A flood disaster chain scenario simulation method and system based on dynamic Bayesian
By constructing a dynamic Bayesian network model, considering the impact of response measures on disaster chain scenarios, the uncertainty and complexity of the disaster chain evolution process are solved, and more accurate disaster chain scenario deduction and scientific response plan selection are achieved.
Patent Information
- Application Number
- CN202411216461.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-02
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2044-09-02
AI Technical Summary
It is difficult for the prior art to effectively predict and control the evolution process of disaster chains, especially under response measures, the uncertainty and complexity of disaster chain scenario evolution are difficult to predict.
Using dynamic Bayesian-based flood disaster chain scenario deduction method, build an ontological model of urban heavy rain and waterlogging disaster chain scenarios, and consider the impact of response measures on disaster chain scenarios through dynamic Bayesian network model to improve the accuracy of disaster chain scenario deduction sex.
By taking into account a variety of influencing factors more comprehensively, the accuracy of disaster chain scenario deduction is improved and scientific disaster response plan selection is supported.
Smart Images

Figure CN118735137B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of natural disaster technology, and in particular to a method and system for simulating flood disaster chain scenarios based on dynamic Bayesian analysis. Background Art
[0002] Compared with a single disaster, disaster chains often have higher uncertainty. On the one hand, the evolution process of the disaster chain itself is uncertain. On the other hand, the evolution process of the disaster chain scenario under the response measures is uncertain. The impact mechanism of different response measures on the evolution of disaster chains is complex, and the evolution and development of disaster chains are difficult to predict and control.
[0003] For the analysis of the evolution process of disaster chains, existing studies mostly start from the perspective of "scenario" and use scenario deduction methods to deduce the dynamic and complex evolution process of disaster chains. The research methods are mainly based on rule reasoning and network analysis. Among them, the network analysis method with Bayesian network as the core is the most common method for disaster chain scenario deduction. At present, most of the disaster scenario deduction research focuses on the path of disaster chain evolution. At the same time, the research content mainly considers various factors that affect the evolution of disaster chains, such as disaster-causing factors, disaster-bearing bodies, and disaster-prone environments. However, there are still deficiencies in the aspects of attention and consideration. Summary of the invention
[0004] In view of this, this application provides a flood disaster chain scenario deduction method and system based on dynamic Bayesian, which aims to consider the impact of response measures on the evolution of disaster chain scenarios, and to more comprehensively consider the impact of various influencing factors on the disaster chain scenario deduction process, thereby improving the accuracy of disaster chain scenario deduction and achieving scientific selection of various disaster response plans.
[0005] The present application embodiment provides a method for deducing a scenario of a rainstorm and waterlogging disaster chain in an urban area based on dynamic Bayesian method, the method comprising:
[0006] Constructing an ontology model for urban rainstorm and waterlogging disaster chain scenarios , where H represents the disaster scenario element, Indicates the scenario elements of the intermediate hazard-bearing body, represents the final disaster-bearing body scenario element, SE represents the sub-event scenario element, and C represents the consequence scenario element;
[0007] According to the ontology model and historical disaster events of urban rainstorm waterlogging, a disaster chain scenario of urban rainstorm waterlogging is constructed;
[0008] According to the urban rainstorm waterlogging disaster chain scenario, the attributes of the disaster-bearing body and the response measures, a dynamic Bayesian network model corresponding to the urban rainstorm waterlogging disaster chain scenario is constructed, wherein the dynamic Bayesian network model includes scenario element nodes, attribute nodes and response measure nodes;
[0009] According to the scenario element nodes and countermeasure nodes with known states in the dynamic Bayesian network model, the probability that the scenario element nodes to be predicted in the dynamic Bayesian network model are in the target state in the time period to be predicted is determined.
[0010] Optionally, based on the ontology model and historical urban rainstorm waterlogging disaster events, a disaster chain scenario of urban rainstorm waterlogging is constructed, including:
[0011] Identify the scenario elements in historical urban rainstorm waterlogging disaster events, including disaster-causing factors, disaster-bearing bodies and disaster conditions;
[0012] According to the response relationship between the hazard-bearing body and the disaster situation, all hazard-bearing bodies are divided into intermediate hazard-bearing bodies and final hazard-bearing bodies, and all disaster situations are divided into sub-events and consequences;
[0013] According to the ontology model, the obtained disaster factor scenario elements, intermediate disaster-bearing body scenario elements, final disaster-bearing body scenario elements, sub-event scenario elements and consequence scenario elements, as well as the relationship between scenario elements, a city rainstorm waterlogging disaster chain scenario is constructed.
[0014] Optionally, according to the urban rainstorm waterlogging disaster chain scenario, the attributes of the disaster-bearing body and the response measures, a dynamic Bayesian network model corresponding to the urban rainstorm waterlogging disaster chain scenario is constructed, including:
[0015] Map the scenario element set in the urban rainstorm waterlogging disaster chain scenario to the node set of the dynamic Bayesian network model;
[0016] According to the attributes and countermeasures of each disaster-prone body scenario element, corresponding attribute nodes and countermeasure nodes are created for the nodes in the node set mapped by the disaster-prone body scenario element, so as to obtain all node sets of the dynamic Bayesian network model;
[0017] Mapping the interaction relationship between scenario elements in the urban rainstorm waterlogging disaster chain scenario into a directed edge set of the dynamic Bayesian network model;
[0018] Constructing an initial network structure of a dynamic Bayesian network model according to the target node set and the directed edge set;
[0019] Determining dynamic nodes in the initial network structure according to whether the states of the nodes in the initial network structure change over time;
[0020] Creating a time-directed edge of the dynamic node according to the state change characteristics of the dynamic node;
[0021] According to the determined dynamic nodes and the created time directed edges, the transfer network structure of the dynamic Bayesian network model is constructed;
[0022] Determining a dynamic Bayesian network structure based on the initial network structure and the transfer network structure;
[0023] Based on historical disaster events of urban rainstorms and waterlogging, determining probability information of each node in the dynamic Bayesian network structure;
[0024] A dynamic Bayesian network model is constructed according to the dynamic Bayesian network structure and the probability information.
[0025] Optionally, determining the dynamic nodes in the initial network structure according to whether the states of the nodes in the initial network structure change over time includes:
[0026] Determining whether the state of the nodes in the initial network structure changes over time;
[0027] In the case where the state of a node in the initial network structure changes over time, the node is determined to be a dynamic node.
[0028] Optionally, creating a time directed edge of the dynamic node according to the state change characteristic of the dynamic node includes:
[0029] Determine whether the state of a dynamic node changes over time due to the influence of other nodes;
[0030] When the state of the dynamic node changes over time without being affected by other nodes, create a first time directed edge between the dynamic node and itself;
[0031] In the case where the state of a dynamic node changes over time due to the influence of other nodes, determining the change state of the dynamic node;
[0032] In the case where the dynamic node is affected by other nodes and changes with time delay, creating a second time-directed edge between the dynamic node and the other nodes;
[0033] In the case where the dynamic node is affected by other nodes and changes slowly over time, creating a third time-directed edge between the dynamic node and the other nodes;
[0034] The step of constructing a transfer network structure of a dynamic Bayesian network model according to the determined dynamic nodes and the created time directed edges includes:
[0035] According to the determined dynamic nodes and the created first-time directed edges, second-time directed edges, and third-time directed edges, a transfer network structure of the dynamic Bayesian network model is constructed.
[0036] Optionally, based on historical disaster events of urban rainstorms and waterlogging, the probability information of each node in the dynamic Bayesian network structure is determined, including:
[0037] According to historical disaster events of urban rainstorm and waterlogging, the number of historical samples of each node in the dynamic Bayesian network structure is determined;
[0038] When a node has a parent node and the number of historical samples corresponding to the node exceeds a set threshold, the conditional probability of the node is determined by performing statistical analysis on the number of historical samples;
[0039] When a node has no parent node and the number of historical samples corresponding to the node exceeds a set threshold, the prior probability of the node is determined by performing statistical analysis on the number of historical samples;
[0040] When a node has a parent node and the number of historical samples corresponding to the node does not exceed the set threshold, the conditional probability of the node being in various states when its parent node is in various states in various time periods is determined by expert scoring of the node;
[0041] When a node has no parent node and the number of historical samples corresponding to the node does not exceed a set threshold, the prior probability of the node is determined by expert scoring of the node.
[0042] Optionally, the conditional probability of the node being in various states when its parent node is in various states in various time periods is determined by expert scoring of the node, including:
[0043] Each expert scores the probability that the node is in the corresponding state when its parent node is in the corresponding state in the corresponding time period, and obtains the scoring result;
[0044] The scoring result is processed by an evidence theory algorithm to obtain the conditional probability that the node is in the corresponding state. The expression of the evidence theory algorithm is: ,in, is a constant in the synthesis rule, used to normalize the synthesis result. is the normalization factor, , it indicates that the evidence is contradictory and the expression is invalid. For Node The status group, Is assigned to the status group The probability group, For the node Conditional probability synthesis.
