A geological disaster risk assessment method

Abstract

The application discloses a geological disaster risk evaluation method, comprising the following steps: in a geological disaster record database, extracting data sequences of dangerousness elements before and after each geological disaster event; performing connectivity analysis on the data sequences of dangerousness elements before and after the geological disaster event, so as to screen out non-motive elements in the dangerousness elements; introducing 3D hollow convolution into a geological disaster potential risk evaluation model to construct a high-sensitivity potential risk evaluation model. The application performs connectivity analysis on the data sequences of dangerousness elements before and after the geological disaster event, screens out non-motive elements in the dangerousness elements, ensures that the feature selection applied to the risk evaluation is more accurate, introduces 3D hollow convolution into the geological disaster potential risk evaluation model to construct a high-sensitivity potential risk evaluation model, and improves the evaluation sensitivity of the geological disaster potential risk evaluation model.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of disaster risk assessment, and particularly relates to a geological disaster risk assessment method. BACKGROUND

[0002] Geological disasters refer to geological actions or geological phenomena caused by natural or human factors, which cause loss of human life and property and damage to the environment. The distribution and change rules of geological disasters in time and space are subject to natural environment and human activities, and are often the result of the interaction between human beings and nature.

[0003] Geological disaster prediction is to predict and evaluate the development and change of geological disasters in a certain period of time and the possible risk and damage degree by professional personnel according to historical geological disaster activity rules, formation conditions, occurrence mechanism and disaster area disaster bearing capacity, and by using logical reasoning, numerical simulation and comprehensive analysis methods. At present, there is a technology for predicting geological disaster risk by using machine learning modeling. However, due to the neglect of model sensitivity and inaccurate selection of evaluation features, the risk evaluation effect is not ideal. SUMMARY

[0004] The present application aims to provide a geological disaster risk assessment method to solve the technical problem of the prior art that the risk evaluation effect is not ideal due to the neglect of model sensitivity and inaccurate selection of evaluation features.

[0005] To solve the above technical problems, the present application specifically provides the following technical solutions:

[0006] A geological disaster risk assessment method comprises:

[0007] Step S1, in a geological disaster record database, data sequences of risk elements before and after each geological disaster event are extracted, and the risk elements are generic features that can be used to evaluate the risk of geological disasters;

[0008] Step S2, connectivity analysis is performed on the data sequences of risk elements before and after the geological disaster event to screen out non-motive elements from the risk elements, and the non-motive elements are real features that can be used to evaluate the risk of geological disasters without disaster dominant motive before the geological disaster occurs;

[0009] Step S3, a CNN neural network is used to construct a geological disaster potential risk assessment model based on the non-motive elements to realize potential risk assessment before the geological disaster occurs;

[0010] Step S4, introducing 3D hollow convolution in the geological disaster potential risk evaluation model to construct a high-sensitivity potential risk evaluation model, so as to improve the evaluation sensitivity of the geological disaster potential risk evaluation model.

[0011] As a preferred scheme of the present application, the risk elements include geological conditions, topographic and geomorphic conditions, climate conditions, hydrological conditions, vegetation conditions and human activity conditions.

[0012] As a preferred scheme of the present application, the connectivity analysis on the data sequence of the risk elements before and after the geological disaster event includes:

[0013] The data connectivity of the geological disaster event before and after the geological disaster event is obtained by performing data connectivity calculation on the data sequence of each risk element before the geological disaster event and the data sequence of each risk element after the geological disaster event.

[0014] The calculation formula of the data connectivity is:

[0015] ;

[0016] In the formula, D k is the data connectivity of the kth risk element before and after the geological disaster event, S k,r is the rth data sequence point value of the kth risk element before the geological disaster, H k,r is the rth data sequence point value of the kth risk element after the geological disaster, n is the sequence point number of the data sequence, and i is a counting variable.

[0017] As a preferred scheme of the present application, the non-motive risk element is screened out from the risk elements, including:

[0018] The data connectivity of each risk element is compared with the connectivity threshold, wherein,

[0019] When the data connectivity of the risk element is higher than or equal to the connectivity threshold, the risk element is taken as the non-motive risk element.

[0020] When the data connectivity of the risk element is lower than the connectivity threshold, the risk element is taken as the non-motive risk element.

