Earthquake influence area determination method and device

Abstract

The embodiment of the invention relates to a method and device for judging an earthquake influence area, and the method comprises the steps: extracting various index data in mobile phone signaling data, and carrying out the statistics of the number of devices of each index data; calculating an equipment change rate according to the equipment number; generating a plurality of earthquake influence area distribution data graphs according to the equipment change rates in the plurality of time periods; according to the distribution data graph, selecting an equipment change rate meeting preset fitting precision and an epicentral distance corresponding to the GeoHash grid, and generating a model data set; constructing a pre-training intensity evaluation model based on a random forest regression algorithm, training and testing the pre-training intensity evaluation model through the model data set, and outputting intensity distribution data; performing model fitting precision calculation according to the intensity distribution data, generating a test evaluation result, and determining an optimal intensity evaluation model; and setting an index type and an evaluation time period for training the optimal intensity evaluation model as an optimal index type and an optimal evaluation time period.

Description

Technical Field

[0001] The present invention relates to the technical field of earthquake disaster monitoring, and in particular to a method and device for determining an earthquake-affected area. Background Art

[0002] After an earthquake occurs, rapid identification of the hardest-hit areas is the key to effectively carrying out post-earthquake emergency response and rescue work, improving emergency rescue efficiency, and reducing casualties. Rapid determination of earthquake-affected areas refers to the earthquake emergency work that helps the government quickly and effectively carry out emergency rescue activities after a destructive earthquake occurs, assesses the intensity of the earthquake impact based on multi-source earthquake information and empirical models, and quickly identifies the extent and scope of the disaster in the disaster area.

[0003] There are currently existing assessment methods for determining earthquake-affected areas based on on-site investigations, instrument intensity, intensity attenuation laws, and remote sensing image interpretation. However, these traditional assessment methods are difficult to meet the actual needs of disaster relief work in terms of timeliness or accuracy.

[0004] How to find a data signal that has a good response relationship with the impact of earthquakes and use it to determine the earthquake-affected area in real time / quasi-real time has become an urgent problem to be solved. Summary of the invention

[0005] The purpose of the present invention is to provide a method and device for determining an earthquake-affected area in view of the defects of the prior art, so as to solve the problem that the prior art determination and evaluation methods are not timely or accurate.

[0006] To achieve the above object, the present invention provides a method for determining an earthquake-affected area in a first aspect, comprising:

[0007] Obtain mobile phone signaling data, extract multiple indicator data from the mobile phone signaling data, and count the number of devices collected by each indicator data in each Geohash grid per minute;

[0008] Calculate the equipment change rate of each indicator in multiple time periods based on the number of equipment;

[0009] According to the equipment change rates in the multiple time periods, a plurality of earthquake impact area distribution data graphs are generated by an interpolation algorithm;

[0010] According to the multiple earthquake impact area distribution data maps, the equipment change rate in different time periods and the epicenter distance corresponding to the GeoHash grid that meet the preset fitting accuracy are selected for each indicator, and a model data set for different time periods is generated for each indicator;

[0011] Constructing a pre-trained intensity assessment model based on a random forest regression algorithm, and training the pre-trained intensity assessment model using the first model data in the model data set to generate an intensity assessment model;

[0012] All intensity assessment models are tested using the second model data in the model data set, and intensity distribution data are output; model fitting accuracy is calculated based on the intensity distribution data, a test evaluation result is generated, and an optimal intensity assessment model is determined based on the test evaluation result;

[0013] The indicator type and evaluation time period used for training the optimal intensity assessment model are set as the optimal indicator type and optimal evaluation time period for earthquake intensity assessment.

[0014] Furthermore, the equipment change rate includes a month-on-month change rate and a year-on-year change rate. The month-on-month change rate is the change rate of the number of equipment in the same time period before and after the earthquake on the day of the earthquake, and the year-on-year change rate is the change rate of the number of equipment after the earthquake on the day of the earthquake relative to the same time on the day before the earthquake.

