40 results about "Possibility distribution" patented technology

A dynamic power consumption controlling method using the depth of stacking for multithreading prediction is suitable for supplying power to the battery. The method includes various idle mode components of low power consumption and uses the devices of multi-task imbedded system. The method includes the following steps: step one: using the cluster mode of operating request to describe the characteristics of the sequence Phi of operating request by a single thread to the components and devices; step 2: using two-stage hashtable to predict the probability distribution of the request interval under single thread mode; step 3, taking the thread as an independent request source and grouping the request source according to the task to which the thread belong and the entry address of the thread; step 4: calculating, according to the possibility distribution of the idle time of the components or devices, the optimum overtime threshold kopt and the optimum power consumption mode sopt under multiple power consumption modes and single thread. The method improves the effective hit rate for prediction and further reduces the power consumption for the idle consumption-manageable components.

The invention discloses a method of constructing a pronouncing dictionary, so as to solve the problem that the quality of the pronouncing dictionary constructed according to the prior art is poor. The method comprises steps: acoustic-phonetic features of a target vocabulary are inputted to a phonetic recognition decoder, wherein the pronouncing dictionary in the phonetic recognition decoder comprises the target vocabulary and a candidate pronouncing phoneme sequence of the target vocabulary; according to the candidate pronouncing phoneme sequence outputted by the phonetic recognition decoder, the possibility distribution of the target vocabulary corresponding to the outputted candidate pronouncing phoneme sequence is determined; according to the possibility distribution, correct pronouncing phoneme sequence for the target vocabulary is selected from the outputted candidate pronouncing phoneme sequence; and according to the correct pronouncing phoneme sequence, the pronouncing dictionary is constructed. The invention also discloses a device for constructing the pronouncing dictionary.

An explicit assumption of continuity is used to generate a fuzzy implication operator, which yields an envelope of possibility for the conclusion. A single fuzzy rule A B entails an infinite set of possible hypothese A′B′ whose degree of consistency with the original rule is a function of the “distance” between A and A′ and the “distance” between B and B′. This distance may be measured geometrically or by set union / intersection. As the distance between A and A′ increases, the possibility distribution B* spreads further outside B somewhat like a bell curve, corresponding to common sense reasoning about a continuous process. The manner in which this spreading occurs is controlled by parameters encoding assumptions about (a) the maximum possible rate of change of B′ with respect to A′ (b) the degree of conservatism or speculativeness desired for the reasoning process (c) the degree to which the process is continuous of chaotic.

The invention relates to a multi-remote sensing image fusion denoising method based on a DS evidence theory. The method particularly comprises the following steps: multiple remote sensing images at the same position at a certain time period are selected, and four noise models are built for each image; statistical analysis of data is carried out on the four noise models, and the possibility that each pixel point under each mode is the noise is obtained as basic possibility distribution for the DS evidence theory; a DS evidence theory fusion rule is used for fusing four evidences into one evidence, and the possibility that each pixel point in each image is the noise is obtained; the DS evidence theory fusion rule is repeatedly used, information of the multiple remote sensing images is fused, and a total evidence for fusion of the multiple remote sensing images is obtained; and finally, according to the evidence, a confidence interval is calculated, a well-designed decision rule is used for denoising, and a fused and denoised image is obtained. The method of the invention has the advantages that with the help of rich information sources, edge texture details of the remote sensing image can be better kept while denoising is realized.

The invention discloses a fuzzy identification method for identifying cracks of a deep drawn part, which comprises the following steps of: firstly preprocessing an acquired original acoustic emission signal; and then extracting a characteristic parameter of the preprocessed acoustic emission signal on the basis of a local wave time frequency domain local energy characteristic extraction method; carrying out automatic recombination by a genetic algorithm to generate an optimal characteristic parameter which is used as a characteristic parameter for identifying the cracks of the deep drawn part; if the optimal characteristic parameter does not comply with the normal distribution, converting the optimal characteristic parameter into a probability variable complying with the normal distribution; solving a possibility distribution function and corresponding possibility, as well as a membership function of the probability variable according to the probability consistency principle and the possibility theory, and finally carrying out fuzzy diagnosis. The fuzzy identification method eliminates the defects in state identification by using a traditional method in which the characteristic parameter is presumed to comply with the normal distribution, can accurately differentiate normal state from crack state, effectively improve the rate of identification on crack characteristics, greatly shorten diagnosis time and realize high-precision diagnosis.

The invention relates to a method and a device for optimizing adaptability of a distribution network scheme. A new automatic adaptability adjustment method suitable for the reasonable scale of a large scale is built, the basic weight of each index is determined, a time probability distribution model is built, a given time scale is used to calculate the time weight vector which is used as a difference weight, the two weights are subjected to linear weighting to obtain a comprehensive weight value of the index; and probability values are assigned to multiple scenes, and the comprehensive scheme score is determined by the comprehensive weight, the index score and the possibility distribution. Finally, the schemes are ranked according to the comprehensive score, and the optimal scheme is determined.

