TWM646347U - Computing device for predicting lightning strike - Google Patents

Abstract

A computing device for predicting lightning strikes. The computing device includes a storage module and a processing module. The storage module stores a lightning prediction model for predicting a lightning strike prediction result related to an area. The processing module The processing module obtains a plurality of cloud discharge data related to cloud discharges that occur in multiple different areas extending from the region and occur in a past time segment earlier than a time point to be predicted. According to each cloud discharge data, the corresponding number of cloud discharges is obtained, and the total number of cloud discharges is calculated based on the number of all cloud discharges. The processing module calculates the total number of cloud discharges based on the number of all cloud discharges and the number of cloud discharges. The total number of medium discharges, and the lightning intensity prediction model is used to obtain the lightning prediction results related to the area at the time point to be predicted.

Description

Computing device for predicting lightning strikes

The present invention relates to a computing device for prediction, in particular to a computing device for predicting lightning strikes.

There are two types of lightning. The first is discharge in the cloud, that is, the air temperature drops above the freezing point, and the liquid water in the cloud turns into ice crystals and supercooled water droplets. Due to the different density of the air, air convection will occur, and these water droplets or ice crystals will Charges will be generated during friction and collision. If there are two strong enough opposite potentials in the cloud, the positively charged area will discharge like the negatively charged area, thereby producing intra-cloud lightning or inter-cloud lightning; the second is cloud pairing. Ground lightning is a lightning strike transmitted from the clouds to the ground. It is a threat to human life and property. If it hits the human body, the water in the body will evaporate instantly, and it can disrupt the heartbeat and kill people. If it hits the electricity The transmission lines set up by the company may damage the lines and the electronic equipment connected to the lines, causing property losses. Due to the different voltages of the transmission lines set up by the power companies, the lightning intensity they can withstand is also different. If lightning strikes can be accurately predicted Levels can provide early warnings for lines with different voltages so that manufacturers or on-site personnel can prepare in advance to reduce losses. Therefore, a solution must be proposed to avoid possible property losses caused by lightning strikes.

Therefore, the purpose of the present invention is to provide a computing device that can automatically predict lightning intensity.

Therefore, the new computing device for predicting lightning strikes includes a storage module and a processing module electrically connected to the storage module.

The storage module is used to store a lightning prediction model for predicting a lightning prediction result related to a region. The lightning prediction result indicates that a first-level lightning strike and a second-level lightning strike will occur in the area at a prediction time point. , and one of the third-level lightning strikes.

The processing module is used to obtain a plurality of cloud discharges related to cloud discharges that occur in multiple different regions extending from the region and occur in a past time segment earlier than a time point to be predicted. data, the processing module obtains the corresponding number of cloud discharges based on each cloud discharge data, and counts the total number of cloud discharges based on all cloud discharges. The processing module calculates the total number of cloud discharges based on all cloud discharge data. Based on the number of discharges and the total number of discharges in the cloud, the lightning intensity prediction model is used to obtain the lightning prediction results related to the region at the time point to be predicted.

The function of this new model is to obtain the cloud discharge data through the processing module, and obtain the corresponding number of cloud discharges and the total number of cloud discharges based on each cloud discharge data, and obtain the cloud discharge data according to all cloud discharge data. The number of medium discharges and the total number of discharges in the cloud use the lightning intensity prediction model to obtain the lightning prediction results related to the area to automatically predict the lightning intensity.

Referring to Figure 1, a computing device 2 is an embodiment of the novel computing device for predicting lightning strikes. The computing device 2 is connected to a wind vector measuring device 3 for measuring wind direction and a wind vector measuring device for collecting relevant information through a communication network 1. Lightning collection device 4 for cloud discharge data in cloud discharge. The computing device 2 includes a storage module 23, a communication module 21, and a processing module electrically connected to the storage module 23 and the communication module 21. Group 22. The implementation form of the computing device 2 is, for example, a personal computer, a server or a cloud host, but is not limited thereto. The implementation form of the wind direction measuring device 3 is, for example, a weather station, but is not limited thereto.

The storage module 23 is used to store a lightning prediction model for predicting a lightning prediction result related to a region, and a plurality of lightning data corresponding to multiple different observation time intervals and related to the region. The lightning prediction result indicates that one of a first-level lightning strike, a second-level lightning strike, and a third-level lightning strike will occur in the area at a predicted time point. Each piece of lightning strike data includes the lightning strike of the area at an observation time point. A lightning strike intensity, related to the observed wind direction in the area at a past observation time point earlier than the observation time point, and multiple lightning strikes related to the occurrence in multiple different areas extending from the area, and occurred in Observed cloud discharge data in the past time period earlier than the observation time point. Wherein, the past wind direction observation time point is the starting time point of the past time section earlier than the observation time point, but is not limited to this.