[0015] Furthermore, the multiple indicator data include mobile device indicator data, active base station device indicator data, active Wifi device indicator data and wireless network networking device indicator data.

[0016] Further, the generating of multiple earthquake impact area distribution data graphs by interpolation algorithm according to the equipment change rates in the multiple time periods specifically includes:

[0017] Calculate the average value of the device change rate in the Geohash grid in each time period;

[0018] Multiple earthquake impact area distribution data maps are generated by plotting the average value of the equipment change rate in each time period through an interpolation algorithm.

[0019] Furthermore, the interpolation algorithm is a Kriging interpolation algorithm.

[0020] Furthermore, the test evaluation results include first test evaluation data and second test evaluation data; the first test evaluation data include mean absolute error, mean square error, root mean square error, and determination coefficient, and the second test evaluation data is prediction error rate.

[0021] Further, the model fitting accuracy calculation is performed according to the intensity distribution data to generate a test evaluation result, and the optimal intensity evaluation model is determined according to the test evaluation result, specifically including:

[0022] Calculate the model fitting accuracy according to the intensity distribution data to generate first test evaluation data;

[0023] Selecting all intensity assessment models that meet preset assessment values ​​in the first test assessment data to generate a suboptimal intensity assessment model;

[0024] Generate second test assessment data based on the intensity distribution data of the suboptimal intensity assessment model;

[0025] The suboptimal intensity assessment model with the smallest prediction error rate of the second test assessment data is selected to generate the optimal intensity assessment model.

[0026] Furthermore, the prediction error rate is the percentage of the number of GeoHash grids with misaligned evaluation to the total number of GeoHash grids.

[0027] A second aspect of the present invention provides a device for determining an earthquake-affected area, comprising:

[0028] A first data processing module is used to obtain mobile phone signaling data, extract multiple indicator data from the mobile phone signaling data, and count the number of devices collected by each indicator data in each Geohash grid per minute;

[0029] A second data processing module is used to calculate the equipment change rate of each indicator in multiple time periods according to the number of equipment;

[0030] A third data processing module is used to generate a plurality of earthquake impact area distribution data graphs by an interpolation algorithm according to the equipment change rates in the plurality of time periods;

[0031] A model data set construction module, for selecting, according to the plurality of earthquake impact area distribution data graphs, the equipment change rate in different time periods and the epicenter distance corresponding to the GeoHash grid for each indicator that meets the preset fitting accuracy, and generating model data sets for different time periods for each indicator;

[0032] A model training module, used to construct a pre-trained intensity assessment model based on a random forest regression algorithm, and train the pre-trained intensity assessment model through the first model data in the model data set to generate an intensity assessment model;

[0033] A model testing module, used to test all intensity assessment models through the second model data in the model data set and output intensity distribution data; perform model fitting accuracy calculation according to the intensity distribution data, generate test evaluation results, and determine the optimal intensity assessment model according to the test evaluation results;

[0034] The model input variable setting module is used to set the indicator type and evaluation time period used for training the optimal intensity assessment model as the optimal indicator type and optimal evaluation time period for earthquake intensity assessment.

[0035] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any one of the methods described in the first aspect.

[0036] The method and device for determining the earthquake affected area provided in the embodiment of the present invention generate an optimal intensity assessment model through mobile phone signaling data training, and determine the optimal indicator type and optimal assessment time period of the mobile phone signaling data according to the optimal intensity assessment model. A data signal that has a good response relationship with the earthquake impact is used to determine the earthquake affected area, and the real-time and accuracy of the determination are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 A flowchart of a method for determining an earthquake-affected area provided in Embodiment 1 of the present invention;

[0038] Figure 2 A schematic diagram of minute-by-minute time series changes of four indicator data of the method for determining the earthquake-affected area provided in the first embodiment of the present invention;

[0039] Figure 3 A distribution data diagram of earthquake-affected areas in the method for determining earthquake-affected areas provided in the first embodiment of the present invention;

[0040] Figure 4 An accuracy evaluation diagram of an intensity assessment model of a method for determining an earthquake-affected area provided in the first embodiment of the present invention;

