As one of the main Internet applications, search engine can retrieve and return effective information from Internet resources according to user needs. However, the obtained returned list often contains noisy information such as advertisements and invalid Web pages, which interfere the user's search and query. Aiming at the complex structural features and rich semantic information of Web pages, a Web page blacklist discrimination method based on attention mechanism and ensemble learning was proposed. And, by using this method, an Ensemble learning and Attention mechanism-based Convolutional Neural Network (EACNN) model was built to filter useless Web pages. First, according to different categories of HTML tag data on Web pages, multiple Convolutional Neural Network (CNN) base learners based on attention mechanism were established. Second, an ensemble learning method based on Web page structural features was used to perform different weight computation to the output results of different base learners to realize the construction of EACNN. Finally, the output result of EACNN was used as the analysis result of Web page content to realize the discrimination of Web page blacklist. The proposed method focuses on the semantic information of Web pages through attention mechanism, and introduces the structural features of Web pages through ensemble learning. Experimental results show that, compared with baseline models such as Support Vector Machine (SVM), K-Nearest Neighbor ( KNN), CNN, Long Short-Term Memory (LSTM) network, Gate Recurrent Unit (GRU) and Attention-based CNN (ACNN), EACNN has the highest accuracy (0.97), recall (0.95) and F ₁ score (0.96) on the geographic information field-oriented discrimination dataset constructed. It verifies the advantages of EACNN in the task of discriminating Web page blacklist.

To find the optimal deployment of allocating the maximum number of virtual Field Programmable Gate Array (vFPGA) in the minimum number of Field Programmable Gate Array (FPGA) in reconfigurable cryptographic resource pool, the traditional Ant Colony Optimization (ACO) algorithm was optimized, and a vFPGA deployment strategy based on optimized ACO algorithm with considering FPGAs’ characteristics and actual requirements was proposed. Firstly, the load balancing among FPGAs was achieved by giving ants the ability of perceiving resource status, at the same time, the frequent migration of vFPGAs was avoided. Secondly, the free space was designed to effectively reduce the Service Level Agreement (SLA) conflicts caused by dynamical demand change of tenants. Finally, CloudSim toolkit was extended to evaluate the performance of the proposed strategy through simulations on synthetic workflows. Simulation results show that the proposed strategy can reduce the usage number of FPGAs by improving the resource utilization under the promise of guaranteeing the system service quality.

For the problem that the existing deep-learning based super-resolution reconstruction methods mainly study on the reconstruction problem of amplifying integer times, not on the cases of amplifying arbitrary times (e.g. non-integer times), a super-resolution reconstruction method with arbitrary magnification based on spatial meta-learning was proposed. Firstly, the coordinate projection was used to find the correspondence between the coordinates of high-resolution image and low-resolution image. Secondly, based on the meta-learning network, considering the spatial information of feature map, the extracted spatial features and coordinate positions were combined as the input of weighted prediction network. Finally, the convolution kernels predicted by the weighted prediction network were combined with the feature map in order to amplify the size of feature map effectively and obtain the high-resolution image with arbitrary magnification. The proposed spatial meta-learning module was able to be combined with other deep networks to obtain super-resolution reconstruction methods with arbitrary magnification. The provided super-resolution reconstruction method with arbitrary magnification (non-integer magnification) was able to solve the reconstruction problem with a fixed size but non-integer scale in the real life. Experimental results show that, when the space complexity (network parameters) is equivalent, the time complexity (computational cost) of the proposed method is 25%-50% of that of the other reconstruction methods, the Peak Signal-to-Noise Ratio (PSNR) of the proposed method is 0.01-5 dB higher than that of the others, and the Structural Similarity (SSIM) of the proposed method is 0.03-0.11 higher than that of the others.

Aiming at the problems of low utilization rate, high error rate, no recognition ability and manual participation in video surveillance processing in coastal and inland waters of China, a new ship tracking and recognition method based on Darknet network model and YOLOv3 algorithm was proposed to realize ship tracking and real-time detection and recognition of ship types, solving the problem of ship tracking and recognition in important monitored waters. In the Darknet network of the proposed method, the idea of residual network was introduced, the cross-layer jump connection was used to increase the depth of the network, and the ship depth feature matrix was constructed to extract advanced ship features for combination learning and obtaining the ship feature map. On the above basis, YOLOv3 algorithm was introduced to realize target prediction based on image global information, and target region prediction and target class prediction were integrated into a single neural network model. Punishment mechanism was added to improve the ship feature difference between frames. By using logistic regression layer for binary classification prediction, target tracking and recognition was able to be realized quickly with high accuracy. The experimental results show that, the proposed algorithm achieves an average recognition accuracy of 89.5% with the speed of 30 frame/s; compared with traditional and deep learning algorithms, it not only has better real-time performance and accuracy, but also has better robustness to various environmental changes, and can recognize the types and important parts of various ships.

