阅读说明：本技术 一种用于手机地震预警的大批量生成手机记录的地震数据的方法 (Method for generating mass seismic data recorded by mobile phone for mobile phone earthquake early warning ) 是由 郑增威 石利飞 孙霖 赵莎 侯建民 董霖 方毅 刘杰 李石坚 潘纲 于 2020-12-01 设计创作，主要内容包括：本发明涉及一种用于手机地震预警的大批量生成手机记录的地震数据的方法,包括：步骤1、使用振动台来模拟地震,构造TMAP所需数据集；步骤2、设计TMAP转换方法。本发明的有益效果是：该方法综合考虑两种造成手机记录地震数据和真实地震数据不一致的因素,并分别构造有限脉冲响应模型和滑移模型来对这些因素进行建模,解释了地震波到达时,桌子上的智能手机的滑动；对转换结果可视化后,可看出本发明生成的手机质量数据和手机在地震场景中采集到的数据高度一致。此外通过计算曲线间相似度衡量参数,发现无论在多么剧烈的地震场景下,本发明采用的方法都能生成高度一致的手机质量数据。(The invention relates to a method for generating mass seismic data recorded by a mobile phone for mobile phone earthquake early warning, which comprises the following steps: step 1, simulating an earthquake by using a vibration table, and constructing a data set required by TMAP; and 2, designing a TMAP conversion method. The invention has the beneficial effects that: the method comprehensively considers two factors causing inconsistency of mobile phone recorded seismic data and real seismic data, and respectively constructs a finite impulse response model and a slippage model to model the factors, so as to explain the slippage of the smart phone on a desk when seismic waves arrive; after the conversion result is visualized, it can be seen that the mobile phone quality data generated by the method is highly consistent with the data acquired by the mobile phone in the earthquake scene. In addition, by calculating similarity measurement parameters between curves, the method disclosed by the invention can generate highly consistent mobile phone quality data no matter how violent the earthquake scene is.)

1. A method for generating seismic data recorded by a mobile phone in a large batch for mobile phone seismic early warning is characterized by comprising the following steps:

step 1, simulating an earthquake by using a vibration table, and constructing a data set required by TMAP;

step 2, designing a TMAP conversion method;

step 3, verifying a finite impulse response model by using data acquired by a fixed scene in a vibration table experiment;

step 4, verifying the slippage model by using data acquired by a free scene in a vibration table experiment;

and 5, testing the performance of the TMAP conversion method designed in the step 2 under various earthquake conditions.

2. The method for mass generation of seismic data recorded by the mobile phone for mobile phone earthquake early warning according to claim 1, wherein the step 1 specifically comprises the following steps:

step 1.1, selecting seismic data recorded by a plurality of original seismic stations and a plurality of manually generated data as input data of a vibrating table simulation experiment; the vibration table simulates an earthquake according to input data of a vibration table simulation experiment;

step 1.2, fixing two strong seismographs on a vibration table, placing the two strong seismographs at two opposite angle positions which are farthest away on the vibration table, and considering vibration data recorded by the strong seismographs as real vibration of the vibration table;

step 1.3, earthquake motion acquisition software suitable for the smart phone is developed, and the earthquake motion acquisition software enables the smart phone to collect three-axis acceleration data of the smart phone at a fixed acquisition frequency in a fixed scene and a free scene respectively; when the vibration table is in operation: in a free scene, multiple smart phones are freely placed on a vibration table; in a fixed scene, a plurality of smart phones are fixed on a vibration table;

step 1.4, simulating input data of a vibration table simulation experiment twice in a free scene; in a fixed scene, input data of a vibration table simulation experiment is simulated once, and differences caused by the precision of the smart phone are analyzed through self triaxial acceleration data collected in the fixed scene.

3. The method for mass production of seismic data recorded by a mobile phone for mobile phone earthquake early warning as claimed in claim 1, wherein the step 2 specifically comprises the following steps:

step 2.1, setting the vertical acceleration of the strong seismograph at the time t as a_av(t) vertical acceleration of smartphone is a_pv(t) the horizontal acceleration of the macroseismic instrument is a_ah(t) the velocity of the macroseisometer is v_ah(t) horizontal acceleration of smartphone is a_ph(t) smartphone speed v_ph(t); wherein a is_av(t) and a_pv(t) are all scalar quantities, a_ah(t)、v_ah(t)、a_ph(t) and v_ph(t) are all two-dimensional vectors; setting:

A_av＝(a_av(1),a_av(2),…,a_av(T))

A_ah＝(a_ah(1),a_ah(2),…,a_ah(T))

A_pv＝(a_pv(1),a_pv(2),…,a_pv(T))

A_ph＝(a_ph(1),a_ph(2),…,a_ph(T)) (1)

(A^TMAP _pv,A^TMAP _ph)＝TMAP(A_av,A_ah) (2)

is provided with (A)_av,A_ah) For recording by the strong seismograph, will (A)_pv,A_ph) Recording as a smart phone record; (A)_av,A_ah) And (A)_pv,A_ph) All contain acceleration information at T moments; data (A) to be converted using the TMAP conversion method^TMAP _pv,A^TMAP _ph) Recording as a TMAP record, wherein the TMAP record is generated quality data of the smart phone;

step 2.2, constructing a finite impulse response model: obtaining a difference between smartphone recorded data and seismograph recorded data using a finite impulse response model; each axis of the smart phone is regarded as a dynamic system, actual ground motion data is used as the input of the dynamic system, and data recorded by the smart phone is used as the output of the dynamic system; respectively training a finite impulse response model for three axes of the smart phone according to the recording data of the strong seismograph and the recording data of the smart phone:

(A^TMAP _pv,A^TMAP _ph)＝FIR(A_av,A_ah) (3)

in the above formula, (A)^TMAP _pv,A^TMAP _ph) For the generated smartphone quality data (TMAP record), FIR () contains the respective finite impulse response models of the three axes of the smartphone, each finite impulse response model only processes the data of its corresponding axis;

step 2.3, constructing a slippage model: the movement of the smart phone in the vertical direction is not influenced by the sliding state of the smart phone, and the vertical acceleration of the smart phone is always equal to the acceleration of the ground:

a_pv(t)＝a_av(t) (4)

in the above formula, a_pv(t) vertical acceleration of smartphone, a_av(t) is the vertical acceleration of the strong seismograph; the static friction coefficient and the dynamic friction coefficient between the ground and the smart phone are respectively assumed to be mu_sAnd mu_dG is the gravity acceleration, m is the mass of smart phone, smart phone plusThe sampling time of the speed sensor is delta t, and the mobile phone is in a static state relative to the ground; in this case, the smartphone records (A)_pv,A_ph) And record of the seismograph (A)_av,A_ah) And the same, satisfy:

a_ph(t)＝a_ah(t) (5)

in the above formula, a_ph(t) horizontal acceleration of smartphone, a_ah(t) is the horizontal acceleration of the macroseism instrument;

step 2.3.1, when a_ah(t)>μ_s(g+a_av(t)), the following equation (6) is no longer true:

m·a_ah(t)<μ_sm(g+a_av(t)) (6)

in the above formula, m is the mass of the smart phone, a_ah(t) is the horizontal acceleration, mu, of the macroseismic instrument_sIs the static friction coefficient between the ground and the smart phone, g is the acceleration of gravity, a_av(t) is the vertical acceleration of the strong seismograph; the smartphone will begin to slide relative to the ground, in the horizontal direction, the smartphone is only affected by the dynamic friction between the smartphone and the ground; horizontal acceleration a of smart phone_phThe direction of the (t) vector is always the same as the direction of the kinetic friction force, and the horizontal acceleration a of the smart phone_phThe magnitude of (t) is always equal to the absolute value of the kinetic friction force:

|m·a_ph(t)|＝|μ_dm(g+a_av(t))| (7)

the derivation is as follows:

in the above formulas (7) to (8), m is the mass of the smartphone, a_ph(t) horizontal acceleration, μ, of smartphone_dIs the dynamic friction coefficient between the ground and the smart phone, g is the acceleration of gravity, a_av(t) is the vertical acceleration of the macroseism instrument,

step 2.3.2, representing the speed and acceleration of the ground relative to the smartphone at time t as Δ v (t) and Δ a (t), respectively:

Δv(t)＝v_ah(t)-v_ph(t) (9)

Δa(t)＝a_ah(t)-a_ph(t) (10)

according to the relation between the acceleration and the speed of the ground relative to the smart phone, the following equation is approximately obtained:

Δv(t+Δt)≈Δv(t)+Δa(t)·Δt (11)

in the above formula, Δ v (t) is the speed of the ground relative to the smartphone, Δ a (t) is the acceleration of the ground relative to the smartphone, and Δ t is the time variation; since the dynamic friction direction of the smart phone is the same as the direction of Δ v (t), the horizontal acceleration a of the smart phone is determined_ph(t) is also in line with the velocity Δ v (t) of the ground relative to the smartphone; then the horizontal acceleration of the smartphone within time t + Δ t is:

in the above formula,. mu._dIs the dynamic friction coefficient between the ground and the smart phone, g is the acceleration of gravity, a_av(t + Δ t) is the vertical acceleration of the seismograph within the time t + Δ t, and Δ v (t + Δ t) is the speed of the surface relative to the smartphone within the time t + Δ t; thus:

Δa(t+Δt)＝a_ah(t+Δt)–a_ph(t+Δt) (13)

in the above formula, Δ a (t + Δ t) is an acceleration of the ground relative to the smartphone within t + Δ t; a is_ah(t + Δ t) is the horizontal acceleration of the macroseismic instrument within the time t + Δ t; a is_ph(t + Δ t) is the horizontal acceleration of the smartphone within time t + Δ t;

step 2.3.3, set the initial values of Δ v (t) and Δ a (t) in equation (9) to 0, knowing a at any time t_ah(t) and a_av(t), then obtaining Δ v (t + Δ t) and Δ a (t + Δ t) further from Δ v (t) and Δ a (t) according to equations (9) to (13), and then obtaining Δ v (t) and Δ a (t) at any time t by repeated iterations;

step 2.3.4, obtaining the acceleration of the smart phone through the following equation:

a_ph(t)＝a_ah(t)-Δa(t) (14)

in the above formula, a_ph(t) horizontal acceleration of smartphone, a_ah(t) is the horizontal acceleration of the seismograph, Δ a (t) is the acceleration of the ground relative to the smartphone.