[0041] Figure 5 A comparison diagram of prediction errors of the intensity field assessment model of the method for determining the earthquake affected area provided in the first embodiment of the present invention;

[0042] Figure 6 This is a schematic diagram of the structure of an apparatus for determining an earthquake-affected area provided in Embodiment 2 of the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] The method and device for determining the earthquake impact area provided by the embodiment of the present invention generate an optimal intensity assessment model through mobile phone signaling data training, and determine the optimal indicator type and optimal assessment time period of the mobile phone signaling data according to the optimal intensity assessment model, utilize a data signal with a good response relationship with the earthquake impact to determine the earthquake impact area, and improve the real-time and accuracy of the determination; generate multiple earthquake impact area distribution data maps through an interpolation algorithm to screen out a part of the data to establish a model data set, reduce the number of pre-trained intensity assessment models, and improve the efficiency of determining the optimal intensity assessment model; screen out a unique optimal intensity assessment model by combining model fitting accuracy and prediction error rate, further improving the screening efficiency of the optimal intensity assessment model; and realize the real-time / quasi-real-time earthquake impact area determination efficiently and accurately through the screened optimal intensity assessment model, optimal indicator type and optimal assessment time period.

[0045] Embodiment 1

[0046] Figure 1 This is a flow chart of the method for determining the earthquake affected area provided in the first embodiment of the present invention. Figure 1 , the technical solution of the present invention is described with specific embodiments.

[0047] Step 110, obtain mobile phone signaling data, extract various indicator data from the mobile phone signaling data, and count the number of devices collected in each Geohash grid per minute for each indicator data.

[0048] Specifically, mobile phone signaling data can continuously track the spatial location, status and other information of mobile phone users through base stations, and record the activities of mobile phone users in a relatively complete manner. It has high timeliness, spatial continuity, sufficient data volume and high accuracy. According to the method of determining location information, mobile phone signaling data can be divided into two categories. One category is to determine the location information of mobile phones based on the base stations or satellites (GPS, Beidou, etc.) of telecom operators, which mainly includes call bill data and signaling data provided by telecom operators. The other category is to determine the location information of mobile phones with the help of software development kits (SDK). This type of data is less affected by the distribution of base stations and directly locates the mobile phone with high accuracy.

[0049] In an optional implementation, the mobile phone signaling data is obtained from the SDK. The SDK data is the mobile phone signaling data counted minute by minute in units of GeoHash grids. The GeoHash grid encodes the two-dimensional spatial longitude and latitude data into a string through address encoding. The longer the string, the higher the accuracy of the representation. For example, a GeoHash grid encoded with 5-bit precision represents an area with a range of 4.9km*4.9km, and a GeoHash grid encoded with 6-bit precision represents an area with a range of 1.2km*0.6km. Each GeoHash grid represented by the code can be converted into longitude and latitude coordinates, and the converted longitude and latitude are the coordinates of the center point of the encoded area. The various indicator data in the mobile phone signaling data include mobile device indicator data, active base station device indicator data, active Wifi device indicator data, and wireless network networking device indicator data. Among them, the mobile device indicator data is the number of mobile devices in each GeoHash grid in the regional range per minute, and the location data of mobile devices served are obtained based on multiple positioning methods; the active base station device indicator data is the number of active base stations in each GeoHash grid per unit time according to the base station information scanned and reported regularly by mobile devices; the active Wifi device indicator data is the number of active Wifi hotspots in each GeoHash grid per unit time according to the base station information scanned and reported regularly by mobile devices; the wireless network networking device indicator data is the number of Wifi networked devices in each GeoHash grid per unit time in the statistical area according to the network connection status reported regularly by the networked mobile devices.

[0050] It should be noted that the following steps in this embodiment will be described based on the obtained mobile phone signaling data, which are four types of indicator data: mobile device indicator data, active base station device indicator data, active Wifi device indicator data, and wireless network networking device indicator data.