[0021] As a preferred scheme of the present application, the geological disaster potential risk evaluation model is constructed based on the non-motive risk element by using the CNN neural network, including:

[0022] The data sequence of the non-motive element before the occurrence of each geological disaster event is taken as the input item of the CNN neural network, and the data value of the disaster evaluation index of each geological disaster event is taken as the output item of the CNN neural network;

[0023] The input item and the output item of the CNN neural network are trained by using the CNN neural network to obtain the geological disaster potential risk evaluation model;

[0024] The model expression of the geological disaster potential risk evaluation model is:

[0025] F=CNN(L)

[0026] In the formula, F is a disaster evaluation index of a geological disaster event, L is a data sequence of a non-motive element before the occurrence of a geological disaster, and CNN is a neural network.

[0027] As a preferred scheme of the present application, a 3D hollow convolution is introduced into the geological disaster potential risk evaluation model to construct a high-sensitivity potential risk evaluation model, comprising:

[0028] Different 3D hollow convolution modules with different hollow rates are combined into a multi-scale hollow convolution module to respectively extract data features from the data sequence of the non-motive element, wherein the data feature information extracted by the different 3D hollow convolution modules with different hollow rates has different feature scales in the data dimension.

[0029] The multi-scale hollow convolution module fuses the data feature information with different feature scales to improve the acquisition of data feature information in the data sequence of the non-motive element with high information entropy.

[0030] The data feature information output by the multi-scale hollow convolution module is taken as the input of the geological disaster potential risk evaluation model to improve the evaluation sensitivity of the geological disaster potential risk evaluation model.

[0031] The relationship between the receptive field of the 3D hollow convolution module and the hollow rate is:

[0032] ;

[0033] wherein represents the receptive field of a single convolution kernel in the 3D hollow convolution module; represents the hollow rate, represents the size of the convolution kernel, and here The default setting is 3.

[0034] The feature extraction operation in the multi-scale hollow convolution module is realized by a 3D convolution network, wherein the first The value at the position of the i-th feature map in the j-th layer is calculated by the following formula:

[0035]

[0036] i,j are count variables.

[0037] The output of the multi-scale hollow convolution module is taken as the input of the geological disaster potential risk evaluation model to obtain the high-sensitivity potential risk evaluation model.

[0038] The specific parameters of the high-sensitivity potential risk evaluation model include that the learning rate is set to 1x10 -3 , the weight decay coefficient is set to 1x10 -6 , the momentum is set to 0.95, epsilon is set to 1x10 -5 , the epoch is set to 50, and the dropout is set to 0.45.

[0039] As a preferred scheme of the present application, each hazard element is normalized before connectivity analysis and construction of the geological disaster potential risk evaluation model.

[0040] As a preferred scheme of the present application, the disaster evaluation index includes disaster activity intensity or scale, disaster activity frequency, disaster distribution density, and disaster hazard intensity.

[0041] As a preferred scheme of the present application, the high-sensitivity potential risk evaluation model uses cross-entropy as a loss function and uses a stochastic gradient descent optimizer for training.

[0042] ​​​​​​​​​​​​​​​​​​​​​​As a preferred scheme of the present application, each sequence point in the data sequence of the risk element is sequentially arranged in time sequence.

[0043] Compared with the prior art, the present application has the following beneficial effects:

[0044] The present application performs connectivity analysis on the data sequence of the risk element before and after the geological disaster event, screens out the non-motive risk element from the risk element, ensures that the feature selection for the risk assessment is more accurate, introduces the 3D hollow convolution into the geological disaster potential risk assessment model to construct a high-sensitivity potential risk assessment model, and improves the evaluation sensitivity of the geological disaster potential risk assessment model, so that the accurate feature selection and the improved model sensitivity jointly guarantee the effect of the risk assessment. BRIEF DESCRIPTION OF DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for those skilled in the art, other drawings can be obtained from the provided drawings without creative labor.