Aiming at the inaccuracy of the methods based on ground textures and space features in detecting new ground buildings, a novel Change Detection model based on Generative Adversarial Networks (CDGAN) was proposed. Firstly, a traditional image segmentation network (U-net) was improved by Focal loss function, and it was used as the Generator (G) of the model to generate the segmentation results of remote sensing images. Then, a convolutional neutral network with 16 layers (VGG-net) was designed as the Discriminator (D), which was used for discriminating the generated results and the Ground Truth (GT) results. Finally, the Generator and Discriminator were trained in an adversarial way to get a Generator with segmentation capability. The experimental results show that, the detection accuracy of CDGAN reaches 92%, and the IU (Intersection over Union) value of the model is 3.7 percentage points higher than that of the traditional U-net model, which proves that the proposed model effectively improves the detection accuracy of new ground buildings in remote sensing images.

In order to solve the problem of low resolution of iamge, a new image super-resolution reconstruction method based on four-channel convolutional sparse coding was proposed. Firstly, the input image was turned over 90° in turn as the input of four channels, and an input image was decomposed into the high frequency part and the low frequency part by low pass filter and gradient operator. Then, the high frequency part and low frequency part of the low resolution image in each channel were reconstructed by convolutional sparse coding method and cubic interpolation method respectively. Finally, the four-channel output images were weighted for mean to obtain the reconstructed high resolution image. The experimental results show that the proposed method has better reconstruction effect than some classical super-resolution methods in Peak Signal-to-Noise Ratio (PSNR), Structural SIMilarity (SSIM) and noise immunity. The proposed method can not only overcome the shortcoming of consistency between image patches destroyed by overlapping patches, but also improve the detail contour of reconstructed image, and enhance the stability of reconstructed image.

In stereo vision measurement system, the distortion caused by the optical system makes imaging of target deviate from the theoretical imaging point, which results in measurement error of system. In order to improve the accuracy of the measuring system, a new measurement method based on stereo vision was proposed. Firstly, a quartic polynomial on the whole imaging plane was fitted through the pixel resolution of each angular point on the calibration board, the coefficient of the fitted polynomial was proportional to the distance from the object to the camera. Then, the longitudinal distance of the detected object was measured by using the measuring distance principle of binocular model. Finally, based on the obtained polynomial, the monocular camera was used to measure the transverse dimension of the detected object. The experimental results show that, when the distance between the object and the camera is within 5 m, the longitudinal distance error of the proposed method can be reduced to less than 5%. And when the object is 1 m away from the camera, the measurement error of transverse width of the proposed method is within 0.5 mm, which approaches to the theoretical highest resolution.

To solve the fault segmentation of the mean shift segmentation algorithm based on the fixed space and color bandwidth, a mean-shift segmentation algorithm based on the density revise with saliency feature was proposed. A region saliency computing method was firstly proposed on the basis of density estimation of main color quantization. Secondly, region saliency was fused with pixel level saliency as density modifying factor, and the fused image was modified as input for mean-shift segmentation. Finally, the scatter regions were merged to obtain the final segmentation results. The experimental results show that for the truth boundaries, the average precision and recall of the proposed segmentation algorithm are 0.64 and 0.78 in 4 scales. Compared with other methods, the accuracy of the proposed segmentation method is significantly improved. It can effectively improve the integrity of the target and the robustness of natural color image segmentation.

