Typhoon prediction method, system and device and storage medium
Through optical fiber sensors, multi-parameter data is collected and timing prediction is used to use LSTM network model to perform timing prediction, which solves the accuracy of typhoon weather monitoring and prediction, and achieves accurate acquisition and early warning of typhoon activity patterns.
Patent Information
- Application Number
- CN202411838030.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-13
- Publication Date
- 2025-04-08
AI Technical Summary
It is difficult for the prior art to achieve real-time monitoring and accurate prediction of typhoon weather through a single physical quantity, especially when sea observation data are scarce.
A variety of environmental data is collected through optical fiber sensors, packaged into multi-parameter timing data signals, and timing prediction is performed using the LSTM network model to calculate the correlation coefficient between the predicted signal and the historical signal or the current signal to output the typhoon prediction result.
It has achieved accurate acquisition of typhoon activity patterns, improved the accuracy and reliability of typhoon prediction, and can warn of abnormal weather in advance.
Smart Images
Figure CN119781085A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a typhoon prediction method, system, equipment and storage medium, belonging to the technical field of meteorological prediction. Background Art
[0002] There are currently many means to monitor typhoons. At present, meteorological satellites use fast imagers and detectors to encrypt scanning observations, achieving regional observations with a resolution of 100 meters within minutes, providing high-resolution day and night continuous monitoring data for short-term extreme weather. Use meteorological stations, meteorological satellites, and meteorological radars to observe and predict typhoons. However, typhoons are usually generated at sea, and there is a lack of observation data at sea, or even blind spots. Existing space-based observation methods (such as) satellites cannot conduct precise "direct" detection of typhoons. Forecasters sometimes cannot accurately know where the center of the typhoon is, and numerical models cannot accurately describe the vortex structure of the typhoon, resulting in large forecast errors. Meteorological stations infer the location and intensity of typhoons by monitoring factors such as temperature, air pressure, wind direction, and wind speed. Meteorological satellites analyze cloud maps and infrared images to understand the structure and dynamics of typhoons. Meteorological radars detect cloud and rain distribution and wind conditions, achieving more precise monitoring.
[0003] Fiber optic sensors have the advantages of anti-electromagnetic interference and distributed monitoring, and have the ability to sense many parameters such as vibration, temperature, pressure, salinity, tide, wind speed, etc., which can meet the environmental monitoring needs closely related to typhoons. By integrating sensor technology in typhoon-prone areas, multiple parameter information can be monitored in real time, abnormal conditions can be discovered in time and early warnings can be issued. In the existing technical solutions, differential interference based on backward Rayleigh scattered spatial light can achieve vibration measurement, which is suitable for large-scale wind speed detection, but the signal-to-noise ratio of this solution is relatively low. After the researchers write gratings in ordinary single-mode optical fibers, they can measure vibration signals based on grating arrays, and preliminarily prove its vibration characteristics under wind vibration. Due to the high sensitivity of fiber optic sensing technology, there is always complex noise interference in the vibration measurement system, which affects the judgment of monitoring events. However, in terms of current technical implementation, it is difficult to accurately obtain the laws of typhoon activities by only using a single physical quantity to realize real-time monitoring of typhoon weather. Even if complex monitoring methods are used to monitor multiple parameters within the region, the prediction of typhoon activities can only be judged by artificial comprehensive experience, so it is difficult to achieve further practical application. Summary of the invention
[0004] In order to solve the above problems existing in the prior art, the present invention proposes a typhoon prediction method, system, device and storage medium.
[0005] The technical solution of the present invention is as follows:
[0006] In one aspect, the present invention provides a typhoon prediction method, comprising the following steps:
[0007] The optical fiber sensor is used to collect different types of environmental data related to the typhoon in the target area and uniformly encapsulates them into multi-parameter time series data signals;
[0008] Divide the multi-parameter time series data signal into a historical signal and a current signal according to time intervals;
[0009] The time series prediction model is trained through historical signals, and the current signal is input into the trained time series prediction model to obtain the prediction signal at the future moment;
[0010] Calculate the correlation coefficient between the predicted signal and the historical signal or the current signal, and output the typhoon prediction result at the future time based on the calculated correlation coefficient and the preset threshold value;
[0011] The correlation coefficient indicates the correlation between two sets of signals.
