Reservoir earthquake monitoring station network observation system

Abstract

The invention relates to the technical field of reservoir earthquake monitoring, in particular to a reservoir earthquake monitoring station network observation system which comprises a server side, a user side and a plurality of earthquake stations used for collecting waveform data around a reservoir. The server side comprises a data acquisition module, a feature extraction module, a processing module, a positioning module and a feedback module, and the processing module is used for extracting key feature data corresponding to the extracted current continuous waveform data based on a preset earthquake event prediction model. Predicting seismic event data corresponding to the current continuous waveform data of each seismic station; and the positioning module is used for positioning an earthquake monitored by the corresponding seismic station according to the predicted seismic event data corresponding to the current continuous waveform data of each seismic station and the current continuous waveform data corresponding to the seismic event data, and outputting seismic positioning result data corresponding to each seismic station.

Description

Technical Field

[0001] The invention relates to the technical field of reservoir earthquake monitoring, and in particular to a reservoir earthquake monitoring station network observation system. Background Art

[0002] Reservoir earthquakes refer to earthquakes caused by changes in crustal stress due to reservoir water storage. Reservoir earthquakes may have serious impacts on reservoir dams, surrounding buildings and the ecological environment, and even threaten people's lives and property. Therefore, it is very necessary to accurately monitor reservoir earthquakes in order to provide timely warnings and take measures to reduce disaster risks.

[0003] At present, reservoir earthquake monitoring mainly relies on seismographs, GPS, inclinometers and other equipment to analyze seismic activities by collecting seismic waveform data, and then use statistical methods or signal processing technology to analyze the monitoring data based on the collected seismic waveform data. Although reservoir earthquake monitoring has been achieved, there are still certain problems in the monitoring process:

[0004] The data collection frequency of the existing monitoring system is not high, and it is difficult to capture the instantaneous information of an earthquake, resulting in insufficient monitoring accuracy. The poor reliability of the collected data leads to low accuracy in earthquake event identification.

[0005] Existing monitoring systems have difficulty accurately locating the source of earthquakes, especially for small earthquakes that occur at or near the bottom of a reservoir. Summary of the invention

[0006] One of the purposes of the present invention is to provide a reservoir earthquake monitoring network observation system that can accurately identify reservoir earthquakes and quickly locate reservoir earthquakes, thereby improving the reliability and effectiveness of reservoir earthquake monitoring.

[0007] In order to achieve the above-mentioned purpose, a reservoir seismic monitoring network observation system is provided, which includes a service end, a user end and a plurality of seismic stations for collecting waveform data around the reservoir;

[0008] The server includes:

[0009] A data acquisition module is used to collect the current continuous waveform data corresponding to the preset time period monitored by each seismic station;

[0010] The feature extraction module is used to extract key feature data corresponding to each current continuous waveform data according to the current continuous waveform data corresponding to the preset time period monitored by each seismic station;

[0011] A processing module, for predicting earthquake event data corresponding to the current continuous waveform data of each seismic station based on the extracted key feature data corresponding to each current continuous waveform data and based on a preset earthquake event prediction model;

[0012] A positioning module is used to locate the earthquake monitored by the corresponding seismic station according to the predicted earthquake event data corresponding to the current continuous waveform data of each seismic station and the current continuous waveform data corresponding to the earthquake event data, and output the earthquake positioning result data corresponding to each seismic station;

[0013] The feedback module is used to send the earthquake location result data corresponding to each seismic station to the user end.

[0014] Technical principles and effects of this scheme: In this scheme, first, the waveform data around the reservoir is monitored in real time through the seismic stations set up around the reservoir, and then the waveform data is cut according to the preset time period through the setting of the data acquisition module to form the current continuous waveform data.

[0015] After that, the current continuous waveform data corresponding to the preset time period monitored by each seismic station is collected, and the key feature data corresponding to each current continuous waveform data is extracted. By extracting the key feature data corresponding to each current continuous waveform data, the difficulty of subsequent data processing is greatly reduced.