[0051] Figure 2 This is a schematic diagram of the minute-by-minute time series changes of four indicator data of the method for determining the earthquake affected area provided in the first embodiment of the present invention, such as Figure 2 As shown, the horizontal axis is time, and the vertical axis is the number of mobile devices, active base stations, active Wifi, and wireless network connected devices. The curves are respectively a schematic diagram of the minute-by-minute time series change of the four indicator data on the day of the earthquake, and a schematic diagram of the minute-by-minute time series change of the four indicator data the day before the earthquake. The earthquake occurred at 21:19 on the day of the earthquake. Figure 2It can be seen that the number of devices collected for the four indicators after the earthquake has changed abnormally, and the number of devices collected on the day of the earthquake is generally less than the day before the earthquake. In different intensity zones, the higher the intensity value, the more obvious the abnormal changes in the four indicators. Therefore, when the number of the four devices collected in some GeoHash grids changes dramatically within a certain time slice, it means that a large number of device data cannot be reported normally or the number of devices increases suddenly within the range and time period. It can be inferred that an earthquake or other sudden disasters may occur in the area.

[0052] Step 120, calculating the equipment change rate of each indicator in multiple time periods according to the number of equipment.

[0053] In one possible implementation, the equipment change rate includes a month-on-month change rate and a year-on-year change rate. The month-on-month change rate is the change rate of the number of equipment in the same time period before and after the earthquake on the day of the earthquake, and the year-on-year change rate is the change rate of the number of equipment after the earthquake on the day of the earthquake relative to the same time on the day before the earthquake.

[0054] The month-on-month change rate is shown in the following formula (1):

[0055] R t =( M t -M 1t ) / M 1t (1)

[0056] Among them, R t M is the rate of change of equipment in the GeoHash grid Δt minutes after the earthquake relative to before the earthquake. t M is the number of devices collected within Δt minutes after the earthquake in the GeoHash grid on the day of the earthquake. 1t is the number of devices collected within Δt minutes before the earthquake on the day of the earthquake in the GeoHash grid.

[0057] The year-on-year change rate is shown in the following formula (2):

[0058] R' t =( M' t -M' 1t ) / M' 1t (2)

[0059] Among them, R' t M' is the rate of change of devices in the GeoHash grid at Δt minutes after the earthquake on the day of the earthquake relative to the same time on the day before the earthquake. t M' is the number of devices collected within the GeoHash grid at Δt minutes after the earthquake on the day of the earthquake. 1t It is the number of devices collected at the same time one day before the earthquake in the GeoHash grid.

[0060] If the month-on-month change rate and the year-on-year change rate are negative, the number of devices after the earthquake is less than that before the earthquake. The closer the change rate is to -1, the greater the decrease in the number of devices after the earthquake relative to before the earthquake, which means that the impact of the earthquake is greater. Areas with a change rate of -1 often appear in extremely severely affected areas, and the number of post-earthquake data collection devices caused by factors such as post-disaster infrastructure damage is 0. If the change rate is positive, the number of devices after the earthquake increases relative to before the earthquake. The greater the change rate, the greater the increase in the number of devices collected after the earthquake, that is, the more devices turn from silent to active after the earthquake. This trend is generally in secondary disaster areas that are less affected by the earthquake. If the change rate tends to 0, it means that the number of devices after the earthquake is equivalent to that before the earthquake, and the impact of the earthquake is small or weakened.

[0061] Step 130, generating a plurality of earthquake impact area distribution data graphs by using an interpolation algorithm according to the equipment change rates in a plurality of time periods.

[0062] In a possible implementation manner, step 130 specifically includes steps A1 to A2:

[0063] Step A1, calculate the average value of the device change rate in the Geohash grid in each time period in turn.

[0064] Specifically, the average value of the device change rate in the Geohash grid in each time period Δt is calculated based on the four indicator data. The average value of the month-on-month change rate in the Geohash grid in each time period Δt can be expressed as The average year-on-year change rate in the Geohash grid within each time period Δt can be expressed as

[0065] Step A2: Generate multiple earthquake impact area distribution data maps by using an interpolation algorithm according to the average value of the equipment change rate in each time period. Optionally, the interpolation algorithm is a Kriging interpolation algorithm.