[0012] As a preferred embodiment, the optical fiber sensor adopts an interferometric optical fiber sensor, and the signal collected by the optical fiber sensor is expressed as:
[0013]
[0014] Among them, I T1 (t) represents the three-way interference signal collected at time T1, A and B are the DC and AC components generated by the photoelectric conversion after interference, and z takes the values of 1, 2, and 3. is the fiber phase at time T1;
[0015] The optical fiber phase variation caused by typhoon weather at different times is obtained by using the grating interferometer phase demodulation algorithm. Input to the time series forecasting model.
[0016] As a preferred implementation, the time series prediction model adopts an LSTM network model, and the state of the LSTM network model is represented as follows:
[0017]
[0018] Among them, h t-1 is the hidden state at the previous moment, G f , G i , G o They are forget gate, input gate, and output gate respectively. is the optical fiber phase change, W f , W i , W o are the weight matrices of the forget gate, input gate, and output gate, σ(·) is the nonlinear sigmoid activation function, G is the second input data, W is the weight matrix of the second input data, tanh(·) is the hyperbolic tangent activation function, b f、b i 、b o , b are the bias parameters of the forget gate, input gate, output gate, and the second input data respectively;
[0019] According to the above status, the LSTM network model is transformed into:
[0020]
[0021] Among them, c t is the cell state at the current moment, c t-1 is the cell state at the previous moment, h t is the hidden state at the current moment, y t Output the prediction result for the current moment.
[0022] As a preferred implementation, in the step of calculating the correlation coefficient between the predicted signal and the historical signal or the current signal, the calculation formula of the correlation coefficient is as follows:
[0023]
[0024] Among them, f cor (x) represents the correlation coefficient, E[·] is the expected value, y t (T1) is the predicted output result of the historical signal or current signal output by the time series prediction model, y t (T3) is the predicted output result of the prediction signal output by the time series prediction model, t is the time step, and N is the upper limit of the time step.
[0025] On the other hand, the present invention also provides a typhoon prediction system, comprising:
[0026] A data acquisition module is used to collect different types of environmental data related to typhoons in the target area through optical fiber sensors and uniformly encapsulate them into multi-parameter time series data signals;
[0027] A data partitioning module is used to partition multi-parameter time series data signals into historical signals and current signals according to time intervals;
[0028] The data prediction module is used to train the time series prediction model through historical signals, input the current signal into the trained time series prediction model, and obtain the prediction signal at the future moment;
[0029] A typhoon prediction module is used to calculate the correlation coefficient between the prediction signal and the historical signal or the current signal, and output the typhoon prediction result at a future moment according to the comparison between the calculated correlation coefficient and a preset threshold value;
[0030] The correlation coefficient indicates the correlation between two sets of signals.
[0031] As a preferred embodiment, the optical fiber sensor adopts an interferometric optical fiber sensor, and the signal collected by the optical fiber sensor is expressed as:
[0032]
[0033] Among them, I T1 (t) represents the three-way interference signal collected at time T1, A and B are the DC and AC components generated by the photoelectric conversion after interference, and z takes the values of 1, 2, and 3. is the fiber phase at time T1;
[0034] The optical fiber phase variation caused by typhoon weather at different times is obtained by using the grating interferometer phase demodulation algorithm. Input to the time series forecasting model.
[0035] As a preferred implementation, the time series prediction model adopts an LSTM network model, and the state of the LSTM network model is represented as follows:
[0036]
[0037] Among them, h t-1 is the hidden state at the previous moment, G f , G i , G o They are forget gate, input gate, and output gate respectively. is the optical fiber phase change, W f , W i , W o are the weight matrices of the forget gate, input gate, and output gate, σ(·) is the nonlinear sigmoid activation function, G is the second input data, W is the weight matrix of the second input data, tanh(·) is the hyperbolic tangent activation function, b f 、b i 、b o , b are the bias parameters of the forget gate, input gate, output gate, and the second input data respectively;
[0038] According to the above status, the LSTM network model is transformed into:
[0039]
[0040] Among them, c t is the cell state at the current moment, c t-1 is the cell state at the previous moment, h t is the hidden state at the current moment, y t Output the prediction result for the current moment.
[0041] As a preferred implementation, in the step of calculating the correlation coefficient between the predicted signal and the historical signal or the current signal, the calculation formula of the correlation coefficient is as follows:
[0042]
[0043] Among them, f cor (x) represents the correlation coefficient, E[·] is the expected value, y t (T1) is the predicted output result of the historical signal or current signal output by the time series prediction model, y t (T3) is the predicted output result of the prediction signal output by the time series prediction model, t is the time step, and N is the upper limit of the time step.
[0044] On the other hand, the present invention further proposes an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the typhoon prediction method as described in any embodiment of the present invention is implemented.