[0045] Optionally, determining the probability that the scenario element node to be predicted in the dynamic Bayesian network model is in a target state in the time period to be predicted according to the scenario element nodes and the countermeasure nodes whose states are known in the dynamic Bayesian network model, includes:
[0046] Determining the scenario element nodes to be predicted, the target state to be predicted and the time period to be predicted in the dynamic Bayesian network model;
[0047] Determine the disaster factor scenario element nodes and the response measure nodes in the dynamic Bayesian network model as nodes with known states;
[0048] The probability that the scenario element node to be predicted is in the target state to be predicted in the time period to be predicted is determined by a target algorithm, and the target algorithm is: ,in, Represents the i-th node in time period t The parent node set The set in the qth state When the i-th node In the time period t, the The conditional probability of Represents the i-th node in time period t The parent node set The set in the qth state The probability of Represents the i-th node in time period t The parent node set includes the nodes in time period t-1 , the node set in time period ts , node set The node set of delayed-action countermeasures is the node set from time period ts to time period t-1. , node set is the node set of slow-acting countermeasures, the node set in time period t , node set A collection of ordinary nodes.
[0049] A second aspect of the embodiment of the present application provides a system for deducing urban rainstorm and waterlogging disaster chain scenarios based on dynamic Bayesian methods, the system comprising:
[0050] Ontology model building module, used to build an ontology model of urban rainstorm and waterlogging disaster chain scenarios , where H represents the disaster scenario element, Indicates the scenario elements of the intermediate hazard-bearing body, represents the final disaster-bearing body scenario element, SE represents the sub-event scenario element, and C represents the consequence scenario element;
[0051] A scenario construction module, used to construct a disaster chain scenario of urban rainstorm waterlogging according to the ontology model and historical disaster events of urban rainstorm waterlogging;
[0052] A dynamic Bayesian network model construction module is used to construct a dynamic Bayesian network model corresponding to the urban rainstorm waterlogging disaster chain scenario according to the urban rainstorm waterlogging disaster chain scenario, the attributes of the disaster-bearing body and the response measures, wherein the dynamic Bayesian network model includes scenario element nodes, attribute nodes and response measure nodes;
[0053] The probability prediction module is used to determine the probability that the scenario element node to be predicted in the dynamic Bayesian network model is in the target state in the time period to be predicted based on the scenario element nodes and countermeasure nodes with known states in the dynamic Bayesian network model.
[0054] Optionally, the scenario building module includes:
[0055] A scenario element determination module is used to identify scenario elements in historical disaster events of urban rainstorm and waterlogging, wherein the scenario elements include disaster-causing factors, disaster-bearing bodies and disaster conditions;
[0056] A scenario element division module is used to divide all hazard-bearing bodies into intermediate hazard-bearing bodies and final hazard-bearing bodies according to the response relationship between the hazard-bearing bodies and the disaster situation, and to divide all disaster situations into sub-events and consequences;
[0057] The scenario construction submodule is used to construct the urban rainstorm waterlogging disaster chain scenario based on the ontology model, the obtained disaster factor scenario elements, the intermediate disaster-bearing body scenario elements, the final disaster-bearing body scenario elements, the sub-event scenario elements and the consequence scenario elements, as well as the relationship between the scenario elements.
[0058] Compared with the prior art, this application has the following advantages:
[0059] The embodiment of the present application provides a method for deducing a chain of urban rainstorm and waterlogging disaster scenarios based on dynamic Bayesian. Constructing an ontology model of a chain of urban rainstorm and waterlogging disaster scenarios , where H represents the disaster scenario element, Indicates the scenario elements of the intermediate hazard-bearing body, represents the final disaster-bearing body scenario element, SE represents the sub-event scenario element, and C represents the consequence scenario element; according to the ontology model and historical disaster events of urban rainstorm waterlogging, an urban rainstorm waterlogging disaster chain scenario is constructed, and the urban rainstorm waterlogging disaster chain scenario includes the above-mentioned various types of scenario elements and the interaction relationship between each scenario element; according to the urban rainstorm waterlogging disaster chain scenario, the attributes of the disaster-bearing body and the response measures, a dynamic Bayesian network model corresponding to the urban rainstorm waterlogging disaster chain scenario is constructed, and the network of the dynamic Bayesian network model includes scenario element nodes, attribute nodes and response measure nodes; according to the scenario element nodes and response measure nodes with known states in the dynamic Bayesian network model, the probability that the scenario element nodes to be predicted in the dynamic Bayesian network model are in the target state during the time period to be predicted is determined. Therefore, when constructing a dynamic Bayesian network model corresponding to the urban rainstorm and waterlogging disaster chain scenario, the present application creates corresponding response measure nodes and attribute nodes for the disaster-prone scenario elements, so as to consider the response measures that will affect the state of each scenario element during the disaster chain scenario deduction process, thereby more comprehensively considering the impact of various influencing factors on the disaster chain scenario deduction process, thereby improving the accuracy of the disaster chain scenario deduction and achieving scientific selection of various disaster response plans.
[0060] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS
[0061] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art are briefly introduced below.
[0062] Figure 1 A flow chart of a method for simulating a chain of urban rainstorm and waterlogging disaster scenarios based on dynamic Bayesian analysis provided in an embodiment of the present application;
[0063] Figure 2 A schematic diagram of an example of a city rainstorm and waterlogging disaster chain scenario constructed by a method for deducing a city rainstorm and waterlogging disaster chain scenario based on dynamic Bayes provided in an embodiment of the present application;
[0064] Figure 3 A schematic diagram of constructing an initial network in a method for deducing a urban rainstorm and waterlogging disaster chain scenario based on dynamic Bayesian analysis provided in an embodiment of the present application;
[0065] Figure 4 A schematic diagram of constructing time-directed edges in a method for deducing a chain of urban rainstorm and waterlogging disaster scenarios based on dynamic Bayesian analysis provided in an embodiment of the present application;
[0066] Figure 5 A schematic diagram of a system for simulating urban rainstorm and waterlogging disaster chain scenarios based on dynamic Bayesian analysis provided in an embodiment of the present application. DETAILED DESCRIPTION
[0067] Exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings.
[0068] Figure 1 A flow chart of a method for simulating a rainstorm and waterlogging disaster chain scenario based on dynamic Bayesian analysis is provided in an embodiment of the present application. Figure 1 As shown, the method includes:
[0069] Step S1: Construct an ontology model of the urban rainstorm waterlogging disaster chain scenario , where H represents the disaster scenario element, Indicates the scenario elements of the intermediate hazard-bearing body, It represents the final disaster-bearing body scenario element, SE represents the sub-event scenario element, and C represents the consequence scenario element.
[0070] In this embodiment, the scenario in the urban rainstorm waterlogging disaster chain scenario is a set of information describing the state and trend of things, where the state refers to the current real-time state, and the trend refers to the future development trend. As a dynamic concept, the scenario is widely used in the field of disasters to describe the complex process of disaster chain evolution. The narrow sense of the disaster chain is centered on disasters and the triggering relationship between disasters, focusing on a single disaster or multiple disasters, and a series of impacts that disasters bring to human society. Since the formation and evolution of the disaster chain is not only the result of the interaction between disasters, but also involves the influence of multiple factors such as disaster-causing factors, disaster-bearing bodies, and disaster-pregnant environments. Therefore, the urban rainstorm waterlogging disaster chain scenario constructed by this application is different from the previous scenarios that focus on the triggering relationship between disasters (such as rainstorms causing waterlogging, and waterlogging causing traffic interruptions). The urban rainstorm waterlogging disaster chain scenario constructed by this application includes at least multiple elements such as disaster-causing factors, disaster-pregnant environments, and disaster-bearing bodies, which can reveal the dynamic evolution process of a series of disaster chain scenarios in which the disaster-causing factors (rainstorms) point to intermediate disaster-bearing bodies, intermediate disaster-bearing bodies point to sub-events, sub-events point to the final disaster-bearing bodies, and the final disaster-bearing bodies point to the consequences. In order to construct such a city rainstorm waterlogging disaster chain scenario in this application, this application first defines an ontology model of the city rainstorm waterlogging disaster chain scenario, which is expressed as Among them, H represents the disaster factor scenario element. For the urban rainstorm waterlogging disaster chain, the disaster factor is rainstorm; The intermediate disaster-bearing body scenario element refers to the disaster-bearing body that will become a potential disaster-causing factor causing new disasters after being hit by a disaster. Common intermediate disaster-bearing bodies include roads, buildings and equipment. It should be understood that intermediate disaster-bearing bodies can also include other entities, which are not specifically limited here; It represents the scenario element of the final disaster-bearing body, which refers to the disaster-bearing body that will not cause new disasters after being affected by the disaster. Common final disaster-bearing bodies include property and residents. It should be understood that the final disaster-bearing body can also include other subjects, which are not specifically limited here; SE represents sub-event, which refers to the human factors caused by the disaster-bearing body that may cause personal injury, property loss and resource damage. Common sub-events of the urban rainstorm disaster chain include road waterlogging, building water intrusion, equipment interruption, etc. It should be understood that sub-events can also include other subjects, which are not specifically limited here; C represents consequence, which refers to all destructive results caused by the final disaster-bearing body. Common consequences in the urban rainstorm disaster chain mainly include property loss and personal injury. It should be understood that the consequence can also include other subjects, which are not specifically limited here.
[0071] Step S2: constructing an urban rainstorm waterlogging disaster chain scenario based on the ontology model and historical urban rainstorm waterlogging disaster events.