[0016] Then, the key feature data corresponding to each extracted current continuous waveform data are used as input data and input into the pre-set earthquake event prediction model, so as to predict the earthquake event data corresponding to the current continuous waveform data collected by each seismic station, so as to predict whether there is a corresponding earthquake event at the seismic stations around the reservoir. Once the corresponding earthquake event data is predicted, the earthquake monitored by the corresponding seismic station is located based on the current continuous waveform data corresponding to the earthquake event data, and the earthquake location result data corresponding to each seismic station is output, so as to accurately locate the earthquake while predicting the earthquake event, greatly improving the reliability and accuracy of reservoir earthquake identification. And the earthquake location result data corresponding to each seismic station is sent to the user end at the first time, that is, the reservoir earthquake can be accurately identified and quickly located at the same time, improving the reliability and effectiveness of reservoir earthquake monitoring.

[0017] Furthermore, the server also includes:

[0018] A model building module, used to build a corresponding earthquake event prediction model;

[0019] A historical data acquisition module, used to acquire the historical continuous waveform data of each seismic station and extract each key feature data corresponding to each historical continuous waveform data based on the historical continuous waveform data of each seismic station;

[0020] It is also used to obtain historical earthquake event data corresponding to each historical waveform data;

[0021] An association module is used to perform multiple associations on the key feature data corresponding to each historical continuous waveform data according to the key feature data corresponding to each historical continuous waveform data extracted, based on the location information of the seismic station corresponding to each continuous waveform data, according to a preset quantity threshold, to form an input set data corresponding to the earthquake event prediction model;

[0022] A weight set setting module, used for randomly generating N initial populations, wherein the individuals of the initial population are weight sets corresponding to the earthquake event prediction model; the weight set includes a plurality of weight combinations;

[0023] The training module is used to input each sub-data in the input set data as input data into the earthquake event prediction model corresponding to each weight combination in the weight set, output the corresponding prediction result data, and calculate the average loss value corresponding to each weight combination in the weight set based on the prediction result data and the corresponding historical earthquake event data and the preset average loss function;

[0024] The average loss function is:

[0025]

[0026] In the formula, is the average loss value, y ij is the magnitude value in the prediction result data corresponding to each key feature data corresponding to the jth historical continuous waveform data in the i-th sub-data, is the magnitude value in the historical earthquake event data corresponding to each key characteristic data corresponding to the jth historical continuous waveform data in the i-th sub-data, m is the total number of historical continuous waveform data in the sub-data, and n is the total number of sub-data in the input set data;

[0027] A selection module is used to select a weight combination whose average loss value is less than a preset first loss value according to the average loss values ​​corresponding to each weight combination in the weight set within a first iteration number;

[0028] When the first number of iterations is exceeded, the weight combination is selected based on a preset selection ratio;

[0029] The hybridization and mutation module is used to perform hybridization and mutation on each selected weight combination based on a preset hybridization and mutation strategy to obtain a progeny population;

[0030] A loop module is used to continue executing the training module after obtaining the offspring population until a preset number of iterations is met;

[0031] The output module is used to output the offspring population as the optimal solution set for multi-objective optimization.

[0032] Beneficial effects: In this scheme, firstly, an earthquake event prediction model is constructed, and then the historical continuous waveform data of each seismic station is obtained and based on the historical continuous waveform data of each seismic station, each key feature data corresponding to each historical continuous waveform data is extracted; it is also used to obtain the historical earthquake event data corresponding to each historical waveform data;

[0033] After that, according to the key feature data corresponding to the extracted historical continuous waveform data and based on the location information of the seismic station corresponding to each continuous waveform data, the key feature data corresponding to each historical continuous waveform data are multiple-associated according to a preset quantity threshold to form an input set data corresponding to the earthquake event prediction model; by associating the relevant data corresponding to several historical continuous waveform data, when the model corresponding to each group of weight combinations is input and output, the adaptability of the weight combination of the model is greatly improved, and the problem of poor adaptability of the trained earthquake event prediction model is avoided.