[0045] On the other hand, the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the typhoon prediction method as described in any embodiment of the present invention.
[0046] Additional aspects and advantages of the present invention will be set forth in the following description, and some of them will be apparent from the description, or may be understood by practicing the present invention. In addition, the various aspects and advantages of the present invention may be realized and obtained by the method steps and combinations particularly pointed out in the appended claims. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] Figure 1 This is a flow chart of a method according to Embodiment 1 of the present invention;
[0048] Figure 2 This is a schematic diagram of simulation detection data in an embodiment of the present invention;
[0049] Figure 3 This is a diagram of the predicted time domain results of the LSTM network model in an embodiment of the present invention. DETAILED DESCRIPTION
[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of 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.
[0051] It should be understood that the step numbers used in this document are only for convenience of description and are not intended to limit the order in which the steps are executed.
[0052] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.
[0053] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.
[0054] The term "and / or" means and includes any and all possible combinations of one or more of the associated listed items.
[0055] Embodiment 1:
[0056] See also Figure 1 This embodiment provides a typhoon prediction method, comprising the following steps:
[0057] S100. Collect different types of environmental data related to the typhoon in the target area through optical fiber sensors and uniformly encapsulate them into multi-parameter time series data signals. For typhoon weather monitoring applications, different types of environmental data include temperature, humidity, wind speed, vibration, etc. in the local area. The above signals can all be realized through optical fiber sensing technology. In order to improve the working efficiency of the monitoring system, the parameters need to be designed in advance. They can be presented in a common optical path or in a unified package, and the output data format of multi-parameter sensor data is unified, thereby forming time series data signals of different vectors.
[0058] S200, dividing the multi-parameter time series data signal collected by the optical fiber sensor into a historical signal and a current signal according to time intervals, and the data interval time is determined according to daily inspections.
[0059] S300, using historical signals as training sets to train the time series prediction model, and the prediction results of historical signals are used to continuously correct the time series prediction model; the current signal is used as a test set to input the trained time series prediction model to obtain the prediction signal for the future moment.
[0060] S400, calculating the correlation coefficient between the predicted signal and the historical signal or the current signal, and outputting the typhoon prediction result at the future time based on the comparison between the calculated correlation coefficient and the preset threshold value; determining the correlation coefficient mutation through the threshold value, when the multi-parameter information is stable, the correlation coefficient of the two groups of signals will continue to remain unchanged, and when the correlation coefficient undergoes a continuous change mutation, it proves that a typhoon or other abnormal weather may be approaching in the monitoring area.
[0061] The correlation coefficient indicates the correlation between two sets of signals.
[0062] As a preferred implementation of this embodiment, the optical fiber sensor adopts an interferometric optical fiber sensor, and the collected signal is expressed as:
[0063]
[0064] Among them, I T1 (t) represents the three-way interference signal collected at [T0, T1], A and B are the DC and AC components generated by the photoelectric conversion after interference, respectively, and z takes the values of 1, 2, and 3. is the fiber phase at time T1.
[0065] The optical fiber phase variation caused by typhoon weather at different times is obtained by using the grating interferometer phase demodulation algorithm. Input to the time series forecasting model.
[0066] As a preferred implementation of this embodiment, the time series prediction model adopts an LSTM network model. Since LSTM has two transmission states, namely, cell state c t and the hidden state h t , fiber phase change After inputting into the LSTM network model, there are four states, which can be expressed as:
[0067]
[0068] Among them, h t-1 is the hidden state at the previous moment, G f , G i , G o They are forget gate, input gate, and output gate respectively. is the optical fiber phase change, W f , W i , W oare the weight matrices of the forget gate, input gate, and output gate respectively; σ(·) is the nonlinear sigmoid activation function, which is normalized to a value between (0,1) through the sigmoid activation function; G is the second input data, W is the weight matrix of the second input data, tanh(·) is the hyperbolic tangent activation function, and G is converted to a value of (-1,1) through the hyperbolic tangent activation function; b f 、b i 、b o , b are the bias parameters of the forget gate, input gate, output gate, and the second input data respectively.
[0069] According to the above status, the LSTM network model can be further transformed into:
[0070]
[0071] Among them, c t is the cell state at the current moment, c t-1 is the cell state at the previous moment, h t is the hidden state at the current moment, y t Output the prediction result for the current moment.