With respect to single-core operating system, multi-core real-time operating system is more functional and complicated. Aiming at the problem that multi-core operating system is difficult for configuration, tailoring and transplantation, a new configuration tool for the application of multi-core real-time operating system, which can improve the application development efficiency of multi-core real-time operating system and reduce the error rate was proposed. First, based on a multi-core real-time operating system named CMOS (Control-oriented Multi-core Operating System), which was independently developed by Chongqing University of Posts and Telecommunications, the configuration tool was hierarchically designed. According to the demand of CMOS, a visualized configuration tool was designed to finish interface generation engine and automatical code generation. Afterwards, in order to ensure the correctness of the configuration logic, configuration correlation detection was proposed. The simulation results show that the CMOS configuration tool is suitable for CMOS operating system because of the short processing time for code generation and low error rate. Compared with the method of troubleshooting by developers manually, correlation detection accelerates the speed of troubleshooting with quickly locating the error code and ensure the generation correctness of configuration file. Thus the configuration tool can promote the application of CMOS multi-core operating system.

In order to overcome the non-person entity's interference, the difficulties in selection of feature words and muti-person influence on target personal relation extraction, this paper proposed person judgment based on decision tree, relation feature word generation based on minimum set cover and statistical approach based on three-layer sentence pattern rules. In the first step, 18 features were extracted from attribute files of China Conference on Machine Learning (CCML) competition 2015, C4.5 decision was used as the classifier, then 98.2% of recall rate and 92.6% of precision rate were acquired. The results of this step were used as the next step's input. Next, the algorithm based on minimum set cover was used. The feature word set covers all the personal relations as the scale of feature word set is maintained at a proper level, which is used to identify the relation type in text headline. In the last step, a method based on statistics of three-layer sentence pattern rules was used to filter small proportion rules and specify the sentence pattern rules based on positive and negative proportions to judge whether the personal relation is correct or not. The experimental result shows the approach acquires 82.9% in recall rate and 74.4% in precision rate and 78.4% in F1-measure, so the proposed method can be applied to personal relation extraction from text headlines, which helps to construct personal relation knowledge graph.

Aiming at the problem of path coverage test data generation, a new test data generation method based on improved Fireworks Xxplosion Optimization (FXO) algorithm was proposed. First, key-point path method was used to represent the program paths, and the hard-covered paths were defined by the theoretical paths, easy-covered paths and infeasible paths; the easy-covered paths adjacent to the hard-covered paths and their testing data were recorded and used as part of the initial fireworks to improve convergence speed, and the remaining initial fireworks were created randomly. Then according to the individuals' fitness values, an adaptive blast radius was designed to improve convergence rate, and the thought of boundary value test was introduced to modify the border-crossing sparkles. Compared with other seven optimization algorithms that generate testing data, including fireworks explosion optimization with adaptive radius and heuristic information (NFEO), FEO, F-method, NF-method, etc, the simulation results show that the proposed algorithm has lower time complexity of calculating level, and better performance in convergence.

Concerning the problem that community detection algorithm based on label propagation in complex networks has a pre-parameter limit in the real network and redundant labels, a community detection model in large scale academic social networks was proposed. The model detected Utmost Maximal Cliques (UMC) in the academic social network and arbitrary intersection between the UMC is the empty set, and then let nodes of each UMC share the unique label by reducing redundant labels and random factors, so the model increased the efficiency and stability of the algorithm. Meanwhile the model completed label propagation of the UMC adjacent nodes using closeness from core node groups (UMC) to spread around, Non-UMC adjacent nodes in the network were updated according to the maximum weight of its neighbor nodes. In the post-processing stage an adaptive threshold method removed useless labels, thereby effectively overcame the pre-parameter limitations in the real complex network. The experimental results on academic social networking platform-SCHOLAT data set prove that the model has an ability to assign nodes with certain generality to the same community, and it provides support of the academic social networks precise personalized service in the future, such as latent friend recommendation and paper sharing.

To overcome slow convergence velocity of Particle Swarm Optimization (PSO) which falls into local optimum easily, the paper proposed a new approach, a PSO algorithm using opposition-based learning and adaptive escape. The proposed algorithm divided states of population evolution into normal state and premature state by setting threshold. If popolation is in normal state, standard PSO algorithm was adopted to evolve; otherwise, it falls into "premature", the algorithm with opposition-based learning strategy and adaptive escape was adopted, the individual optimal location generates the opposite solution by opposition-based learning, increases the learning ability of particle, enhances the ability to escape from local optimum, and raises the optimizing rate. Experiments were conducted on 8 classical benchmark functions, the experimental results show that the proposed algorithm has better convergence velocity and precision than classical PSO algorithm, such as Fully Imformed Particle Swarm optimization (FIPS), self-organizing Hierarchical Particle Swarm Optimizer with Time-Varying Acceleration Coefficients (HPSO-TVAC), Comprehensive Learning Particle Swarm Optimizer (CLPSO), Adaptive Particle Swarm Optimization (APSO), Double Center Particle Swarm Optimization (DCPSO) and Particle Swarm Optimization algorithm with Fast convergence and Adaptive escape (FAPSO).