[0034] Then, the prediction results of the model corresponding to each weight combination are evaluated according to the calculation of the average loss value corresponding to the model, and then each weight combination is optimized in combination with the optimization algorithm, so as to quickly and accurately determine the weight combination corresponding to the earthquake event prediction model. Specifically, within the first number of iterations, according to the average loss value corresponding to each weight combination in the weight set, a weight combination with an average loss value less than the preset first loss value is selected; when the first number of iterations is exceeded, the weight combination is selected based on the preset selection ratio; based on the preset hybridization and mutation strategy, each selected weight combination is hybridized and mutated to obtain a progeny population; after obtaining the progeny population, the model training iteration is continued until the preset number of iterations is met; the progeny population is output as the optimal solution set for multi-objective optimization.

[0035] In the process of training the constructed earthquake event prediction model, the optimization algorithm is combined to realize the rapid and accurate determination of the weight combination in the model, which greatly improves the training reliability of the earthquake event prediction model.

[0036] Furthermore, the preset hybridization mutation strategy includes the following steps:

[0037] During hybridization, the selected weight combinations are divided into several hybridization sets; the hybridization set includes three weight combinations;

[0038] Determine the hybridization fragments Z1, Z2, and Z3 for the three weight combinations in the hybridization set;

[0039] Generate hybrid random numbers p1, P2, and p3 for the three weight combinations in the hybrid set, and calculate the corresponding random averages based on the preset random average calculation formula. If the random average is greater than the preset average threshold, the hybrid fragments corresponding to the three weight combinations are exchanged in sequence; otherwise, the hybrid fragments of the three weight combinations in the corresponding hybrid set are not exchanged;

[0040] The random average calculation formula is:

[0041]

[0042] is the random mean;

[0043] After the crossover is completed, mutation is performed, and during the mutation, a mutation random number P4 corresponding to each weight combination is randomly generated;

[0044] According to the mutation random numbers corresponding to each weight combination, calculate the mutation random differences between each weight combination, as well as the ratio of the mutation random differences between each weight combination and other weight combinations that are greater than the preset difference, and judge whether the corresponding ratio is greater than the preset ratio threshold. If so, select the weight combination corresponding to the smallest mutation random difference between the corresponding weight combination and other weight combinations, and perform mutation operation on the mutation fragments corresponding to the corresponding two weight combinations.

[0045] Beneficial effects: In this scheme, during hybridization, the selected weight combinations are divided into several hybridization sets containing three weight combinations, and the respective hybridization fragments are determined for the three weight combinations in each hybridization set. The respective hybridization random numbers are generated for the three weight combinations in each hybridization set. Based on the preset random average calculation formula, the corresponding random average is calculated. If the random average is greater than the preset average threshold, the hybridization fragments corresponding to the three weight combinations are exchanged in turn; otherwise, no exchange is performed. During mutation, whether to mutate the weight combination is determined based on the ratio of the random difference between the mutation of each weight combination and other weight combinations greater than the preset difference. The diversity of the population is increased by setting the hybridization and mutation strategies, which helps to explore more solution spaces. At the same time, reasonable hybridization and mutation strategies can prevent premature entrapment in local optimal solutions. Through random number generation and average value calculation, the hybridization and mutation operations are dynamically adjusted, which improves the search efficiency of the algorithm.

[0046] Furthermore, the server also includes an alarm module for sending an alarm signal to the seismic station, the user end and the server end corresponding to the earthquake location result after determining the earthquake event data and the corresponding earthquake location result.

[0047] Beneficial effects: In this solution, the alarm module can be set up to send alarm signals to seismic stations, user terminals and service terminals as soon as an earthquake is detected, thereby realizing a full range of alarms, which greatly improves the vigilance of each person and the timeliness of the corresponding early warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a logic block diagram of the reservoir earthquake monitoring network observation system in Example 1 of the present invention.

