8 results about "Fcm clustering" patented technology

The invention relates to the technical field of video content analysis, in particular to a mine surveillance video key frame extraction method, which comprises the following steps: extracting multi-dimensional features of each frame of a candidate frame set, obtaining a fusion value, and obtaining a candidate key frame set; based on an Euclidean distance method, calculating an Euclidean distance between continuous frames of the candidate key frame set, and determining a total clustering number according to a relationship between the Euclidean distance and a preset clustering threshold value; and clustering the frames in the candidate key frame set by using a preset mixed GWO-FCM clustering algorithm and the total clustering number to obtain a key frame set. According to the method, a hybrid clustering algorithm GWO-FCM combining grey wolf optimization and fuzzy C-means is adopted, global search and local optimization capabilities are considered, and the accuracy and representativeness of key frame clustering are ensured. The high-quality extraction of the key frames is realized, the number of redundant frames is obviously reduced, and the compactness and readability of the video abstract are improved.

The invention discloses an SAR (Synthetic Aperture Radar) image change detection method combining convolution and mixed attention. The SAR image change detection method comprises the following implementation steps of: firstly, generating a difference chart for two SAR images by using a composite neighborhood intensity difference method; then, a hierarchical FCM clustering algorithm is used for carrying out pre-classification processing on the difference image, a pseudo label matrix is generated, variable and invariable high-probability sample pixels in pseudo label pixels are selected, spatial positions of the pixels are extracted, and on the pixels of the corresponding spatial positions of the two original SAR images, a pseudo label matrix is generated; pixel blocks with the pixel points as the centers are taken as a training set, and pixel blocks with all the pixel points as the centers are extracted from the two original SAR images to serve as a test set; and then training a neural network combining convolution and mixed attention by using the training sample set, and then carrying out change detection analysis on a test set by using the trained network to generate a final change detection result graph. The method has clear advantages in the aspects of SAR speckle noise suppression and change detection precision.

The application provides a traffic cell division method based on resident travel characteristics. Administrative division boundaries and hierarchical road network vector data are acquired and data preprocessing is performed; basic clustering units are divided based on the road network and the administrative division; a clustering index system is established, and resident travel characteristics in the units are acquired through mobile phone signaling in combination with the clustering units; according to the location of the clustering units in the administrative division, a k-means clustering algorithm with spatial characteristics is adopted for preliminary clustering, and clustering results and clustering centers corresponding to each clustering number are obtained; clustering effectiveness indexes of the clustering results of each clustering number are calculated; the clustering effectiveness indexes of each clustering number are compared to determine a proper clustering number; final clustering division is performed through a FCM clustering algorithm with comprehensive spatial characteristics; the clustering results are modified in combination with main hierarchical roads, and traffic cell division is completed. The application improves the robustness of the traffic cell division results and can support the construction of subsequent traffic planning models.

This invention claims protection for a task offloading method for selecting and optimizing energy consumption of trusted edge servers, comprising the following main steps: S1, constructing a system model based on server and device related information data; S2, standardizing task information and MEC cache task classes and performing FCM clustering, followed by encoding matching to obtain the MEC server to be preferentially selected for offloading for each task class; S3, using set pair analysis theory to measure and analyze the reliability relationship in S2; S4, based on the results of S2, dividing the task into subtasks and offloading them in a multi-access point network while calculating energy consumption. Using the WOA algorithm, energy consumption is optimized within the tolerable latency range to obtain the offloading decision for each subtask; S5, based on S4, offloading the subtasks to the corresponding edge nodes or processing them locally. This invention considers issues such as trusted edge server selection and energy consumption optimization during task offloading, meeting user needs and reducing system energy consumption within the tolerable latency range.

The invention discloses a combined model photovoltaic output short-term prediction method and model based on self-adaptive FCM clustering, and aims to solve the problems that in a traditional prediction model, the clustering number is difficult to determine autonomously, feature correlation mining is insufficient, parameter optimization is insufficient, prediction precision is limited and the like, and the method comprises the steps that photovoltaic related data such as solar irradiance, temperature and relative humidity are collected; performing z-score standardized preprocessing to remove abnormal values and missing values; high-correlation feature vectors are screened through Pearson's correlation coefficient analysis, and data dimension reduction is achieved; a self-adaptive FCM clustering algorithm is adopted, Markov-included angle cosine comprehensive distance optimization membership calculation is introduced, the optimal clustering number is autonomously determined through a self-adaptive function L (c), and precise clustering is carried out on the data after dimension reduction; and inputting the divided clusters into a CNN-BiLSTM-SA combination model constructed based on a CNN, BiLSTM, an improved Kepler optimization algorithm and a self-attention mechanism, and finally outputting a photovoltaic output short-term prediction result by using key parameters of an IKOA optimization model and important information of an SA dynamic focusing sequence.

The invention discloses an EMD-BiGRU photovoltaic power generation power prediction method based on FCM clustering and SSA optimization. Photovoltaic related data is obtained and preprocessed, the photovoltaic related data is divided into three types of weather data sets through FCM clustering, EMD decomposition parameters and BiGRU hyper-parameters are optimized through SSA, decomposed subsequences and meteorological characteristics are input into optimized BiGRU prediction, and a time sequence is superposed to obtain a result. According to the method, hyper-parameter empirical configuration is replaced, the data complexity is reduced, the photovoltaic power frequency characteristics are accurately captured, the prediction decision coefficient is higher than 0.96, and the method is suitable for short-term photovoltaic power generation power prediction.

The invention discloses an expressway traffic state monitoring method and system based on full-trajectory tracking data, and the method comprises the steps: obtaining standardized data through data preprocessing and time-space consistency conversion, carrying out the refined preprocessing of full-trajectory data, and completing the key feature screening through employing a PCC and MI combined two-stage strategy. Extracting multi-dimensional candidate features through a sliding window, rejecting linear redundant features by adopting a Pearson's correlation coefficient method, dividing traffic states through fuzzy C-means clustering based on the screened features, generating state labels, and realizing traffic state fuzzy division conforming to physical characteristics by utilizing FCM clustering; according to the method, the pain point of data imbalance is solved through the auto-encoder, the finally constructed recognition model has excellent performance in various indexes, and reliable technical support can be provided for accurate monitoring and active management and control of expressway traffic.

This invention discloses a bridge health monitoring method based on fuzzy clustering envelope generative adversarial network (GAN). The method uses FCEGAN to augment imbalanced samples in bridge structural health monitoring. FCEGAN converts time-series data of different modes from sensor data acquired in the bridge health monitoring system into time-frequency image data. Then, based on the time-frequency image data, a multi-level clustering consensus mechanism is employed to jointly optimize multi-level FCM clustering, sample distribution consistency, and dimensionality, generating data samples with better separability and diversity, i.e., envelope samples. Next, the envelope samples are integrated into a GAN for training. Finally, the trained envelope GAN is used to generate new data samples to balance the different mode categories in structural health monitoring. The method of this invention can generate highly separable and diverse samples, thereby improving the ability to identify and classify anomalies.