion slots. A/d card is used to collect data and control the hardware in the water circulation system through relays. LCD touch screen can not only display water quality parameters, but also directly operate the analyzer. The structure principle of the instrument is shown in Figure 1

3 system software design

as an intelligent analysis and testing system, it can not only detect water quality parameters automatically and in real time, but also update and display the historical data of water samples at any time through its own database. This is of great practical significance for monitoring the changing trend of water quality

3.1 control of water circulation system and acquisition of water quality data

the software controls the water pump and solenoid valve through the relay to form a water circulation system. When the measured sewage enters the sample pool, the system turns on the light source and obtains spectral data after full preheating. Other water quality parameters are sent through the probe in the sample pool and the transmitter. The data acquisition process is shown in Figure 2. Finally, discharge sewage and wash

3.2 processing and display of water quality data

impact testing machine Jinan new era Gold Testing Instrument Co., Ltd., the largest manufacturer in China, informs you of all the operations and questions of the testing machine

the spectral data and other water quality physical and chemical parameters obtained by the software are sent to the model for analysis after pretreatment, and the final results are displayed on the LCD screen, and the analysis data are stored in the database at the same time, It is used to display historical trends and update water quality models. The data processing and storage procedures are shown in Figure 3

3.3 update of water quality model

the core technology of intelligent analysis software is to model the relationship between UV spectrum and water quality parameters. The model is not unchanging. When the properties of water samples differ greatly, the model based on the original water sample can not predict the new water sample well. The software has the function of re modeling. The water quality model can be modified by inputting new water samples that have been measured according to the standard, so as to solve the inconsistency of the model when the properties of water samples are greatly different

4 introduction to intelligent analysis model

intelligent analysis software adopts a hybrid neural network model, which is composed of a polynomial model and a multi-layer forward Shen meridian superimposed in parallel. Such a structural design can ensure the universal adaptability of the hybrid model to function fitting. For the determination of the number of hidden nodes and weights of multi-layer forward networks, the training algorithm [8] used can be automatically determined according to the requirements of fitting accuracy. With the improvement of fitting accuracy requirements, the number of hidden nodes of forward networks will gradually increase until the fitting requirements are met. Even when the fitting accuracy is too high, the algorithm can also determine whether to continue to increase hidden nodes according to the improvement of fitting accuracy. Such a design can ensure the generation of forward divine meridians with the minimum number of hidden nodes, which can ensure the generation of a hybrid neural network model with the best generalization ability. At the same time, because the training algorithm is based on the least square algorithm, it is suitable for real-time measurement

the training algorithm adopted by the hybrid model is briefly described as follows:

1) the raw data are processed by coarse error elimination, data filtering, normalization and correlation analysis, and the sample data set is divided into training samples and test samples

2) obtain the linear regression model of the object by using the least square method

3) generate sample error set

4) take the error set of training samples and test samples as the training samples and test samples of multi-layer forward network respectively

5) use the least square method based collateral training algorithm to generate the most simplified forward God meridians

6) combine the forward-looking meridian and polynomial model to form a hybrid neural network model

5 actual operation analysis

this water quality parameter analyzer can be used to measure a variety of parameters, such as cod, BOD, TOC, etc. here, only the COD value (chemical oxygen demand) test is taken as an example to illustrate its test accuracy and effectiveness

the sample data used in this paper are from the urban sewage sampled 54 times, and the corresponding absorbance data are obtained by UV spectrometer, as shown in Figure 4; The standard COD test method (potassium dichromate method) [9] is used to obtain the actual COD value of each water sample

according to the processing results of the software preprocessing module, the spectral absorbance of three specific bands (absorbance data of 254 nm, 265 nm and 360 nm) is selected from the full band as the input data. In order to prevent model deviation in the modeling process, the intelligent analysis software can use the absorbance data of two or three specific spectral bands as input at the same time, and compare various modeling results, so as to automatically determine the absorbance data of the characteristic band with the best prediction effect as the modeling input data

due to the small number of data points (only 54), the left one test [10] is used to compare the accuracy of each model prediction. That is, take one of the 54 sample data points in turn as the test data, use the remaining 53 data as the training data, and then use the trained model to predict the test data. Repeat this 54 times to get the estimated value of 54 samples

Figure 5 reflects the results of calculating the hybrid model with the two characteristic wavelengths (254nm and 265nm) with the highest correlation coefficient. The y-axis represents the true value of COD, the x-axis represents the predicted value of CO calculated by the model to remove the polished damaged layer D back and forth at a large polishing rate, and the dot is the distribution of COD samples on the graph according to the true value and the predicted value, in which a 45 ° straight line measures the deviation of the predicted value of COD from the true value of COD. It can be seen from the figure that the sample training results are distributed near the straight line, which shows the good prediction accuracy of the hybrid model

6 conclusion

the experimental results show that the prediction model has good correlation and high analysis accuracy; The algorithm is fast and suitable for monitoring. from

Copyright © 2011 JIN SHI