[0049] Figure 2 This is a logic block diagram of the server of the reservoir earthquake monitoring network observation system in the first embodiment of the present invention. DETAILED DESCRIPTION

[0050] The following is further described in detail through specific implementation methods:

[0051] Embodiment 1

[0052] The reservoir earthquake monitoring network observation system is basically as follows Figure 1 and Figure 2 As shown, it includes a server, a user, and several seismic stations for collecting waveform data around the reservoir;

[0053] The server includes:

[0054] A data acquisition module is used to collect the current continuous waveform data corresponding to the preset time period monitored by each seismic station;

[0055] The feature extraction module is used to extract key feature data corresponding to each current continuous waveform data according to the current continuous waveform data corresponding to the preset time period monitored by each seismic station;

[0056] A processing module, for predicting earthquake event data corresponding to the current continuous waveform data of each seismic station based on the extracted key feature data corresponding to each current continuous waveform data and based on a preset earthquake event prediction model;

[0057] A positioning module is used to locate the earthquake monitored by the corresponding seismic station according to the predicted earthquake event data corresponding to the current continuous waveform data of each seismic station and the current continuous waveform data corresponding to the earthquake event data, and output the earthquake positioning result data corresponding to each seismic station;

[0058] The feedback module is used to send the earthquake location result data corresponding to each seismic station to the user end.

[0059] The server also includes:

[0060] A model building module, used to build a corresponding earthquake event prediction model;

[0061] The historical data acquisition module is used to acquire the historical continuous waveform data of each seismic station and extract the key feature data corresponding to each historical continuous waveform data based on the historical continuous waveform data of each seismic station; in this embodiment, the key feature data include amplitude, energy, periodicity data, average value, kurtosis and kurtosis of waveform data, cross-correlation between each continuous waveform data, etc.

[0062] It is also used to obtain the historical earthquake event data corresponding to each historical waveform data; in this embodiment, the earthquake event prediction model is constructed using the BP neural network technology, wherein the earthquake event prediction model includes an input layer, a hidden layer, and an output layer, wherein the key characteristic data such as amplitude, energy, periodicity data, average value of waveform data, degree of dispersion, kurtosis and kurtosis, and the mutual correlation between each continuous waveform data are used as input data of the input layer, and the corresponding historical earthquake event data are used as output data of the output layer. For the hidden layer, this embodiment uses the following formula to determine the number of hidden layer nodes: Wherein l is the number of nodes in the hidden layer, n is the number of nodes in the input layer, m is the number of nodes in the output layer, and a is a number between 1 and 10, which is 8 in this embodiment, so there are 11 nodes in the hidden layer.

[0063] BP neural network usually uses Sigmoid differentiable function and linear function as the network's activation function. This paper selects the S-type tangent function tansig as the activation function of the hidden layer neurons. The prediction model selects the S-type logarithmic function tansig as the activation function of the output layer neurons.

[0064] The association module is used to associate the key feature data corresponding to each historical continuous waveform data extracted, based on the location information of the seismic station corresponding to each continuous waveform data, with each key feature data corresponding to each historical continuous waveform data according to a preset number threshold to form an input set data corresponding to the earthquake event prediction model; in this embodiment, according to the location information of each seismic station, the distance value between the corresponding seismic station and other seismic stations is calculated and arranged in order from small to large to form a distance arrangement table corresponding to each seismic station, and then based on the preset number threshold, the corresponding seismic station is selected from the distance arrangement table, and each seismic station can be selected multiple times. Finally, the key feature data corresponding to each historical continuous waveform data corresponding to each seismic station selected with the preset number threshold are associated to form the corresponding input set data.