[0072] In this embodiment, after the ontology model of the urban rainstorm waterlogging disaster chain scenario is constructed in step S1, based on the ontology model, various scenario elements in the ontology model are extracted from historical urban rainstorm waterlogging disaster events, and then the interaction relationship between the various scenario elements is created to obtain the corresponding urban rainstorm waterlogging disaster chain scenario. For example, Figure 2 As shown, Figure 2 A schematic diagram of a simple urban rainstorm waterlogging disaster chain scenario constructed based on a dynamic Bayesian urban rainstorm waterlogging disaster chain scenario deduction method provided by an embodiment of the present application is shown. First, a historical urban rainstorm waterlogging disaster event is obtained. Then, the historical urban rainstorm waterlogging disaster event is a rainstorm disaster event. The rainstorm disaster event causes road waterlogging, and the road waterlogging causes travel obstruction for some residents and traps some residents outdoors. Therefore, based on the constructed ontology model, the rainstorm that is a disaster-causing factor can be extracted from the historical urban rainstorm waterlogging disaster event, and the intermediate disaster-bearing body "road" that will cause new disasters after the disaster is extracted, and the "road waterlogging" sub-event caused by the disaster of the intermediate disaster-bearing body is extracted, and the final disaster-bearing body "residents" that will not cause new disasters after the disaster is extracted, and the consequences of "travel obstruction" and "trapped outdoors" caused by the final disaster-bearing body are extracted, and then the action relationship between each scenario element is created to obtain the final urban rainstorm waterlogging disaster chain scenario.
[0073] Step S3: According to the urban rainstorm waterlogging disaster chain scenario, the attributes of the disaster-bearing body and the response measures, a dynamic Bayesian network model corresponding to the urban rainstorm waterlogging disaster chain scenario is constructed, wherein the dynamic Bayesian network model includes scenario element nodes, attribute nodes and response measure nodes.
[0074] In this embodiment, after the corresponding urban rainstorm waterlogging disaster chain scenario is determined in step S2, in order to refine the impact of the disaster-causing factor or sub-event on the disaster-prone body, when creating the corresponding dynamic Bayesian model based on the determined urban rainstorm waterlogging disaster chain scenario, the present application adds an attribute node representing the attributes of the disaster-prone body to the disaster-prone body node in the urban rainstorm waterlogging disaster chain scenario, and the attribute node corresponding to the disaster-prone body is the parent node of the disaster-prone body, wherein the disaster-prone body includes an intermediate disaster-prone body and a final disaster-prone body. For example, Figure 3 As shown in the figure, for the road, an intermediate disaster-bearing body in the urban rainstorm waterlogging disaster chain scenario, the rainstorm affects the road by affecting its drainage capacity. The road's drainage capacity and ground elevation are important attributes that affect the sub-event of road waterlogging. Therefore, the two attributes of "drainage capacity" and "ground elevation" are added as parent nodes for the road, an intermediate disaster-bearing body.
[0075] At the same time, considering that different response measures will also affect the evolution process of the disaster chain scenario, this application also adds response measure nodes corresponding to the response measures that can be taken for the disaster-bearing body nodes in the urban rainstorm waterlogging disaster chain scenario. The response measure node corresponding to the disaster-bearing body is the parent node of a certain attribute of the disaster-bearing body, wherein the disaster-bearing body includes the intermediate disaster-bearing body and the final disaster-bearing body. The response measure nodes corresponding to the disaster-bearing body can be determined based on historical disposal experience and various plan texts. For example, the plan text describes that "when the yellow warning signal for heavy rain is issued... the drainage network operating unit and the sanitation operation unit are on duty on the flood-prone road section... all places are prepared for defense, especially the property management agency... remind residents to make defense preparations through broadcasting, posting flood prevention reminders and other measures." The relevant response measures extracted from the plan text are "checking and clearing the drainage network" and "issuing warnings", and the corresponding attributes of the intermediate disaster-bearing body of the road are extracted as "drainage capacity" and the corresponding attribute of the final disaster-bearing body of the resident is "resident exposure".
[0076] In this embodiment, after the corresponding urban rainstorm waterlogging disaster chain scenario is determined in step S2, the corresponding attributes of each disaster-bearing body and the corresponding countermeasures of each disaster-bearing body are constructed for each disaster-bearing body in the urban rainstorm waterlogging disaster chain scenario. Then, the node set consisting of the attributes and countermeasures of each scenario element and the disaster-bearing body scenario element in the urban rainstorm waterlogging disaster chain scenario is determined as the node set in the dynamic Bayesian network model, and the action relationship between the attributes and countermeasures of each scenario element and the disaster-bearing body scenario element in the urban rainstorm waterlogging disaster chain scenario is determined as the directed edge set in the dynamic Bayesian network model, so as to construct the initial dynamic Bayesian network model. Finally, it is determined that under the condition that the state of the disaster factor and the states of all the response measure nodes are constant, the prior probability of each node without a father node in the initial dynamic Bayesian network model being in various states in various time periods and the conditional probability of each node with a father node in the initial dynamic Bayesian network model being in various states in various time periods. After determining the conditional probability and prior probability of each node in the initial dynamic Bayesian network model being in various states in various time periods, a dynamic Bayesian network model consisting of the node set of the dynamic Bayesian network model, the directed edge set and the conditional probability and prior probability of each node being in various states in various time periods is obtained.
[0077] Step S4: According to the scenario element nodes and countermeasure nodes with known states in the dynamic Bayesian network model, determine the probability that the scenario element nodes to be predicted in the dynamic Bayesian network model are in the target state in the time period to be predicted.
[0078] In this embodiment, after the dynamic Bayesian network model is determined in step S3, the probability that the scenario element node to be predicted is in the target state in the time period to be predicted is determined based on the conditional probability and prior probability of each node in the dynamic Bayesian network according to the known state of the disaster-causing factor scenario element node and the known state of the response measure node of each disaster-prone body in the dynamic Bayesian network model. Among them, the scenario element node to be predicted can be any scenario element node other than the disaster-causing factor scenario element node in the dynamic Bayesian network model (such as road waterlogging), and the time period to be predicted can also be any time period after the response measures for each disaster-prone body in the dynamic Bayesian network model are in the corresponding state, and the target state can also be any state that the scenario element node to be predicted can have (such as shallow road waterlogging, deep road waterlogging, and medium depth of road waterlogging).
[0079] The embodiment of the present application provides a method for deducing a chain of urban rainstorm and waterlogging disaster scenarios based on dynamic Bayesian. Constructing an ontology model of a chain of urban rainstorm and waterlogging disaster scenarios , where H represents the disaster scenario element, Indicates the scenario elements of the intermediate hazard-bearing body, represents the final disaster-bearing body scenario element, SE represents the sub-event scenario element, and C represents the consequence scenario element; according to the ontology model and historical disaster events of urban rainstorm waterlogging, an urban rainstorm waterlogging disaster chain scenario is constructed, and the urban rainstorm waterlogging disaster chain scenario includes the above-mentioned various types of scenario elements and the interaction relationship between each scenario element; according to the urban rainstorm waterlogging disaster chain scenario, the attributes of the disaster-bearing body and the response measures, a dynamic Bayesian network model corresponding to the urban rainstorm waterlogging disaster chain scenario is constructed, and the network of the dynamic Bayesian network model includes scenario element nodes, attribute nodes and response measure nodes; according to the scenario element nodes and response measure nodes with known states in the dynamic Bayesian network model, the probability that the scenario element nodes to be predicted in the dynamic Bayesian network model are in the target state during the time period to be predicted is determined. Therefore, when constructing a dynamic Bayesian network model corresponding to the urban rainstorm waterlogging disaster chain scenario, the present application creates corresponding response measure nodes and attribute nodes for the disaster-prone body scenario elements, so as to consider the response measures that will affect the state of each scenario element in the process of disaster chain scenario deduction, so as to more comprehensively consider the impact of various influencing factors on the disaster chain scenario deduction process, thereby improving the accuracy of disaster chain scenario deduction and realizing the scientific selection of various disaster response plans. For example, by adjusting the response measures for the disaster-prone body, the probability of each scenario element node in the dynamic Bayesian network being in each state in each time period can be determined, so that it can be predicted which corresponding response measures can be adopted for the disaster-prone body in a city rainstorm waterlogging disaster chain state to make the state of each disaster-prone body safer, thereby making a more scientific response measure selection for the urban rainstorm waterlogging disaster chain state.
[0080] In combination with the above embodiments, in one implementation, the present application embodiment further provides a method for deducing a chain of urban rainstorm and waterlogging disaster scenarios based on dynamic Bayesian. In the method for deducing a chain of urban rainstorm and waterlogging disaster scenarios based on dynamic Bayesian, step S2 may include:
[0081] Step S21: Identify scenario elements in historical disaster events of urban rainstorm and waterlogging, wherein the scenario elements include disaster-causing factors, disaster-bearing bodies and disaster conditions.
[0082] In this embodiment, historical disaster events of urban rainstorms and waterlogging that have occurred are obtained, and then scenario elements in the historical disaster events are identified. The scenario elements include the disaster-causing factors, various disaster-bearing bodies, and various disaster conditions in the historical disaster events.
[0083] Step S22: According to the response relationship between the disaster-susceptible body and the disaster situation, all the disaster-susceptible bodies are divided into intermediate disaster-susceptible bodies and final disaster-susceptible bodies, and all the disaster situations are divided into sub-events and consequences.
