US-20260125439-A1 - VEGETATION MANAGEMENT SYSTEM AND VEGETATION MANAGEMENT METHOD

US20260125439A1US 20260125439 A1US20260125439 A1US 20260125439A1US-20260125439-A1

Abstract

A vegetation management system includes a vegetation classification unit that classifies vegetation photographed in remote sensing data, a long-term change prediction unit that predicts, based on a classification result of the vegetation and the remote sensing data, long-term wide-area fluctuation that is range fluctuation of the vegetation in a predetermined long-term time sequence, a short-term change prediction unit that predicts, based on the classification result of the vegetation and the remote sensing data, short-term wide-area fluctuation that is range fluctuation of the vegetation in a predetermined short-term time sequence, a risk determination unit that determines, based on a prediction result by long-term change prediction unit and a prediction result by short-term change prediction unit, a risk due to contact of the vegetation and a facility, and a visualization unit that visualizes a determination result of the risk.

Inventors

Yu Zhao

Assignees

HITACHI ENERGY LTD

Dates

Publication Date: 20260507
Application Date: 20260102
Priority Date: 20210819

Claims (14)

1 . A method, for preventing contact between vegetation and a power facility, that is computationally scalable for a wide geographical area, the method comprising using at least one hardware processor to: train a tree-height estimation model using a value of a digital surface model as an objective variable and using spectral data, meteorological data, altitude data, and vegetation information data as explanatory variables; receive two-dimensional image data obtained by a remote sensing sensor that observes a ground surface from above; apply a mask to the two-dimensional image data to extract a target region from the two-dimensional image data; generate a vegetation classification map from the target region based on spectra feature values of the target region, wherein the vegetation classification map represents a classification of each type of vegetation in the target region; generate a short-term vegetation wide-area map based on the vegetation classification map and a sequence of past two-dimensional image data obtained by the remote sensing sensor for a plurality of past short-term periods, wherein the short-term vegetation wide-area map represents a predicted fluctuation in area of the vegetation in the target region for at least one future short-term period; generate a short-term tree-height map by applying the tree-height estimation model to the short-term vegetation wide-area map, wherein the short-term tree-height map represents a predicted fluctuation in height of the vegetation in the target region for the at least one future short-term period; generate a long-term vegetation wide-area map based on the vegetation classification map and a sequence of past two-dimensional image data obtained by the remote sensing sensor for a plurality of past long-term periods, wherein the long-term vegetation wide-area map represents the predicted fluctuation in area of the vegetation in the target region for at least one future long-term period; generate a long-term tree-height map by applying the tree-height estimation model to the long-term vegetation wide-area map, wherein the long-term tree-height map represents a predicted fluctuation in height of the vegetation in the target region for the at least one future long-term period; determine one or more growth risk spots between the vegetation in the target region and one or more power facilities in the target region, in three dimensions, by comparing boundaries of the vegetation and the one or more power facilities in a first dimension and second dimension using the short-term and long-term vegetation wide-area maps, and in a third dimension using the short-term and long-term tree-height maps; determine one or more contact risk spots from the one or more growth risk spots, wherein each of the one or more contact risk spots represents a growth risk spot at which the vegetation is likely to contact the one or more power facilities; at only each of the one or more contact risk spots, perform a damage simulation to determine a damage risk at the contact risk spot; and generate a visualization of the damage risk at each of the one or more contact risk spots.
2 . The method of claim 1 , wherein the tree height estimation model is a random forest machine-learning model.
3 . The method of claim 1 , wherein the remote sensing sensor comprises a satellite, and wherein the two-dimensional image data comprise a photograph captured by the satellite.
4 . The method of claim 1 , wherein each short-term period is no more than three months.
5 . The method of claim 1 , wherein each long-term period is at least one year.
6 . The method of claim 1 , wherein the visualization comprises a map visualization of one or more of the short-term vegetation wide-area map, the short-term tree-height map, the long-term vegetation wide-area map, or the long-term tree-height map.
7 . The method of claim 6 , wherein the map visualization comprises moving images at a designated time resolution during a designated time interval that includes one or both of the at least one future short-term period or the at least one future long-term period.
8 . The method of claim 7 , wherein the moving images are two-dimensional images.
9 . The method of claim 7 , wherein the moving images are three-dimensional images.
10 . The method of claim 1 , further comprising using the at least one hardware processor to, based on the damage risk at each of the one or more contact risk spots, determine and output an instruction content that specifies a schedule for two or more maintenance places and a route between the two or more maintenance places, wherein at least one of the two or more maintenance places represents one of the one or more contact risk spots.
11 . The method of claim 10 , wherein at least one of the two or more maintenance places represents a material warehouse.
12 . The method of claim 1 , wherein one or both of the short-term vegetation wide-area map or the long-term vegetation wide-area map are generated by a neural network.
13 . A system comprising: at least one hardware processor; and software configured to, when executed by the at least one hardware processor, perform the method of claim 1 .
14 . A non-transitory computer-readable medium having instructions stored thereon, wherein the instructions, when executed by a processor, cause the processor to perform the method of claim 1 .