[0066] The Kriging interpolation method is shown in the following formula (3):

[0067]

[0068] in, is the estimated value of the point Z(x0,y0), Z(x i ) is the actual value of point Z(x0,y0), n is the total number of sample points, λ i is the weight.

[0069] Step 140, selecting the equipment change rate in different time periods and the epicenter distance corresponding to the GeoHash grid that meet the preset fitting accuracy for each indicator according to the multiple earthquake impact area distribution data maps, and generating a model data set for different time periods for each indicator.

[0070] In a specific example A, Figure 3 This is a diagram of distribution data of earthquake-affected areas of the method for determining earthquake-affected areas provided in the first embodiment of the present invention, such as Figure 3 As shown, the first row of each indicator is from left to right C 11 -C 15 This is a data map of the earthquake-affected area distribution drawn based on the average value of the month-on-month change rate 1-5 minutes after the earthquake. The second row of each indicator is from left to right C 21 -C 25 This is a data map of the earthquake-affected area distribution drawn based on the average year-on-year change rate 1-5 minutes after the earthquake. Figure 3 The Kriging interpolation algorithm is used, where Gid is based on mobile device indicator data, Status is based on active base station device indicator data, Wifi is based on active Wifi device indicator data, and Logi nmac is based on wireless network networking device indicator data. Figure 3 It can be seen that the four indicators have unclear divisions between the hardest-hit areas and the second hardest-hit areas in the first 1 minute and 2 minutes, and the fitting accuracy cannot be achieved. The average month-on-month change rate and the average year-on-year change rate in the first 1 minute and 2 minutes are eliminated, and the average month-on-month change rate and the average year-on-year change rate in the first 3-5 minutes and the epicenter distance corresponding to the GeoHash grid are selected to generate the model data set of each indicator in the first 3-5 minutes after the earthquake.

[0071] Step 150: construct a pre-trained intensity assessment model based on a random forest regression algorithm, and train the pre-trained intensity assessment model using the first model data in the model data set to generate an intensity assessment model.

[0072] Specifically, the data in the model data set is divided into first model data and second model data in a preset ratio, wherein the first model data is used as a training set for training the pre-trained intensity assessment model, and the second model data is used as a test set for performance testing and evaluation of the trained intensity assessment model. In a possible implementation, the ratio of the first model data is 70%, and the ratio of the second model data is 30%.

[0073] The model built based on the random forest regression algorithm is an integrated learning model composed of multiple decision trees. The final prediction value is obtained by averaging the prediction results of multiple trees, thereby improving the accuracy and stability of the prediction. The intensity assessment model is pre-trained through the first model data training. The model fits the data by continuously splitting and combining decision trees, and finally generates an intensity assessment model.

[0074] Step 160, testing all intensity assessment models through the second model data in the model data set, and outputting intensity distribution data; calculating the model fitting accuracy based on the intensity distribution data, generating test evaluation results, and determining the optimal intensity assessment model based on the test evaluation results.

[0075] Among them, the intensity distribution data can be the intensity zone where the mobile phone signaling data is located, for example, the output of intensity zone VI is represented by 6, and the output of intensity zone VIII is represented by 8.

[0076] The test evaluation results include first test evaluation data and second test evaluation data, which include mean absolute error MAE, mean square error MSE, root mean square error RMSE and determination coefficient R2, and the second test evaluation data is the prediction error rate.

[0077] In a possible implementation manner, step 160 specifically includes steps B1 to B4:

[0078] Step B1, calculate the model fitting accuracy based on the intensity distribution data to generate the first test evaluation data.

[0079] The calculation formula of the first test evaluation data is shown in the following formula (4) to formula (7):

[0080]

[0081] Among them, y i is the actual intensity of the GeoHash grid, is the GeoHash grid prediction intensity, n is the total number of GeoHash grids, is the average value of the actual crack degree of the GeoHash grid.

