River flood routing simulation and flood prevention emergency decision support system and method

Abstract

The invention relates to the technical field of water conservancy informatization and flood prevention and disaster reduction, in particular to a river flood routing simulation and flood prevention emergency decision support system and method.The system comprises edge space-time reference nodes deployed at key flood prevention nodes and a central decision correction server, and the data acquisition module is used for acquiring state data such as sluice opening and personnel transfer and executing a control instruction, and the control instruction is used for aligning heterogeneous data through a high-precision timestamp, constructing a digital twinning situation under a unified space-time reference, dynamically correcting a hydrodynamic model boundary condition based on an instruction execution deviation, and triggering resimulation and emergency decision updating. The invention provides a river flood routing simulation and flood prevention emergency decision support system and a river flood routing simulation and flood prevention emergency decision support method, which solve or at least alleviate the problem that a simulation result is disjointed from reality due to the lack of closed-loop feedback and unified time reference in simulation deduction and physical execution in an existing system.

Description

Technical Field

[0001] This invention belongs to the field of water conservancy informatization and flood control and disaster reduction technology, and in particular relates to a river flood evolution simulation and flood control emergency decision support system and method. Background Technology

[0002] With the intensification of global climate change and the increasing frequency of extreme weather events, river floods, as a major natural disaster, require scientific and timely early warning, simulation, and emergency decision-making. This is crucial for safeguarding people's lives and property and maintaining social stability. Therefore, river flood evolution simulation and flood control emergency decision support systems, integrating hydrology, hydrodynamics, computer science, and communication technologies, have emerged and become a key technological support in modern flood control systems. These systems aim to construct mathematical models by real-time monitoring of hydrological conditions such as rainfall and upstream inflows, combined with watershed geographic information and hydraulic structure parameters, to deduce the propagation and evolution of floods in rivers. This provides forward-looking predictions for key indicators such as the inundation range and peak arrival time in downstream areas, assisting decision-makers in formulating scientific flood control scheduling plans and emergency response plans.

[0003] In the development of existing technologies, researchers have developed two main technical paths to improve the accuracy of simulation predictions. The first is the real-time inversion technology based on remote sensing data. For example, the scheme disclosed in Chinese Patent CN112084712B, by fusing active and passive microwave remote sensing information, can accurately obtain the flood inundation range at a specific moment, providing valuable macro-scale measured data for the model and effectively solving the problems of model initialization and boundary condition setting. The second is numerical simulation technology based on physical mechanisms, such as the dam-break flood evolution prediction method based on the wave-breaking principle disclosed in Chinese Patent CN116663223B. This type of technology, by solving fundamental hydrodynamic equations such as the Saint-Venant equations, can theoretically characterize the complex motion laws of flood waves and achieve dynamic estimation of the flood evolution process. In practical applications, the two are usually combined, using remote sensing data to calibrate and verify the numerical model, forming the current mainstream monitoring-simulation-prediction technology paradigm.

[0004] However, as flood control operations place increasingly higher demands on the system's response timeliness and the scientific nature of decision-making, a deep-seated contradiction inherent in the aforementioned technological paradigm has begun to emerge. The root cause is that these systems are all based on an implicit assumption: the open-loop assumption of perfect command execution. Simulations based on remote sensing correction and hydrodynamic models are essentially observations and forward-looking deductions of physical reality in a twin digital space. When the decision support system issues flood control commands based on model simulation results, such as "ordering a certain sluice gate to open its gate to 50% flood discharge within 15 minutes" or "organizing the evacuation of all residents of a certain village within one hour," the model assumes that these commands are executed without errors within the issued time and are used as new boundary conditions to re-simulate the working conditions of the next period. This means that the simulation and deduction in the digital space and the command execution in the physical space are fundamentally separated, forming a one-way, non-feedback closed-loop control architecture. In actual flood control operations, the execution of instructions is fraught with uncertainty. For example, sluice gates may fail to open due to malfunctions, pumping stations may shut down due to power outages, and personnel evacuations may be delayed due to poor organization and weather conditions. Deviations may occur in the execution from the physical space to the virtual space. The fundamental reason is not the lack of a simple status feedback channel; the technical bottleneck lies in whether a mechanism or architecture can be found to guarantee the "time consistency" of status feedback information. In a widely distributed flood control system, different hydrological monitoring points, sluice gate actuators, and personnel resettlement points all have their own local clocks, and drift and errors are unavoidable. When these status data with timestamps containing certain errors reach the center, the delay in their convergence is unpredictable, making it impossible for the center to determine the true time of a status feedback message. If the center does not receive confirmation of a sluice gate execution status feedback after a preset time, it could mean that the sluice gate did not actually execute the opening command (execution failure); it could also be due to the time delay of the opening action itself (execution delay); or it could be that the opening was successfully completed but the status reporting information was stuck in network congestion (reporting delay). Without microsecond-level time synchronization, these three distinct properties are difficult to distinguish at the message level, and the central system cannot qualitatively and quantitatively classify deviations, thus failing to correct errors. Due to the lack of time consistency, any feedback-based correction cannot be implemented correctly, and may even introduce new errors due to error correction. Therefore, in the existing technological system, once a deviation occurs between the simulated world and physical reality, it will be continuously accumulated and amplified in the open-loop simulation chain over time, which may eventually lead to the simulation results deviating significantly from the actual water conditions, thereby weakening or even undermining the reliability of the entire decision support system.

