Abstract: Embodiments of the invention provide a method and system for accelerating convergence of Recurrent Neural Network (RNN) for machine failure prediction. The method comprises: setting initial parameters in RNN wherein the initial parameters include an initial learning rate which is determined based on a standard deviation of a plurality of basic memory depth values identified from a machine failure sequence; training RNN based on the initial parameters and at the end of each predetermined time period, calculating current pattern error based on a vector distance between the machine failure sequence and current predicted sequence; and if the current pattern error is less than or not greater than a predetermined error threshold value, determining, by the processor, an updated learning rate based on the current pattern error, and updating weight values between input and hidden units in RNN based on the updated learning rate.

Abstract: There is provided an electric power system monitoring apparatus that can estimate dynamic behavior constants using measurement signals and can obtain the accuracy of an estimation result. An apparatus according to the present invention that monitors an electric power system includes a dynamic behavior constant estimation unit to estimate a predetermined dynamic behavior constant from a measurement value obtained from the electric power system, a relative accuracy calculation unit to calculate, from the measurement value, a relative accuracy index that indicates accuracy of a dynamic behavior constant estimation value estimated at the dynamic behavior constant estimation unit, and an estimation error calculation unit to calculate an estimation error from reference dynamic behavior information, the dynamic behavior constant estimation value, and the relative accuracy index. Thus, after the accuracy of a dynamic behavior constant estimation value is grasped, a process using the estimation value can be executed.

Abstract: There is provided a failure mode specifying system or the like that suitably specifies a failure mode of an apparatus. A failure mode specifying system 10 includes a communication portion 11 that obtains data including a detection value of a sensor 22 which is installed in an apparatus 20, an apparatus structural parameter estimating portion 14 that estimates a predetermined apparatus structural parameter based on a structure and properties of the apparatus 20, based on the data which is obtained by the communication portion 11, a failure mode specifying portion 16 that specifies a failure mode indicating a kind of failure or failure sign of the apparatus 20, based on the apparatus structural parameter, and an input-output portion 18 that presents the failure mode which is specified by the failure mode specifying portion 16.

Abstract: An analog circuit fault feature extraction method based on a parameter random distribution neighbor embedding winner-take-all method, comprising the following steps: (1) collecting a time-domain response signal of an analog circuit under test, wherein the input of the analog circuit under test is excited by using a pulse signal, a voltage signal is sampled at an output end, and the collected time-domain response signal is an output voltage signal of the analog circuit; (2) applying a discrete wavelet packet transform for the collected time-domain response signal to acquire each wavelet node signal; (3) calculating energy values and kurtosis values of the acquired wavelet node signals to form an initial fault feature data set of the analog circuit; and (4) analyzing the initial fault feature data by the parameter random distribution neighbor embedding winner-take-all method, to acquire optimum low-dimensional feature data.

Abstract: A control unit (21) determines an allocation rate of computational resources to be allocated to each of a plurality of sensing processes of analyzing sensor data output from a plurality of sensors that observe an area around a moving body (100), based on a moving environment of the moving body (100), such as the type of a road where the moving body (100) travels, behavior of the moving body (100), and visibility from the moving body (100). A detection unit (22) detects an object in the area around the moving body (100) by using, for a corresponding sensing process, computational resources of an allocated amount specified based on the allocation rate determined by the control unit (21).

Abstract: A machine learning device observes a state variable of the inside and the outside of a laser device including time-series data of light output, which is detected by an output light sensor, and a light output command through a control unit of the laser device, and acquires a determination result on correctness with respect to a quantitative failure occurrence mechanism outputted for each failure in the laser device. The machine learning device learns the quantitative failure occurrence mechanism corresponding to each failure while associating the quantitative failure occurrence mechanism with the state variable and the determination result on correctness with respect to the quantitative failure occurrence mechanism, and decides a quantitative failure occurrence mechanism which is to be outputted when an occurrence of each failure is detected.

Abstract: An arc detection method includes classifying whether an arc fault is present in the power system by, for each of a plurality of bins of a current frame of a signal, marking the bin as a candidate bin if a magnitude spectrum of the bin meets first criteria; determining a number of candidate bins in the current frame; marking the number of candidate bins as candidate cluster bins if the number of candidate bins exceeds a minimum cluster size; for each of the candidate cluster bins, determining whether the candidate cluster bin is also a candidate cluster bin of a previous frame of the first signal and if so, identifying the current frame as a candidate frame and incrementing a candidate frame count; and if the candidate frame count exceeds a candidate frame count threshold, determining that an arc fault is present in the power system.

Abstract: A technique is provided for measuring attitude of an aerial vehicle without using a highly accurate IMU. A total station (TS) for tracking a UAV includes a laser scanner. The laser scanner is made to emit laser scanning light to the UAV that is flying. The UAV has four identifiable targets, and identification and locating of the four targets are performed by means of laser scanning. The attitude of the UAV is calculated on the basis of the locations of the four targets.

Abstract: A method can include receiving sensor data from at least three different types of sensor situated in the geographic area, the types of sensors including an air temperature sensor, relative humidity sensor, dewpoint sensor, soil moisture sensor, soil temperature sensor, average wind speed sensor, maximum wind speed sensor, and a rainfall sensor, producing a feature vector including a time series of values corresponding to the received sensor data, and using a neural network, estimating the physical characteristics, the physical characteristics including at least one of (a) a leaf wetness, (b) a solar radiation, (c) an evapotranspiration, (d) a future soil moisture, and (e) a future soil temperature.

Abstract: Provided is a method of estimating a speed of a vehicle includes obtaining time domain acoustic data from an acoustic storage apparatus when the vehicle passes over a horizontally grooved road; calculating frequency domain acoustic data from the time domain acoustic data by using Fourier transformation; calculating, from the frequency domain acoustic data, a resonance frequency of sound generated between tires of the vehicle and horizontal groovings in the road; and estimating the speed when the vehicle passes over the horizontally grooved road by multiplying the resonance frequency by an interval of the horizontal groovings.