[0072] As a preferred implementation of this embodiment, in the step of calculating the correlation coefficient between the predicted signal and the historical signal or the current signal:
[0073] When the monitoring area remains normal, the current signal and the future forecast signal have similar characteristics. When the forecast signal is different from the historical signal, the monitoring area is very likely to experience drastic weather changes. We determine whether an abnormal state exists by the correlation between the forecast signal and the historical signal or the current signal. The calculation formula for the correlation coefficient between the two can be obtained by calculating the correlation coefficient at [1, N] discrete moments as follows:
[0074]
[0075] Among them, f cor (x) represents the correlation coefficient, E[·] is the expected value, y t (T1) is the predicted output result of the historical signal or current signal output by the time series prediction model, y t (T3) is the predicted output result of the prediction signal output by the time series prediction model, t is the time step, and N is the upper limit of the time step.
[0076] When the weather in the monitored area remains basically unchanged, the historical or current signals and future signals are well consistent, and the correlation coefficient remains basically unchanged. When there is abnormal extreme weather such as typhoons, the correlation between historical or current signals and predicted signals will continue to change over a period of time. By setting the corresponding alarm thresholds, continuous high-accuracy predictions of abnormal weather can be achieved, thus realizing early warning of typhoon weather.
[0077] In order to verify the effectiveness of the method provided in this embodiment, a specific simulation case is provided below:
[0078] The simulation is based on the large changes in vibration signals in a local area under typhoon weather. In order to reduce the amount of data in long-term monitoring, this case selects one point per second as data storage for 24 hours. Figure 2 As shown in the figure, in order to consider the impact of weather factors on the results, this case selects data from 00:00 to 12:00 as the training set, and data from 12:00 to 24:00 as the test set. The number of data points in the sample set and the test set is the same, and the same season is selected for testing. The gradient descent algorithm is used, and the maximum number of iterations is 100. After learning, the learning rate is 0.001, and the data set is scattered after each training. After iteration, the error converges quickly and finally stabilizes. The complete training cycle lasts for 100 times, and the number of iterations in each cycle reaches 80,000 times.
[0079] like Figure 3 As shown, Figure 3 The data from 00:00 to 12:00 is used as the training set, and the test set from 12:00 to 24:00 is obtained. The unstable test state time at the beginning and end is ignored. The total number of data points is 24 hours, of which Figure 3 (a) is the result obtained by training the training set. Figure 3 (b) is the result of the test set prediction. It can be seen that the RMSE in the training result is 0.86136, which is a low error. The predicted data in the test set has the same trajectory as the actual data, and the root mean square error is 0.8729. Further analysis shows that since the training set and the test set are both on the same day of the same season, and Figure 3 (a) from 00:00 to 12:00 noon, Figure 3 (b) is from 12:00 noon to 24:00 at night. The data continuity is good. During the test, the abnormal vibration is more obvious in the typhoon weather at noon and decreases continuously at night. Expanding the data points to a longer time range will help to achieve abnormal vibration prediction in a longer time range.
[0080] In order to compare the early warning capability of the LSTM network model under abnormal typhoon weather, this embodiment takes a section of abnormal vibration data for analysis, selects historical data as a test set, takes the data section at time T0 as a reference, uses the common neural network algorithm and the LSTM network model to train the data at this time, predicts the results at time T1, and inputs them into the algorithm. By calculating f cor To judge the warning accuracy of different algorithms.
[0081] Table 1
[0082]
[0083] The comparison results are shown in Table 1. The average PSD of the signal at time T0 is -38.9dB. After training with the common neural network and LSTM network models, the average PSD of the signal obtained is -38.2dB and -38.7dB respectively, which proves that the training results of the two algorithms are closer to the actual value and can realize signal reconstruction. At this time, the correlation coefficients of the signals obtained by the two algorithms are 95.31% and 98.01% respectively compared with the original signals, which proves that the prediction accuracy of the LSTM network model is higher. After predicting the future time T1 based on the current data, the two algorithms obtained an average PSD of -35.2dB and -35.7dB respectively. Compared with the time T0, the signal energy has increased. The predicted signal is correlated with the signal at the time T0, and the correlation estimation results of 81.59% and 77.62% are obtained respectively. The correlation of the subsequent signal with the time T0 begins to decrease. The decrease in correlation proves that a large-scale abnormal vibration may have occurred in the local area under typhoon weather, which is consistent with the actual results. Quantitative analysis shows that the results predicted by the LSTM network model are more distinguishable from the original signal, proving that the LSTM network model is effective and more advantageous in large-scale and long-period warnings in typhoon weather.