[0005] Therefore, how to construct a closed-loop feedback mechanism that can incorporate the physical execution status of flood control instructions into the model evolution, and ensure the timeliness and reliability of feedback data through a unified high-precision time benchmark, so as to achieve dynamic coupling and continuous correction between the simulated world and physical reality, has become a key challenge and a technical problem that needs to be solved urgently by those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art, solve or at least alleviate the defects in the existing river flood evolution simulation and flood control decision support system, which are caused by the inability to provide real-time and accurate feedback of the physical execution status of flood control dispatching instructions to the digital simulation model, and the lack of a unified high-precision time reference, resulting in a disconnect between simulation prediction and reality and reduced decision reliability. The invention provides a river flood evolution simulation and flood control emergency decision support system and method.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a river flood evolution simulation and flood control emergency decision support system, comprising multiple edge spatiotemporal reference nodes and a central decision correction server; The edge spatiotemporal reference nodes are distributed and deployed at preset locations in the flood control area. Each edge spatiotemporal reference node is configured to synchronize with a unified time reference and is responsible for collecting the actual status information of the physical facilities connected to it, generating status data with high-precision timestamps, and executing control commands from the outside. The central decision correction server communicates bidirectionally with the multiple edge spatiotemporal reference nodes. The central decision correction server internally includes: an instruction issuance and tracking module, used to construct and issue flood control dispatch instructions carrying expected execution time windows to the target edge spatiotemporal reference nodes; a data aggregation and alignment module, used to receive and perform time-series alignment processing on the status data reported by the edge spatiotemporal reference nodes based on the high-precision timestamps; a hydrodynamic simulation engine, which integrates a hydrodynamic numerical model for simulating the evolution of river floods, and the boundary conditions or initial conditions of the numerical model are configured to be dynamically modified externally; a deviation identification and quantification engine, used to identify and quantify instruction execution deviations by comparing the flood control dispatch instructions with the time-series aligned status data, and generate deviation events; a model coupling interface module, used to, upon detecting the deviation event, convert the quantified instruction execution deviations into specific numerical updates to the boundary conditions or initial conditions of the hydrodynamic simulation engine, and trigger the hydrodynamic simulation engine to re-simulate; and an emergency decision generation module, used to receive the updated prediction results generated by the hydrodynamic simulation engine's re-simulation, and generate new flood control dispatch instructions accordingly.

[0008] To further realize the present invention, the following technical solutions may be preferred: Preferably, the hardware structure of the edge spatiotemporal reference node includes: a central processing unit, serving as the control core of the node; a high-precision time synchronization module, used to receive external standard time signals and generate hardware synchronization signals strictly aligned with the standard time signals; a data acquisition interface module, used to connect various types of hydrological sensors and status feedback devices to collect the actual status information of the physical facilities; an actuator control interface module, used to output control signals to external physical actuators; and a wireless communication module, used to realize bidirectional data exchange with the central decision correction server.

[0009] Preferably, the high-precision time synchronization module is a receiver module that supports receiving signals from a global navigation satellite system and has a second pulse signal output function; the output pin of the second pulse signal is directly connected to the external interrupt request pin of the central processing unit through a hardware circuit, and the rising edge of the second pulse signal is used to trigger the central processing unit to execute the highest priority interrupt service routine to achieve time synchronization.