[0084] In this embodiment, after identifying the disaster-causing factors, various disaster-bearing bodies and various disaster situations in historical disaster events of urban rainstorms and waterlogging, the disaster-bearing bodies are further divided, and the disaster-bearing bodies that will become potential disaster-causing factors that cause new disasters after being affected by the disaster are determined as intermediate disaster-bearing bodies, such as roads, buildings and equipment. The disaster-bearing bodies that will not cause new disasters after being affected by the disaster are determined as final disaster-bearing bodies, such as property and residents. At the same time, the disaster situation is further divided, and the disaster situation caused by the damage to the disaster-bearing body that may cause personal injury, property loss and resource damage is determined as a sub-event, such as waterlogging on roads, water inflow in buildings, equipment interruption, etc. All destructive results caused by the final disaster-bearing body are determined as consequences, such as property loss and personal injury.
[0085] Step S23: constructing an urban rainstorm waterlogging disaster chain scenario based on the ontology model, the obtained disaster factor scenario elements, the intermediate disaster-bearing body scenario elements, the final disaster-bearing body scenario elements, the sub-event scenario elements and the consequence scenario elements, and the relationship between the scenario elements.
[0086] In this embodiment, after identifying the disaster factor scenario elements, intermediate disaster-bearing body scenario elements, final disaster-bearing body scenario elements, sub-event scenario elements and consequence scenario elements in historical urban rainstorm waterlogging disaster events based on the ontology model, a corresponding urban rainstorm waterlogging disaster chain scenario is constructed based on the interaction relationship between various scenario elements. The urban rainstorm waterlogging disaster chain scenario is composed of nodes corresponding to various types of scenario elements and the interaction relationship between each node.
[0087] In combination with the above embodiments, in one implementation, the present application embodiment further provides a method for deducing a chain of urban rainstorm and waterlogging disaster scenarios based on dynamic Bayesian. In the method for deducing a chain of urban rainstorm and waterlogging disaster scenarios based on dynamic Bayesian, step S3 may include steps S31 to S39:
[0088] Step S31: Mapping the scenario element set in the urban rainstorm waterlogging disaster chain scenario to a node set of the dynamic Bayesian network model.
[0089] In this embodiment, the composition structure of the Bayesian network model includes a node set, a directed edge set formed by the relationship between nodes, and probability information of each node, the probability information including the prior probability of a node without a parent node and the conditional probability of a node with a parent node. After determining the urban rainstorm waterlogging disaster chain scenario, the present application maps the scenario element set composed of all scenario elements in the urban rainstorm waterlogging disaster chain scenario to a node set of the dynamic Bayesian network model.
[0090] Step S32: according to the attributes and countermeasures of each disaster-prone body scenario element, corresponding attribute nodes and countermeasure nodes are created for the nodes in the node set mapped by the disaster-prone body scenario element, so as to obtain all node sets of the dynamic Bayesian network model.
[0091] In this embodiment, the present application further determines the attributes of each disaster-prone scenario element and the corresponding coping measures that can be taken in the urban rainstorm and waterlogging disaster chain scenario based on historical disaster events of urban rainstorm and waterlogging.
[0092] In this embodiment, if Figure 3 As shown in the figure, after determining the attributes of each disaster-prone scenario element and the corresponding countermeasures that can be taken for each disaster-prone scenario element in the urban rainstorm waterlogging disaster chain scenario, the disaster-prone scenario element nodes mapped by the disaster-prone scenario elements in the determined dynamic Bayesian network model node set are created with attribute nodes corresponding to their own attributes and countermeasure nodes corresponding to the countermeasures that can be taken for themselves, so as to obtain all the node sets of the dynamic Bayesian network model, which is the node set needed to finally construct the dynamic Bayesian network model.
[0093] Step S33: Mapping the interaction relationship between scenario elements in the urban rainstorm waterlogging disaster chain scenario into a directed edge set of the dynamic Bayesian network model.
[0094] In this embodiment, if Figure 3 As shown, after the present application constructs the urban rainstorm waterlogging disaster chain scenario, the urban rainstorm waterlogging disaster chain scenario includes the interaction relationships between scenario elements and between scenario elements and attribute nodes and response measure nodes. At the same time, for the created attribute nodes and response measure nodes, the interaction relationships between attribute nodes and response measure nodes and each scenario element in the urban rainstorm waterlogging disaster chain scenario are created in the urban rainstorm waterlogging disaster chain scenario. Then, the present application will determine the interaction relationships between scenario elements and between scenario elements and attribute nodes and response measure nodes in the urban rainstorm waterlogging disaster chain scenario and map them into a directed edge set of the dynamic Bayesian network model, which is a directed edge set between each node in the target node set.
[0095] Step S34: constructing an initial network structure of a dynamic Bayesian network model according to the target node set and the directed edge set.
[0096] In this embodiment, the initial network of the dynamic Bayesian network model corresponding to the urban rainstorm waterlogging disaster chain scenario is constructed based on the target node set and directed edge set determined in the above steps S31 to S33. Figure 3 As shown, Figure 3The initial network structure of a dynamic Bayesian network model constructed in this application is shown.
[0097] Step S35: determining the dynamic nodes in the initial network structure according to whether the states of the nodes in the initial network structure change over time.
[0098] In this embodiment, after the initial network of the Bayesian network model is obtained, it is determined whether the state of each node in the initial network changes with time, and the dynamic nodes in the initial network are determined according to whether the state of the node changes with time.
[0099] In the present application, step S35 may include: determining whether the state of the node in the initial network structure changes over time; if the state of the node in the initial network structure changes over time, determining the node as a dynamic node.
[0100] In this embodiment, it is determined whether the state of the node in the initial network changes with time. If the state of the node in the initial network changes with time, the node is determined to be a dynamic node.
[0101] Step S36: creating a time directed edge of the dynamic node according to the state change characteristics of the dynamic node.
[0102] In this embodiment, a time-directed edge related to the dynamic node and time is created according to the state change characteristics of the dynamic node.
[0103] In the present application, step S36 may include: determining whether the state of the dynamic node is affected by other nodes and changes over time; when the state of the dynamic node is not affected by other nodes and changes over time, creating a first time directed edge between the dynamic node and itself; when the state of the dynamic node is affected by other nodes and changes over time, determining the changed state of the dynamic node; when the dynamic node is affected by other nodes and changes with time delay, creating a second time directed edge between the dynamic node and the other nodes; when the dynamic node is affected by other nodes and changes slowly over time, creating a third time directed edge between the dynamic node and the other nodes.
[0104] In this embodiment, if Figure 4As shown, after determining the dynamic nodes in the initial network, determine whether the state of the dynamic node changes over time due to the influence of other nodes. If the state of the dynamic node is not affected by other nodes and changes over time, it indicates that the state of the dynamic node itself changes over time. At this time, a first time directed edge pointing from the dynamic node to itself is created. Such nodes are mainly rainfall intensity nodes, road surface waterlogging nodes, etc. If the state of the dynamic node is affected by other nodes and changes over time, further determine the change state of such dynamic nodes to determine whether such dynamic nodes are affected by other nodes and change with time delay or change slowly over time. For such dynamic nodes, when the change state changes with time delay, create a second time directed edge between the dynamic node and the parent node that affects the change of the dynamic node with time delay. The parent node is mainly a countermeasure node. For such dynamic nodes, when the change state changes slowly over time, create a third time directed edge between the dynamic node and the parent node that affects the change of the dynamic node slowly over time. The parent node is mainly a countermeasure node. For example, as Figure 4 As shown, Figure 4 For the intermediate disaster-bearing body of the road surface drainage, when the emergency drainage measure is implemented on the road surface, the road surface drainage capacity will not be improved immediately, but it is necessary to continuously take emergency drainage measures. At the same time, when the emergency drainage is continuously carried out, it will affect the road surface drainage capacity after two time periods, that is, improve the road surface drainage capacity. Therefore, it is determined that the road surface drainage node is affected by the emergency drainage node and changes slowly over time, thereby creating a third time directed edge between the emergency drainage response measure node and the road surface drainage node. Figure 4 The number 2 in the box on the third time directed edge of the emergency drainage pointing to road surface drainage indicates that the number of time periods of the directed edge is 2. Figure 4 For the attribute node of residents' exposure, when the countermeasure of assisting transfer is implemented for residents, the residents' exposure will not be improved immediately after the countermeasure of assisting transfer is implemented, but the residents' exposure will be improved after a period of time after the countermeasure is collected. For example, it will have an impact on the residents' exposure after 2 time periods, that is, it will improve the residents' exposure. Therefore, it is determined that the attribute node of residents' exposure is affected by the parent node of assisting transfer and changes with time delay, thereby creating a second time directed edge between the countermeasure node of assisting transfer and the exposure attribute node. Figure 4 The directed edges from the first to the third time in the table record the number of corresponding cross-time periods. Figure 4 The number 2 in the box on the second time directed edge of the assisted transfer pointing to the exposure indicates that the number of time periods spanned by the directed edge is 2. Figure 4The numerical value in the box on each time-directed edge in represents the number of time periods spanned by the directed edge. The time step for the scenario element to change from one state to another is The implementation time of the measures is The time required for the measures to take effect is According to the time when the countermeasures take effect, the alternative countermeasures are divided into three categories, namely, slow-acting, immediate-acting and delayed-acting. After the implementation of the measures, it takes a long time for them to take full effect ( > ,exist It has been acting on the disaster-stricken body before the moment, and For delayed-action measures, these measures are After the implementation of the measures, it takes a long time for them to take effect ( > , the measures cannot take effect immediately. In this application, a time period is preferably 1 hour.
