132 results about "Fcm clustering" patented technology

The invention discloses a method for segmenting a multi-dimensional texture image on the basis of fuzzy C-means FCM clustering and spatial information and mainly solves the problem of poor quality of image segmentation. The realizing process comprises the following steps of: inputting the texture image to be segmented, carrying out two-dimensional discrete wavelet transformation to the image, and calculating the characteristic vector corresponding to each wavelet coefficient; segmenting the coarsest scale of wavelet transformation; calculating spatial coordinate factors corresponding to the coefficients of the coarsest scale, adding the spatial coordinate factors into an objective function of a traditional FCM clustering algorithm and obtaining the segmenting result marker mapping and the marking field of the scale; obtaining the segmenting result marker mapping of the next scale by adopting the multiple dimensional segmenting method determined by an adaptive scale until the obtained segmenting result marker mapping is at the finest scale; and outputting the segmenting result of the finest scale as the final segmenting result. The method has the advantages of accurate segmenting edge and good consistency of segmenting regions and can be used for segmenting texture images, SAR images including texture information, remote sensing images and medical images.

The invention discloses an image segmentation method based on an immune clonal selection cluster, and relates to the technical field of an image processing. The purpose of the invention is to solve the disadvantages that the robustness is lower due to sensitivity of a FCM cluster segmentation method to an initial clustering center and the noise; and spatial relationship between pixels of the image is not considered by the FCM cluster segmentation method. An implementation procedure of the method is as follows: an initial population is created at random according to a setup parameter; adaptation degree of each individual in the present population is calculated to judge whether a halt condition is met; a transitional population is created by a recurrence formula of the FCM; the adaptation degree of each individual in the transitional population is calculated; based on the adaptation degree, a cloning operation is made to the transitional population; a mutating operation is made to the individual in the cloned population; after the mutating operation, a roulette wheel selection is carried on to get a new population to carry out the second step; finally, an optimum individual is selected; and the image of a segmentation result corresponding to the optimum individual is output. The image segmentation method based on the immune clonal selection cluster can be used for the cluster segmentation of a pixel level of the image.

The invention discloses an accurate classification method of the zone area load based on fuzzy C-means clustering analysis. The method comprises steps of S1: selecting a 48-point load curve of one day which is usual to carry out preprocessing; S2: using the FCM clustering algorithm to realize the division of the zone area and adopting the cluster validity function P '(U; c) to determine the best result of load classification; and S3, determining the typical daily load curve of each type of the load and the zone area included in its classification. In the accurate classification method of the zone area load based on fuzzy C-means clustering analysis, based on the comparative analysis of a variety of clustering effectiveness functions, the most suitable one is selected to judge the load classification result, furthermore, the key load eigenvectors are extracted from the user daily load curve of the actual measurement, and the cluster analysis is carried out by using the fuzzy C-means clustering algorithm to ensure a high consistency among the loads classified into the same category.

The invention discloses a driving behavior evaluation method based on FCM clustering and a BP neural network, belonging to the technical field of the Internet of vehicles. The method comprises the following steps: querying a driving satellite positioning signal set of a sample vehicle from an Internet-of-vehicles database as a sample data set; extracting the driving behavior characteristic parameters of the sample vehicle from the sample data set; building a cluster platform based on Spark of Hadoop to run an FCM clustering algorithm, and getting a clustering result of the driving behavior characteristic parameters according to the preset number of clusters; creating a BP neural network, normalizing the clustering result of the driving behavior characteristic parameters, and taking the normalized clustering result as a training sample to train the created BP neural network; and using the trained BP neural network to evaluate a normalized driving behavior to be evaluated. By mining and analyzing Internet-of-vehicles data, driving behavior evaluation is more accurate and scientific.

The invention relates to a photovoltaic power plant output power prediction method based on a weighted FCM clustering algorithm. The method provided by the invention comprises the steps that a weather data sample which matches a meteorological data sample to be predicted and the corresponding photovoltaic power plant output power are selected from the existing photovoltaic power plant operation database and are used as a reference sample; through knowledge evaluation, a typical data matrix is selected and is combined with the meteorological data sample to be predicted; after normalization, a final standard sample matrix is formed and is used as an input variable of the algorithm; and after property-weighted FCM clustering algorithm iteration, output power corresponding to the meteorological data sample to be predicted is acquired. According to the invention, the shortcomings of complex meteorological factors, unbalanced influence on the output power, meteorological data randomness and uncertainty and the like are overcome; the method has the advantages of fast prediction and high accuracy; a prediction result provides a data support for rational resource dispatching and scientific overall planning of the power industry; and good economic and social benefits are acquired.