For example, if the area is Taichung Science Park, the different areas (see Figure 6) can be 0~3 kilometers, 3~5 kilometers, 5~10 kilometers, and 10~ extending outward from Taichung Science Park. 20 kilometers, taking the Taichung Science Park corresponding to the observation time interval from 1:00 to 2:00 pm on August 6, 2016 as an example. If the observation time point is 2:00 pm on August 6, 2016, then the The past time range is from 1:00 pm to 1:30 pm on August 6, 2016, and the observed wind direction is 1:00 pm on August 6, 2016 (the time point of the past wind direction observation) relative to the Taichung Science Park wind direction, but not limited to this; that is, if the time of lightning strike in Taichung Science Park is 2:00 pm on August 6, 2016 (the observation time), then the observed cloud discharge data are related to The Taichung Science Park observed clouds at 0~3 km, 3~5 km, 5~10 km, and 10~20 km from 1:00 to 1:30 pm on August 6, 2016 (the past time period) Medium discharge, but not limited to this; and the current range of the first-level lightning strike is 0~10kA, the current range of the second-level lightning strike is 10~20kA, and the current range of the third-level lightning strike is greater than 20kA, but Not limited to this.

The operation of each component between the computing device 2 , the wind vector measuring device 3 and the lightning collecting device 4 will be explained below through a lightning prediction method, and a model building procedure for establishing the lightning prediction model will be included in sequence. , and a lightning prediction program used to predict lightning strikes in the area.

Referring to Figure 1 and Figure 2, the model establishment procedure includes steps 61 to 66.

In step 61, the processing module 22 obtains a plurality of mine data to be trained based on the lightning intensity and a lightning intensity threshold (for example, 5kA) of the lightning data stored in the storage module 23, each of which is to be trained. The lightning intensity of the training lightning data is higher than the lightning intensity threshold.

In step 62, for each piece of mine data to be trained, the processing module 22 obtains a pair of mine marks of the mine data to be trained based on the mine intensity of the mine data to be trained. The mine marks include the first level mine, One of the second-level lightning strike and the third-level lightning strike.

In step 63, for each minefield data to be trained, the processing module 22 filters each observed cloud discharge data in the minefield data to be trained according to the region and the observation wind direction to obtain a corresponding filtered data. Observed discharge data in clouds. It is worth mentioning that in other embodiments of the present invention, the filtering action of step 63 can also be omitted, and the model can be directly established based on the observed cloud discharge data in the lightning data to be trained. This is not a limitation. .

Referring to Figures 1 and 3, this step 63 also includes the following sub-steps.

In step 631, for each minefield data to be trained, the processing module 22 selects the area corresponding to the discharge data in each observed cloud in the minefield data to be trained based on the area and the observed wind direction corresponding to the minefield data to be trained. range to obtain a corresponding filtered observation range, in which the processing module 22 expands the observation wind direction corresponding to the to-be-trained mine data passing through the area to both sides by a preset angle to obtain two extended observation wind directions, and according to the Region, and the area range corresponding to each observed cloud discharge data in the lightning strike data to be trained by the two extended observation wind directions defines the corresponding screening observation range.

In step 632, for each minefield data to be trained, the processing module 22 filters out the clouds within the filtered observation range according to the filtered observation range corresponding to the discharge data in each observed cloud in the minefield data to be trained. medium discharge to obtain a corresponding filtered observed cloud discharge data.

For example, if the observed wind direction is blowing northwest to the area, and the preset angle is, for example, 30 degrees, the processing module 22 will expand to both sides in the opposite direction of the observed wind direction 5 (see Figure 7) of the area. The preset angle is used to obtain two extended observation wind directions 51 (see Figure 7). Then, the region and the two extended observation wind directions are located in the area corresponding to the discharge data in each observation cloud in the minefield data to be trained. (0~3 km, 3~5 km, 5~10 km, and 10~20 km) define the corresponding filtered observation range (0~3 km filtered observation range 52, 3~5 km filtered observation range 53, 5 ~10 km screening observation range 54, 10~20 km screening observation range 55) (see Figure 7).

In step 64, for each minefield data to be trained, the processing module 22 obtains a plurality of discharge numbers in the observation cloud based on the filtered discharge data in the observation cloud in the minefield data to be trained, and based on the discharge data in the observation cloud, The number of discharges in all observation clouds in the lightning data to be trained is calculated to calculate the total number of discharges in an observation cloud.

Continuing the example in Figure 7 above, the number of discharges in these observation clouds is respectively 52 in the 0~3 km filtered observation range, 53 in the 3~5 km filtered observation range, 54 in the 5~10 km filtered observation range, and 10~20 km in the filtered observation range. The number of discharges in clouds within the observation range of 55 is screened, and the total number of discharges in the observed clouds is the sum of the number of discharges in the observed clouds, but is not limited to this.