4. The method for mass production of seismic data for mobile phone recordings for mobile phone seismic forewarning as claimed in claim 2, wherein: the number of seismic data recorded by the original seismic stations in step 1.1 is 46, and the number of manually generated data is 3.

5. The method for mass production of seismic data for mobile phone recordings for mobile phone seismic forewarning as claimed in claim 2, wherein: in step 1.3, the earthquake motion acquisition software enables the smart phone to collect own three-axis acceleration data in a fixed scene and a free scene respectively, and the acquisition frequency is 100 Hz.

Technical Field

The invention belongs to the field of mobile phone earthquake early warning, and particularly relates to a method for generating mass seismic data recorded by a mobile phone for mobile phone earthquake early warning.

Background

Major earthquakes cause many casualties and property losses, especially in densely populated cities. An earthquake early warning system (EEW) may detect an earthquake and calculate the time and place of the earthquake. The system may then send an alert to the area that may be affected. The system plays an important role in reducing seismic damage. However, since the main functions of seismic detection and localization of EEW systems require a sufficiently dense seismic network, which requires a large construction cost, only a few countries are able to build EEW systems. To avoid this problem, more and more research is using crowd sourcing as another way to warn of earthquakes. Seismic capture networks (QCNs) (e.s.cochran, j.f.lawrence.the Quake-Catcher Network: Citizen science expanding seismic networks.seismol.res.lett., 2009,80, 26-30) and Community Seismic Networks (CSNs) (r.w.classification, t.heaton, and m.aivazis.community seismic networks.annals of Geophysics,2012,54 (6)) are representative of this approach. They provide seismic information using any networked computer with an internal or external MEMS accelerometer. These early warning systems are of limited size due to the need to communicate the relevant hardware from the network operator to the user of the system.

Smartphones have found widespread use in the EEW field due to their wide distribution, computing power and sensors that can measure ambient parameters (Lee, s., Suh, j., Choi, y.view of smart applications for the geographic issues: current status, limits, and future perspectives. earth Science information, 2018,11(4), 463-. Such early warning systems use signal processing techniques and various machine learning algorithms to analyze the earthquake to predict and prevent damage caused by the earthquake. The iShake project at the berkeley division of university of california (j.reilly, s.dasti, m.ervasist, j.bray, s.glaser, and a.bayer. mobile phones as a sensor: Automation data extraction for the same system.journal of IEEE Transactions on Automation Science and Engineering,2013,10 (2)) designed a system architecture that used a smartphone and its built-in sensor to detect ground vibrations. The project is subjected to a vibration table experiment to prove that the smart phone sensor has the capability of detecting earthquakes. The MyShake project (Kong, q., y., w.kwony, l.schreierz, s.allen, r.allen, and j.strauss.smartphone-based networks for earth detection, 201515 th International Conference on Innovations for public Services (I4CS), IEEE, numberg, Germany,2015) is a new form of smartphone-based global earthquake warning system. The project developed an application for Android and iPhone users to distinguish between earthquake and human activity, and then uploaded relevant information (including time to trigger, Peak Ground Acceleration (PGA) and GPS position) to a Central Processing Center (CPC). The CPC uses this information to detect earthquakes and calculate the magnitude, epicenter, and start time of the earthquake.

Smart phone based seismic warning Systems (SEEW) are receiving increasing attention and are becoming a promising area of research. To better train the models associated with SEEW, enough representative handset-recorded seismic data needs to be collected. But this is difficult to achieve due to seismic unpredictability. Thus, some researchers use a vibration table (otherwise known as a seismic simulator) to simulate the earthquake and place a smartphone on the vibration table to collect the data. However, only modest earthquakes can be simulated due to the high cost of operating a vibration table. Furthermore, the number of smartphones used to collect data is limited due to the high cost of smartphones. Thus, the amount of data collected by the vibration table is very limited. Kong et al collected only 241 three-component recordings from 45 shaker runs.

In view of the above, considering the low accuracy of the cell phone sensor, a system identification (Ljung, l., and t. glad. modeling of Dynamic systems. ptr precision Hall, Upper saddleriver, NJ,1994) model can be used to interpret the difference between the data recorded by the smart phone and the actual ground motion data. Another type of difference may also result from a smartphone not being fixed to a desktop. For example, in the real world, when a seismic wave arrives, the smartphone on the table starts to slide, especially in larger seismic events. This slippage results in a large difference between the handset quality data and the real seismic data.

In summary, it is very important to provide a method for generating a large amount of seismic data recorded by a mobile phone for mobile phone earthquake early warning.

Disclosure of Invention

The invention aims to overcome the defects and provides a method for generating seismic data recorded by a mobile phone in a large batch for mobile phone seismic early warning.

The method for generating the seismic data recorded by the mobile phone in large batch for mobile phone seismic early warning comprises the following steps:

step 1, because of unpredictability of earthquake, it is difficult to collect a large amount of earthquake data, so a shaking table is used for simulating earthquake, and a data set required by TMAP is constructed;

step 2, designing a TMAP conversion method; because two factors causing inconsistency between the mobile phone and real vibration are considered in the TMAP conversion method, an FIR model and a slippage model are constructed to explain the two factors;

step 3, verifying a finite impulse response model by using data acquired by a fixed scene in a vibration table experiment;

step 4, verifying the slippage model by using data acquired by a free scene in a vibration table experiment;

and 5, testing the performance of the TMAP conversion method designed in the step 2 under various earthquake conditions.