The invention proposes a SAR image change detection system and a method based on a sparse auto-encoder and a convolution neural network, which belong to the field of SAR image processing. The method comprises: extracting characteristics from difference graphs through the sparse auto-encoder (SAE) and then using the FCM clustering to perform difference graph clustering according to the image characteristics to obtain an initial classification result; in combination with the difference graphs and the initial classification result, training the convolution neural network; and through the well-trained CNN, fine-tuning the initial classification result to obtain a final classification result graph. This method makes full use of the characteristics information and the neighborhood information of image pixels to further improve the accuracy of the change detection results. The simulation result shows that compared with the traditional algorithms such as KI and FCM, the SAR image change detection method based on the sparse auto-encoder and the convolution neural network obviously has a marked increase in detection efficiency and the Kappa coefficient as well.

The invention relates to an energy-saving method of a central air-conditioning water system. Historical running data of the central air-conditioning water system are utilized, a central air-conditioning water system energy-saving optimization model is established by fusing characteristic recognition and BP neural network based on an FCM clustering analysis method; under the condition of knowing load requirements and ambient temperature, an optimized model controllable input variable is used as an optimized operation parameter, and energy-saving optimization is carried out on the central air-conditioning water system to obtain the optimal operation parameters of the central air-conditioning water system; under the condition that the load and the environment are changed dynamically, all operation parameters of the central air-conditioning water system are optimized in real time, equipment is controlled in time, and it is guaranteed that the running efficiency of the central air-conditioning water system reaches the maximum degree while the load requirements are met; on the other hand, according to the energy-saving method, an energy-saving optimization model combines the advantages of the characteristic recognition and the BP neural network, the operation data of the central air-conditioning water system can be obtained in real time to correct the model, the central water-conditioning water system can be continuously operated in an efficient state, and the energy-saving optimization of the central air-conditioning water system is realized.

The invention relates to a method for identifying an activity type of a stay node of a dangerous goods transportation vehicle based on GPS data, comprising extracting the stay node. According to the mean longitude (shown in the description) and the mean latitude (shown in the description) of the stay node, clustering the stay node into an active hot area Hu; constructing the space-time characteristics according to the active hot area Hu, wherein the space-time characteristics comprise an average retention time avgTu, an average starting point containing rate avgFu and a mixed entropy index Entropyu; defining loading activity, unloading activity, returning activity and non-freight activity, and obtaining the actual value tendency of the space-time characteristics corresponding to the four activity types; acquiring the alternative vehicle activity types of the stay nodes according to the space-time characteristics and the defined four types of activity types by adopting an FCM clusteringmethod; setting an interest point category for each stay node, and judging the final vehicle activity type from the two alternative vehicle activity types.

The present invention discloses a dynamic state peak-valley time-of-use tariff method for improving new energy absorption capability. The method provided by the invention is characterized by dynamically guiding user rational electricity consumption, building a demand response assessment model taking a dynamic state peak-valley price into account and simulating that a user predicate the changing of force according to the new energy to dynamically respond so as to prompt the new energy absorption capability. The method comprises: (1) performing cluster analysis of the current payload of a system and dynamically dividing a peak balka period to obtain the peak-valley time-of-use tariff; (2) determining the classification of each data sample according to the maximum membership principle through adoption of a FCM cluster algorithm to perform effective peak balka fuzzy classification of the system payload of each period; and (3) building a demand response assessment model taking the dynamic state peak-valley price into account to take the set operation cost and the minimum wind abandoning as an optimal object, and introducing corresponding constraints to obtain corresponding system wind abandoning electric quantity. The dynamic state peak-valley time-of-use tariff method for improving the new energy absorption capability has a wide range of application.