[0082] Step B2, selecting all intensity assessment models that meet preset assessment values ​​in the first test assessment data to generate a suboptimal intensity assessment model.

[0083] Step B3, generating second test assessment data based on the intensity distribution data of the suboptimal intensity assessment model.

[0084] Specifically, the second test evaluation data is the prediction error rate. The prediction error rate is the percentage of the number of GeoHash grids with evaluation misalignment to the total number of GeoHash grids. The prediction error rate PER calculation formula is shown in formula (8):

[0085]

[0086] Step B4, selecting the suboptimal intensity assessment model with the smallest prediction error rate of the second test assessment data to generate the optimal intensity assessment model.

[0087] Specifically, the prediction accuracy of the model is quantified by the PER value. The lower the PER value, the higher the prediction accuracy of the model, that is, the model's prediction is very close to the actual earthquake intensity, and the model is more conducive to accurately assessing the impact of an earthquake.

[0088] Combined with Example A, in a specific example B, the first test evaluation data is calculated, and the data is the result of the evaluation and calculation of the intensity assessment model trained according to the average value of the month-on-month change rate and the average value of the year-on-year change rate in the first 3, 4, and 5 minutes after the earthquake. As shown in Table 1, 6 models are trained for each indicator. In the model number, the first indicator Gid after the letter C is represented by 1, Status is represented by 2, Wifi is represented by 3, Logi nmac is represented by 4, the second digit after the letter C is represented by 1 for the month-on-month change rate average, and the year-on-year change rate average is represented by 2, and the third digit after the letter C is 3, 4, and 5 for 3, 4, and 5 minutes, respectively. For example, in model C113, the first digit 1 after C represents Gid, the second digit 1 after C represents based on the average value of the month-on-month change rate, and the third digit 3 after C represents that the time value is 3 minutes.

[0089]

[0090] Table 1

[0091] Based on the accuracy evaluation map, an intensity assessment model that meets the preset assessment value is selected to generate a suboptimal intensity assessment model. Figure 4 The accuracy evaluation diagram of the intensity assessment model of the earthquake impact area determination method provided in the first embodiment of the present invention is combined with Table 1, as shown in FIG. Figure 4 As shown, the default evaluation values ​​are set to MAE<0.25, MSE<0.15, RMSE<0.35 and R 2 >75%, extraction Figure 4 The model that satisfies the conditions at the same time, that is, the preset evaluation values ​​of MAE, MSE and RMSE fall within the set radius circle, R 2 The models whose preset evaluation values ​​fall outside the set radius circle are C113, C123, C125, C224, C225, C314, C315 and C324.

[0092] Figure 5 This is a comparison diagram of the prediction errors of the intensity field assessment model of the method for determining the earthquake affected area provided in the first embodiment of the present invention, such as Figure 5 As shown in the figure, the PER value of the intensity assessment model C315 based on the average month-on-month anomaly rate of the active Wifi equipment indicator data in the first 5 minutes after the earthquake and the corresponding epicenter distance is the lowest, which is 11.29%, indicating that the prediction accuracy of model C315 is the highest and it is determined to be the optimal intensity assessment model.

[0093] Step 170, setting the indicator type and evaluation time period used for optimal intensity assessment model training as the optimal indicator type and optimal evaluation time period for earthquake intensity assessment.

[0094] Specifically, after a new earthquake occurs, optimal indicator data in mobile phone signaling data determined according to the optimal intensity assessment model is obtained, and the earthquake impact area is determined in real time through the optimal intensity assessment model based on the optimal indicator data and the optimal assessment time period.

[0095] Embodiment 2

[0096] A second embodiment of the present invention provides a device for determining an earthquake-affected area. Figure 6 The schematic diagram of the structure of the device for determining the earthquake affected area provided in the second embodiment of the present invention is as follows: Figure 6 As shown, the earthquake impact zone determination device 200 includes a first data processing module 201, a second data processing module 202, a third data processing module 203, a model data set construction module 204, a model training module 205, a model testing module 206, and a model input variable setting module 207.

