JP2014174071A - Calculation device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve the problem that the hitherto known movement calculation of an object makes an intermediate point between two points of an estimated position calculated by a drift calculation as a start point of a calculation for a next estimated position, and therefore, diffusion of a probable existence distribution calculated at this point cannot be incorporated when a next time probable existence distribution is calculated.SOLUTION: A calculation process 204 of a probable existence is provided, and a circle showing a spread of a diffusion effect from a calculated estimation position is divided into meshes, and a probable existence of an object to be a target in each mesh is calculated, and after that, the probable existence of an object to be a target reflecting the probable existence at a start point of an estimation position calculation is calculated for each mesh, and the probable existence of a whole region is calculated by superposing the probable existences from all start points at an identical time. Thereby, a region within the circle of the probable existence from all start points in a theoretical time tis constantly arranged with the start point in time tat a next calculation pitch width by an arrangement process 207 of a next start point.

Description

  The present invention relates to a calculation device that performs prediction calculation of movement of an object based on weather data.

  When performing drift calculation that predicts the movement trajectory of drifting ships, drifters, and other marine objects using weather data such as wind and sea data, the step size of the simulation logical time (hereinafter, logical time) Movement calculation for calculating the estimated position of the object is performed for each calculation time interval (see, for example, Patent Document 1). In this object movement calculation method, the start point of the object movement calculation at the next logical time after the calculation time step size has elapsed is determined from the estimated position of the calculated object, and this start point is used to further calculate the next logical time. The estimated position of the object is calculated.

  In the conventional object movement calculation method, two circles indicating the distribution of the estimated position of the object calculated for each calculation time step are obtained, and the midpoint of the center of the two circles is determined as the starting point for the calculation of the next estimated position. Then, the estimated position at the next logical time is calculated with this starting point as a reference (see, for example, Non-Patent Document 1).

JP 2012-159344 A

"U.S. Coast Guard Addendum to the United States National Search and Rescue Supplement (NSS)", COMDTINST M16130.2E (2009), Page H-54

  In the above-described conventional object movement calculation method, an intermediate point between two calculated estimated positions of the object is used as a starting point for calculating the estimated position at the next logical time. However, the estimated position of the object calculated at this point is limited to one of the intermediate points, and diffusion of the estimated position based on the estimated distribution of the estimated positions is not considered. For this reason, even if the existence probability distribution is calculated when calculating the estimated position of the object, the calculated existence probability distribution is not included in the next calculation of the existence probability distribution. There is a problem that diffusion from the position is not sufficiently considered.

  In addition, as another drift calculation method, there is a method of calculating the next estimated position using each of the two center positions of the calculated estimated position of the object as a starting point. In this different drift calculation method, the center position of the two points calculated sequentially is the starting point, and the estimated position of the object is obtained sequentially. Therefore, the movement calculation of the object in consideration of the estimated distribution of existence of the estimated position and diffusion It can be performed. However, since two start points at the next logical time are generated for each start point, the amount of calculation increases by the square of the number of calculations for each calculation time step size, and simulation of object movement calculation There was a problem that the execution speed decreased.

  The present invention has been made to solve such a problem, and calculates the next existence probability distribution using the start point in consideration of diffusion from the estimated position of the object, and the number of calculations at each logical time. The purpose is to suppress the increase of.

  The calculation device according to the present invention performs a drift calculation for predicting the movement trajectory of a marine object such as a drifting ship and a drifter based on weather data such as wind image data and sea condition data, and moves after the movement for each calculation time step. Based on the estimated position calculation process for calculating the estimated position of the object, the flow error obtained from the weather data, the wind pressure flow velocity, and the wind pressure declination, a diffusion region in which the existence probability of the estimated position of the object is greater than or equal to a predetermined value is determined. A calculation process of the spread of the desired diffusion effect and a plurality of calculation grids in the above diffusion area, and the existence probability of the estimated position of the object in each grid is calculated for each start point, and the existence for each stored start point at each time Presence calculation processing to calculate the existence probability of the object by superimposing the probability, and the object obtained by the calculation processing of the existence probability in the diffusion region of the estimated position of the object from all the start points The presence of By selecting a certain number of points with the highest calculation as the starting point at the next calculation time step, the selected starting point is placed without any bias, and the position of the selected starting point and the existence probability of the selected starting point are And a start point arrangement process for storing the stored start point as the stored start point.