[0046] Figure 1 The geological disaster risk assessment method flowchart provided by the present application is shown in the figure. DETAILED DESCRIPTION

[0047] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0048] Geological disaster prediction is to understand and predict the development and change of geological disasters in a certain period of time and the possible risk and damage loss degree by professional personnel according to the historical geological disaster activity law, formation condition, occurrence mechanism and disaster area disaster bearing capacity and other factors, and by using logical reasoning, numerical simulation and comprehensive analysis method. At present, the technology of predicting geological disaster risk by using machine learning modeling has appeared. However, due to the neglect of model sensitivity and the inaccurate selection of evaluation characteristics, the risk evaluation effect is not ideal. Therefore, the present application provides a geological disaster risk evaluation method, which fully mines the geological feature data before and after the geological disaster, realizes the selection of data characteristics for evaluating geological disasters, improves the accuracy of geological risk evaluation, and at the same time introduces 3D hollow convolution when constructing the risk evaluation model, improves the sensitivity of risk evaluation, so as to improve the timeliness of risk evaluation while improving the risk evaluation, and is more suitable for geological disaster evaluation which requires high timeliness, avoids more disaster losses caused by low sensitivity.

[0049] As shown in Figure 1 The present application provides a geological disaster risk evaluation method, comprising:

[0050] Step S1, in the geological disaster record database, the data sequence of the risk degree elements before and after each geological disaster event is extracted, and the risk degree element is a generic feature that can be used to evaluate the risk of geological disasters;

[0051] Step S2, the data sequence of the risk degree elements before and after the geological disaster event is analyzed to filter out the non-motive elements in the risk degree elements, and the non-motive elements are real features that can be used to evaluate the risk of geological disasters without disaster dominant motive before the occurrence of geological disasters;

[0052] Step S3, a CNN neural network is used to construct a geological disaster potential risk evaluation model based on the non-motive elements, so as to realize the potential risk evaluation before the occurrence of geological disasters;

[0053] Step S4, a 3D hollow convolution is introduced into the geological disaster potential risk evaluation model to construct a high-sensitivity potential risk evaluation model, so as to improve the evaluation sensitivity of the geological disaster potential risk evaluation model.

[0054] The application screens out the characteristic feature elements (i.e., non-motive elements) for the geological disaster risk assessment according to the connectivity of the risk elements before and after the geological disaster, can mine the real feature elements for evaluating the geological disaster risk under the condition that the disaster dominant motive does not exist before the geological disaster, can realize the assessment of the potential risk of the geological disaster, the non-motive elements realize the targeted matching before the geological disaster, the risk assessment scene is highly targeted, and therefore the risk assessment accuracy of the subsequent model construction is improved.

[0055] The application constructs the risk assessment model (the potential risk assessment model of the geological disaster) by using the non-motive elements, inherits the accuracy of the risk assessment effect and the targeted matching of the non-motive elements and the risk assessment scene, can ensure the timeliness of the scene matching after the accuracy is ensured, i.e., the risk of the disaster needs to be predicted more quickly before the geological disaster, the high timeliness can give a longer preparation time for disaster prevention, and therefore the application introduces the 3D hollow convolution to improve the risk assessment sensitivity of the potential risk assessment model of the geological disaster (the high-sensitivity potential risk assessment model).

[0056] The application uses the non-motive elements and the 3D hollow convolution to make the constructed high-sensitivity potential risk assessment model have the accuracy of the risk assessment and the timeliness of the risk assessment,

[0057] The risk elements include the geological conditions, the topographic and geomorphic conditions, the climate conditions, the hydrological conditions, the vegetation conditions and the human activity conditions.

[0058] The generic feature refers to the data feature with weak scene targeting but can be used for evaluating the risk of the geological disaster, and the real feature refers to the data feature with strong scene targeting and can be used for evaluating the risk of the geological disaster.

[0059] The application determines the data feature that can be used for evaluating the risk of the geological disaster under the condition that the disaster dominant motive does not exist before the geological disaster, i.e., determines the non-motive elements, obtains the data sequence of the risk elements before and after the geological disaster by the connectivity analysis, quantizes the connectivity analysis by using the data dispersion, the data sequence of the risk elements before and after the geological disaster has high connectivity, the dispersion degree of the data sequence of the risk elements before and after the geological disaster is high, the data time sequence regularity is in the split feature before and after the disaster, and therefore it is indicated that the occurrence of the geological disaster leads to a great change of the risk element, the risk of the disaster can be evaluated by using the risk element, i.e., the risk element is the data feature that can be used for evaluating the risk of the geological disaster under the condition that the disaster dominant motive does not exist before the geological disaster, and can be determined as the non-motive element, and the specific method for determining the non-motive element is as follows:

[0060] The data sequence of the dangerousness elements before and after the geological disaster event is analyzed for connectivity, including:

[0061] The data sequence of each dangerousness element before the geological disaster event is calculated for data connectivity with the data sequence of each dangerousness element after the geological disaster event, to obtain the data connectivity of each dangerousness element before and after the geological disaster event;

[0062] The calculation formula of the data connectivity is:

[0063] ;

[0064] In the formula, D k is the data connectivity of the kth dangerousness element before and after the geological disaster event, S k,r is the rth data sequence point value of the kth dangerousness element before the geological disaster, H k,r is the rth data sequence point value of the kth dangerousness element after the geological disaster, n is the number of sequence points of the data sequence, and i is a counting variable.

[0065] The data connectivity is quantified using data dispersion, wherein high data sequence connectivity of the dangerousness elements before and after the geological disaster indicates that the dispersion degree of the data sequence of the dangerousness elements before and after the geological disaster is high, and the data time sequence regularity is fragmented before and after the disaster, thus indicating that the occurrence of the geological disaster leads to a very large change in the dangerousness element, which can be used to evaluate the disaster risk.

[0066] Similarly, low data sequence connectivity of the dangerousness elements before and after the geological disaster indicates that the dispersion degree of the data sequence of the dangerousness elements before and after the geological disaster is low, and the data time sequence regularity is indistinguishable before and after the disaster, thus indicating that the occurrence of the geological disaster leads to no change in the dangerousness element, and it is difficult to determine whether the disaster has occurred or not when evaluating the disaster risk using the dangerousness element, i.e., the dangerousness element cannot be used to evaluate the disaster risk.

[0067] The non-motive elements are screened out from the dangerousness elements, including:

[0068] The data connectivity of each dangerousness element is compared with the connectivity threshold, wherein,

[0069] When the data connectivity of the dangerousness element is higher than or equal to the connectivity threshold, the dangerousness element is regarded as a non-motive element;

[0070] When the data connectivity of the dangerousness element is lower than the connectivity threshold, the dangerousness element is regarded as a non-non-motive element.

[0071] The CNN neural network is used to construct a geological disaster potential risk evaluation model based on non-motive elements, including:

[0072] The data sequence of the non-motive element before the occurrence of each geological disaster event is taken as the input item of the CNN neural network, and the data value of the disaster evaluation index of each geological disaster event is taken as the output item of the CNN neural network;

[0073] The input item of the CNN neural network and the output item of the CNN neural network are learned and trained by using the CNN neural network to obtain a geological disaster potential risk evaluation model;

[0074] The model expression of the geological disaster potential risk evaluation model is:

[0075] F=CNN(L)

[0076] In the formula, F is the disaster evaluation index of the geological disaster event, L is the data sequence of the non-motive element before the occurrence of the geological disaster, and CNN is the neural network.

[0077] In the process of geological disaster risk analysis, due to the different degrees of geological disaster occurrence and the resistance of geological structure, the data time sequence of the selected non-motive element contains a large amount of data feature information, which makes it difficult to obtain sufficient data feature information by using a fixed-size convolution kernel to construct a detection model, thereby reducing the recognition sensitivity. Therefore, the three-dimensional hollow convolution module is introduced into the geological disaster potential risk evaluation model to expand the receptive field of the data dimension, and different scales of data features are extracted by using the three-dimensional hollow convolution module, and finally fused calculation is performed to obtain more rich data information and improve the model detection performance. Specifically, the 3D hollow convolution is introduced into the geological disaster potential risk evaluation model to construct a high-sensitivity potential risk evaluation model, including:

[0078] Different three-dimensional hollow convolution modules with different hollow rates are combined into a multi-scale hollow convolution module to extract data features from the data sequence of the non-motive element, wherein the data feature information extracted by the three-dimensional hollow convolution modules with different hollow rates has different feature scales in the data dimension.

[0079] The multi-scale hollow convolution module fuses the data feature information of different feature scales obtained to improve the acquisition of data feature information in the high-information-entropy data sequence of the non-motive element.