[0084] Embodiment 2:
[0085] This embodiment provides a typhoon prediction system, including:
[0086] A data acquisition module is used to collect different types of environmental data related to the typhoon in the target area through an optical fiber sensor, and uniformly encapsulate them into a multi-parameter time series data signal; this module is used to implement the function of step S100 in the first embodiment, and will not be described in detail here;
[0087] A data division module, used to divide the multi-parameter time series data signal into a historical signal and a current signal according to a time interval; this module is used to implement the function of step S200 in the first embodiment, and will not be described in detail here;
[0088] A data prediction module is used to train the time series prediction model through historical signals, input the current signal into the trained time series prediction model, and obtain the prediction signal at the future moment; this module is used to implement the function of step S300 in the first embodiment, and will not be repeated here;
[0089] A typhoon prediction module is used to calculate the correlation coefficient between the prediction signal and the historical signal or the current signal, and output the typhoon prediction result at the future time according to the comparison between the calculated correlation coefficient and the preset threshold value; this module is used to implement the function of step S400 in the first embodiment, and will not be repeated here;
[0090] The correlation coefficient indicates the correlation between two sets of signals.
[0091] As a preferred implementation of this embodiment, the optical fiber sensor adopts an interferometric optical fiber sensor, and the collected signal is expressed as:
[0092]
[0093] Among them, I T1 (t) represents the three-way interference signal collected at time T1, A and B are the DC and AC components generated by the photoelectric conversion after interference, and z takes the values of 1, 2, and 3. is the fiber phase at time T1;
[0094] The optical fiber phase variation caused by typhoon weather at different times is obtained by using the grating interferometer phase demodulation algorithm. Input to the time series forecasting model.
[0095] As a preferred implementation of this embodiment, the time series prediction model adopts an LSTM network model, and the state of the LSTM network model is represented as follows:
[0096]
[0097] Among them, h t-1 is the hidden state at the previous moment, G f , G i , G o They are forget gate, input gate, and output gate respectively. is the optical fiber phase change, W f , W i , W o are the weight matrices of the forget gate, input gate, and output gate, σ(·) is the nonlinear sigmoid activation function, G is the second input data, W is the weight matrix of the second input data, tanh(·) is the hyperbolic tangent activation function, b f 、b i 、b o, b are the bias parameters of the forget gate, input gate, output gate, and the second input data respectively;
[0098] According to the above status, the LSTM network model is transformed into:
[0099]
[0100] Among them, c t is the cell state at the current moment, c t-1 is the cell state at the previous moment, h t is the hidden state at the current moment, y t Output the prediction result for the current moment.
[0101] As a preferred implementation of this embodiment, in the step of calculating the correlation coefficient between the predicted signal and the historical signal or the current signal, the calculation formula of the correlation coefficient is as follows:
[0102]
[0103] Among them, f cor (x) represents the correlation coefficient, E[·] is the expected value, y t (T1) is the predicted output result of the historical signal or current signal output by the time series prediction model, y t (T3) is the predicted output result of the prediction signal output by the time series prediction model, t is the time step, and N is the upper limit of the time step.
[0104] Embodiment three:
[0105] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the typhoon prediction method as described in any embodiment of the present invention is implemented.
[0106] Embodiment 4:
[0107] This embodiment provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the typhoon prediction method as described in any embodiment of the present invention is implemented.
[0108] In the embodiments of the present application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Among them, A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c can be represented by: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, c can be single or multiple.
[0109] Those of ordinary skill in the art will appreciate that the various units and algorithm steps described in the embodiments disclosed herein can be implemented in a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel 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 this application.
[0110] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.
[0111] In several embodiments provided in the present application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory; hereinafter referred to as: ROM), random access memory (Random Access Memory; hereinafter referred to as: RAM), disk or optical disk, and other media that can store program codes.
[0112] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.
Claims
1. A typhoon prediction method, characterized in that: The following steps are involved: The optical fiber sensor is used to collect different types of environmental data related to the typhoon in the target area and uniformly encapsulates them into multi-parameter time series data signals; Divide the multi-parameter time series data signal into a historical signal and a current signal according to time intervals; The time series prediction model is trained through historical signals, and the current signal is input into the trained time series prediction model to obtain the prediction signal at the future moment; Calculate the correlation coefficient between the predicted signal and the historical signal or the current signal, and output the typhoon prediction result at the future time based on the calculated correlation coefficient and the preset threshold value; The correlation coefficient indicates the correlation between two sets of signals.