[0105] In this embodiment, the probability dependence between each response measure node and the node reflecting the attribute state of the disaster-prone body will change over time, and different response measures have different mechanisms of action, and there are also differences in the change of probability dependence between nodes over time. For example, taking "emergency drainage measures" on the road will improve the "road drainage capacity", and the probability of "emergency drainage → road drainage capacity" will change as the response measures gradually take effect. When the emergency drainage is just taken, the road drainage capacity may still be in an "overloaded" state. As the emergency drainage continues, the road drainage capacity gradually changes to "not overloaded". Other dynamic nodes include "building entrance and exit height", "resident exposure", "equipment status", etc. Among them, placing flood control sandbags and setting up flood control baffles will gradually increase the height of building entrances and exits, emergency evacuation will gradually reduce the exposure of residents, and emergency repair equipment will gradually change the equipment status. For the directed edge between the "response measure node" and the "disaster-prone body attribute node", the mechanism of the effectiveness of the response measures is further considered.
[0106] Step S37: constructing a transfer network structure of a dynamic Bayesian network model according to the determined dynamic nodes and the created time directed edges.
[0107] In this embodiment, after determining each dynamic node in the initial network and the created time directed edges, a transfer network structure of the dynamic Bayesian network model is constructed based on the determined dynamic nodes and all the created time directed edges.
[0108] In the present application, step S37 may include: constructing a transfer network structure of a dynamic Bayesian network model according to the determined dynamic nodes and the created first time directed edges, second time directed edges, and third time directed edges.
[0109] In this embodiment, when the first-time directed edges, second-time directed edges, and third-time directed edges of the dynamic nodes in the book publishing network are constructed through another implementation method of step S36, another implementation method of step S37 constructs the transfer network structure of the dynamic Bayesian network model based on the determined dynamic nodes and the created first-time directed edges, second-time directed edges, and third-time directed edges.
[0110] Step S38: Determine a dynamic Bayesian network structure based on the initial network structure and the transfer network structure.
[0111] In this embodiment, the initial network structure and the transfer network structure determined by steps S31 to S7 are merged to construct a dynamic Bayesian network result, such as Figure 4 As shown, Figure 4 A simple dynamic Bayesian network structure is shown.
[0112] Step S39: Based on historical disaster events of urban rainstorms and waterlogging, the probability information of each node in the dynamic Bayesian network structure is determined.
[0113] In this embodiment, based on a large number of historical disaster events of urban rainstorms and waterlogging that have occurred in the past, the probability information of each node in the constructed dynamic Bayesian network structure is determined. The probability information refers to the conditional probability of a node with a parent node being in various states in various time periods when the disaster factor scenario element node and the response measure node are each in a certain state, and the prior probability of a node without a parent node being in various states in various time periods.
[0114] Step S310: constructing a dynamic Bayesian network model according to the dynamic Bayesian network structure and the probability information.
[0115] In this embodiment, a dynamic Bayesian network model corresponding to the urban rainstorm waterlogging disaster chain scenario is constructed based on the probability information of each node in the determined dynamic Bayesian network structure.
[0116] In combination with the above embodiments, in one implementation, the embodiment of the present application further provides a method for deducing a chain of urban rainstorm and waterlogging disaster scenarios based on dynamic Bayesian. In the method for deducing a chain of urban rainstorm and waterlogging disaster scenarios based on dynamic Bayesian, step S39 may include steps S391 to S395:
[0117] Step S391: Determine the number of historical samples of each node in the dynamic Bayesian network structure based on historical disaster events of urban rainstorms and waterlogging.
[0118] In this embodiment, for the historical disaster events of urban rainstorms and waterlogging that occurred in the past, the number of occurrences of some nodes may be large, while the number of occurrences of other nodes may be relatively small (for example, the number of occurrences of the node of road waterlogging is generally large, while the number of occurrences of the node of emergency evacuation is generally small). For the nodes with a large number of occurrences, this application directly determines the conditional probability of the node and the prior probability of the node by statistical analysis of the historical disaster event data of urban rainstorms and waterlogging. For the nodes with a small number of occurrences, this application determines the conditional probability of the node and the prior probability of the node by expert scoring. Specifically, according to a large number of historical disaster events of urban rainstorms and waterlogging that occurred in the past, the number of occurrences of each node in the dynamic Bayesian network structure in the historical disaster events of urban rainstorms and waterlogging in the past is determined. According to the number of occurrences of each node in the historical disaster events of urban rainstorms and waterlogging in the past, the number of historical samples of each node is determined. For example, there are 100 historical disaster events of urban rainstorms and waterlogging that occurred in the past, of which 70 occurred in road waterlogging, and the number of historical samples of the node of road waterlogging in the corresponding determination network is 70.
[0119] Step S392: When a node has a parent node and the number of historical samples corresponding to the node exceeds a set threshold, the conditional probability of the node is determined by performing statistical analysis on the number of historical samples.
[0120] In this embodiment, when a node has a parent node and the number of historical samples corresponding to the node exceeds a set threshold, it is determined that the number of historical samples corresponding to the node is sufficient. At this time, based on the number of historical samples, the conditional probability that the parent node of the node is in any state under various state combinations can be determined through statistical analysis. For example, node a includes two states a1 and a2, and includes parent nodes A1 and A2. Parent node A1 includes two states A11 and A12. Parent node A2 includes two states A21 and A22. By performing statistical analysis on the historical sample data, it can be determined that the conditional probability that node a is in state a1 when parent node A1 is in state A11 and parent node A2 is in state A21; the conditional probability that node a is in state a2 when parent node A1 is in state A11 and parent node A2 is in state A21 can be determined, and so on.
[0121] Step S393: When a node has no parent node and the number of historical samples corresponding to the node exceeds a set threshold, the prior probability of the node is determined by performing statistical analysis on the number of historical samples.
[0122] In this embodiment, when a node has no parent node and the number of historical samples corresponding to the node exceeds a set threshold, it is determined that the number of historical samples corresponding to the node is sufficient. At this time, the prior probability of the node can be determined through statistical analysis based on the number of historical samples.
[0123] Step S394: When the node has a parent node and the number of historical samples corresponding to the node does not exceed the set threshold, the conditional probability of the node being in various states when its parent node is in various states in various time periods is determined by expert scoring of the node.
[0124] In this embodiment, when a node has a parent node and the number of historical samples corresponding to the node does not exceed the set threshold, it is determined that the number of historical samples corresponding to the node is not large enough. At this time, expert scoring is used to determine the conditional probability of the node being in various states in each time period when its parent node is in various states in each time period.
[0125] For example, the parent node of node a is only parent node A, the time period includes 2 time periods, parent node A has two different states A1 and A2, and node a has two states a1 and a2. The conditional probability of node a being in state a1 and the conditional probability of node a being in state a2 in the first time period when parent node A is in state A1 are determined by expert scoring; the conditional probability of node a being in state a1 and the conditional probability of node a being in state a2 in the second time period when parent node A is in state A1 are determined by expert scoring; the conditional probability of node a being in state a1 and the conditional probability of node a being in state a2 in the first time period when parent node A is in state A2 are determined by expert scoring; the conditional probability of node a being in state a1 and the conditional probability of node a being in state a2 in the second time period when parent node A is in state A2 are determined by expert scoring.
[0126] Step S395: When the node has no parent node and the number of historical samples corresponding to the node does not exceed a set threshold, the prior probability of the node is determined by expert scoring of the node.
[0127] In this embodiment, when a node has no parent node and the number of historical samples corresponding to the node does not exceed the set threshold, it is determined that the number of historical samples corresponding to the node is not large enough. At this time, the prior probability of the node is determined by expert scoring.
[0128] In combination with the above embodiments, in one implementation, the present application embodiment further provides a method for deducing a chain of urban rainstorm and waterlogging disaster scenarios based on dynamic Bayesian. In the method for deducing a chain of urban rainstorm and waterlogging disaster scenarios based on dynamic Bayesian, step S394 may include steps S3941 to S3942:
[0129] Step S3941: each expert scores the probability that the node is in the corresponding state when its parent node is in the corresponding state in the corresponding time period to obtain a scoring result.
[0130] In this embodiment, for any node, the conditional probability of the node being in various states when its parent node is in various states in various time periods is determined in the same manner, and the determination of the conditional probability of a node being in various states when its parent node is in a certain state in a certain time period is taken as an example for explanation. First, multiple experts score the conditional probability of the node being in various states when its parent node is in a certain state in a certain time period to obtain the scoring results of the multiple experts.
[0131] Step S3942: Process the scoring result by using an evidence theory algorithm to obtain the conditional probability that the node is in the corresponding state. The expression of the evidence theory algorithm is: ,in, is a constant in the synthesis rule, used to normalize the synthesis result. is the normalization factor, , it indicates that the evidence is contradictory and the expression is invalid. For Node The status group, Is assigned to the status group The probability group, For the node Conditional probability synthesis.
[0132] In this embodiment, after obtaining the scoring result in step S3941, the scoring result is processed by the evidence theory algorithm to obtain the conditional probability of the node being in various specific states when the parent node of the node is in a certain state in a certain time period. The expression of the evidence theory algorithm is: ,in, is a constant in the synthesis rule, used to normalize the synthesis result; is the normalization factor, When , it indicates that the evidence is contradictory and the expression is invalid; For Node A state group, which includes a set of states of all parent nodes of node A in a time period; Is assigned to the status group The probability group includes the probability that the i-th expert has an opinion on the node A in the state group Scoring of the conditional probabilities of being in various states; For the node The conditional probability synthesis is performed, that is, node A is in the state group The conditional probability of being in various states.