[0010] Preferably, the method for establishing a high-precision time scale at the edge spatiotemporal reference node is as follows: the central processing unit is configured with a hardware timer; when the external interrupt request pin detects the rising edge of the second pulse signal, the central processing unit clears and resets the hardware timer in the interrupt service routine; the precise timestamp of any local event is composed of the global second-level timestamp recorded by the central processing unit and the count value of the hardware timer since the clearing.

[0011] Preferably, the data aggregation and alignment module of the central decision correction server uses the high-precision timestamp carried in the status data as the primary key to achieve accurate temporal alignment of heterogeneous data from different edge spatiotemporal reference nodes.

[0012] A method for simulating river flood evolution and supporting flood control emergency decision-making includes the following steps: constructing a distributed spatiotemporal reference network, the network consisting of multiple edge spatiotemporal reference nodes deployed at preset locations in the flood control area, and establishing a high-precision time scale for all edge spatiotemporal reference nodes in the network that traces back to a unified time reference; The central decision correction server issues flood control dispatch instructions with precise time constraints to the target edge spatiotemporal reference nodes. The precise time constraints define the expected execution time window of the instructions. After receiving the flood control dispatch instruction, the target edge spatiotemporal reference node drives the connected physical execution mechanism to perform corresponding actions, and uses the high-precision time scale to collect data on the actual execution status of the physical execution mechanism, generate a status data packet carrying a high-precision timestamp, and report the status data packet to the central decision correction server in real time. The central decision correction server aggregates and aligns the received status data packets in real time based on the high-precision timestamp, and compares them with the issued flood control dispatch instructions to identify and quantify the execution deviation between the actual execution status and the expected execution status of the instructions, and generates a structured deviation event report. Based on the identified and quantified execution deviation, the central decision correction server dynamically corrects the mathematical model parameters of its internal hydrodynamic simulation engine, which are the boundary conditions or initial conditions used by the hydrodynamic simulation engine to describe the state of the physical actuator. The hydrodynamic simulation engine is triggered to perform a rolling re-simulation based on the corrected mathematical model parameters, generating an updated set of flood evolution prediction results, and generating new flood control emergency decisions based on the updated flood evolution prediction results.

[0013] Preferably, the flood control dispatch instructions issued by the central decision correction server include a unique instruction identifier, a target node identifier, an action code, an action parameter set, an instruction issuance timestamp, an expected execution window start timestamp, and an expected execution window end timestamp.

[0014] Preferably, the status data packet reported by the edge spatiotemporal reference node includes a unique node identifier, event type encoding, status data payload, global second-level timestamp, and microsecond-level offset, wherein the global second-level timestamp and the microsecond-level offset together constitute the high-precision timestamp.

[0015] Preferably, the specific process of the deviation identification and quantification engine identifying and quantifying the execution deviation is as follows: for each issued flood control dispatch instruction, continuously monitor the status data stream reported by the target edge spatiotemporal reference node; after the expected execution window deadline of the instruction, compare the timestamp and data value of the latest received status data with the target status required by the instruction in time series; if the actual status does not reach the target status, it is determined as a deviation event, and the difference between the actual status and the target status in the time dimension or numerical dimension is calculated as the deviation quantification result.

[0016] Preferably, the process of the central decision correction server dynamically correcting the model and rolling resimulation based on the execution deviation also includes a graded triggering mechanism: according to the degree of impact of the execution deviation event on the global hydrological evolution, it is divided into different levels; if the deviation level is low, the hydrodynamic model of the local river section is recalculated; if the deviation level is high, the current global simulation process is interrupted, the global hydrological field at the time of the deviation occurs is used as the new initial condition, the corrected model parameters are used as the new boundary condition, and the hydrodynamic model recalculation covering the entire downstream risk area is started.

[0017] The beneficial effects of this invention are: The edge spatiotemporal reference node of this invention exchanges data bidirectionally with the central decision correction server via a wireless communication network. The server issues commands, the node executes them, and reports its status. This interaction process, guaranteed by the high-precision time synchronization mechanism, ensures absolute consistency of information flow in the time dimension, thus enabling closed-loop correction based on execution deviations. By incorporating the execution uncertainties of the physical world into the iterative evolution of the digital twin model, the entire system fundamentally solves the inherent defects of traditional open-loop systems, greatly improving the accuracy of flood simulation and prediction, and the scientific rigor and timeliness of emergency decision-making. Attached Figure Description

