Traditional fault detection methods for batch process based on data-driven often need to make assumptions about the distribution of process data, and often lead to false positives and false negatives when dealing with non-linear data and other complex data. To solve this problem, a supervised learning algorithm based on Long Short-Term Memory (LSTM) network and Batch Normalization (BN) was proposed, which does not need to make assumptions about the distribution of original data. Firstly, a preprocessing method based on variable-wise unfolding and continuous sampling was applied to the batch process raw data, so that the processed data could be input to the LSTM unit. Then, the improved deep LSTM network was used for feature learning. By adding the BN layer and the representation method of cross entropy loss, the network was able to effectively extract the characteristics of the batch process data and learned quickly. Finally, a simulation experiment was performed on a semiconductor etching process. The experimental results show that compared with Multilinear Principal Component Analysis (MPCA) method, the proposed method can identify more faults types, which can effectively identify various faults, and the overall detection rate of faults reaches more than 95%. Compared with the traditional single-LSTM model, it has higher recognition speed, and its overall detection rate of faults is increased by more than 8%, and it is suitable for dealing with fault detection problems with non-linear and multi-case characteristics in the batch process.

When the semi-supervised composite kernel Support Vector Machine (SVM) constructing cluster kernel, the universal existence problem is high complexity and not suitable for large-scale image classification. In addition, when using K-means algorithm for image clustering, the parameter is difficult to estimate. In allusion to the above problems, semi-supervised composite kernel SVM image classification method based on adaptive parameters of Mean-Shift was proposed. This method combined with Mean-Shift to make a cluster analysis of the pixel to avoid the limitations of K-means algorithm for image clustering, determined the parameters adaptively by using the structure feature of the image to avoid the volatility of the algorithm, and constructed Mean Map cluster kernel with Mean-Shift image clustering results to enhance the possibility of the same clustering samples belong to the same category, so as to make the composite kernel function guide SVM image classification better. The experimental results show that the improved clustering algorithm and parameter selection method can obtain the image clustering information better, the classification rate of the proposed method to ordinary and noise image can generally increase more than 1-7 percentage points compared with the other semi-supervised methods, and it has some applicability for the larger scale images, make the image classification more efficiently and stably.

Semi-Supervised Support Vector Machine using label mean (meanS3VM) for image classification selects a small number of unlabeled instances randomly to train the classifier, and the classification accuracy is low; meanwhile, the parameter's determination always derives much oscillation of the results. In allusion to the above problems, meanS3VM image classification method based on mean shift was proposed. The smoothed image acquired by mean shift was used as original segmented image to reduce diversities of image features; an instance in each smoothed area was randomly selected as unlabeled instance to ensure that it carried useful information for classification and had a more efficient classifier; and the parameters value were also investigated and improved, the grid search method was used for sensitive parameters, the parameter ep was estimated by combining with Support Vector Machine (SVM) mean shift results, so that there will be a better and more stable result. The experimental results indicate that the classification rate of the proposed method to ordinary and noise image can be averagely increased more than 1% and 5%, and it has higher efficiency and avoids the oscillation of the results effectively, which is suitable for image classification.

Through analyzing the application status of Vehicle Ad Hoc NETwork (VANET) in road transportation field, according to the characteristics of VANET and challenges in news transmission process, concerning the problems of previous algorithms being difficult to establish spatial model accurately and hardly considering the regularity characteristics of social behavior, a routing scheme named HBSR was proposed based on the historical behavior statistics of vehicles, including nodes connected algorithm calculating the connectivity between vehicles, topological overlap algorithm calculating the number of periods between the source node and destination node, paths selected algorithm selecting messages forwarding paths and loss strategy. Compared with several typical routing algorithms on ONE simulation platform, the simulation results prove that HBSR can find news forwarding paths more effectively, and reduces message delivery delay obviously while delivery rate increases significantly, and performance is relatively stable in VANET.

For the dynamic characteristic of vehicle Ad-Hoc network's topology, the file is difficult to completely downloaded within the communication of a single vehicle road Access Point (AP) using the existing schemes of information dissemination, having the long time delay limitation of waiting for the next AP to document communication. A method that downloaded and spread the file fragmentation using multiple vehicles within the range of some free APs was proposed, the message delivery delay was divided into direct and indirect encounter delays, and discussion was made on them respectively, and a specific option was given to choose message carried vehicles. The experimental results of the message loss rate and delay show that the environment joined with the proposed scheme can effectively improve the reliability of message downloading, shorten the delay of downloading message to the purpose vehicle, without significant additional load to the network.