The invention relates to a transformer station intelligent alarm model method for big data and a cloud environment. The method is technically characterized by comprising the following steps: a data obtaining phase; data cleaning; data aligning; data calculation and data display. The data calculation comprises inputting historical data after the cleaning and aligning to multiple calculation algorithms, calculating a probability of failure of each device, and by use of corresponding relations of families where the devices belong to, calculating probabilities of other devices of the families, wherein relations between input data and device defects and device faults can be automatically learned through each algorithm. According to the invention, through a mode for the big data and the cloud environment, an alarm information total amount is automatically counted and monitored, an alarm information trend is analyzed and monitored, causes of alarm amount abnormal growth points are determined, and distribution of failure probability of each device of a whole network is analyzed. The method provided by the invention can be widely applied to an electric power company and a transformer station.

A dynamic power consumption controlling method using the depth of stacking for multithreading prediction is suitable for supplying power to the battery. The method includes various idle mode components of low power consumption and uses the devices of multi-task imbedded system. The method includes the following steps: step one: using the cluster mode of operating request to describe the characteristics of the sequence Phi of operating request by a single thread to the components; step 2: using two-stage hashtable to predict the probability distribution of the request interval under single thread mode; step 3, taking the thread as an independent request source and grouping the request source according to the task to which the thread belong and the entry address of the thread; step 4: calculating, according to the possibility distribution of the idle time of the components, the optimum overtime threshold kopt and the optimum power consumption mode sopt under multiple power consumption modes and single thread. The method improves the effective hit rate for prediction and further reduces the power consumption for the idle consumption-manageable components.

A timing calibration circuit and a timing calibration method for a time-interleaved analog-to-digital converter (ADC) are provided. The timing calibration circuit includes a correlation unit, an adaptive filter and a delay cell. The correlation unit generates a first correlation coefficient according to a first zero-crossing possibility distribution between a first digital data and a second digital data, and generates a second correlation coefficient according to a second zero-crossing possibility distribution between the second digital data and a third digital data. The adaptive filter generates a predicted time skew according to a difference between the first correlation coefficient and the second correlation coefficient. The delay cell calibrates a clock signal of the ADC according to the predicted time skew.

The invention discloses an improved method of a fault tree analysis method for safety assessment in an electrical power system. The improved method comprises the following steps: 1) in accordance with the fault tree analysis method, firstly establishing a top event, and then analyzing each reason event layer by layer, including an elementary event and a non-elementary event, and establishing a fault tree on the basis; 2) performing the membership value assessment on all elementary events and performing the value assessment on the characteristic value of the top event through research, statistics and analysis; 3) giving the membership of each none-elementary event corresponding to each elementary event by utilizing a weighted average method according to the membership assessed value of each elementary event obtained in step 2); 4) giving the membership of the top event by utilizing the weighted average method according to the membership of each non-elementary event obtained in step 3); 5) comparing the membership of the top event in step 4) with the characteristic value of the top event in step 2), and judging whether the top event occurs or not and the likelihood of occurrence of the top event. The fault tree analysis method is improved by applying possibility distribution in fuzzy mathematics, and all possible factors are better taken into account, so that the results are scientific, accurate and comprehensive and specific.

Methods and systems for improving the uniformity of a tire are provided. More particularly, a magnitude of a uniformity parameter can be obtained for each tire in a set of a plurality of tires. The magnitudes associated with the set of tires can be transformed according to a probability distribution function (e.g. a Weibull distribution function) to obtain a set of transformed magnitudes. Parameters associated with a probability distribution function can be estimated based at least in part on the transformed magnitudes. Parameters associated with the probability distribution function can be used to determine data indicative of a measurement error in the uniformity measurements attributable to a measurement process harmonic. The data indicative of the measurement error can be used to correct uniformity measurements obtained for tires and to modify tire manufacture based at least in part on the corrected measurements.

The invention discloses a fuzzy volume Kalman filtering-based single-station passive navigation method. The method comprises the steps of: establishing a fuzzy passive navigation model through a fuzzypossibility distribution method; and on the basis of the fuzzy passive navigation mode, importing a volume Kalman filtering algorithm to form a fuzzy volume Kalman filtering algorithm, and carrying out filtering correction on system state to carry out high-precision positioning navigation. According to the method, the fuzzy passive navigation model is established through importing the fuzzy method; the model preferably utilizes expert experiences and requires fewer priori knowledges, so that an exact analysis model does not need to be established; through further combining the advantages of the volume Kalman filtering algorithms in the aspects of high dimension and non-linearity, the problems that systems are uncertain and system models cannot be correctly established can be effectively solved, and the problem of high nonlinearity and high-order state in passive navigation can be effectively solved, thereby improving the applicability of passive navigation under complicated conditions.

The invention relates to a multi-remote sensing image fusion denoising method based on a DS evidence theory. The method particularly comprises the following steps: multiple remote sensing images at the same position at a certain time period are selected, and four noise models are built for each image; statistical analysis of data is carried out on the four noise models, and the possibility that each pixel point under each mode is the noise is obtained as basic possibility distribution for the DS evidence theory; a DS evidence theory fusion rule is used for fusing four evidences into one evidence, and the possibility that each pixel point in each image is the noise is obtained; the DS evidence theory fusion rule is repeatedly used, information of the multiple remote sensing images is fused, and a total evidence for fusion of the multiple remote sensing images is obtained; and finally, according to the evidence, a confidence interval is calculated, a well-designed decision rule is used for denoising, and a fused and denoised image is obtained. The method of the invention has the advantages that with the help of rich information sources, edge texture details of the remote sensing image can be better kept while denoising is realized.