The invention discloses a combustion monitoring and diagnosis method based on feature extraction and fuzzy C-means cluster, comprising steps of utilizing time-frequency domain analysis and comentropy analysis technology to extract a flame characteristic value reflecting the combustion condition in flame detection signals, utilizing the fuzzy C-means cluster algorithm to perform FCM cluster analysis on the flame characteristic value and calculating a fuzzy combustion index through the fuzzy recognition. The invention can obtain a quantitative value of a combustion state through the fuzzy mode recognition and calculation of the combustion index, can classify the combustion states to realize quantitative monitoring of the combustion state, can promptly and accurately reflect boiler operation performance attributes, can provide new thoughts and methods for optimizing the performance and monitoring the state of the boiler, and can provide a reference model to the advanced modules of the power plant monitoring information system (boiler operation optimization, state monitoring and fault diagnosis).

The invention provides an electricity usage behavior analysis method based on an FCM clustering algorithm, comprising steps of (1) copying electricity usage data from a relation database into a distributed file system HDFS to determine the clustering number c and a stopping field Epsilon, (2) determining an initial clustering center according to the clustering result of the last time, and transmitting the data to data nodes participating distributed calculation, (3) performing pre-processing on the electricity usage data and producing a key value pair <user, profile>, (4) dividing all the key value pairs <user, profile> into a plurality of data subsets and transmitting the data subsets to a Map function for calculation, (5) transmitting the Map function calculation result to an Reduce node, wherein the Reduce task combines the middle key values produced by the Map according clustering numbers and then performs calculation to obtain a new clustering center, and (6) repeating the steps (2)-(5) until a membership grade matrix satisfies the conditions of the stopping field, finishing the algorithm and outputting the clustering result. The electricity usage behavior analysis method performs direction calculation based on the file massive history electricity usage data and obtains the electricity usage behavior characteristics.

The invention discloses a self-adaptive financial time sequence prediction method based on k-line clustering and reinforcement learning. The method comprises the following steps: firstly acquiring financial data, performing K linearization processing on the financial data, and computing the data after the K linearization processing, thereby obtaining the K line data in the current matching period;clustering various sub-components of the K line by using the Kmeans clustering algorithm, the FCM clustering algorithm or the online clustering method based on the data density; inputting the clustering result in the deep reinforcement learning model to perform the parameter training, and performing the financial transaction by using the trained deep reinforcement learning. The K-linearization isperformed on the financial data, and various sub-parts of the K line are clustered, the clustering result is input into the deep reinforcement learning model to obtain the deep reinforcement learningmodel based on the decomposing k-line clustering, and the online self-adaptive prediction of the real-time financial transaction price is realized.

The invention relates to a network intrusion detection method based on machine learning. Use of conventional processing methods, such as: for example, in a data pre-processing stage carrying out symbol value conversion and normalization processing; artificially synthesizing samples with a small data volume by using an SMOTE algorithm, carrying out undersampling on the samples with the multiple data volumes by using a simple resample undersampling method, sorting the features by using an FCM clustering algorithm according to membership degrees, and selecting the features with high membership degrees as main features for extraction; in a model construction phase, firstly, classifying each large exception into a plurality of types of small exceptions by using kmeans clustering, establishing the four-layer nested XGboost model through establishment of the classification refinement on the XGboost model, and finally, explaining whether the better performance exists or not by comparing the trained optimal model with other models and comparing the detection rate and the false detection rate of intrusion detection.

The present invention discloses a data flow detection method based on a fuzzy C-means clustering algorithm and an entropy theory. An FCM clustering algorithm is introduced into the clustering analysis of a data flow, and data flow data is subjected to fuzzy C-means clustering analysis. The information entropy of the data flow is calculated by using the membership of the obtained data. Through analyzing the change trend of the entropy of the data flow, the detection of the concept drift with attribute change is carried out. The calculation of the membership and the entropy of the data flow are included. According to the method, the entropy theory is introduced, by using the membership of data to a class, the entropy of the data flow is calculated, the change of an entropy value is expressed in a time axis, and the concept drift with the attribute change is detected through the trend of a curve. The detection of the concept drift with the attribute change is visually carried out through observing the trend of an entropy value curve. The detection is mainly applied to timely prompt a system to update a parameter or not so as to ensure the correct clustering analysis possible of continuous influx of data streams.