In step 65, for each piece of lightning data to be trained, the processing module 22 uses the number of discharges in the observation cloud, the total number of discharges in the observation cloud, and the corresponding lightning marks as a set of training data.

In step 66, the processing module 22 uses a machine learning algorithm to establish the lightning prediction model based on the training data. Among them, the machine learning algorithm is a General Regression Neural Networks (GRNN) calculation model.

Referring to Figures 1 and 4, the lightning prediction program includes steps 71 to 74.

In step 71 , the processing module 22 obtains from the lightning collection device 4 through the communication module 21 through the communication network 1 a plurality of transactions related to occurrences in the different areas extending from the region, and occurrences Cloud discharge data in a past time segment earlier than a time point to be predicted, and the past wind direction of the area measured by the wind direction measuring device 3 earlier than the predicted time point One point in time is the wind direction. Wherein, the past wind direction time point is the starting time point of the past time section earlier than the predicted time point, but is not limited to this.

For example, if the time point to be predicted is 3:00 pm on a certain day, the past time period earlier than the time point to be predicted is 2:00~2:30 pm on the same day, and the wind direction is the same day measured at 2:00 pm (the past wind direction time point), but is not limited to this.

In step 72, for each piece of cloud discharge data, the processing module 22 obtains filtered cloud discharge data based on the region, the wind direction and the cloud discharge data.

Referring to Figure 1 and Figure 5, this step 72 also includes the following sub-steps.

In step 721, for each piece of cloud discharge data, the processing module 22 obtains a filtering range from the area corresponding to the cloud discharge data according to the region and the wind direction. For each piece of cloud discharge data, the processing module 22 expands the opposite direction of the wind direction passing through the region to both sides by a preset angle to obtain two expanded wind directions, and based on the region and the two expanded wind directions, the The area range corresponding to the discharge data in the cloud defines the filtering range.

In step 722, for each piece of cloud discharge data, the processing module 22 filters out the cloud discharges within the filtering range according to the filtering range corresponding to the cloud discharge data to obtain the filtered cloud discharge data. .

In step 73 , the processing module 22 obtains the number of cloud discharges corresponding to each filtered cloud discharge data, and counts the total number of cloud discharges based on all the cloud discharge numbers.

In step 74, the processing module 22 uses the lightning intensity prediction model stored in the storage module 23 to obtain the lightning intensity prediction model related to the region at the time to be predicted based on the number of all cloud discharges and the total number of cloud discharges. Click on the thunder prediction results.

To sum up, the new computing device for predicting lightning strikes obtains the cloud discharge data and the wind direction through the processing module 22 after establishing the lightning strike prediction model, and based on each cloud discharge data and The wind direction obtains the number of discharges in the cloud and the total number of discharges in the cloud within the corresponding screening range, and based on the number of discharges in the cloud, the total number of discharges in the cloud and the wind direction, the lightning intensity prediction model is used to obtain The lightning prediction results related to the area at the time point to be predicted can automatically predict the lightning situation in the area at the time point to be predicted, so as to warn in advance and avoid lives and items from being harmed by lightning strikes, so the cost can indeed be achieved. A new type of purpose.

However, the above are only examples of the present invention, and should not be used to limit the scope of the present invention. All simple equivalent changes and modifications made based on the patent scope of the present invention and the content of the patent specification are still within the scope of the present invention. Within the scope covered by this new patent.

1: Communication network 2:Computing device 21:Communication module 22: Processing module 23:Storage module 3: Wind direction measurement device 4: Lightning collection device 5: Observe the wind direction 51:Extended observation of wind direction 52: 0~3 kilometers filtered observation range 53: 3~5 kilometers screening observation range 54:5~10km filtered observation range 55: 10~20 kilometers screening observation range 61~66: Steps 631~632: Steps 71~74: Steps 721~722: Steps

Other features and functions of the present invention will be clearly presented in the embodiments with reference to the drawings, in which: Figure 1 is a block diagram illustrating an embodiment of the new computing device for predicting lightning strikes; Figure 2 is a flow chart illustrating a model training procedure for implementing a lightning prediction method using an embodiment of the novel computing device for predicting lightning strike; Figure 3 is a flow chart illustrating the detailed process of using an embodiment of the new computing device for predicting lightning strike to implement the model training program to obtain multiple filtered observation cloud discharge data based on multiple lightning strike data to be trained; Figure 4 is a flow chart illustrating the implementation of a lightning prediction procedure of the lightning prediction method using an embodiment of the novel computing device for predicting lightning strikes; Figure 5 is a flow chart illustrating the detailed process of using an embodiment of the novel computing device for predicting lightning strike to implement the lightning strike prediction program to obtain multiple filtered cloud discharge data based on multiple cloud discharge data; Figure 6 is a schematic diagram illustrating discharges in multiple different areas and multiple observed clouds extending over an area; and Figure 7 is a schematic diagram illustrating multiple filtered observation ranges defined according to an observation wind direction.