[0065] A weight set setting module, used for randomly generating N initial populations, wherein the individuals of the initial population are weight sets corresponding to the earthquake event prediction model; the weight set includes a plurality of weight combinations;

[0066] The training module is used to input each sub-data in the input set data as input data into the earthquake event prediction model corresponding to each weight combination in the weight set, output the corresponding prediction result data, and calculate the average loss value corresponding to each weight combination in the weight set based on the prediction result data and the corresponding historical earthquake event data and the preset average loss function;

[0067] The average loss function is:

[0068]

[0069] In the formula, is the average loss value, y ij is the magnitude value in the prediction result data corresponding to each key feature data corresponding to the jth historical continuous waveform data in the i-th sub-data, is the magnitude value in the historical earthquake event data corresponding to each key characteristic data corresponding to the jth historical continuous waveform data in the i-th sub-data, m is the total number of historical continuous waveform data in the sub-data, and n is the total number of sub-data in the input set data;

[0070] A selection module is used to select a weight combination whose average loss value is less than a preset first loss value according to the average loss values ​​corresponding to each weight combination in the weight set within a first iteration number;

[0071] When the first number of iterations is exceeded, the weight combination is selected based on a preset selection ratio;

[0072] The hybridization and mutation module is used to perform hybridization and mutation on each selected weight combination based on a preset hybridization and mutation strategy to obtain a progeny population;

[0073] The preset hybridization mutation strategy includes the following steps:

[0074] During hybridization, the selected weight combinations are divided into several hybridization sets; the hybridization set includes three weight combinations;

[0075] Determine the hybridization fragments Z1, Z2, and Z3 for the three weight combinations in the hybridization set;

[0076] Generate hybrid random numbers P1, P2, and P3 for the three weight combinations in the hybrid set, and calculate the corresponding random averages based on the preset random average calculation formula. If the random average is greater than the preset average threshold, the hybrid fragments corresponding to the three weight combinations are exchanged in sequence; otherwise, the hybrid fragments of the three weight combinations in the corresponding hybrid set are not exchanged;

[0077] The random average calculation formula is:

[0078]

[0079] is the random mean;

[0080] After the crossover is completed, mutation is performed, and during the mutation, a mutation random number P4 corresponding to each weight combination is randomly generated;

[0081] According to the mutation random numbers corresponding to each weight combination, calculate the mutation random differences between each weight combination, as well as the ratio of the mutation random differences between each weight combination and other weight combinations that are greater than the preset difference, and judge whether the corresponding ratio is greater than the preset ratio threshold. If so, select the weight combination corresponding to the smallest mutation random difference between the corresponding weight combination and other weight combinations, and perform mutation operation on the mutation fragments corresponding to the corresponding two weight combinations.

[0082] A loop module is used to continue executing the training module after obtaining the offspring population until a preset number of iterations is met;

[0083] The output module is used to output the offspring population as the optimal solution set for multi-objective optimization.

[0084] It also includes an alarm module, which is used to send an alarm signal to the seismic station, user end and service end corresponding to the earthquake positioning result after determining the earthquake event data and the corresponding earthquake positioning result.

[0085] The above is only an embodiment of the present invention. The common sense such as the known specific structure and characteristics in the scheme is described too much here. The ordinary technicians in the relevant field know all the common technical knowledge in the technical field of the invention before the application date or priority date, can know all the existing technologies in the field, and have the ability to apply the conventional experimental means before that date. The ordinary technicians in the relevant field can improve and implement this scheme in combination with their own abilities under the enlightenment given by this application. Some typical known structures or known methods should not become obstacles for ordinary technicians in the relevant field to implement this application. It should be pointed out that for those skilled in the art, without departing from the structure of the present invention, several deformations and improvements can be made, which should also be regarded as the scope of protection of the present invention, which will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. Reservoir earthquake monitoring network observation system, characterized by: It includes a server, a user, and several seismic stations for collecting waveform data around the reservoir; The server includes: A data acquisition module is used to collect the current continuous waveform data corresponding to the preset time period monitored by each seismic station; The feature extraction module is used to extract key feature data corresponding to each current continuous waveform data according to the current continuous waveform data corresponding to the preset time period monitored by each seismic station; A processing module, for predicting earthquake event data corresponding to the current continuous waveform data of each seismic station based on the extracted key feature data corresponding to each current continuous waveform data and based on a preset earthquake event prediction model; A positioning module is used to locate the earthquake monitored by the corresponding seismic station according to the predicted earthquake event data corresponding to the current continuous waveform data of each seismic station and the current continuous waveform data corresponding to the earthquake event data, and output the earthquake positioning result data corresponding to each seismic station; The feedback module is used to send the earthquake location result data corresponding to each seismic station to the user end.