The domain concepts are complex, various and hard to capture the development of concepts in software engineering. It's difficult for students to understand and remember. A new effective method which extracts the historical evolution information on software engineering was proposed. Firstly, the candidate sets included entities and entity relationships from Wikipedia were extracted with the Nature Language Processing (NLP) and information extraction technology. Secondly, the entity relationships which being closest to historical evolution from the candidate sets were extracted using TextRank; Finally, the knowledge base was constructed by quintuples composed of the neighboring time entities and concept entities with concerning the key entity relationship. In the process of information extraction, TextRank algorithm was improved based on the text semantic features to increase the accuracy rate. The results verify the effectiveness of the proposed algorithm, and the knowledge base can organize the concepts in software engineering field together according to the characteristics of time sequence.

A bridge crack measurement system based on binocular stereo vision technology was proposed considering the low efficiency, high cost and low precision of bridge cracks measurement at home and abroad. The system realized by using some binocular stereo vision methods like camera calibration, image matching and three dimensional coordinates reconstruction to calculate the width and the length of bridge cracks. The measured results by binocular vision and by monocular vision system under the same conditions were compared, which show that using the binocular vision measurement system made width relative error keep within 10% and length relative error keep within 1% steadily, while the results measured by monocular vision were changed widely in different angles with a maximum width relative error 19.41% and a maximum length relative error 54.35%. The bridge crack measurement system based on binocular stereo vision can be used in practical well with stronger robustness and higher precision.

Since the face images might be not over-complete and they might be also corrupted under different viewpoints or different lighting conditions with noise, an efficient and effective method for Face Recognition (FR) was proposed, namely Robust Principal Component Analysis with Collaborative Representation based Classification (RPCA_CRC). Firstly, the face training dictionary D₀ was decomposed into two matrices as the low-rank matrix D and the sparse error matrix E; Secondly, the test image could be collaboratively represented based on the low-rank matrix D; Finally, the test image was classified by the reconstruction error. Compared with SRC (Sparse Representation based Classification), the speed of RPCA_CRC on average is 25-times faster. Meanwhile, the recognition rate of RPCA_CRC increases by 30% with less training images. The experimental results show the proposed method is fast, effective and accurate.

Aiming at the problem that virtual-real registered accuracy and real-time performance are influenced by image texture and uneven illumination in Augmented Reality (AR), a method based on improved ORB (Oriented FAST (Features from Accelerated Segment Test) and Rotated BRIEF (Binary Robust Independent Elementary Features)) algorithm was proposed to solve it. The method firstly optimized the dense region of image feature points by setting the number and distance threshold of it and used parallel algorithm to reserve N points of greater eigenvalue; Then, the method adopted discrete difference feature to enhance the stability of uneven illumination changes and combined the improved ORB with BOF (Bag-of-Features) model to realize quick retrieval of Benchmark image. Finally, it realized the virtual-real registration by using the homographics between images. Comparative experiments among the proposed method, original ORB, Scale Invariant Feature Transform (SIFT) and Speed Up Robust Features (SURF) algorithms were performed from the aspects of accuracy and efficiency, and the proposed method reduced the registration time to about 40% and reached the accuracy more than 95%. The experimental results show that the proposed method can get a better real-time performance and higher accuracy in different texture and uneven illumination.

To tackle the higher requirement of mobile network for movie service system and the lack of description of movie domain knowledge, the necessity and feasibility of establishing the Movie Ontology (MO) were illustrated. Firstly, the objects and components of MO were summarized, and the principle and method for building the MO model were also put forward, with using the Web Ontology Language (OWL) and Protege 4.1 to build the model. After that, the concrete representation of the class, property, individual, axioms and inference rules in the MO were explained. Finally, the consistency of MO was analyzed, including the consistency analysis of relationship between classes and the consistency analysis based on axioms.