The invention provides an air quality prediction method, which can automatically determine a hidden state number of an HMM. The method comprises the steps of obtaining observation sequences; clustering the obtained observation sequences by utilizing an FCM clustering algorithm to obtain an optimal clustering number and a corresponding clustering result; building the HMM for each category according to the clustering result, and determining an optimal clustering number of each category through the FCM clustering algorithm; taking the determined optimal clustering number of each category as the hidden state number of the built HMM; and training the built HMM according to the clustering result, and based on the trained HMM, predicting air quality. The method is suitable for the technical field of environmental detection.

The invention discloses a photovoltaic output time sequence simulation method based on a multi-scene state transition matrix and conditional probability sampling. The photovoltaic output time sequencesimulation method is used for simulating and generating photovoltaic time sequence output considering seasonal characteristics, daily characteristics, weather characteristics and fluctuation characteristics. The method comprises the following steps: firstly, aiming at a monthly photovoltaic output sequence, taking FCM clustering as internal optimization, and taking DB (-)clustering effectivenessindex as external optimization to form an original photovoltaic output sequence scene with clearer data characteristics. Secondly, establishing photovoltaic output state transfer matrixes of differentscenes, generating a photovoltaic output time sequence through a Markov chain Monte Carlo method, in the process, carrying conditional probability sampling through the Copula theory, generating a photovoltaic output state value at the next moment, and superposing the fluctuation amount conforming to the original probability distribution characteristic. Compared with an existing model, the probability statistics characteristic and the time sequence characteristic of the data are more accurate, and the implementation process is simple and easy to implement.

The invention relates to an FCM-GASVM-based industrial control system intrusion detection method, in particular, a CM-GASVM-based industrial control system application layer network intrusion detection method. According to the method, unsupervised fuzzy C-mean clustering and a supervised support vector machine are combined so as to extract the communication traffic data of the Modbus / TCP of an industrial control system; FCM clustering is performed on the communication data; and a part of the data, which meet a threshold condition, are classified through a genetic algorithm optimized-support vector machine. Since unsupervised learning and supervised learning are combined together, training time can be effectively reduced with category labels not required to be known in advance, and classification accuracy can be improved.

The present invention relates to one kind of mobile communication terminal with cartoon creating function and its cartoon creating method. The mobile communication terminal includes a memory for maintaining Gabor wavelet algorithm and FCM clustering algorithm and basic face model; a cartoon creating unit for extracting the face characteristics by means of the said algorithms to form the cartoon face; and a control unit for receiving the relevant image input and controlling the image converting process. The present invention extracts the main face characteristics from the input face image and performs the basic face model deformation to create the user's cartoon face, and the created cartoon face may be used as user's specific virtual character for free use in the terminal and capable of being sent to other communication terminals.

The invention discloses a new energy typical scene construction method based on an improved FCM clustering algorithm. The method comprises the following steps that S1, the FCM clustering algorithm isimproved by establishing a clustering effectiveness index function; S2, clustering division is performed on the new energy output historical data by utilizing an improved FCM clustering algorithm; andS3, a new energy output typical scene is selected in each category after clustering. According to the method, the new energy output characteristic of a new energy rich area is taken as a research object, the improved FCM algorithm is utilized to carry out clustering analysis on the historical time sequence output data of the new energy, original large-scale scenes are reduced and combined to obtain a plurality of representative new energy output scene sets, and the method has certain practical application value.

An embodiment of the invention discloses a method and a device for classifying network traffic on the basis of grey wolf algorithms. The method for classifying the network traffic on the basis of the grey wolf algorithms in the embodiment of the invention includes S1, carrying out multilevel and multi-scale preprocessing decomposition on network traffic data by means of wavelet packet transformation; S2, optimizing FCM (fuzzy C-means) clustering algorithm models by the aid of novel swarm intelligence algorithms-grey wolf horizontal-vertical multi-dimensional chaos optimization algorithms and classifying decomposed network traffic data by the aid of the FCM clustering algorithm models. The method and the device have the advantage that the technical problems of low decomposition accuracy and influence on network traffic clustering and classification in late periods due to the fact that existing network traffic is only decomposed on low-frequency portions by means of wavelet transformation when multi-scale decomposition is carried out on the existing network traffic by the aid of existing methods for classifying the existing network traffic can be solved by the aid of the method and the device, and the technical problems of low network traffic classification and identification accuracy and efficiency of models due to the fact that the clustering optimization models on the basis of swarm intelligence algorithms are easy to fall into local optimal solution and are low in convergence speed also can be solved by the aid of the method and the device.