  According to this invention, since the next existence probability distribution is calculated using the start point in consideration of the diffusion from the estimated position of the object, it is possible to calculate the existence probability distribution in consideration of diffusion in each calculation. In addition, it is possible to prevent a decrease in the execution speed of the simulation by limiting the number of calculation times to a certain number.

1 is a block diagram illustrating a configuration of a calculation device that performs an object movement calculation according to Embodiment 1. FIG. It is a figure which shows the processing flow of the estimated position calculation in the calculation apparatus by Embodiment 1. FIG. It is a figure explaining the breadth of the spreading | diffusion effect in the estimated position calculation by Embodiment 1. FIG.

Embodiment 1 FIG.
Hereinafter, a calculation apparatus for calculating the movement of an object according to the first embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a diagram illustrating a configuration of a calculation apparatus that performs an object movement calculation according to the first embodiment. FIG. 2 is a diagram showing a processing flow of estimated position calculation in the calculation apparatus according to the first embodiment. The calculation device according to the first embodiment is used to search for a target marine object (hereinafter referred to as an object) such as a drifting ship or a drifter, and after the movement at each time for each calculation time step by drift current calculation. The estimated position and existence probability distribution of the target object are calculated, and the movement of the object is calculated.

  First, the configuration of a calculation apparatus 101 that performs object movement calculation according to the first embodiment will be described with reference to FIG. The calculation device 101 includes an object movement calculation unit 102, a start point creation unit 103, and an existence probability calculation unit 104. The object movement calculation unit 102 receives weather data 105 given from the outside in advance, and holds the inputted weather data 105. The start point creation unit 103 holds start point information 106 composed of the estimated position and existence probability of the object at each time. The existence probability calculation unit 104 holds the existence probability distribution 107 of the target object at each time, and outputs it to the outside after the calculation is completed. The object movement calculation unit 102 calculates the distribution of the estimated position of the object after movement at each time based on the weather data 105 and the start point information 106 from the start point creation unit 103. Based on the estimated position distribution of the object calculated by the object movement calculation unit 102 and the start point information 106 from the start point creation unit 103, the existence likelihood calculation unit 104 is a target object existence probability distribution 107 at each time. Is calculated.

Subsequently, a processing procedure of estimated position calculation in object movement calculation will be described with reference to FIG.
The estimated position calculation processing procedure includes an estimated position calculation process 201, a diffusion effect spread calculation process 202, an existence probability calculation process 203, an existence probability calculation determination process 204, an existence probability calculation process 205 for all regions, and an end. The process includes a determination process 206 and a next start point arrangement process 207.

  First, in the estimated position calculation processing 201, the object movement calculation unit 102 performs target object information such as position, time, and drift resistance, and wind data such as ocean current data, wave data, and tidal current data. On the basis of the meteorological data 105 including the oceanographic data, various flows on the sea such as ocean currents, blowing currents, waves, tidal currents and wind pressure flows are calculated.

Next, the object movement calculation unit 102, in the estimated position calculation process 201, sums the vectors of various flows at the logical time t _n (n is an integer equal to or greater than 1), that is, an ocean current vector + inflow current vector + tidal current vector. + The wind pressure flow vector is calculated, and the target at the next logical time t _{n + 1} at the preset calculation time step Δt is calculated for a plurality of start points calculated at the logical time t _n stored in the start point information 106. The distribution of the estimated position of the object to be calculated is calculated. The calculation method of the distribution of the estimated position of the target object is shown in, for example, Patent Document 1 and Non-Patent Document 1, and detailed description thereof is omitted here.

Based on the estimated position distribution calculated in the estimated position calculation processing 201, the start point creation unit 103 uses the processing result of the existence calculation unit 104 to further calculate the logical time t _{n + 2} at the next calculation time step size. The next starting point for calculating the estimated position at (= t _{n + 1} + Δt) is determined.

  The existence probability calculation unit 104 obtains the presence probability distribution of the target object at the time of the output step size time from the estimated position calculation result in the diffusion effect spread calculation processing 202 and the existence probability calculation processing 203 described later. .

  Further, the existence probable calculation unit 104 repeats the above processing 201, 202, and 203 a plurality of times until it is determined that the existence probabilities from all starting points have been calculated in the existence probabilistic determination process 204. Calculate the probability distribution of the target object at the time of the output step size time of the calculation result. Thereafter, in the calculation process 205 of the existence probability of all the areas, the existence probability distribution 107 is stored. The above series of processing is repeated until it is determined in the end determination processing 206 that the calculation end time has passed.