2. The reservoir earthquake monitoring network observation system according to claim 1, characterized in that: The server also includes: A model building module, used to build a corresponding earthquake event prediction model; A historical data acquisition module, used to acquire the historical continuous waveform data of each seismic station and extract each key feature data corresponding to each historical continuous waveform data based on the historical continuous waveform data of each seismic station; It is also used to obtain historical earthquake event data corresponding to each historical waveform data; An association module is used to perform multiple associations on the key feature data corresponding to each historical continuous waveform data according to the key feature data corresponding to each historical continuous waveform data extracted, based on the location information of the seismic station corresponding to each continuous waveform data, according to a preset quantity threshold, to form an input set data corresponding to the earthquake event prediction model; A weight set setting module, used for randomly generating N initial populations, wherein the individuals of the initial population are weight sets corresponding to the earthquake event prediction model; the weight set includes a plurality of weight combinations; The training module is used to input each sub-data in the input set data as input data into the earthquake event prediction model corresponding to each weight combination in the weight set, output the corresponding prediction result data, and calculate the average loss value corresponding to each weight combination in the weight set based on the prediction result data and the corresponding historical earthquake event data and the preset average loss function; The average loss function is: In the formula, is the average loss value, y ij is the magnitude value in the prediction result data corresponding to each key feature data corresponding to the jth historical continuous waveform data in the i-th sub-data, is the magnitude value in the historical earthquake event data corresponding to each key characteristic data corresponding to the jth historical continuous waveform data in the i-th sub-data, m is the total number of historical continuous waveform data in the sub-data, and n is the total number of sub-data in the input set data; A selection module is used to select a weight combination whose average loss value is less than a preset first loss value according to the average loss values ​​corresponding to each weight combination in the weight set within a first iteration number; When the first number of iterations is exceeded, the weight combination is selected based on a preset selection ratio; The hybridization and mutation module is used to perform hybridization and mutation on each selected weight combination based on a preset hybridization and mutation strategy to obtain a progeny population; A loop module is used to continue executing the training module after obtaining the offspring population until a preset number of iterations is met; The output module is used to output the offspring population as the optimal solution set for multi-objective optimization.

3. The reservoir earthquake monitoring network observation system according to claim 2, characterized in that: The preset hybridization mutation strategy includes the following steps: During hybridization, the selected weight combinations are divided into several hybridization sets; the hybridization set includes three weight combinations; Determine the hybridization fragments Z1, Z2, and Z3 for the three weight combinations in the hybridization set; Generate hybrid random numbers P1, P2, and P3 for the three weight combinations in the hybrid set, and calculate the corresponding random averages based on the preset random average calculation formula. If the random average is greater than the preset average threshold, the hybrid fragments corresponding to the three weight combinations are exchanged in sequence; otherwise, the hybrid fragments of the three weight combinations in the corresponding hybrid set are not exchanged; The random average calculation formula is: is the random mean; After the crossover is completed, mutation is performed, and during the mutation, a mutation random number P4 corresponding to each weight combination is randomly generated; According to the mutation random numbers corresponding to each weight combination, calculate the mutation random differences between each weight combination, as well as the ratio of the mutation random differences between each weight combination and other weight combinations that are greater than the preset difference, and judge whether the corresponding ratio is greater than the preset ratio threshold. If so, select the weight combination corresponding to the smallest mutation random difference between the corresponding weight combination and other weight combinations, and perform mutation operation on the mutation fragments corresponding to the corresponding two weight combinations.

4. The reservoir earthquake monitoring network observation system according to claim 3 is characterized in that: The server also includes an alarm module, which is used to send an alarm signal to the seismic station, the user end and the server end corresponding to the earthquake positioning result after determining the earthquake event data and the corresponding earthquake positioning result.

Images