2. A typhoon prediction method according to claim 1, characterized in that: The optical fiber sensor adopts an interferometric optical fiber sensor, and the collected signal is expressed as: Among them, I T1 (t) represents the three-way interference signal collected at time T1, A and B are the DC and AC components generated by the photoelectric conversion after interference, and z takes the values of 1, 2, and 3. is the fiber phase at time T1; The optical fiber phase variation caused by typhoon weather at different times is obtained by using the grating interferometer phase demodulation algorithm. Input to the time series forecasting model.
3. A typhoon prediction method according to claim 1, characterized in that: The time series prediction model adopts the LSTM network model, and the state of the LSTM network model is expressed as: Among them, h t-1 is the hidden state at the previous moment, G f , G i , G o They are forget gate, input gate, and output gate respectively. is the optical fiber phase change, W f , W i , W o are the weight matrices of the forget gate, input gate, and output gate, σ(·) is the nonlinear sigmoid activation function, G is the second input data, W is the weight matrix of the second input data, tanh(·) is the hyperbolic tangent activation function, b f 、b i 、b o , b are the bias parameters of the forget gate, input gate, output gate, and the second input data respectively; According to the above status, the LSTM network model is transformed into: Among them, c t is the cell state at the current moment, c t-1 is the cell state at the previous moment, h t is the hidden state at the current moment, y t Output the prediction result for the current moment.
4. A typhoon prediction method according to claim 1, characterized in that: In the step of calculating the correlation coefficient between the predicted signal and the historical signal or the current signal, the calculation formula of the correlation coefficient is as follows: Among them, f cor (x) represents the correlation coefficient, E[·] is the expected value, y t (T1) is the predicted output result of the historical signal or current signal output by the time series prediction model, y t (T3) is the predicted output result of the prediction signal output by the time series prediction model, t is the time step, and N is the upper limit of the time step.
5. A typhoon prediction system, characterized in that: include: A data acquisition module is used to collect different types of environmental data related to typhoons in the target area through optical fiber sensors and uniformly encapsulate them into multi-parameter time series data signals; A data partitioning module is used to partition multi-parameter time series data signals into historical signals and current signals according to time intervals; The data prediction module is used to train the time series prediction model through historical signals, input the current signal into the trained time series prediction model, and obtain the prediction signal at the future moment; A typhoon prediction module is used to calculate the correlation coefficient between the prediction signal and the historical signal or the current signal, and output the typhoon prediction result at a future moment according to the comparison between the calculated correlation coefficient and a preset threshold value; The correlation coefficient indicates the correlation between two sets of signals.
6. A typhoon prediction system according to claim 5, characterized in that: The optical fiber sensor adopts an interferometric optical fiber sensor, and the collected signal is expressed as: Among them, I T1 (t) represents the three-way interference signal collected at time T1, A and B are the DC and AC components generated by the photoelectric conversion after interference, and z takes the values of 1, 2, and 3. is the fiber phase at time T1; The optical fiber phase variation caused by typhoon weather at different times is obtained by using the grating interferometer phase demodulation algorithm. Input to the time series forecasting model.
7. A typhoon prediction system according to claim 5, characterized in that: The time series prediction model adopts the LSTM network model, and the state of the LSTM network model is expressed as: Among them, h t-1 is the hidden state at the previous moment, G f , G i , G o They are forget gate, input gate, and output gate respectively. is the optical fiber phase change, W f , W i , W o are the weight matrices of the forget gate, input gate, and output gate, σ(·) is the nonlinear sigmoid activation function, G is the second input data, W is the weight matrix of the second input data, tanh(·) is the hyperbolic tangent activation function, b f 、b i 、b o , b are the bias parameters of the forget gate, input gate, output gate, and the second input data respectively; According to the above status, the LSTM network model is transformed into: Among them, c t is the cell state at the current moment, c t-1 is the cell state at the previous moment, h t is the hidden state at the current moment, y t Output the prediction result for the current moment.
8. A typhoon prediction system according to claim 5, characterized in that: In the step of calculating the correlation coefficient between the predicted signal and the historical signal or the current signal, the calculation formula of the correlation coefficient is as follows: Among them, f cor (x) represents the correlation coefficient, E[·] is the expected value, y t (T1) is the predicted output result of the historical signal or current signal output by the time series prediction model, y t (T3) is the predicted output result of the prediction signal output by the time series prediction model, t is the time step, and N is the upper limit of the time step.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the typhoon prediction method according to any one of claims 1 to 4 is implemented.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the typhoon prediction method according to any one of claims 1 to 4 is implemented.