116 results about "Fuzzy clustering analysis" patented technology

The invention discloses a photovoltaic power generation capacity/power prediction device which comprises a monitoring system, a data reading module, a data base, a data processing module, a weather information processing module and a prediction module. The data reading module reads photovoltaic power station data information monitored by the monitoring system in real time. The data base stores data information needed by prediction. The data processing module performs classification and classification result assessment on historical power generation capacity/power data and various weather data in the data base in a fuzzy cluster analysis method to obtain a similar day sample set. The weather information processing module processes weather forecast data. The prediction module comprises a BP prediction sub module, a grey prediction sub module and a support vector machine prediction sub module, and prediction results of all the prediction sub modules are combined to obtain the final prediction result. The photovoltaic power generation capacity/power prediction device combines functions of information monitoring and prediction to obtain the prediction result and provides a complete technical solution.

The invention discloses a method of monitoring faults in sections for an intermittent control system and relates to a fault monitoring method. Firstly, a plurality of batches of collected intermittent process data are standardized in a way of expanding variables, and a data matrix on each sampling time is subjected to principal component analysis; secondly, a fuzzy C-means clustering is a fuzzy clustering analysis method which is suitable for soft partition and is generated through combining a fuzzy set theory and a k-means clustering; and thirdly, after segmentation is finished, an improved MPCA (Multiway Principal Component Analysis) model with a time varying principal element covariance on the basis of expanding variables is established on each subphase, then when on-line monitoring is carried out, which phase a new batch of data belongs to is judged, whether the data exceeds the fault monitoring control limit or not is calculated and judged, if so, a fault occurs, and the fault monitoring in sections ends. According to the invention, process multi-phase partition is more accurate, misinformation and missing report rates in monitoring are reduced, and the practical application and operability are strong.

The invention discloses a multisource partial discharge detection method and device of a transformer substation based on time-frequency characteristic parameters. The method comprises the following steps of: collecting an ultra-high-frequency signal received by a UHF (Ultra-High Frequency) antenna in pre-set time; obtaining a segment of a pulse waveform of the ultra-high-frequency signal from the UHF antenna to form a waveform segment sequence; calculating a time-frequency parameter according to the waveform segment sequence and determining a time-frequency characteristic vector; carrying out fuzzy clustering analysis on the time-frequency characteristic vector; comparing a characteristic of each classified pulse waveform in a result of the clustering analysis with a characteristic of a pre-set ultra-high-frequency pulse waveform; and determining the type of the pulse waveform. With the adoption of the invention, ultra-high-frequency discharge signals are clustered by using a fuzzy clustering method through that the characteristics of the ultra-high-frequency discharge signals generated based on different discharge power supplies in time domains and frequencies are different, so as to separate the partial discharge UHF signals generated by the different discharge power supplies.

The invention provides a method of discovering zones of different functions based on point of interest data. The method comprises a step 1, map segmentation: map rasterization; a step 2, searching of a base station of a point of interest: the finding of the base station closest to the point of interest; step 3, the calculation of the distribution features of the points of interest of the various base stations; a step 4, clustering: execution of fuzzy clustering analysis of a matrix in the step 3 to acquire different clustering results; a step 5, discovering of zones of different functions: the calculation of the distribution overlap of the points of interests having category features and the different clustering results acquired in the step 4 in order to discover the base stations after the clustering. The method of discovering the zones of the different functions based on the point of interest data is advantageous in that the zones of the different functions such as tourist zones, working zones, and residential zones are discovered, and results are basically in accordance with reality, and effect is completed in a conclusive way.

The invention discloses a dynamic subarea division method based on C-mean fuzzy clustering analysis, wherein the method is used for area signal coordinated control. The method comprises that a control subarea division index system is constructed from the aspects of crossing distance, signal control period, traffic flow continuity and traffic flow discreteness; and a control subarea is divided at two levels, wherein at the first level, crossings of the subarea are selected initially to eliminate crossings whose period and distance are long; and at the second level, related indexes, which include static indexes related to distance and dynamic indexes such as the signal period, the traffic flow discreteness and the traffic flow continuity, of each crossing are calculated, and crossings are clustered in real time based on the C-mean fuzzy clustering analysis. The method divides the control subarea in real time based on the C-mean fuzzy clustering analysis, is not limited by the number and types of indexes of the subarea, and meets requirements of real-time traffic control.

The invention discloses a fuzzy clustering analysis method for detecting SAR image changes based on non-local means. The method is implemented through the processes of inputting a difference chart composed of two SAR images in a same region at different times; correcting pixels of the difference chart according to similarity measure indexes in a fast global fuzzy C-Means clustering (FGFCM) algorithm to obtain a local spatial information pixel matrix; performing non-local mean processing on the difference chart to generate a pixel matrix of non-local filtering waves; weighting and summing up the two matrixes and generating a complete pixel matrix; clustering the complete pixel matrix through the FGFCM algorithm to generate a change detection binary result image and complete the change detection of the two SAR images integrally. According to the fuzzy clustering analysis method for detecting SAR image changes based on non-local means, local spatial information and non-local mean information of images are considered simultaneously and combined organically, so that noise influences are overcome effectively and image details are kept in an image analysis clustering process, and accurate difference chart analysis results are obtained.