The invention relates to an image labeling method, comprising the following steps: clustering each label sample in a data set by using an improved FCM clustering algorithm, dividing data sets with different semantic into different classes, and obtaining a cluster center set of each class; calculating an Euclidean distance from an image to the cluster center of each class, calculating to obtain theaverage distance from the image to each class, solving the class which has the smallest distance from the image, using the class as a labeling class of the image; finding the class having the nearestdistance from the image to the cluster center of the class, counting a labeling word which have highest occurrence number in the class, using the labeling words as the labeling word of the image. Theimage labeling method uses an improved FCM clustering algorithm to cluster each semantic label class, and a new distance measurement algorithm is usedd in the improved FCM clustering algorithm, thereby greatly improving the accuracy of image labeling. The method has good labeling effect and can preferably meet needs of practical application.

The invention provides methods for constructing and predicting the leaf trait of a woody plant and a photosynthetic characteristic model based on a DNA methylation level, and belongs to the technicalfield of biological analysis. The predicting method comprises selecting important characteristic variable embodying a geographic position difference based on a random forest, screening out 7 leaf characteristic variables, determining an optimal cluster number, and obtaining each group of cluster leaf samples by using an improved FCM clustering algorithm; according to the correlation between variables and the importance of Enzyme digestion combination obtained by a gradient boosted tree, obtaining an important enzyme digestion combination in each group of cluster leaf samples; by using the DNAmethylation level of the enzyme digestion combination as a regression variable, constructing LS-SVM regression prediction model based on Gaussian radial basis function; inputting the DNA methylation level of important enzyme digestion combination to accurately predict a leaf shape factor, leaf area and a net photosynthetic rate. The method is used for predicting the phenotypic characteristic and the photosynthetic characteristic of the woody plant, and screening individuals of woody plants with excellent traits.

The invention discloses a method based on multiple correspondence and K. The method comprises the following steps: (1) according to the obtained traffic accident data set, selecting and classifying the variables that affect the occurrence of traffic accidents; (2) Through the statistics of the number of categories of each variable and the corresponding accident number in the database, the variablecategories of the merged abnormal values are screened to obtain the accident data table; (3) processing the obtained accident data table to obtain a binary index matrix; (4) Multiple correspondence analysis is carried out by taking accident type as the variable representing accident characteristics, and the coordinates of multiple correspondence analysis of each variable type are obtained; 5) uselocal linear embedding algorithm to reduce that dimension of the variable category coordinate obtained from the multi-correspondence analysis of the accident data, and obtaining the LLE reduced dimension coordinate; (6) Use of K-Means clustering algorithm is used to cluster the variables, and the results are analyzed according to the clustering results. According to the clustering result, the invention comprehensively probes into the causes of traffic accidents from multiple dimensions, and not only analyzes two-dimensional correspondence analysis diagrams.

The invention discloses a cluster segmentation method for ancient architecture wall inscription contaminated writing brush character images, which belongs to the field of ancient architecture digital repair. The cluster segmentation method comprises the steps of: constructing a partial differential model for denoising an acquired image, and carrying out block-based enhancement according to illumination characteristics of the inscription image; segmenting the enhanced image by utilizing a maximum between-class variance method, and carrying out morphological processing on the image; carrying out regional positioning on the processed image to obtain minimum enclosing rectangles of character regions, and marking the corresponding character regions in the enhanced image; and finally, carrying out first FCM clustering on the character regions to determine a clustering central matrix, restraining a membership degree by utilizing an average grey degree similarity and a distance punishment function, and carrying out NKFCM clustering and deblurring processing to obtain a final cluster segmentation image. The cluster segmentation method can effectively eliminate influence of noise on clustering, can maintain the segmentation integrity, and can extract inscription characters with high quality. The cluster segmentation method is mainly used for clustering segmentation of ancient architecture wall inscription contaminated writing brush characters.