Here, the start point creation unit 103 uses the above processing 202 to 205 by the calculation unit 104 of the existence probability to calculate the start point of the estimated position calculation of the target object at the logical time t _{n + 2} at the next calculation time step size. Calculate according to the following procedure.

First, after calculating the estimated position distribution at the next logical time t _{n + 1} in the estimated position calculation process 201, the diffusion error spread calculation process 202 calculates the total error obtained from the ocean current, blowing current, and wind pressure flow errors, Based on the wind pressure declination, the spread of the diffusion effect, which is a distribution of estimated positions where the existence probability of the estimated position of the object is greater than or equal to a predetermined value, is obtained. The total error d is calculated from, for example, the square root of the sum of the squares of the error d _c caused by the ocean current, the error d _s caused by the blowing flow, and the error d _w caused by the wind pressure flow as shown in Equation 1.

A method for calculating the spread of the diffusion effect in the diffusion effect spread calculation process 202 will be described with reference to FIG. Since the wind pressure flow is biased to the right or left toward the leeward side, all the target objects are calculated by drifting from a single starting point 301 and two estimated positions 302 and estimated points serving as reference points. The position 303 is calculated as the estimated position distribution. Here, reference numeral 304 in FIG. 3 denotes a vector sum of various flows, reference numeral 305 denotes a wind pressure flow vector, and reference numeral 306 denotes a wind pressure deflection angle. The diffusion effect spread 309 considering the total error 307 and the wind pressure declination 306 is a circle with a radius D (hereinafter, circle P) centered on the midpoint 308 of the line segment connecting the two estimated positions of the wind pressure flow. . The radius D is calculated based on the wind pressure flow velocity V _w , the total error d, and the wind pressure deviation angle θ as shown in, for example, Equation 2.

Subsequently, the calculation process 203 is executed for the existence probability giving the existence probability of the estimated position of the object. The existence probability of the target object in the circle P of the diffusion effect spread 309 considering the total error and the wind pressure declination is highest at the center of the circle P of the diffusion effect spread 309, and is 50% on the circle P. According to the probability f having a normal distribution (that is, the closer to the center of the circle P, the higher the probability of existence). Incidentally, the circle P is in contact with a circle where the existence probability of the estimated position 302 is 50% and a circle where the existence probability of the estimated position 303 is 50%. In the calculation process 204 of existence, according to the probability f, the inside of the circle P of the diffusion effect spread 309 is divided into meshes (square mesh-like calculation grids), and the target object existence probabilities in each mesh (each calculation grid) Is calculated. Further, from the stored data of the start point information 106, the next logical time is determined from the data of each start point including the estimated position and existence probability of the object at each logical time t _k (k = 1, 2,..., N). Data serving as starting points for calculating the estimated position at t _n is extracted. Further, the product of the existence probability of the object in the region including the start points of the estimated position extraction (circle P obtained at the previous logical time) and the existence probability of each mesh (each calculation grid) The object existence probability reflecting the existence probability of the starting point is calculated for each mesh in the circle P of the diffusion effect spread 309.

Next, after determining that the existence probabilities of the target objects from all starting points in the calculation time increment are calculated in the existence probabilities determination processing 204, the same time is determined in the existence probabilities calculation processing 205 of all regions. Existence probabilities from all start points as base points at (time t _{n + 1} ) are overlapped, and the existence probabilities distribution of all regions is calculated. Here, when a certain region is included in the circle P of the probability distribution of existence from a plurality of start points, the probability of existence of the region is calculated as the sum of the probability of existence calculated from each start point. The existence probabilities calculated by weighting in this way are stored as the existence probable distribution 107.

If it is determined in the end determination process 206 that the calculation end time has not elapsed, the existence from all the start points at the logical time t _{n + 1} stored in the existence probable distribution 107 in the next start point arrangement process 207 The start point (next start point) of the logical time t _{n + 2} at the next calculation step size is arranged without deviation in the region within the circle P of the probability distribution. Here, the number m (m is a positive integer of 2 or more) of the next start point to be rearranged is set in advance from the outside, and the mesh points with the highest probability of existence are arranged in descending order, and the probability of existence is the most. A mesh point within the mth from a high mesh point is selected as a start point and arranged. On the other hand, the m + 1th mesh point from the mesh point with a high probability of existence is regarded as having a relatively low probability of existence, and is not selected and arranged as the next start point. In addition, when the next start point is arranged, the target object existence probability (start point existence probability) in the region is registered in the start point information 106 in association with the next start point. The next start point registered in the start point information 106 and its existence probability are used to calculate the existence probability of each region after calculating the circle of the existence probability distribution at the logical time t _{n + 3} at the next calculation time step size. .