The present invention relates to a method of making video search of scene based on contents. It uses the fuzzy aggregation analysis method in the scene search. As compared with existent method it canobtain higher accuracy and quick searching speed.

The invention provides an abnormal road identification method. The detailed process of the abnormal road identification method includes steps: receiving seismic data of a plurality of seismic channels generated through a single shot, wherein the seismic data is seismic amplitude at a preset sampling interval; extracting attributive character data of the seismic channels from the seismic data, and building an attributive characteristic matrix; obtaining the main direction of the attributive character matrix; calculating distances from vectors which respectively represent attributive characters of the seismic channels to the main direction in the attributive character matrix, and if one distance is larger than a set threshold value, considering the corresponding seismic channel is an abnormal channel. The abnormal road identification method has the advantages of starting from multiple characteristics of the seismic channels, being capable of identifying the abnormal channel well by studying the different attributive characters of the seismic data and performing fuzzy clustering analysis on the attributive characters, and achieving automation, precision and practicability for an abnormal channel identification technology.

The invention discloses a multisource partial discharge detection method and device of a transformer substation based on space characteristic parameters. The method comprises the following steps of: collecting an ultra-high-frequency signal received by a UHF (Ultra-High Frequency) antenna in pre-set time; obtaining a segment group of a pulse waveform of the ultra-high-frequency signal from the UHF antenna to form a waveform segment group sequence; calculating a time delay of each antenna according to the waveform segment group sequence, calculating a space characteristic vector to form a sequence and carrying out fuzzy clustering analysis on a space characteristic vector sequence; comparing a characteristic of each classified pulse waveform in a result of the clustering analysis with a characteristic of a pre-set ultra-high-frequency pulse waveform; and determining the type of the pulse waveform. With the adoption of the invention, ultra-high-frequency discharge signals are clustered by using a fuzzy clustering method through that the characteristics of the ultra-high-frequency discharge signals generated based on different discharge power supplies in time domains and frequencies are different, so as to separate the partial discharge UHF signals generated by the different discharge power supplies.

The invention provides a Ka-band satellite channel modeling method integrated with meteorological factors and aims at solving the problems that data classification is lack of theoretical basis and simulation data are difficult to obtain in a traditional Ka-band satellite channel modeling process. The method comprises the following steps of: firstly, analyzing the propagation characteristic of a Ka-band satellite channel and the influences of multipath propagation, rainfall, atmospheric absorption, atmospheric scintillation and other factors to satellite channel modeling; and secondly, establishing a multistate Markov model of the Ka-band satellite channel by using a principal component analysis and fuzzy clustering analysis method. Through comparing the electrical level crossing-over rate and the average decay time of the established model with the electrical level crossing-over rate and the average decay time of a former model, the accuracy and the effectiveness of the channel model are checked, the theoretical basis is provided for resisting the rain attenuation of the Ka-band satellite channel, and meanwhile, Ka-band satellite channel modeling method has great practical significance in technical research on the aspects of signal modulation, coding mode and power control of Ka-band satellite communication.

The invention provides a method for performing texture segmentation on an image, wherein the image is provided with a plurality of kinds of textures. The method comprises the following steps: converting the image to a grayscale image; according to the size of a largest texture unit in the image, dividing the grayscale image to a plurality of areas with same size, and extracting a plurality of gradient characteristic vectors which are in one-to-one correspondence with the plurality of areas; and performing fuzzy cluster analysis on the plurality of extracted gradient characteristic vectors, and classifying the plurality of areas, thereby classifying the parts with same pattern in the image into a same kind. The invention further provides a device for performing texture segmentation on the image. The method and the device for performing texture segmentation on the image according to the invention can be adapted for multiple directions and multiple gray scales and furthermore have small computing amount.

A user preference based data cleaning method is characterized by adopting a semi-supervised learning algorithm and using a K-means fuzzy clustering analysis method to carry out semantic content marking on the preferred information of users, thereby forming a corresponding user preference data area in a data storage area; meanwhile, using the monitoring service in the user preference data area to monitor the user preference data area in real time, analyzing the change of the data in the data area and predicting the possible results, thereby deciding the operation on the next step. In a data cleaning module, the dirty data recognition service is an important part of data cleaning, and the dirty data are efficiently and accurately recognized and marked by adopting the data analysis method of optimal data location prediction. The data cleaning service gets rid of the dirty data and the error data in the system and inputs the clean data via an external interface of a bottom hardware interface.

The invention discloses a method for extracting road surface characteristic parameters based on modern time series of vertical dynamic load. Firstly, a road spectrum signal is collected by a wheel force sensor; the signal is pretreated; an auto-regressive (AR) model is established; self-adaptive Kalman filtering is performed by a state-space model to further extract more real parameters; finally, the parameters are selected and confirmed by fuzzy clustering analysis to obtain the result to judge the road identification information. The method has the beneficial effects of high automation degree, good instantaneity, accurate measurement, low cost and positive influence on improving measurement level of road quality and construction of high-quality road and reducing domestic and overseas difference.