[0097] The first data processing module 201 is used to obtain mobile phone signaling data, extract various indicator data from the mobile phone signaling data, and count the number of devices collected by each indicator data in each Geohash grid every minute.

[0098] Among them, the various indicator data include mobile device indicator data, active base station equipment indicator data, active Wi fi equipment indicator data and wireless network networking equipment indicator data.

[0099] The second data processing module 202 is used to calculate the equipment change rate of each indicator in multiple time periods according to the number of equipment.

[0100] Among them, the equipment change rate includes the month-on-month change rate and the year-on-year change rate. The month-on-month change rate is the change rate of the number of equipment in the same time period before and after the earthquake on the day of the earthquake, and the year-on-year change rate is the change rate of the number of equipment after the earthquake on the day of the earthquake relative to the same time on the day before the earthquake.

[0101] The third data processing module 203 is used to generate a plurality of earthquake impact area distribution data graphs through an interpolation algorithm according to the equipment change rates in a plurality of time periods.

[0102] Specifically, the average value of the equipment change rate in the Geohash grid in each time period is calculated in turn, and multiple earthquake impact area distribution data maps are generated by interpolation algorithm according to the average value of the equipment change rate in each time period. Optionally, the interpolation algorithm is a Kriging interpolation algorithm.

[0103] The model data set construction module 204 is used to select the equipment change rate in different time periods and the epicenter distance corresponding to the GeoHash grid in each indicator that meets the preset fitting accuracy according to multiple earthquake impact area distribution data maps, and generate model data sets for different time periods for each indicator.

[0104] The model training module 205 is used to construct a pre-trained intensity assessment model based on a random forest regression algorithm, and train the pre-trained intensity assessment model through the first model data in the model data set to generate an intensity assessment model.

[0105] The model testing module 206 is used to test all intensity assessment models through the second model data in the model data set and output intensity distribution data; calculate the model fitting accuracy according to the intensity distribution data, generate test evaluation results, and determine the optimal intensity assessment model according to the test evaluation results.

[0106] The test evaluation results include first test evaluation data and second test evaluation data; the first test evaluation data include mean absolute error, mean square error, root mean square error, and determination coefficient; the second test evaluation data is the prediction error rate, and the prediction error rate is the percentage of the number of GeoHash grids with evaluation misalignment to the total number of GeoHash grids.

[0107] Specifically, the model fitting accuracy is calculated based on the intensity distribution data to generate first test evaluation data; all intensity assessment models that meet preset assessment values ​​in the first test evaluation data are selected to generate a suboptimal intensity assessment model; second test evaluation data are generated based on the intensity distribution data calculation of the suboptimal intensity assessment model; the suboptimal intensity assessment model with the smallest prediction error rate of the second test evaluation data is selected to generate an optimal intensity assessment model.

[0108] The model input variable setting module 207 is used to set the indicator type and evaluation time period used for optimal intensity assessment model training as the optimal indicator type and optimal evaluation time period for earthquake intensity assessment.

[0109] The device for determining an earthquake-affected area provided in the second embodiment of the present invention is used to execute the steps of the method provided in the first embodiment of the present invention. Its implementation principle and technical effects are similar and will not be described in detail here.

[0110] Embodiment 3

[0111] Embodiment 3 of the present invention provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the method for determining any earthquake-affected area provided in Embodiment 1 is implemented.

[0112] The professionals should further realize that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to the function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0113] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0114] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for determining an earthquake-affected area, characterized in that: The method comprises: Obtain mobile phone signaling data, extract multiple indicator data from the mobile phone signaling data, and count the number of devices collected in each Geohash grid per minute for each indicator data; Calculate the equipment change rate of each indicator in multiple time periods based on the number of equipment; Generate multiple earthquake impact area distribution data graphs through an interpolation algorithm according to the equipment change rates in the multiple time periods; According to the multiple earthquake impact area distribution data maps, the equipment change rate in different time periods and the epicenter distance corresponding to the GeoHash grid that meet the preset fitting accuracy are selected for each indicator, and a model data set for different time periods is generated for each indicator; Constructing a pre-trained intensity assessment model based on a random forest regression algorithm, and training the pre-trained intensity assessment model using the first model data in the model data set to generate an intensity assessment model; All intensity assessment models are tested using the second model data in the model data set, and intensity distribution data are output; model fitting accuracy is calculated based on the intensity distribution data, a test evaluation result is generated, and an optimal intensity assessment model is determined based on the test evaluation result; The indicator type and evaluation time period used for training the optimal intensity assessment model are set as the optimal indicator type and optimal evaluation time period for earthquake intensity assessment.