[0133] For example, as shown in Table 1 below, Table 1 shows that for the node of pavement drainage capacity, its parent nodes include heavy rain, inspection and dredging, and emergency drainage. When the status of the heavy rain parent node is "moderate rain", the status of the inspection and dredging parent node is "yes", the emergency drainage parent node is "yes" in the t-1 time period, and "no" in the t time period, the three experts scored the status of the pavement drainage capacity node as overloaded and non-overloaded, and the first expert scored , that is, the conditional probability of being in an overloaded state is 0.45, and the conditional probability of being in a non-overloaded state is 0.55; the second expert's score is , that is, the conditional probability of being in an overloaded state is 0.35, and the conditional probability of being in a non-overloaded state is 0.65; the score of the third expert is , that is, the conditional probability of being in an overload state is 0.35, and the conditional probability of being in a non-overload state is 0.65. Then the scoring result is calculated using the evidence theory algorithm to obtain , the expression is valid. At this time, when the parent node is in the various states mentioned in the above example, the conditional probability of the emergency extraction node being in the overload state is , the conditional probability that the emergency drainage node is in a non-overload state is When the status of the rainstorm parent node is "moderate rain", the status of the inspection and dredging parent node is "yes", and the emergency drainage parent node is "yes" in both time periods t-1 and t, the three experts scored the status of the road drainage capacity node as overloaded and non-overloaded, and the first expert scored , that is, the conditional probability of being in an overloaded state is 0, and the conditional probability of being in a non-overloaded state is 1; the score of the second expert is , that is, the conditional probability of being in an overloaded state is 0.05, and the conditional probability of being in a non-overloaded state is 0.95; the score of the third expert is , that is, the conditional probability of being in an overload state is 0.05, and the conditional probability of being in a non-overload state is 0.95. Then the scoring result is calculated using the evidence theory algorithm to obtain , the expression is valid. At this time, when the parent node is in the various states mentioned in the above example, the conditional probability of the emergency extraction node being in the overload state is , the conditional probability that the emergency drainage node is in a non-overload state is . From this we can find that Under the condition that emergency drainage measures are implemented in time period t, whether emergency drainage measures are continued in time period t will affect the probability of overload of pavement drainage capacity.
[0134] Table 1
[0135]
[0136] In combination with the above embodiments, in one implementation, the present application embodiment further provides a method for deducing a chain of urban rainstorm and waterlogging disaster scenarios based on dynamic Bayesian. In the method for deducing a chain of urban rainstorm and waterlogging disaster scenarios based on dynamic Bayesian, step S4 may include steps S41 to S43:
[0137] Step S41: Determine the scenario element nodes to be predicted, the target state to be predicted and the time period to be predicted in the dynamic Bayesian network model.
[0138] In this embodiment, first determine which scenario element node in the dynamic Bayesian network model the scenario element node to be predicted is, and at the same time determine what state of the node to be predicted is, determine the state as the target state, and at the same time determine which time period needs to be predicted to be in the target state.
[0139] Step S42: Determine the disaster factor scenario element nodes and the response measure nodes in the dynamic Bayesian network model as nodes with known states.
[0140] In this embodiment, after determining the scenario element node to be predicted in the dynamic Bayesian network model and determining the target state to be predicted and the time period to be predicted for the scenario element node to be predicted, the disaster factor scenario element node and the response measure node in the dynamic Bayesian network model are determined as nodes with known states.
[0141] For example, Figure 4 As shown in the figure, it is assumed that the scenario element node to be predicted is the node of road drainage capacity, the target state to be predicted is not overloaded, and the disaster factor scenario element node of rainstorm intensity is determined as a node with a known state, and the known state is moderate rain. Figure 4 As shown, since the countermeasures that can affect the state of the scenario element node to be predicted, which is the pavement drainage capacity, are only emergency pumping and inspection and dredging, only the countermeasure nodes that can affect the state of the scenario element node to be predicted can be determined as nodes with known states. At this time, the two countermeasure nodes, emergency pumping and inspection and dredging, are determined as nodes with known states, and the known state of inspection and dredging is "yes". Since there is a time-directed edge between the countermeasure node, emergency pumping, and the node, which is the pavement drainage capacity, the state of the node, which is the pavement drainage capacity, will change over time under the influence of the countermeasure node, which is the emergency pumping. Therefore, in order to know the state of the countermeasure node, it is necessary not only to know the state of the countermeasure node, but also to know the time period in which the countermeasure node takes measures. At this time, the known state of emergency pumping is determined to be that emergency pumping was not taken in the current time period t, but was taken in the time period t before the current time period. Among them, Figure 4 The value s of the time period on the second time directed edge represents that the state of the response parent node in the second time directed edge in the past time period ts will affect the state of the child node in the current time period t. Figure 4 The value s of the time period on the third time directed edge represents that the state of the response measure parent node in the third time directed edge in the past s consecutive time periods from ts to t-1 will affect the state of the child node in the current time period t.
[0142] Step S43: The probability that the scenario element node to be predicted is in the target state to be predicted in the time period to be predicted is determined by a target algorithm, and the target algorithm is: ,in, Represents the i-th node in time period t The parent node set The set in the qth state When the i-th node In the time period t, the The conditional probability of Represents the i-th node in time period t The parent node set The set in the qth state The probability of Represents the i-th node in time period t The parent node set includes the nodes in time period t-1 , the node set in time period ts , node set is the set of nodes for delayed-action countermeasures, from time period ts to time period t-1 , node set is the node set of slow-acting countermeasures, the node set in time period t , node set A collection of ordinary nodes.
[0143] In this embodiment, for the determined scenario element node to be predicted, the parent node of the scenario element node to be predicted may include multiple parent node sets having first-time directed edges, second-time directed edges, third-time directed edges, and common directed edge relationships, and the parent node of the scenario element node to be predicted may also include parent nodes having one or more of the four types of directed edge relationships. Figure 4 As shown, for the node of road surface water, it has a parent node (that is, itself) with a first-time directed edge relationship, and also has a parent node of the road surface with a normal directed edge relationship. Figure 4The number 1 in the box on the first time directed edge of the middle road surface water pointing to itself indicates that the number of time periods spanned by the directed edge is 1; for the road surface drainage node, it has a third time directed edge relationship, the emergency drainage response measure parent node, and at the same time, there are two parent nodes of the ordinary directed edge relationship, the inspection and dredging and the rainfall intensity. Figure 4 The number 2 in the box on the third time directed edge of the emergency drainage pointing to the road surface drainage indicates that the number of time periods spanning the directed edge is 2. Therefore, after determining the scenario element node to be predicted, the present application determines the probability of the scenario element node to be predicted being in the target state to be predicted in the time period to be predicted through the target algorithm, wherein the target algorithm is: ,in, Represents the i-th node in time period t The parent node set The set in the qth state When the i-th node In the time period t, the The conditional probability of Represents the i-th node in time period t The parent node set The set in the qth state The probability of Represents the i-th node in time period t The parent node set includes the nodes in time period t-1 , the node set in time period ts , node set The node set of delayed-action countermeasures is the node set from time period ts to time period t-1. , node set is the node set of slow-acting countermeasures, the node set in time period t , node set A collection of ordinary nodes.
[0144] In this embodiment, the target algorithm represents the parent node set. It includes a set of parent nodes that have a first-time directed edge with the node to be predicted, a set of parent nodes that have a second-time directed edge with the node to be predicted, a set of parent nodes that have a third-time directed edge with the node to be predicted, and a set of parent nodes that have a common directed edge with the node to be predicted. When performing actual calculations, the target algorithm is used to perform calculations based on the various parent node types of the scenario element node to be predicted in the dynamic Bayesian network model. When the parent node of the scenario element node to be predicted does not have a certain or certain parent node type in the dynamic Bayesian network, the corresponding target algorithm does not perform calculations on the certain or certain parent node types. For example, if the parent node of the scenario element node to be predicted in the dynamic Bayesian network model does not have a parent node with a first-time directed edge relationship, the nodes in the time period t-1 included in the parent node set of the target algorithm are excluded. If the parent node of the scenario element node to be predicted in the dynamic Bayesian network model does not have a parent node with a directed edge relationship at the second time, the node set under the time period ts included in the parent node set of the target algorithm is excluded. If the parent node of the scenario element node to be predicted in the dynamic Bayesian network model does not have a parent node with a third time directed edge relationship, the node set from the time period ts to the time period t-1 included in the parent node set of the target algorithm is excluded. ; If the parent node of the scenario element node to be predicted in the dynamic Bayesian network model does not have a parent node with a common directed edge relationship, the node set in time period t included in the parent node set of the target algorithm is excluded During the calculation process, the various conditional probabilities that are accumulated and summed in the target algorithm and the probabilities of the target algorithm being in various states have been determined through the implementation of the above steps S391 to S395.
