Search

In this paper, we use the electronic tongue as a detection tool and Normal Pulse voltammetry (NPV) as the measuring principle to identify the sulfate-reducing bacteria and iron bacteria. The main task is to classify the two bacteria on the reference of broth culture medium. We adopted non-supervised (principal component analysis PCA) and supervised (partial least squares PLS and artificial neural network ANN) pattern recognition to process data, and compared the performance of model for supervised identification, the experimental results showed that ANN was better. The recognition rate of ANN in the training set came to 100%, and in the test set came to 91.67%. The results prove that the electronic tongue with ANN pattern recognition system is a good choice to identify the two types of bacteria.

The principle on the electrochemical noise detected by using wavelet transform was described briefly. In the electrochemical noise measurements, there is always having noise more or less. If they are analyzed directly, the useful information and data can not be got from them, so they must be de-noised. Wavelet analysis has many advantages that other signal processing tools do not have, it can provide characteristics of the signal both in time domain and frequency domain, it has high resolution, and high resulting accuracy. The electrochemical noise of 304 stainless steel corrosion which is measured in the solution of sulfate-reducing bacteria were de- noised by using a wavelet threshold de-noising method, which made the quadratic biorthogonal spline wavelet as the mother wavelet and adopted an soft threshold processing function. The results showed that the quadratic biorthogonal spline wavelet much smoother and it can remove the noise from the electrochemical noise effectively, and can effectively identify the location of the sudden changes in the signal and accurately reflect the useful information of the signal.

Support vector machine (SVM) is applied for classification in this paper. The SVM operates on the principle of structure risk minimization; hence better generalization ability is guaranteed. This paper discussed the basic principle of the SVM at first, and then we chose SVM classifier with polynomial kernel and the Gaussian radial basis function kernel (RBFSVM) to recognize the cancer samples (benign and malignant). Selecting some value for parameters to know different performance each parameter produces to outputs. The simulations of the recognizing of two class samples have been presented and discussed. Results show the RBF SVM can classify complicated patterns and achieve higher recognition rate. SVM overcomes disadvantages of the artificial neural networks. The results indicate that the SVM classifier exhibits good generalization performance and the recognition rate above 93.33% for the testing samples. This means the support vector machines are effective for classification.

The application of wavelet neural network based on Levenberg-Marquardt Optimization to predict heat exchanger fouling is reported in this paper. We construct a 6-6-1 network according to the fouling monitor principle and parameters, the modeling of the wavelet neural network programmed with MATLAB, and trained with Levenberg-Marquarde Optimization algorithm, all training data came from the Automatic Dynamic Simulator of Fouling and input the network after normalized processing and reclassification. Simulations show that the relative error of fouling prediction is less than 0.71 percent, and wavelet neural network can be used to predict heat exchanger fouling, and has the rapid convergence rate and perfect prediction precision, the Levenberg-Marquarde Optimization algorithm can also improve convergence rate of wavelet neural network to a certain extent.

Heterotrophic bacteria is detected and counted by National Standard Method of China at present, i.e. flat dish numberation, which has the disadvantages of subjectivity, big error and low efficiency because of relying on manual counting. A method of rapid determination of heterotrophic bacteria in industrial cooling water is developed. After converting the original image into a greyscale one, we removed the edge image of culture dish on the basis of the principle that three unco-linear points can determine a circle. Then binarization with the greyscale-weighted threshold method, distance transform and watershed algorithm to segment overlapping colonies are applied, and finally four-connected searching algorithm was used to mark and count the colony. It has been proved that the method greatly improves the accuracy and efficiency of the heterotrophic bacteria counting and the counting results are not affected by shapes or sizes of the colony. It has been revealed that colony counting based on image processing method can be an effective way.

In allusion to the present heterotrophic bacteria colony counting method having the disadvantages of subjectivity, big error and low efficiency, we proposed a recognition method for heterotrophic bacteria colony based on SVM. After a series of pre-processing and segmentation to the acquired colony image, 6 feature parameters such as: area, perimeter, equivalent diameters of colony individual and non-colony ones were extracted. Then we adopted SVM to recognize them and the recognition rate of 98.7% was obtained. This means the effectiveness of the feature extraction method and the feasibility of support vector machines used for heterotrophic bacteria colony recognition.

To adopt protective measure and guarantee safe and economic operation for heat exchanger, it's necessary to accurate monitor and evaluate fouling and corrosion of heat exchanger. Point at the imperfection of present method to monitor fouling and corrosion of heat exchanger, A new practical on-line equipment was developed based on the soft-measure technology for fouling with easily measured parameters and the mental corrosion rate monitoring technology with the weak polarization method, and introduces mainly the measuring principle, the structure, the performance and the application of the monitoring equipment based on the above principle.