2. The determination method according to claim 1, characterized in that: The equipment change rate includes a month-on-month change rate and a year-on-year change rate. The month-on-month change rate is the change rate of the number of equipment in the same time period before and after the earthquake on the day of the earthquake, and the year-on-year change rate is the change rate of the number of equipment after the earthquake on the day of the earthquake relative to the same time one day before the earthquake.

3. The determination method according to claim 1, characterized in that: The multiple indicator data include mobile device indicator data, active base station device indicator data, active Wifi device indicator data and wireless network networking device indicator data.

4. The determination method according to claim 1, characterized in that: The generating of multiple earthquake impact area distribution data graphs by interpolation algorithm according to the equipment change rates in the multiple time periods specifically includes: Calculate the average value of the device change rate in the Geohash grid in each time period; Multiple earthquake impact area distribution data maps are generated by plotting the average value of the equipment change rate in each time period through an interpolation algorithm.

5. The determination method according to claim 1, characterized in that: The interpolation algorithm is a Kriging interpolation algorithm.

6. The determination method according to claim 1, characterized in that: The test evaluation result includes first test evaluation data and second test evaluation data; the first test evaluation data includes mean absolute error, mean square error, root mean square error, and determination coefficient, and the second test evaluation data is the prediction error rate.

7. The determination method according to claim 6, characterized in that: The step of calculating the model fitting accuracy according to the intensity distribution data, generating a test evaluation result, and determining the optimal intensity evaluation model according to the test evaluation result specifically includes: Calculate the model fitting accuracy according to the intensity distribution data to generate first test evaluation data; Selecting all intensity assessment models that meet preset assessment values ​​in the first test assessment data to generate a suboptimal intensity assessment model; Generate second test assessment data based on the intensity distribution data of the suboptimal intensity assessment model; The suboptimal intensity assessment model with the smallest prediction error rate of the second test assessment data is selected to generate the optimal intensity assessment model.

8. The determination method according to claim 6, characterized in that: The prediction error rate is the percentage of the number of GeoHash grids with misaligned evaluation to the total number of GeoHash grids.

9. A device for determining an earthquake-affected area, characterized in that: The determining device comprises: A first data processing module is used to obtain mobile phone signaling data, extract multiple indicator data from the mobile phone signaling data, and count the number of devices collected by each indicator data in each Geohash grid per minute; A second data processing module is used to calculate the equipment change rate of each indicator in multiple time periods according to the number of equipment; A third data processing module is used to generate a plurality of earthquake impact area distribution data graphs by an interpolation algorithm according to the equipment change rates in the plurality of time periods; A model data set construction module, for selecting, according to the plurality of earthquake impact area distribution data graphs, the equipment change rate in different time periods and the epicenter distance corresponding to the GeoHash grid for each indicator that meets the preset fitting accuracy, and generating model data sets for different time periods for each indicator; A model training module, used to construct a pre-trained intensity assessment model based on a random forest regression algorithm, and train the pre-trained intensity assessment model through the first model data in the model data set to generate an intensity assessment model; A model testing module, used to test all intensity assessment models through the second model data in the model data set and output intensity distribution data; perform model fitting accuracy calculation according to the intensity distribution data, generate test evaluation results, and determine the optimal intensity assessment model according to the test evaluation results; The model input variable setting module is used to set the indicator type and evaluation time period used for training the optimal intensity assessment model as the optimal indicator type and optimal evaluation time period for earthquake intensity assessment.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to claims 1 to 8 are implemented.

Images