Preferably, step 1 specifically comprises the following steps:

step 1.1, selecting seismic data recorded by a plurality of original seismic stations and a plurality of manually generated data as input data of a vibrating table simulation experiment; to represent as many earthquakes as possible, the magnitude and epicenter distance distributions of the input data are as dispersed as possible; the vibration table simulates an earthquake according to input data of a vibration table simulation experiment;

step 1.2, fixing two strong vibration instruments on a vibration table, wherein the vibration table can roll during vibration simulation, so that the motion conditions of each point on the vibration table are different; therefore, two strong seismographs are placed at two opposite angle positions which are farthest away from the vibration table, and vibration data recorded by the strong seismographs are regarded as real vibration of the vibration table; when the vibration table generates obvious rolling motion, the vibration condition of any point on the vibration table is obtained by interpolating data recorded by the two strong vibration instruments;

step 1.3, seismic motion acquisition software suitable for smart phones (including Android and iPhone) is developed, and the seismic motion acquisition software enables the smart phones to collect triaxial acceleration data of the smart phones at a fixed acquisition frequency in a fixed scene and a free scene respectively; when the vibration table is in operation: in a free scene, multiple smart phones are freely placed on a vibration table; in a fixed scene, a plurality of smart phones are fixed on a vibration table;

step 1.4, simulating input data of a vibration table simulation experiment twice in a free scene; in a fixed scene, input data of a vibration table simulation experiment is simulated once, and differences caused by the precision of the smart phone are analyzed through self triaxial acceleration data collected in the fixed scene.

Preferably, the step 2 specifically comprises the following steps:

step 2.1, setting the vertical acceleration of the strong seismograph at the time t as a_av(t) vertical acceleration of smartphone is a_pv(t) the horizontal acceleration of the macroseismic instrument is a_ah(t) the velocity of the macroseisometer is v_ah(t) horizontal acceleration of smartphone is a_ph(t) smartphone speed v_ph(t); wherein a is_av(t) and a_pv(t) are all scalar quantities, a_ah(t)、v_ah(t)、a_ph(t) and v_ph(t) are all two-dimensional vectors; setting:

A_av＝(a_av(1),a_av(2),…,a_av(T))

Aah＝(aah(1),a_ah(2),…,aah(T))

A_pv＝(a_pv(1),a_pv(2),…,a_pv(T))

A_ph＝(a_ph(1),a_ph(2),…,a_ph(T)) (1)

(A^TMAP _pv,A^TMAP _ph)＝TMAP(A_av,A_ah) (2)

is provided with (A)_av,A_ah) For recording by the strong seismograph, will (A)_pv,A_ph) Recording as a smart phone record; (A)_av,A_ah) And (A)_pv,A_ph) All contain acceleration information at T moments; data (A) to be converted using the TMAP conversion method^TMAP _pv,A^TMAP _ph) The TMAP record is recorded as TMAP record, and the TMAP record is generated quality data of the smart phone, which is as corresponding to the hand as possibleMachine records are similar;

step 2.2, constructing a finite impulse response model (FIR model): the accuracy of the sensor of the smart phone is lower than that of the strong seismograph, so that the recorded data of the sensor and the recorded data of the strong seismograph are different; obtaining a difference between smartphone recorded data and seismograph recorded data using a finite impulse response model; each axis of the smart phone is regarded as a dynamic system, actual ground motion data is used as the input of the dynamic system, and data recorded by the smart phone is used as the output of the dynamic system; respectively training a finite impulse response model for three axes of the smart phone according to the recording data of the strong seismograph and the recording data of the smart phone:

(A^TMAP _pv,A^TMAP _ph)＝FIR(A_av,A_ah) (3)

in the above formula, (A)^TMAP _pv,A^TMAP _ph) For the generated smartphone quality data (TMAP record), FIR () contains the respective finite impulse response models (FIR models) of the three axes of the smartphone, each finite impulse response model only processing the data of its corresponding axis; the finite impulse response model is a mathematical model, the internal structure of the dynamic system can be deduced according to the measured values of the input and output signals of the dynamic system, and the acceleration sensor of each mobile phone can record the acceleration of the mobile phone on three axes;

step 2.3, the difference can also be caused when the smart phone is not fixed on the ground; when an earthquake comes, the ground starts to vibrate; due to friction between the smartphone and the ground, the smartphone will start to vibrate accordingly; however, when an earthquake is serious, the smart phone slides relative to the ground, so that great difference is caused; establishing a slip model to account for the slip; constructing a slippage model: the movement of the smart phone in the vertical direction is not influenced by the sliding state of the smart phone, and the vertical acceleration of the smart phone is always equal to the acceleration of the ground:

a_pv(t)＝a_av(t) (4)