The invention discloses a recommendation system noise filtering method based on information entropies and fuzzy C-means clustering. The method comprises steps that first, user historical scoring dataof a target recommendation system is collected and arranged; second, Monte Carlo stochastic simulation is utilized to construct sub data sets of the user scoring data, a recommendation algorithm is utilized to acquire information entropies and recommendation precision of different sub data sets; third, the information entropies are classified according to uncertainty levels, recommendation precision is classified according to recommendation precision levels, and an empirical model is constructed to determine the potential natural noise data proportion; fourth, fuzzy clustering analysis on allthe user scoring data sets is carried out, and noise data is identified and deleted; and fifth, a recommendation algorithm operates for all the scoring data sets, and a recommendation precision indexis utilized to evaluate recommendation quality. The method is advantaged in that quantization measurement of the user scoring information can be realized, and the proposed natural noise data filteringtechnology has certain universality and portability.

The invention relates to a multi-data fusion power plant fault diagnosis method based on fuzzy clustering analysis. Sample data of a variety of sensors in one device is standardized, and optimally classified through fuzzy clustering. Then, classification information is fused based on a D-S evidence theory to get a credibility value describing the state of the device. Thus, a novel fault diagnosis method is obtained. The D-S evidence theory, a fuzzy algorithm and a clustering analysis method are combined efficiently and reasonably, and the advantages thereof are integrated. For a complex power plant operation system, the diagnosis result is more accurate and efficient. The algorithms are highly coherent and correlated. Comprehensive diagnosis based on multi-sensor data is faster, and the result obtained is more accurate. The method, which is of strong applicability, is applicable to all kinds of complex, coupling and random systems, and can also be used in thermal, nuclear and other power plant systems.

The present invention relates to a commodity recommendation method based on an association analysis algorithm. Commodity recommendation is performed according to user behavior logs in shopping websites. Carry out fuzzy clustering analysis with products to generate multiple user groups and product categories, and each user group’s evaluation score for each product category; S2, build recommendation matrix M; S3, obtain new user behavior logs, use recommendation matrix M recommends products to users in the newly added user behavior log; S4, judges whether the data volume of the newly added user behavior log has reached the set value, if so, returns to step S1, and counts the data volume of the newly added user behavior log value, if not, return to step S3. Compared with the prior art, the present invention has the advantages of small amount of calculation, high reliability and strong pertinence, and has adaptability to constantly changing market consumption trends.

The invention discloses a single-phase grounding on-line fault locating method based on a distribution automation main station. The single-phase grounding on-line fault locating method based on a distribution automation main station includes the steps: extracting each fault characteristic quantity various signals when a single-phase earth fault occurs; utilizing a fuzzy clustering analysis methodto fuse each fault characteristic quantity; and fully applying the fault information, accurately positioning the fault position, thus not only improving accuracy and reliability of fault detection, and also improving adaptability and accuracy of fault determination.

The provided auto-grouping method for digital family network devices comprises: determining the minimal group number for all on-line devices; obtaining the system group parameters, extracting device feature index to generate an equal feature matrix; analyzing the matrix with fuzzy clustering technology; outputting the initial group result. Compared with prior art, this invention can automatic group by ISODATA algorithm when on-line device number up to some value, and can add a new device into some group without re-grouping.

The invention relates to a spatial self-adaptive block-matching image denoising method based on a fuzzy set theory, which comprises the following steps of: 1, setting the size of an initial similar block search window deltai,1; 2, calculating the mean and square-normalized symmetric distance between an image block y(Ni) of a pixel i to be treated and an image block y(Nj) of a pixel j in the search window deltai,1; 3, calculating the similarity of the image blocks according to the distance between the image blocks by utilizing fuzzy clustering analysis and performing weighted average on pixel values in the search window to obtain an estimated value of the pixel i to be treated; 4, correcting the pixel value of residual noise; and 5, increasing the size of a similar block search window deltai,n, and repeating the step 2 to the step 4 until an iterative termination condition is met. The spatial self-adaptive block-matching image denoising method based on the fuzzy set theory is reasonable in design; the effectiveness of the similarity division of the pixels is ensured; the accuracy of the estimated value is enhanced, and the performance of the block-based image denoising method is effectively improved.

The embodiment of the invention discloses a conference summary generation method and device based on artificial intelligence, equipment and a medium, and relates to the technical field of communication information recording. The method comprises the steps of obtaining an initial conference record of a target conference; identifying non-text information and first text information in the initial conference record, converting the non-text information into second text information, and jointly marking the first text information and the second text information as to-be-processed text information; calling a preset fuzzy clustering model, performing fuzzy clustering processing on the to-be-processed text information through the fuzzy clustering model, and extracting conference content from the to-be-processed text information; and generating a conference summary according to the conference content. The content of the conference record is quickly sorted in a fuzzy clustering analysis mode, so that the waste of human resources can be reduced, the induction difficulty is reduced, the content omission can be effectively avoided when the conference summary is output, the accuracy of the conference summary is improved, and the efficiency of sorting the conference content is improved.