The invention discloses a handwritten numeral recognition method based on point density weighting online FCM clustering. The method is used for processing the large-scale offline handwritten numeral recognition problem. The method includes the steps that (1), all handwritten numeral image sets are preprocessed; (2), clustering centers are initialized, and data points are made to sequentially enter processing procedures; (3), the membership degree of the current data point and all the clustering centers is calculated; (4), if the membership degree reaches a threshold value, the position of the nearest clustering center is updated; (5), if the membership degree does not reach the threshold value, the current data point is not processed and is temporarily placed in a to-be-processed region; (6), when the to-be-processed region reaches certain standards, data in the to-be-processed region are clustered through a point density weighting FCM algorithm, and the clustering centers are updated; (7), circulation continues until all the data points are processed; (8), the membership degrees of all the data points are calculated through acquired clustering center blocks, the data points are divided into different classes, and data classification is finished through scanning at a time. According to the method, the space complexity and the time complexity can be lowered from the aspect of processing the large-scale handwritten numeral recognition problem.

The invention discloses an electric spindle unbalance fault diagnosis method which comprises the following steps: acquiring vibration data of an electric spindle at different unbalance states, carrying out denoising pretreatment on the vibration data, converting one-dimensional vibration signals of the vibration data into two-dimensional snowflake images by using a symmetric coordinate method, andsimply and visibly judging a serious degree of an electric spindle unbalance state according to change rules of the two-dimensional snowflake images; extracting image characteristic parameters from snowflake images generated from samples with known faults by using a gray level co-occurrence matrix so as to obtain a characteristic parameter matrix, inputting as FCM clustering so as to obtain a clustering center, extracting image characteristic parameters from a sample to be tested, and calculating a nearness degree of the image characteristic parameters with the clustering center according toa Hamming nearness degree, so as to obtain fault types of the sample to be tested. By adopting the method, a function of electric spindle unbalance fault diagnosis is achieved, the difficulty of faultdiagnosis is reduced, and the intelligence degree and the accuracy rate of diagnosis are increased.

The invention discloses a method for recognizing an abnormal condition of bridge monitoring data based on fuzzy clustering. The method comprises the following steps: 1) taking the monitoring data under the normal running state of a bridge as a to-be-analyzed training sample set; 2) classifying the monitoring data according to different temperature monitoring conditions; 3) adopting a pauta criterion for primarily screening all the classifications and removing noise and isolated points; 4) endowing all the samples with different fuzzy membership values; 5) performing fuzzy clustering on the primarily screened classifications according to different fuzzy membership; 6) respectively treating the training samples under different temperatures, thereby acquiring class centers and class boundaries under different temperatures; 7) comparing the class centers and class boundaries, thereby judging the running condition of the bridge. According to the invention, the fuzzy clustering technique is adopted for analyzing and processing mass bridge monitoring data. The alarm threshold value at a measured point is corrected on the basis of a statistics principle, so that the alarm threshold value can be more suitable for the practical running state of the bridge.

The invention provides a high-pressure heater state analysis method based on a genetic simulated annealing algorithm. The method comprises the steps that firstly, data of the operation state of the high-pressure heater are acquired, data loss error processing is conducted on data problems caused by equipment heat, and accidental error processing is conducted on data problems caused by unstable environments or unstable instruments and signals; Main variables influencing power generation operation are found by adopting a main element analysis method, so that the calculation data is reduced, theoperation reaction speed is increased, and representative measuring points are convenient to select; and mining the operation state data of the high-pressure heater based on a genetic annealing algorithm, and finally calculating the membership degree of the high-pressure heater and obtaining the optimal operation state of the high-pressure heater. The FCM clustering mining of the high-pressure heater operation state data is carried out based on the genetic simulated annealing algorithm, and the algorithm not only overcomes the defect that the FCM algorithm is caught in a local optimal solution, but also strengthens the global search function, thereby having better convergence and global search capability.

The invention discloses a power consumer industry dimension power utilization mode identification and analysis method and system based on a biclustering method, and belongs to the technical field of power system load characteristic analysis. The method comprises: forming typical daily load data of each power consumer in the same industry through a method of calculating an average value; carrying out first load clustering by adopting a Ward clustering algorithm, and carrying out second clustering by taking a result of the Ward clustering algorithm as an initial value of an FCM clustering algorithm to obtain different power consumption modes of power consumers in the industry, thereby finishing power consumption mode identification analysis of the power consumer industry dimension. Accordingto the method, the reasonable initial value is generated through the Ward clustering method and substituted into the FCM clustering algorithm to analyze the power consumption mode of the user, and the problem that the clustering effect is poor due to initial value sensitivity of a traditional clustering method is solved.