in the above formula, a_pv(t) vertical acceleration of smartphoneDegree of a_av(t) is the vertical acceleration of the strong seismograph; the static friction coefficient and the dynamic friction coefficient between the ground and the smart phone are respectively assumed to be mu_sAnd mu_dG is gravity acceleration, m is mass of the smart phone, delta t is sampling time of an acceleration sensor of the smart phone, and the smart phone is in a static state relative to the ground; in this case, the smartphone records (A)_pv,A_ph) And record of the seismograph (A)_av,A_ah) And the same, satisfy:

a_ph(t)＝a_ah(t) (5)

in the above formula, a_ph(t) horizontal acceleration of smartphone, a_ah(t) is the horizontal acceleration of the macroseism instrument;

step 2.3.1, when a_ah(t)>μ_s(g+a_av(t)), the following equation (6) is no longer true:

m·a_ah(t)<μ_sm(g+a_av(t)) (6)

in the above formula, m is the mass of the smart phone, a_ah(t) is the horizontal acceleration, mu, of the macroseismic instrument_sIs the static friction coefficient between the ground and the smart phone, g is the acceleration of gravity, a_av(t) is the vertical acceleration of the strong seismograph; the smartphone will begin to slide relative to the ground, in the horizontal direction, the smartphone is only affected by the dynamic friction between the smartphone and the ground; horizontal acceleration a of smart phone_phThe direction of the (t) vector is always the same as the direction of the kinetic friction force, and the horizontal acceleration a of the smart phone_phThe magnitude of (t) is always equal to the absolute value of the kinetic friction force:

|m·a_ph(t)|＝|μ_dm(g+a_av(t))| (7)

the derivation is as follows:

in the above formulas (7) to (8), m is the mass of the smartphone, a_ph(t) horizontal acceleration, μ, of smartphone_dIs the dynamic friction coefficient between the ground and the smart phone, g is the acceleration of gravity, a_av(t) is the vertical acceleration of the macroseism instrument,

step 2.3.2, representing the speed and acceleration of the ground relative to the smartphone at time t as Δ v (t) and Δ a (t), respectively:

Δv(t)＝v_ah(t)-v_ph(t) (9)

Δa(t)＝a_ah(t)-a_ph(t) (10)

according to the relation between the acceleration and the speed of the ground relative to the smart phone, the following equation is approximately obtained:

Δv(t+Δt)≈Δv(t)+Δa(t)·Δt (11)

in the above formula, Δ v (t) is the speed of the ground relative to the smartphone, Δ a (t) is the acceleration of the ground relative to the smartphone, and Δ t is the time variation; since the dynamic friction direction of the smart phone is the same as the direction of Δ v (t), the horizontal acceleration a of the smart phone is determined_ph(t) is also in line with the velocity Δ v (t) of the ground relative to the smartphone; then the horizontal acceleration of the smartphone within time t + Δ t is:

in the above formula,. mu._dIs the dynamic friction coefficient between the ground and the smart phone, g is the acceleration of gravity, a_av(t + Δ t) is the vertical acceleration of the seismograph within the time t + Δ t, and Δ v (t + Δ t) is the speed of the surface relative to the smartphone within the time t + Δ t; thus:

Δa(t+Δt)＝a_ah(t+Δt)–a_ph(t+Δt) (13)

in the above formula, Δ a (t + Δ t) is an acceleration of the ground relative to the smartphone within t + Δ t; a is_ah(t + Δ t) is the horizontal acceleration of the macroseismic instrument within the time t + Δ t; a is_ph(t + Δ t) is the horizontal acceleration of the smartphone within time t + Δ t;

step 2.3.3, set the initial values of Δ v (t) and Δ a (t) in equation (9) to 0, knowing a at any time t_ah(t) and a_av(t), then obtaining Δ v (t + Δ t) and Δ a (t + Δ t) further from Δ v (t) and Δ a (t) according to equations (9) to (13), and then obtaining Δ v (t) and Δ a (t) at any time t by repeated iterations;

step 2.3.4, obtaining the acceleration of the smart phone through the following equation:

a_ph(t)＝a_ah(t)-Δa(t) (14)

in the above formula, a_ph(t) horizontal acceleration of smartphone, a_ah(t) is the horizontal acceleration of the seismograph, Δ a (t) is the acceleration of the ground relative to the smartphone.

Preferably, in step 1.1, the number of seismic data recorded by the original seismic stations is 46, and the number of manually generated data is 3; because the shaker table experiment is expensive, only the 46 seismic data are used.

Preferably, in the step 1.3, the earthquake motion acquisition software enables the smart phone to respectively collect own three-axis acceleration data in a fixed scene and a free scene, and the acquisition frequency is 100 Hz; the larger the acquisition frequency is, the better the acquisition frequency is, so that during the later data processing, the data with lower frequency can be obtained through down sampling.

The invention has the beneficial effects that the invention provides a method for rapidly generating mass mobile phone quality seismic data, the method (TMAP conversion method) comprehensively considers two factors causing inconsistency of mobile phone recorded seismic data and real seismic data, respectively constructs a finite impulse response model and a slippage model to model the factors, and explains the sliding of a smart mobile phone on a desk when seismic waves arrive; after the conversion result is visualized, it can be seen that the mobile phone quality data generated by the method is highly consistent with the data acquired by the mobile phone in the earthquake scene. In addition, by calculating similarity measurement parameters between curves, the method disclosed by the invention can generate highly consistent mobile phone quality data no matter how violent the earthquake scene is.