From the Nth time (logical time t _{N, where} N is a positive integer equal to or greater than 2), the existence probabilities of each region are obtained by performing the same calculation starting from each point arranged at the (N−1) th time. . When it is determined in the determination process 206 that the calculation end time has passed, the calculation is ended, and the existence probability distribution at each time is output.

As described above, the estimated position at the logical time t _{n + 1 and} the estimated position at the logical time t _{n + 2} are calculated from the start points arranged without deviation in the circle P of the diffusion effect spread 309 calculated at the logical time t _n . Since the starting point serving as the base point is calculated, it is possible to calculate the existence probable distribution in consideration of the spread of the diffusion effect calculated at the logical time t _n at each calculation step size. This can improve the accuracy of the existence probability distribution.

  Also, considering the existence probability of the starting point itself based on the existence probability, the estimated position calculated from the starting point with the highest existence probability is assumed to have a higher existence probability, and the existence probability distribution of the target object in each area is calculated. As a result, the accuracy of the existence probability distribution can be further improved.

  In addition, since the number m of starting points to be arranged in each calculation step size is made constant, the calculation amount by the number of calculations becomes constant, and the number of calculations can be limited to a certain number, so that the simulation speed is not reduced. Can be calculated.

  As described above, the calculation apparatus according to the first embodiment performs the drift calculation for predicting the movement trajectory of marine objects such as drifting ships and drifters based on weather data such as wind image data and sea condition data, and calculates the calculation time. Presence probability of the estimated position of the object based on the estimated position calculation processing 201 for calculating the estimated position of the object after movement for each step size and the flow error, wind pressure flow velocity, and wind pressure declination obtained from weather data A diffusion effect spread calculation process 202 for obtaining a diffusion region having a predetermined value or more, and a plurality of calculation grids are provided in the diffusion region, and the existence probabilities of the estimated positions of the objects in each grid are calculated for each starting point, In the diffusion region of the estimated position of the object from all the start points, the calculation process 203 of the existence probability that calculates the existence probability of the object by superimposing the existence probabilities for each start point stored at the time, By selecting a certain number of points with the highest probability of existence of the object obtained in the calculation process of the current count as the start point at the next calculation time step, the selected start point is placed without bias and the selection is made And a start point arrangement process 207 for associating the position of the selected start point with the existence probability of the selected start point and storing it as the stored start point.

  Thereby, using the start point considering diffusion from the estimated position of the object and calculating the next existence probability distribution, it is possible to calculate the existence probability distribution considering diffusion in each calculation, By limiting the number of calculations to a certain number, it is possible to prevent a decrease in the simulation execution speed.

  DESCRIPTION OF SYMBOLS 101 Calculation apparatus, 102 Object movement calculation part, 103 Start point creation part, 104 Presence existence calculation part, 105 Weather data, 106 Start point information, 107 Presence probability distribution, 201 Estimated position calculation process, 202 Diffuse effect spread calculation Processing, 203 Presence calculation processing, 204 Presence determination processing, 205 Presence calculation processing of all regions, 206 End determination processing, 207 Next start point placement processing.

Claims (1)

Based on meteorological data such as wind and sea data, drift calculation that predicts the movement trajectory of drifting ships, drifters, and other marine objects is performed, and the estimated position of the moved object is calculated for each calculation time step. Estimated position calculation processing;
Based on flow error obtained from weather data, wind pressure flow velocity, and wind pressure declination, a calculation process of the spread of the diffusion effect to obtain a diffusion region where the existence probability of the estimated position of the object is equal to or greater than a predetermined value,
Multiple calculation grids are provided in the above diffusion area, the existence probabilities of the estimated positions of the objects in each grid are calculated for each start point, and the existence probabilities for each start point stored at each time are overlapped to calculate the object existence probabilities The calculation process of the existence probability to calculate
Within a diffusion region of the estimated positions of the object from all the start points, a certain number of points having the highest object existence probability obtained by the calculation process of the existence probability are calculated by the following calculation time step size. The selected starting point is arranged without bias by selecting it as the starting point, and the position of the selected starting point is associated with the existence probability of the selected starting point, and the starting point arrangement processing for saving as the saved starting point;
A computing device with

Images