As standard Particle Swarm Optimization (PSO) algorithm has some shortcomings, such as getting trapped in the local minima, converging slowly and low precision in the late of evolution, a new improved PSO algorithm based on Gaussian disturbance (GDPSO) was proposed. Gaussian disturbance was put into in the personal best positions, which could prevent falling into local minima and improve the convergence speed and accuracy. While keeping the same number of function evaluations, the experiments were conducted on eight well-known benchmark functions with dimension of 30. The experimental results show that the GDPSO algorithm outperforms some recently proposed PSO algorithms in terms of convergence speed and solution accuracy.

In Dynamically Tuned Gyroscope (DTG) system, traditional identification methods, including least square identification method and traditional frequency domain identification method, could not achieve acceptable identification fitness degree. To deal with this problem, outlier-eliminated frequency identification method was proposed. In consideration of the characteristics of DTG model structure and intrinsic colored noise, outlier-eliminated method was applied to DTG frequency domain identification. The experimental results indicate that outlier-eliminated frequency identification method, with a fitness degree above 90%, compared with both least square identification method and traditional frequency domain identification method, has a better performance. In addition, outlier-eliminated frequency identification method possesses of good repeatability and stability. Outlier-eliminated frequency identification method could improve the identification fitness degree of DTG system.

For the control allocation problem of flexible fly-wing aircraft with multi-control surfaces, the machine vibration force index was put forward to measure the elastic vibration. Total control allocation model was established, the superior performance of the Estimation of Distribution Algorithm (EDA) was used for solving the model. Firstly the rudder structure was designed, the way of work and control capability of every aerodynamic rudder were analyzed, and the rudder functional configuration was built in accordance with the rudder control efficiency of redundant rudder, elevator aileron and aileron rudder in aerodynamic data. During the control allocation, main performance indices of control allocation were analyzed, the overall multi-objective optimal evaluation function was established, which combined with the equality and inequality constraints, and solved by EDA. The true distribution was estimated by establishing a probability model, during the evolutionary process of EDA, the rudder would be allocated according to the deflection efficiency, the optimal solution was got by combining with the optimization function. At last, the impact of aero wing flexibility on static control performance of the system was analyzed. After considering aeroelasticity, the overshoot and transition time are decreases. The flying quality of flying wing aircraft is significantly improved, the system efficiency is improved by at least 10% after optimization. The simulation results show that the EDA can better solve the control allocation problem, and can improve the dynamic quality of the system, verifying the effectiveness of multi-control surfaces to control allocation.

After analyzing and comparing the existing security situation assessment methods, a network security situation assessment method was proposed based on time dimension, which focused on the necessity of using different methods for short-term and long-term assessment respectively. Based on the alarm information which came from security device such as firewall and Intrusion Detection Systems (IDS), the whole short-term situation was got according to the score of destination host. Combining the result of short-term assessment and static index, the weight of long-term assessment system was determined by entropy method. The proposed assessment method divides network security situation into short-term and long-term, and makes up for the lack of setting situation assessment boundaries in terms.

In order to further improve the efficiency of the population diversity in the implementation process of the Particle Swarm Optimization (PSO), an Adaptive PSO (APSO) algorithm based on diversity feedback was proposed. APSO adopted a new population diversity evaluation strategy which enabled the automatic control of the inertia weight with population diversity in the search process to balance exploration and the exploitation's process. In addition, an elite learning strategy was used in the globally best particle to jump out of local optimal solution. It not only ensured the convergence rate of the algorithm, but also adaptively adjusted the search direction to improve the accuracy of solutions. The simulation results on a set of typical test functions verify the validity of APSO.

Considering the problems caused by the traditional association rules visualization approaches, including being unable to display the frequent pattern and relationships of items, unitary express, especially being not conducive to represent multi-schema association rules, a new visualizing algorithm for multi-valued association rules mining was proposed. It introduced the redefinition and classification of multi-valued attribute data by using conceptual lattice and presented the multi-valued attribute items of frequent itemset and association rules with concept lattice structure. This methodology was able to achieve frequent itemset visualization and multi-schema visualization of association rules, including the type of one to one, one to many, many to one, many to many and concept hierarchy. At last, the advantages of these new methods were illustrated with the help of experimental data obtained from demographic data of a province, and the source data visualization, frequent pattern and association relation visual representation of the demographic data were also achieved. The practical application analysis and experimental results prove that the schema has more excellent visual effects for frequent itemset display and authentical multi-schema association rules visualization.