[0080] The data feature information output by the multi-scale hollow convolution module is taken as the input of the geological disaster potential risk evaluation model to improve the evaluation sensitivity of the geological disaster potential risk evaluation model.

[0081] The relationship between the receptive field of the three-dimensional dilated convolution module and the dilated rate is as follows:

[0082] ;

[0083] wherein represents the receptive field of a single convolution kernel in the three-dimensional dilated convolution module; represents the dilated rate, represents the size of the convolution kernel, and here is set as 3 by default;

[0084] The feature extraction operation in the multi-scale dilated convolution module is realized by a 3D convolution network, wherein the calculation formula of the value of a position on the th feature map in the th layer of the 3D convolution network is as follows:

[0085] ;

[0086] wherein, is the value of the position on the th feature map in the th layer, , and are the height, width and depth of the convolution kernel in the th layer, is the activation function of the th layer, is the weight parameter connected to the position of the convolution kernel of the th feature map in the previous layer, is the value of the position on the th feature map in the previous layer, is the bias of the th feature map in the th layer, and i and j are counting variables;

[0087] The output of the multi-scale dilated convolution module is taken as the input of the geological disaster potential risk evaluation model to obtain a high-sensitivity potential risk evaluation model;

[0088] wherein, the specific parameters of the high-sensitivity potential risk evaluation model include that the learning rate is set as 1×10 -3 , the weight decay coefficient is set as 1×10 -6 , the momentum is set as 0.95, ε is set as 1×10 -5 , the epoch is set as 50, and the dropout is set as 0.45.

[0089] The various risk elements are normalized before connectivity analysis and construction of a geological disaster potential risk evaluation model.

[0090] Disaster evaluation indexes include disaster activity intensity or scale, disaster activity frequency, disaster distribution density and disaster damage intensity.

[0091] The high-sensitivity potential risk evaluation model uses cross-entropy as a loss function and uses a stochastic gradient descent optimizer for training.

[0092] The sequence points in the data sequence of the risk elements are sequentially arranged in time sequence.

[0093] The application performs connectivity analysis on the data sequence of the risk elements before and after the geological disaster event, screens out non-motive elements from the risk elements, ensures more accurate selection of features for risk evaluation, introduces 3D hollow convolution into the geological disaster potential risk evaluation model to construct a high-sensitivity potential risk evaluation model, improves the evaluation sensitivity of the geological disaster potential risk evaluation model, and the accurate selection of features and the improvement of the model sensitivity jointly guarantee the effect of the risk evaluation.

[0094] The above examples are only exemplary embodiments of the application and are not used to limit the application, and the protection scope of the application is defined by the claims. Those skilled in the art can make various modifications or equivalent replacements to the application within the spirit and protection scope of the application, and such modifications or equivalent replacements are also regarded as falling within the protection scope of the application.

Claims

1. A method for evaluating a geological disaster risk, characterized by, The method comprises the following steps: Step S1, extracting data sequences of each geological disaster event before and after the geological disaster in a geological disaster record database, wherein the data sequences are dangerous element data that can be used to evaluate the geological disaster risk; Step S2, performing connectivity analysis on the data sequences of the dangerous element before and after the geological disaster event, to screen out non-motive elements from the dangerous elements, wherein the non-motive elements are real-type features that can be used to evaluate the geological disaster risk without disaster dominant motive before the geological disaster occurs; Step S3, constructing a geological disaster potential risk evaluation model based on the non-motive elements by using a CNN neural network, to realize potential risk assessment before the geological disaster occurs; Step S4, introducing a 3D hollow convolution into the geological disaster potential risk evaluation model to construct a high-sensitivity potential risk evaluation model, to improve the evaluation sensitivity of the geological disaster potential risk evaluation model; The connectivity analysis on the data sequences of the dangerous element before and after the geological disaster event comprises: performing data connectivity calculation on the data sequences of each dangerous element before and after the geological disaster event, to obtain the data connectivity of each dangerous element before and after the geological disaster event; The calculation formula of the data connectivity is: ； In the formula, D k is the data connectivity of the kth hazard element before and after the geological disaster event, S k,r is the rth data sequence point value of the kth hazard element before the geological disaster, H k,r is the rth data sequence point value of the kth hazard element after the geological disaster, n is the number of sequence points of the data sequence, and i is a count variable.