Drawings

FIG. 1 is a diagram of the magnitude-to-center distribution of input data in a vibrating table experiment;

FIG. 2 is a visualization result diagram of the transformation of the TMK method in a fixed scene;

FIG. 3 is a diagram illustrating the visual result of the TMAP conversion method under a fixed scenario;

FIG. 4 is a visualization result diagram of the transformation of the TMK method in a free scene;

FIG. 5 is a diagram illustrating the visual result of the transformation of the TMAP transformation method in a free scene;

FIG. 6 is a graph comparing the conversion effect of TMAP and TMK conversion methods under different shocks.

Detailed Description

The present invention will be further described with reference to the following examples. The following examples are set forth merely to aid in the understanding of the invention. It should be noted that, for those skilled in the art, it is possible to make various improvements and modifications to the present invention without departing from the principle of the present invention, and those improvements and modifications also fall within the scope of the claims of the present invention.

In the field of earthquake early warning by using a mobile phone, a large amount of earthquake data recorded by the mobile phone needs to be acquired. But this is difficult due to the unpredictability of the occurrence of earthquakes. The invention constructs a method for directly generating the mobile phone quality data according to the data of the strong seismograph by analyzing and learning the difference between the seismic data recorded by the strong seismograph and the data recorded by the mobile phone of the traditional seismic table network. The method can help to design a better earthquake early warning system.

First, the overall idea of the invention:

the main idea of the invention is to comprehensively consider two factors causing the difference between the data recorded by the smart phone and the real earthquake motion data: the accuracy of the built-in sensor of the smart phone is low and the smart phone is not fixed on the ground, so a Finite Impulse Response (FIR) model is used to model a first cause of difference and a second cause of difference is modeled by explaining the slip phenomenon of the smart phone. The TMAP conversion method adopted by the invention comprises the two models. Experimental results prove that the TMAP conversion method can well obtain the quality data of the mobile phone.

Secondly, the invention comprises the following specific steps:

1. construction of data sets required for the TMAP conversion method:

the difference research needs to collect real acceleration data recorded by a strong seismograph and mobile phone data of a smart phone at the same time when an earthquake occurs. But it is difficult to collect such data in large quantities due to the unpredictability of earthquakes. The seismic was therefore simulated using a vibrating table (a large instrument which, depending on the input to the vibrating table, can vibrate a 3.5m by 3.5m table like an earthquake). To simulate an earthquake, 46 data (and 3 manually generated data) were selected as input data for the shaker table simulation experiment. To represent as many earthquakes as possible, the magnitude and epicenter distance distributions of these 46 selected seismic data are as dispersed as possible. Fig. 1 shows the distribution of input data of a vibration table simulation experiment.

Two strong seismographs are fixed on the vibration table; because the macroseisms are fixed and have high precision, the vibration data recorded by the macroseisms are considered as the real vibration of the vibration table; software for smart phones (including Android and iPhone) has also been developed that enables smart phones to collect their own three-axis acceleration data at 100 HZ. The present invention collected 48 smartphones (with software installed) and placed on a vibrating table to obtain its own vibration data.

Table 1 below lists the experimental setup parameters for the shaker table experiment, including the number of experimental materials and the number of experimental runs. As shown on the right side of table 1, the present invention performed experiments in both the fixed and free scenarios. A free scene means that when the vibrating table is running, selected 48 smartphones are placed freely on the vibrating table; whereas a fixed scenario means that these smartphones are fixed on a vibrating table. The vibrating table simulates the earthquake from its 49 input data (the selected 46 data and 3 manually generated data). Therefore, when an earthquake is simulated, the smart phone is freely placed, and 49 input data are completely simulated twice. Because the data recorded by the smart phone sensor is inconsistent with the real data due to the limitation of the precision of the smart phone sensor, in the vibration table experiment, the smart phone is fixed on the vibration table, and 49 input data are completely simulated. And then analyzing the difference of the smart phone caused by the precision of the smart phone through the data collected by the fixed scene.

Table 1 experiment setup parameter table for shaking table experiment

Experimental Material	Number of	Experimental scenario	Number of times
				Strong vibration instrument	2	Fixed scene	49×1
Mobile phone	48	Free scene	49×2
				Input data of vibration table	46+3

2. Designing a TMAP conversion method:

because the TMAP conversion method considers two factors causing inconsistency between the smartphone and the real vibration, two models are constructed to explain the two factors. Before introducing these two models separately, conventions are made for some symbols: when an earthquake occurs, the strong seismograph of the earthquake platform net is parallelly installed on the ground, and the placement direction of the smart phone is assumed to be parallel to the ground, so that the state of the smart phone in daily life is met; at time t, the vertical acceleration of the strong vibration instrument and the smart phone is a_av(t) and a_pv(t) the horizontal acceleration and the speed of the strong seismograph are respectively a_ah(t) and v_ah(t) the horizontal acceleration of the mobile phone is a_ph(t) and v_ph(t); wherein a in the vertical direction_av(t) and a_pv(t) are all scalars. A in the horizontal direction_ah(t),v_ah(t),a_ph(t) and v_ph(t) are both two-dimensional vectors; then define:

A_av＝(a_av(1),a_av(2),…,a_av(T))

A_ah＝(a_ah(1),a_ah(2),…,a_ah(T))

A_pv＝(a_pv(1),a_pv(2),…,a_pv(T))

A_ph＝(a_ph(1),a_ph(2),…,a_ph(T)) (1)

(A^TMAP _pv,A^TMAP _ph)＝TMAP(A_av,A_ah) (2)

will (A)_av,A_ah) Called the record of the seismograph, will (A)_pv,A_ph) Referred to as smartphone records; because the strong seismographs of the earthquake table net are all installed on the ground and are more accurate than the intelligent mobile phone, the strong seismographs are recorded as real ground motion data; given a seismograph recording and a cell phone recording, the present invention aims to find the difference between these two data,and constructing a conversion method by learning such differences; in equation (2), the conversion method is expressed by using TMAP (-) and the converted data (a) is converted^TMAP _pv,A^TMAP _ph) Referred to as TMAP record, TMAP record is generated smartphone quality data that should be as similar as possible to its corresponding smartphone record. In 2.1 and 2.2, two models constituting the TMAP conversion method will be described in detail.

2.1 construction of FIR model

The precision of the sensor of the smart phone is lower than that of the strong vibration instrument. Resulting in a difference between the two recorded data. The difference is obtained using a Finite Impulse Response (FIR) model. The model is a mathematical model, each axis of the handset can be considered as a dynamic system, and the internal structure of the system can be deduced according to the measured values of the input and output signals of the dynamic system. The acceleration sensor of each smart phone records the acceleration of the smart phone on three axes, the actual ground motion data is regarded as the input of the dynamic system, and the data recorded by the smart phone is regarded as the output of the system. Respectively training an FIR model for three axes of the mobile phone according to the record of the strong seismograph and the corresponding record of the mobile phone:

(A^TMAP _pv,A^TMAP _ph)＝FIR(A_av,A_ah) (3)

the FIR () in equation (3) contains these three trained FIR models, each model processing only the data for its respective axis.

2.2 construction of slip model

Discrepancies can also result if the smartphone is not fixed to the ground. When an earthquake comes, the ground starts to vibrate. Due to friction between the smartphone and the ground, the smartphone will also start to vibrate accordingly; however, when an earthquake is serious, the smart phone can slide relative to the ground, so that great difference is caused; a slip model is therefore built to account for this slip.

The movement of the smartphone in the vertical direction is not affected by the sliding state of the smartphone. The vertical acceleration of the smartphone is therefore always equal to the acceleration of the ground, as defined by equation (4). The static friction coefficient and the dynamic friction coefficient between the ground and the mobile phone are respectively assumed to be mu s and mu d, the gravity acceleration is g, the mass of the mobile phone is m, and the sampling time of an acceleration sensor of the mobile phone is delta t. The mobile phone is in a static state relative to the ground. In this case, the handset recording and the seismograph recording are almost the same as defined in equation (5).

a_pv(t)＝a_av(t) (4)

In the above formula, a_pv(t) vertical acceleration of smartphone, a_av(t) is the vertical acceleration of the strong seismograph;

a_ph(t)＝a_ah(t) (5)

in the above formula, a_ph(t) horizontal acceleration of smartphone, a_ah(t) is the horizontal acceleration of the macroseism instrument;

m·a_ah(t)<μ_sm(g+a_av(t)) (6)

in the above formula, m is the mass of the smart phone, a_ah(t) is the horizontal acceleration, mu, of the macroseismic instrument_sIs the static friction coefficient between the ground and the smart phone, a_av(t) is the vertical acceleration of the strong seismograph;

when a is_ah(t)>μ_s(g+a_av(t)), this means that equation (6) no longer holds:

m·a_ah(t)<μ_sm(g+a_av(t)) (6)

in the above formula, m is the mass of the smart phone, a_ah(t) is the horizontal acceleration, mu, of the macroseismic instrument_sIs the static friction coefficient between the ground and the smart phone, g is the acceleration of gravity, a_av(t) is the vertical acceleration of the strong seismograph; the smartphone will begin to slide relative to the ground; at the moment, in the horizontal direction, the smartphone is only influenced by the dynamic friction between the smartphone and the ground; thus a_phThe direction of the (t) vector should always be the same as the direction of the kinetic friction force, and a_phThe magnitude of (t) should always be equal to the absolute value of the kinetic friction force:

|m·a_ph(t)|＝|μ_dm(g+a_av(t)) | (7) is derived as:

in the above formulas (7) to (8), m is the mass of the smartphone, a_ph(t) horizontal acceleration, μ, of smartphone_dIs the dynamic friction coefficient between the ground and the smart phone, g is the acceleration of gravity, a_av(t) is the vertical acceleration of the macroseism instrument,

the velocity and acceleration of the ground relative to the handset at time t is represented as:

Δv(t)＝v_ah(t)-v_ph(t) (9)

Δa(t)＝a_ah(t)-a_ph(t) (10)

from the relationship between acceleration and velocity, the following equation can be approximated:

Δv(t+Δt)≈Δv(t)+Δa(t)·Δt (11)

since the direction of the dynamic friction force of the smartphone is the same as the direction of Δ v (t), a_phThe direction of (t) should also coincide with Δ v (t). Therefore, the horizontal acceleration of the handset at time t + Δ t is:

in the above formula,. mu._dIs the dynamic friction coefficient between the ground and the smart phone, g is the acceleration of gravity, a_av(t + Δ t) is the vertical acceleration of the seismograph within the time t + Δ t, and Δ v (t + Δ t) is the speed of the surface relative to the smartphone within the time t + Δ t; thus:

Δa(t+Δt)＝a_ah(t+Δt)–a_ph(t+Δt) (13)

in the above formula, Δ a (t + Δ t) is an acceleration of the ground relative to the smartphone within t + Δ t; a is_ah(t+Δt) is the horizontal acceleration of the seismograph within the time t + delta t; a is_ph(t + Δ t) is the horizontal acceleration of the smartphone within time t + Δ t;

setting the initial values of Δ v (0) and Δ a (0) to 0, a at any time t is also known_ah(t) and a_avThe value of (t). Then, according to the formulas (9) to (13), Δ v (t + Δ t) and Δ a (t + Δ t) can be further obtained from Δ v (t) and Δ a (t); through repeated iterations, Δ v (t) and Δ a (t) at any time t can be obtained; the acceleration of the smartphone can then be obtained by the following equation:

a_ph(t)＝a_ah(t)-Δa(t) (14)

in the above formula, a_ph(t) horizontal acceleration of smartphone, a_ah(t) is the horizontal acceleration of the seismograph, Δ a (t) is the acceleration of the ground relative to the smartphone.

Thirdly, experiments and results:

since the TMAP conversion method is composed of two models, a finite impulse response model (FIR model) and a slip model, the validity of the two models is verified respectively using data collected in two experimental scenarios in a shaking table experiment. In parallel, the FIR () model is compared to the conversion method proposed by Kong et al, which is referred to as the TMK method. The key idea of TMK is to convert the record of 24-bit strong seismograph into 16-bit record and increase the inherent noise of some smart phones; the mobile phone quality data obtained by the TMK method is called TMK record.

The FIR model was validated using data collected from a fixed scene in a shaking table experiment. As shown in fig. 2 and 3, FIR records, TMK records and handset records are visualized directly on the X-axis; it can be seen that: the vibration amplitude recorded by the TMK is larger than that recorded by the mobile phone, because the TMK keeps the sharp vibration sensing capability of the strong seismograph, but the mobile phone does not have the sharp vibration sensing capability as the strong seismograph; fig. 3 shows that FIR recordings and handset recordings are highly consistent. This shows that the TMAP conversion method adopted by the invention can generate the mobile phone quality data well.

The data collected from the free scene in the shaking table experiment was used to validate the slip model. As shown in fig. 4 and 5, comparing the TMAP record and the TMK record with the handset record, it can be seen that they are almost the same on the Z-axis. On the X axis and the Y axis in the horizontal direction, the TMAP record and the mobile phone record are also highly consistent, and a great difference exists between the TMK record and the mobile phone record, especially when the vibration is serious. This shows that the TMAP conversion method adopted by the invention can generate the mobile phone quality data well.

The performance of the TMAP conversion method needs to be tested under the condition of earthquakes as much as possible to verify the generalization capability of the TMAP conversion method; all records of a certain cell phone and a macroseism instrument were collected from the shaking table experiment. These seismograph records are converted to TMAP records and TMK records. Then, the similarity parameter defined in the following equation (15) is used to measure the consistency between each handset record and its corresponding conversion record (including TMAP record and TMK record). X and Y in equation (15) represent two time series data. mean (X) represents the average value of X. The resulting goodness from the formula can be used to measure the similarity between X and Y. The closer Goodness is to 1, the greater the similarity between X and Y.

As shown in fig. 6, the dark dots represent the similarity between TMK records and mobile phone records, and the light dots represent the similarity between TMAP records and mobile phone records; the x-axis represents the PGA values recorded by the corresponding seismographs, while the y-axis represents the average similarity in the two horizontal directions; PGA represents the maximum absolute amplitude from the three-component acceleration. It can be seen that as the PGA increases, the value of the dark dots drops rapidly due to the sliding of the phone. However, even if the vibration simulated by the vibration table is very serious, the TMAP conversion method can enable the value of the light-colored dots to be kept close to 1 all the time; this fact proves that under different intensity earthquakes, TMAP conversion method can perfectly generate the mobile phone quality records.

Fourthly, experimental conclusion:

the invention provides a method for quickly generating a large amount of mobile phone quality seismic data. The method comprehensively considers two factors causing inconsistency of mobile phone recorded seismic data and real seismic data, and respectively constructs a model to model the factors. The experimental results show that: the seismic data recorded by the mobile phone is quite consistent with the mobile phone quality data generated by the TAMP conversion method adopted by the invention. Through visualization of the conversion result, the generated mobile phone quality data is highly consistent with the data acquired by the mobile phone in the earthquake scene. In addition, by calculating similarity measurement parameters between curves, the TMAP conversion method adopted by the invention can generate very consistent mobile phone quality data no matter how violent the earthquake scene is.