Description

CROSS-REFERENCE TO RELATED APPLICATIONS The present application is a Divisional Application of U.S. patent application Ser. No. 18/684,631, filed Feb. 16, 2024, which is a national stage entry of International Patent App. No. PCT/JP2022/012910, filed on Mar. 18, 2022, which claims priority to Japanese Patent App. No. 2021-134011, filed on Aug. 19, 2021, which are all hereby incorporated herein by reference as if set forth in full. TECHNICAL FIELD The present invention relates to a vegetation management system and a vegetation management method and is suitably applied to a vegetation management system and a vegetation management method for supporting maintenance work for a power facility against contact of vegetation using measurement information of remote sensing. BACKGROUND ART In conventional power facility maintenance work, routes on which power facilities such as distribution lines and power transmission lines are disposed have been periodically investigated by manpower and work such as removal of tree branches and use of a herbicide has been performed for places where problems such as contact of trees with the power facilities are likely to occur. In contrast, in recent years, an approach for automating investigation has been advanced because of labor shortage and the like. In this approach for the automation of the investigation work, a remote sensing technology that can remotely monitor the power facilities and trees (vegetation) considering that most of the power facilities such as the distribution lines and the power transmission lines are installed in places where it is difficult for people to enter such as mountainous areas has been attracting attention. As representative means of the remote sensing technology, there is use of an artificial satellite, an airplane, a drone, and the like. Further, as a method of determining contact of vegetation and a facility, research and development of vegetation contact determination by three-dimensional measurement utilizing a LiDAR (Light Detection and Ranging) sensor has been advanced. For example, Patent Literature 1 discloses a system that analyzes growth of plants based on remote sensing images photographed by the remote sensing technology. CITATION LIST Patent Literature PTL 1: Japanese Patent Laying-Open No. 2016-123369 SUMMARY OF INVENTION Technical Problem However, in the related art, when contact of a facility and vegetation is analyzed using remote sensing data, it is necessary to perform vegetation contact determination by three-dimensional measurement utilizing a remote sensing sensor such as a LiDAR sensor. There is a problem in that cost extremely increases because a photographing range is expanded and photographing is performed frequently in order to perform accurate determination. In order to support a user in visualization and intuitive operation, conversion of the facility and the vegetation into three dimensions is an indispensable process. Therefore, a large amount of heterogeneous and time sequence geographical information is necessary, which is also a factor of the increase in cost. The present invention has been devised in view of the above points and proposes a vegetation management system and a vegetation management method capable of suppressing an increase in cost and accurately performing an analysis of a risk that vegetation and a facility come into contact. Solution to Problem In order to solve such a problem, the present invention provides a vegetation management system that analyzes, using remote sensing data obtained by photographing, with remote sensing, a facility and vegetation that are analysis targets, a risk that the vegetation comes into contact with the facility, the vegetation management system including: a vegetation classification unit that classifies the vegetation photographed in the remote sensing data; a long-term change prediction unit that predicts, based on a classification result of the vegetation by the vegetation classification unit and the remote sensing data, long-term wide-area fluctuation that is range fluctuation of the vegetation in a predetermined long-term time sequence; a short-term change prediction unit that predicts, based on the classification result of the vegetation by the vegetation classification unit and the remote sensing data, short-term wide-area fluctuation that is range fluctuation of the vegetation in a predetermined short-term time sequence; a risk determination unit that determines, based on a prediction result by the long-term change prediction unit and a prediction result by the short-term change prediction unit, a risk due to contact of the vegetation and the facility; and a visualization unit that visualizes a determination result of the risk by the risk determination unit. In order to solve such a problem, the present invention provides a vegetation management method by a vegetation management system that analyzes, using remote sensing data obtained by photographing, with remo