1: Communication network

2:Computing device

21:Communication module

22: Processing module

23:Storage module

3: Wind direction measurement device

4: Lightning collection device

Claims (8)

A computing device for predicting lightning strikes, including: A storage module for storing a lightning prediction model for predicting a lightning prediction result related to a region. The lightning prediction result indicates that a first-level lightning strike and a second-level lightning strike will occur in the area at a prediction time point. Lightning strikes, and one of the third-level lightning strikes; and a processing module electrically connected to the storage module; Among them, the processing module obtains a plurality of cloud discharges related to cloud discharges that occur in multiple different areas extending from the region and occur in a past time segment earlier than a time point to be predicted. data, the processing module obtains the corresponding number of cloud discharges based on each cloud discharge data, and counts the total number of cloud discharges based on all cloud discharges. The processing module calculates the total number of cloud discharges based on all cloud discharge data. Based on the number of discharges and the total number of discharges in the cloud, the lightning intensity prediction model is used to obtain the lightning prediction results related to the region at the time point to be predicted. The computing device as described in claim 1, wherein the processing module also obtains a wind direction in the region at a past wind direction time point earlier than the predicted time point, and for each cloud discharge data, the processing module calculates the wind direction according to For the region, the wind direction and the cloud discharge data, a filtered cloud discharge data is obtained, and based on the filtered cloud discharge data, the corresponding number of cloud discharges is obtained, and based on all cloud discharge data The number of medium discharges counts the total number of discharges in the cloud. The computing device as described in claim 2, wherein for each piece of cloud discharge data, the processing module obtains a filtering range from the area corresponding to the cloud discharge data based on the region and the wind direction, and based on the cloud discharge data The filtering range corresponding to the medium discharge data is used to filter out the cloud discharges within the filtering range to obtain filtered cloud discharge data. The computing device as described in claim 2, wherein for each cloud discharge data, the processing module expands the reverse direction of the wind direction passing through the area to both sides by a preset angle to obtain two expanded wind directions, and based on The area corresponding to the discharge data in the cloud in the region and the two extended wind directions defines the screening range. The computing device according to claim 1, wherein the storage module is also used to store multiple pieces of lightning data corresponding to multiple different observation time intervals and related to the region, and each piece of lightning data includes an observation time point in the region. A lightning intensity, and a plurality of observation cloud discharge data related to discharges in the observation cloud that occurred in these different areas and occurred in a past time section earlier than the observation time point, the processing module According to the lightning intensity of the lightning data and a lightning intensity threshold, a plurality of mine data to be trained are obtained. The lightning intensity of each mine data to be trained is higher than the lightning intensity threshold. For each mine data to be trained, the lightning intensity is The processing module obtains a pair of lightning marks for the mine data to be trained based on the lightning intensity of the mine data to be trained. The mine marks include one of the first-level lightning, the second-level lightning, and the third-level lightning. Or, obtain the discharge numbers in multiple observation clouds based on the discharge data in the observation clouds in the lightning strike data to be trained, and calculate an observation cloud based on the discharge numbers in all observation clouds in the lightning strike data to be trained. The total number of medium discharges, the module here uses a machine learning algorithm to establish the lightning prediction model based on the training data. The computing device as described in claim 5, wherein each piece of lightning data also includes an observed wind direction related to the area at one of the past wind direction time points earlier than the observation time point, and for each piece of lightning data to be trained, the The processing module filters each observed cloud discharge data in the lightning strike data to be trained according to the region and the observation wind direction to obtain a corresponding filtered observation cloud discharge data, and based on the lightning strike data to be trained, The discharge data in the observation cloud are screened to obtain the number of discharges in the observation cloud, and the total number of discharges in the observation cloud is calculated based on the number of discharges in all observation clouds in the lightning strike data to be trained. The computing device as described in claim 6, wherein for each piece of mine data to be trained, the processing module extracts data from each observation cloud in the mine data to be trained based on the region and observation wind direction corresponding to the mine data to be trained. For the area corresponding to the discharge data, a corresponding screening observation range is obtained, and based on the screening observation range corresponding to the discharge data in each observation cloud in the lightning data to be trained, the clouds within the screening observation range are screened out discharge to obtain a corresponding filtered discharge data in the observed cloud. The computing device as described in claim 6, wherein for each minefield data to be trained, the processing module expands the reverse direction of the observed wind direction corresponding to the minefield data to be trained through the area to both sides by a preset angle. Two extended observation wind directions are obtained, and the corresponding screening observation range is defined based on the region and the area range corresponding to each observation cloud discharge data in the two extended observation wind directions in the lightning strike data to be trained.

Images