[0145] The second aspect of the present application provides a scenario deduction system for urban rainstorm and waterlogging disaster chain based on dynamic Bayesian, such as Figure 5 As shown, the system 500 includes:
[0146] Ontology model building module 501, used to build an ontology model of urban rainstorm waterlogging disaster chain scenario , where H represents the disaster scenario element, Indicates the scenario elements of the intermediate hazard-bearing body, represents the final disaster-bearing body scenario element, SE represents the sub-event scenario element, and C represents the consequence scenario element;
[0147] A scenario construction module 502 is used to construct a disaster chain scenario of urban rainstorm waterlogging according to the ontology model and historical disaster events of urban rainstorm waterlogging;
[0148] A dynamic Bayesian network model construction module 503 is used to construct a dynamic Bayesian network model corresponding to the urban rainstorm waterlogging disaster chain scenario according to the urban rainstorm waterlogging disaster chain scenario, the attributes of the disaster-bearing body and the response measures, wherein the dynamic Bayesian network model includes scenario element nodes, attribute nodes and response measure nodes;
[0149] The probability prediction module 504 is used to determine the probability that the scenario element node to be predicted in the dynamic Bayesian network model is in the target state in the time period to be predicted based on the scenario element nodes and countermeasure nodes with known states in the dynamic Bayesian network model.
[0150] Optionally, the scenario construction module 502 includes:
[0151] A scenario element determination module is used to identify scenario elements in historical disaster events of urban rainstorm and waterlogging, wherein the scenario elements include disaster-causing factors, disaster-bearing bodies and disaster conditions;
[0152] A scenario element division module is used to divide all hazard-bearing bodies into intermediate hazard-bearing bodies and final hazard-bearing bodies according to the response relationship between the hazard-bearing bodies and the disaster situation, and to divide all disaster situations into sub-events and consequences;
[0153] The scenario construction submodule is used to construct the urban rainstorm waterlogging disaster chain scenario based on the ontology model, the obtained disaster factor scenario elements, the intermediate disaster-bearing body scenario elements, the final disaster-bearing body scenario elements, the sub-event scenario elements and the consequence scenario elements, as well as the relationship between the scenario elements.
[0154] Optionally, the dynamic Bayesian network model building module 503 includes:
[0155] The first mapping module is used to map the scenario element set in the urban rainstorm waterlogging disaster chain scenario into a node set of the dynamic Bayesian network model;
[0156] A node construction module is used to create corresponding attribute nodes and countermeasure nodes for nodes mapped by the disaster-prone body scenario elements in the node set according to the attributes and countermeasures of each disaster-prone body scenario element, so as to obtain all node sets of the dynamic Bayesian network model;
[0157] A second mapping module is used to map the action relationship between the scenario elements in the urban rainstorm waterlogging disaster chain scenario into a directed edge set of the dynamic Bayesian network model;
[0158] An initial network construction module, used to construct an initial network structure of a dynamic Bayesian network model according to a target node set and the directed edge set;
[0159] A dynamic node determination module, used to determine the dynamic nodes in the initial network structure according to whether the states of the nodes in the initial network structure change over time;
[0160] A time directed edge determination module, used for creating a time directed edge of a dynamic node according to a state change characteristic of the dynamic node;
[0161] A transfer network construction module is used to construct a transfer network structure of a dynamic Bayesian network model according to the determined dynamic nodes and the created time directed edges;
[0162] A network determination module, used to determine a dynamic Bayesian network structure based on the initial network structure and the transfer network structure;
[0163] A probability information determination module, used to determine the probability information of each node in the dynamic Bayesian network structure based on historical disaster events of urban rainstorms and waterlogging;
[0164] The dynamic Bayesian network model construction submodule is used to construct a dynamic Bayesian network model according to the dynamic Bayesian network structure and the probability information.
[0165] Optionally, the dynamic node determination module includes:
[0166] A first determination module is used to determine whether the state of the nodes in the initial network structure changes over time;
[0167] The second determination module is configured to determine a node as a dynamic node when the state of a node in the initial network structure changes over time.
[0168] Optionally, a time directed edge determination module includes:
[0169] A third determination module is used to determine whether the state of the dynamic node is affected by other nodes and changes over time;
[0170] A first time directed edge determination module, configured to create a first time directed edge between the dynamic node and itself when the state of the dynamic node changes over time without being affected by other nodes;
[0171] A change state determination module, used to determine the change state of the dynamic node when the state of the dynamic node is affected by other nodes and changes over time;
[0172] A second time directed edge determination module, configured to create a second time directed edge between the dynamic node and the other nodes when the dynamic node is affected by other nodes and changes with time delay;
[0173] A third time directed edge determination module, configured to create a third time directed edge between the dynamic node and the other nodes when the dynamic node is affected by other nodes and changes slowly over time;
[0174] The transfer network building module comprises:
[0175] The transfer network construction submodule is used to construct the transfer network structure of the dynamic Bayesian network model according to the determined dynamic nodes and the created first-time directed edges, second-time directed edges, and third-time directed edges.
[0176] Optionally, the probability information determination module includes:
[0177] A historical sample quantity determination module is used to determine the number of historical samples of each node in the dynamic Bayesian network structure according to historical disaster events of urban rainstorm and waterlogging;
[0178] A first probability information determination module, configured to determine the conditional probability of the node by performing statistical analysis on the number of historical samples when the node has a parent node and the number of historical samples corresponding to the node exceeds a set threshold;
[0179] A second probability information determination module is used to determine the prior probability of the node by performing statistical analysis on the number of historical samples when the node has no parent node and the number of historical samples corresponding to the node exceeds a set threshold;
[0180] A third probability information determination module is used to determine the conditional probability of the node being in various states when its parent node is in various states in various time periods by expert scoring the node when the node has a parent node and the number of historical samples corresponding to the node does not exceed a set threshold;
[0181] The fourth probability information determination module is used to determine the prior probability of the node by performing expert scoring on the node when the node has no parent node and the number of historical samples corresponding to the node does not exceed a set threshold.
[0182] Optionally, the third probability information determination module includes:
[0183] The expert scoring module is used to score the probability of a node being in a corresponding state when its parent node is in a corresponding state in a corresponding time period through each expert, and obtain a scoring result;
[0184] The third probability information determination submodule is used to process the scoring result through an evidence theory algorithm to obtain the conditional probability that the node is in a corresponding state. The expression of the evidence theory algorithm is: ,in, is a constant in the synthesis rule, used to normalize the synthesis result. is the normalization factor, , it indicates that the evidence is contradictory and the expression is invalid. For Node The status group, Is assigned to the status group The probability group, For the node Conditional probability synthesis.
[0185] Optionally, the probability prediction module 504 includes:
[0186] A module for determining information to be predicted, used to determine the scenario element nodes to be predicted, the target state to be predicted and the time period to be predicted in the dynamic Bayesian network model;
[0187] A known node determination module, used to determine the disaster factor scenario element nodes and the response measure nodes in the dynamic Bayesian network model as nodes with known states;
[0188] Probabilistic prediction submodule for The probability that the scenario element node to be predicted is in the target state to be predicted in the time period to be predicted is determined by a target algorithm, and the target algorithm is: ,in, Represents the i-th node in time period t The parent node set The set in the qth state When the i-th node In the time period t, the The conditional probability of Represents the i-th node in time period t The parent node set The set in the qth state The probability of Represents the i-th node in time period t The parent node set includes the nodes in time period t-1 , the node set in time period ts , node set The node set of delayed-action countermeasures is the node set from time period ts to time period t-1. , node set is the node set of slow-acting countermeasures, the node set in time period t , node set A collection of ordinary nodes.
[0189] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website site, a computer, a server or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or a data center that includes one or more available media integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive Solid State Disk (SSD)), etc.
[0190] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.
[0191] Each embodiment in this specification is described in a related manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.
[0192] The above description is only a preferred embodiment of the present application and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application are included in the protection scope of the present application.
Claims
1. A method for simulating urban rainstorm and waterlogging disaster chain scenarios based on dynamic Bayesian, characterized in that: The method comprises: Constructing an ontology model for urban rainstorm and waterlogging disaster chain scenarios , where H represents the disaster scenario element, Indicates the scenario elements of the intermediate hazard-bearing body, represents the final disaster-bearing body scenario element, SE represents the sub-event scenario element, and C represents the consequence scenario element; According to the ontology model and historical disaster events of urban rainstorm waterlogging, a disaster chain scenario of urban rainstorm waterlogging is constructed; According to the urban rainstorm waterlogging disaster chain scenario, the attributes of the disaster-bearing body and the response measures, a dynamic Bayesian network model corresponding to the urban rainstorm waterlogging disaster chain scenario is constructed, wherein the dynamic Bayesian network model includes scenario element nodes, attribute nodes and response measure nodes; Determine the probability that the scenario element node to be predicted in the dynamic Bayesian network model is in a target state in the time period to be predicted according to the scenario element nodes and the countermeasure nodes whose states are known in the dynamic Bayesian network model; According to the urban rainstorm waterlogging disaster chain scenario, the attributes of the disaster-bearing body and the response measures, a dynamic Bayesian network model corresponding to the urban rainstorm waterlogging disaster chain scenario is constructed, including: Map the scenario element set in the urban rainstorm waterlogging disaster chain scenario to the node set of the dynamic Bayesian network model; According to the attributes and countermeasures of each disaster-prone body scenario element, corresponding attribute nodes and countermeasure nodes are created for the nodes in the node set mapped by the disaster-prone body scenario element, so as to obtain all node sets of the dynamic Bayesian network model; Mapping the interaction relationship between scenario elements in the urban rainstorm waterlogging disaster chain scenario into a directed edge set of the dynamic Bayesian network model; Constructing an initial network structure of a dynamic Bayesian network model according to the target node set and the directed edge set; Determining whether the state of the nodes in the initial network structure changes over time; In the case where the state of a node in the initial network structure changes over time, determining the node as a dynamic node; Determine whether the state of a dynamic node changes over time due to the influence of other nodes; When the state of the dynamic node changes over time without being affected by other nodes, create a first time directed edge between the dynamic node and itself; In the case where the state of a dynamic node changes over time due to the influence of other nodes, determining the change state of the dynamic node; In the case where the dynamic node is affected by other nodes and changes with time delay, creating a second time-directed edge between the dynamic node and the other nodes; In the case where the dynamic node is affected by other nodes and changes slowly over time, creating a third time-directed edge between the dynamic node and the other nodes; According to the determined dynamic nodes and the created first-time directed edges, second-time directed edges, and third-time directed edges, a transfer network structure of the dynamic Bayesian network model is constructed; Determining a dynamic Bayesian network structure based on the initial network structure and the transfer network structure; Based on historical disaster events of urban rainstorms and waterlogging, determining probability information of each node in the dynamic Bayesian network structure; A dynamic Bayesian network model is constructed according to the dynamic Bayesian network structure and the probability information.