For the poor readability and understandability with the existing display form for genealogical data, this paper presented visualization for genealogical data with metagraph. In the metagraph representation of genealogy, the generating set comprised of all persons in the family; each edge represented only "parents-child" relationship. An edge in the metagraph representation of genealogy was a pair consisting of an invertex and an outvertex, the invertex consisted of two nodes of the marital relationship, and the outvertex represented a single child node set. The experimental results show that the number of the edges in the metagraph form is almost half of common form in the case of the same data, and the visualizing effect is significantly improved. At the same time, the proposed methodology has a guiding role in the mathematical modeling of genealogy, the research of genealogy visualization and the improvement of genealogical information system.

Concerning the poor robustness of the state-of-art spectrum-based algorithms when the outliers and noises exist，a new and robust point pattern matching algorithm based on Quasi Laplacian spectrum and Point Pair Topological Characteristic (QL-PPTC) was proposed. In this paper, firstly, a signless Laplacian matrix was constructed by using the minimal spanning tree of weighted graph, and then the eigenvalues and eigenvectors obtained from the spectrum decomposition were used to represent the point’s feature, which made it possible to calculate the matching probability. Secondly, the similarity measurement of point pair topological characteristic was computed to define local compatibility between the point pairs, and then correct matching results were achieved by using the method of probabilistic relaxation. The contrast experimental results show that the proposed algorithm is robust when the outliers and noises exist in point matching.

For resolving the problem of curvature calculation for scattered point cloud data with strong noise, a robust statistics approach to curvature estimation was presented. Firstly the local shape at a sample point in 3D space was fitted by a quadratic surface. In addition,the fitting was performed at multiple times with randomly sampled subsets of points, and the best fitting result evaluated by variable-bandwidth maximum kernel density estimator was obtained. At last, the sample point was projected onto the best fitted surface and the curvatures of the projected point was estimated. The experimental results demonstrate that the proposed method is robust to noise and outliers. Especially with increasing noise variance, the proposed method is significantly better than the traditional parabolic fitting method.

The advantages of simplicity and easy implementation of Particle Swarm Optimization (PSO) algorithm have been validated in science and engineering fields. However, the weaknesses of PSO algorithm are the same as that of other evolutionary algorithms, such as being easy to fall into local minimum, premature convergence. The causes of these disadvantages were analyzed, and an improved algorithm named Cosine PSO (CPSO) was proposed, in which the inertia weight of the particle was nonlinearly adjusted based on cosine functions and the learning factor was symmetrically changed, as well as population diversity was maintained based on bacterial chemotaxis. Therefore, CPSO algorithm is better than the Standard PSO (SPSO) in a certain degree. Simulation comparison of the three algorithms on five standard test functions indicates that, CPSO algorithm not only jumps out of local optimum and effectively alleviates the problem of premature convergence, but also has fast convergence speed.

With regard to the problem that recognition software does not work when the Quick Response (QR) code image is contaminated, damaged or obscured, a QR code recognition method based on sparse representation was proposed. Forty categories QR code images were used as research subjects and each category has 13 images. Three images were randomly selected from each category and thus a total of 120 images were got as the training sample and the remaining 400 as test sample. Sparse representation dictionary was composed of all training samples. The test samples were a sparse linear combination of the training samples and the coefficients were sparse. The projection of each test sample in the dictionary was calculated, so category with the smallest residual was classification category. Finally, comparison and analysis were done between the recognition results of the proposed method and the QR code recognition software PsQREdit. The experimental results show that, the proposed method is able to correctly identify for partially contaminated, damaged and obscured image, and it has good robustness. It is a new effective means for the recognition of QR code.

The soft fault samples of the elevator control system are difficult to obtain and their generated time is always short. To solve the problem, this paper proposed a fault diagnosis method based on Finite State Machine (FSM). The switch variables and analog variables of elevator were taken as the state characteristics of FSM. The diagnostic system collected and recorded various states and transitions between states during the normal operation of elevator. Base on this operation a specification model of elevator control system could be built. To detect and diagnose faults of elevator control system, the passive test error detect algorithm based on FSM was adopted and optimized. The last step was to verify new faults and complete the specification model. Experimental results show this method can not only detect unknown situations timely, but also diagnose known faults effectively; the method plays a good supervisory role to the transient soft faults of elevator control system.