2. The method according to claim 1, wherein: The dangerous elements include geological conditions, topographic and geomorphic conditions, climate conditions, hydrological conditions, vegetation conditions, and human activity conditions.

3. The method according to claim 2, wherein: The screening of the non-motive elements from the dangerous elements comprises: comparing the data connectivity of each dangerous element with a connectivity threshold value, wherein, when the data connectivity of the dangerous element is higher than or equal to the connectivity threshold value, the dangerous element is regarded as a non-motive element; when the data connectivity of the dangerous element is lower than the connectivity threshold value, the dangerous element is regarded as a non-motive element.

4. The geological disaster risk assessment method according to claim 3, characterized in that: The construction of the geological disaster potential risk evaluation model based on the non-motive elements by using the CNN neural network comprises: taking the data sequences of the non-motive elements before the geological disaster of each geological disaster event as the input items of the CNN neural network, and taking the data values of the disaster evaluation indexes of each geological disaster event as the output items of the CNN neural network; training the input items and the output items of the CNN neural network by using the CNN neural network to obtain the geological disaster potential risk evaluation model; The model expression of the geological disaster potential risk evaluation model is: F=CNN(L) In the formula, F is the disaster evaluation index of the geological disaster event, L is the data sequence of the non-motive elements before the geological disaster, and CNN is the neural network.

5. The geological disaster risk assessment method according to claim 1, characterized in that: The introduction of the 3D hollow convolution into the geological disaster potential risk evaluation model to construct a high-sensitivity potential risk evaluation model comprises: The three-dimensional hollow convolution module with different void ratios is combined as a multi-scale hollow convolution module to extract data features from the data sequence of the non-motive element, wherein the data feature information extracted by the three-dimensional hollow convolution module with different void ratios has different feature scales in the data dimension; The multi-scale hollow convolution module fuses the data feature information with different feature scales to improve the acquisition of data feature information in the data sequence of the non-motive element with high information entropy; The data feature information output by the multi-scale hollow convolution module is used as the input of the geological disaster potential risk evaluation model to improve the evaluation sensitivity of the geological disaster potential risk evaluation model; The relationship between the receptive field of the three-dimensional hollow convolution module and the void ratio is: ； wherein represents the receptive field of a single convolution kernel in a three-dimensional dilated convolution module; represents the dilated rate, represents the size of the convolution kernel, here is set to 3 by default; The feature extraction operation in the multi-scale cavity convolution module is realized by a 3D convolution network, wherein the calculation formula of the value of a position on the first layer of the 3D convolution network is as follows: The calculation formula of the value of a position on the first layer of the 3D convolution network is as follows: ； in, For the first Layer Location on each feature map The value at that location, , and It is the first The height, width, and depth of the convolutional kernel. For the first The activation function of the layer, It is connected to the previous level. Convolution kernel positions of feature maps The weight parameters at that location, It is the next level Location on each feature map The value at that location, It is the first Layer The deviation of each feature map, where i and j are count variables; The output of the multi-scale hollow convolution module is used as the input of the geological disaster potential risk evaluation model to obtain the high-sensitivity potential risk evaluation model; wherein the specific parameters of the high-sensitivity potential danger evaluation model include a learning rate set to 1x10 -3 , a weight decay coefficient set to 1x10 -6 , a momentum set to 0.95, an epsilon set to 1x10 -5 , an epoch set to 50, and a dropout set to 0.

45.

6. The method according to claim 1, wherein, Each hazard element is normalized before connectivity analysis and construction of the geological disaster potential risk evaluation model.

7. The method according to claim 4, wherein the method is characterized by, The disaster evaluation index includes disaster activity intensity or scale, disaster activity frequency, disaster distribution density, and disaster damage intensity.

8. The geological disaster risk assessment method according to claim 5, characterized in that, The high-sensitivity potential risk evaluation model uses cross-entropy as the loss function and uses the stochastic gradient descent optimizer for training.

9. The method according to claim 2, wherein, The sequence points in the data sequence of the hazard element are sequentially arranged in time sequence.

Images