2. The method for simulating urban rainstorm and waterlogging disaster chain scenarios based on dynamic Bayesian analysis according to claim 1, characterized in that: Based on the ontology model and historical urban rainstorm waterlogging disaster events, a disaster chain scenario of urban rainstorm waterlogging is constructed, including: Identify the scenario elements in historical urban rainstorm waterlogging disaster events, including disaster-causing factors, disaster-bearing bodies and disaster conditions; According to the response relationship between the hazard-bearing body and the disaster situation, all hazard-bearing bodies are divided into intermediate hazard-bearing bodies and final hazard-bearing bodies, and all disaster situations are divided into sub-events and consequences; According to the ontology model, the obtained disaster factor scenario elements, intermediate disaster-bearing body scenario elements, final disaster-bearing body scenario elements, sub-event scenario elements and consequence scenario elements, as well as the relationship between scenario elements, a city rainstorm waterlogging disaster chain scenario is constructed.
3. The method for simulating urban rainstorm and waterlogging disaster chain scenarios based on dynamic Bayesian analysis according to claim 1, characterized in that: Based on historical disaster events of urban rainstorms and waterlogging, the probability information of each node in the dynamic Bayesian network structure is determined, including: According to historical disaster events of urban rainstorm and waterlogging, the number of historical samples of each node in the dynamic Bayesian network structure is determined; When a node has a parent node and the number of historical samples corresponding to the node exceeds a set threshold, the conditional probability of the node is determined by performing statistical analysis on the number of historical samples; When a node has no parent node and the number of historical samples corresponding to the node exceeds a set threshold, the prior probability of the node is determined by performing statistical analysis on the number of historical samples; When a node has a parent node and the number of historical samples corresponding to the node does not exceed the set threshold, the conditional probability of the node being in various states when its parent node is in various states in various time periods is determined by expert scoring of the node; When a node has no parent node and the number of historical samples corresponding to the node does not exceed a set threshold, the prior probability of the node is determined by expert scoring of the node.
4. The method for simulating urban rainstorm and waterlogging disaster chain scenarios based on dynamic Bayesian analysis according to claim 3 is characterized in that: By performing expert scoring on the node, the conditional probability of the node being in various states when its parent node is in various states in various time periods is determined, including: Each expert scores the probability that the node is in the corresponding state when its parent node is in the corresponding state in the corresponding time period, and obtains the scoring result; The scoring result is processed by an evidence theory algorithm to obtain the conditional probability that the node is in the corresponding state. The expression of the evidence theory algorithm is: ,in, is a constant in the synthesis rule, used to normalize the synthesis result. is the normalization factor, , it indicates that the evidence is contradictory and the expression is invalid. For Node The status group, Is assigned to the status group The probability group, For the node Conditional probability synthesis.
5. The method for simulating urban rainstorm and waterlogging disaster chain scenarios based on dynamic Bayesian analysis according to claim 1, characterized in that: According to the scenario element nodes and the countermeasure nodes whose states are known in the dynamic Bayesian network model, determining the probability that the scenario element nodes to be predicted in the dynamic Bayesian network model are in the target state in the time period to be predicted, including: Determining the scenario element nodes to be predicted, the target state to be predicted and the time period to be predicted in the dynamic Bayesian network model; Determine the disaster factor scenario element nodes and the response measure nodes in the dynamic Bayesian network model as nodes with known states; The probability that the scenario element node to be predicted is in the target state to be predicted in the time period to be predicted is determined by a target algorithm, and the target algorithm is: ,in, Represents the i-th node in time period t The parent node set The set in the qth state When the i-th node In the time period t, the The conditional probability of Represents the i-th node in time period t The parent node set The set in the qth state The probability of Represents the i-th node in time period t The parent node set includes the nodes in time period t-1 , the node set in time period ts , node set The node set of delayed-action countermeasures is the node set from time period ts to time period t-1. , node set is the node set of slow-acting countermeasures, the node set in time period t , node set A collection of ordinary nodes.
6. A scenario simulation system for urban rainstorm and waterlogging disaster chain based on dynamic Bayesian, characterized in that: The system comprises: Ontology model building module, used to build an ontology model of urban rainstorm and waterlogging disaster chain scenarios , where H represents the disaster scenario element, Indicates the scenario elements of the intermediate hazard-bearing body, represents the final disaster-bearing body scenario element, SE represents the sub-event scenario element, and C represents the consequence scenario element; A scenario construction module, used to construct a disaster chain scenario of urban rainstorm waterlogging according to the ontology model and historical disaster events of urban rainstorm waterlogging; A dynamic Bayesian network model construction module is used to construct a dynamic Bayesian network model corresponding to the urban rainstorm waterlogging disaster chain scenario according to the urban rainstorm waterlogging disaster chain scenario, the attributes of the disaster-bearing body and the response measures, wherein the dynamic Bayesian network model includes scenario element nodes, attribute nodes and response measure nodes; A probability prediction module, used to determine the probability that the scenario element node to be predicted in the dynamic Bayesian network model is in a target state in the time period to be predicted based on the scenario element nodes and the response measure nodes whose states are known in the dynamic Bayesian network model; Among them, the dynamic Bayesian network model building module includes: The first mapping module is used to map the scenario element set in the urban rainstorm waterlogging disaster chain scenario into a node set of the dynamic Bayesian network model; A node construction module is used to create corresponding attribute nodes and countermeasure nodes for nodes mapped by the disaster-prone body scenario elements in the node set according to the attributes and countermeasures of each disaster-prone body scenario element, so as to obtain all node sets of the dynamic Bayesian network model; A second mapping module is used to map the action relationship between the scenario elements in the urban rainstorm waterlogging disaster chain scenario into a directed edge set of the dynamic Bayesian network model; An initial network construction module, used to construct an initial network structure of a dynamic Bayesian network model according to a target node set and the directed edge set; A first determination module is used to determine whether the state of the nodes in the initial network structure changes over time; A second determination module is used to determine the node as a dynamic node when the state of the node in the initial network structure changes over time; A third determination module is used to determine whether the state of the dynamic node is affected by other nodes and changes over time; A first time directed edge determination module, configured to create a first time directed edge between the dynamic node and itself when the state of the dynamic node changes over time without being affected by other nodes; A change state determination module, used to determine the change state of the dynamic node when the state of the dynamic node is affected by other nodes and changes over time; A second time directed edge determination module, configured to create a second time directed edge between the dynamic node and the other nodes when the dynamic node is affected by other nodes and changes with time delay; A third time directed edge determination module, configured to create a third time directed edge between the dynamic node and the other nodes when the dynamic node is affected by other nodes and changes slowly over time; A transfer network construction submodule, used to construct a transfer network structure of a dynamic Bayesian network model according to the determined dynamic nodes and the created first-time directed edges, second-time directed edges, and third-time directed edges; A network determination module, used to determine a dynamic Bayesian network structure based on the initial network structure and the transfer network structure; A probability information determination module, used to determine the probability information of each node in the dynamic Bayesian network structure based on historical disaster events of urban rainstorms and waterlogging; The dynamic Bayesian network model construction submodule is used to construct a dynamic Bayesian network model according to the dynamic Bayesian network structure and the probability information.
7. The urban rainstorm waterlogging disaster chain scenario deduction system based on dynamic Bayesian analysis according to claim 6 is characterized in that: The scenario building module comprises: A scenario element determination module is used to identify scenario elements in historical disaster events of urban rainstorm and waterlogging, wherein the scenario elements include disaster-causing factors, disaster-bearing bodies and disaster conditions; A scenario element division module is used to divide all hazard-bearing bodies into intermediate hazard-bearing bodies and final hazard-bearing bodies according to the response relationship between the hazard-bearing bodies and the disaster situation, and to divide all disaster situations into sub-events and consequences; The scenario construction submodule is used to construct an urban rainstorm waterlogging disaster chain scenario based on the ontology model, the obtained disaster factor scenario elements, the intermediate disaster-bearing body scenario elements, the final disaster-bearing body scenario elements, the sub-event scenario elements and the consequence scenario elements, as well as the relationship between the scenario elements.
Citation Information
Patent Citations
Typhoon storm surge disaster chain catastrophe risk assessment method based on Bayesian network
CN117808302A