[0018] Figure 1 This is a block diagram of the overall architecture of the system of the present invention; Figure 2 This is a hardware structure block diagram of the edge spatiotemporal reference node of the present invention; Figure 3 This is a block diagram showing the internal functional modules of the central decision correction server of the present invention. Figure 4 This is a schematic diagram illustrating the principle of high-precision time synchronization achieved by the edge spatiotemporal reference node of the present invention. Figure 5 This is an overall flowchart of the method of the present invention; Figure 6 A detailed flowchart of the closed-loop correction and re-simulation process performed by the central decision correction server of this invention; Figure 7 This is a timing diagram illustrating the information interaction between the central decision correction server and the edge spatiotemporal reference node of the present invention. Figure 8 This is a schematic diagram of the closed-loop control principle of the present invention. Detailed Implementation

[0019] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Example 1

[0021] This embodiment discloses a river flood evolution simulation and flood control emergency decision support system, referring to... Figure 1 and Figure 2 Logically, the system consists of multiple edge spatiotemporal reference nodes deployed on-site and a central decision correction server located in the flood control command center. The edge spatiotemporal reference nodes, acting as the system's sensing and execution endpoints, are distributed across predetermined physical locations along the river, including key hydrological sections, sluice gates, pumping stations, control points of hydraulic structures, vulnerable sections of dikes, and emergency evacuation assembly points. The central decision correction server, as the system's decision-making hub, is responsible for data aggregation, analysis, simulation, decision generation, and command issuance. Both systems exchange data bidirectionally via highly reliable wireless communication networks, such as cellular networks based on Long Term Evolution (LTE-M / NB-IoT) or long-range low-power networks based on Linear Frequency Modulation (LoRa) technology.

[0022] Specifically, the hardware structure of the edge spatiotemporal reference node can be referred to Figure 3 As shown, each node is physically an embedded system unit, integrating a central processing unit (CPU), a high-precision time synchronization module, a data acquisition interface module, an actuator control interface module, and a wireless communication module on its internal circuit board. As a preferred implementation, the CPU is a 32-bit microcontroller based on the ARM Cortex-M4 core. This CPU is the control core of the entire node, responsible for running the embedded real-time operating system and scheduling various data acquisition, instruction execution, time synchronization, and communication tasks.

[0023] The high-precision time synchronization module is crucial for achieving a unified time base across the entire network. In one embodiment, this module utilizes the ZED-F9P Global Navigation Satellite System (GNSS) receiver module from U-blox, Switzerland. This module can simultaneously receive signals from multiple satellite systems, including GPS, GLONASS, Galileo, and BeiDou, and possesses raw observation output capabilities. More importantly, it provides a pulse-of-seconds (PPS) hardware signal output pin. The rising edge of this PPS signal is strictly aligned with the Coordinated Universal Time (UTC) second boundary, and its time jitter is strictly controlled within 30 nanoseconds. This PPS signal output pin is directly connected to the general-purpose input / output pins of the central processing unit, which support external interrupts and event capture, via a trace on an impedance-matched printed circuit board, thus providing the foundation for subsequent hardware-level timestamp synchronization.

[0024] The data acquisition interface module serves as a bridge for nodes to perceive the state of the physical world. It includes, but is not limited to, a 12-bit precision multi-channel analog-to-digital converter (ADC) for connecting sensors that output analog signals, such as pressure level gauges or radar flow meters that output 4-20mA standard industrial current signals. The module also includes multiple digital input / output ports for connecting switch signal sources, such as the operating status feedback dry contacts of a water pump motor or the switching signals of a dike seepage alarm. Furthermore, the module integrates multiple Universal Asynchronous Receiver / Transmitter (UARTs), a Serial Peripheral Interface (SPI), and internal integrated circuits for serial communication interfaces, enabling data exchange with devices possessing digital communication capabilities. For example, it can connect multiple gate opening sensors via an RS-485 bus or connect a radio frequency identification (RFID) reader for personnel verification via a UART interface.

[0025] The actuator control interface module is the output from which the node exerts control over the physical world. Depending on the command type, it can be configured in different hardware forms. For the start-stop control of equipment such as pumps and valves, this module can integrate a high-power solid-state relay array driven by optocoupler isolation to achieve safe control of the high-voltage circuit. For equipment requiring fine-tuning, such as control gates that need precise control of the opening angle, this module can integrate a 16-bit precision digital-to-analog converter (DAC), whose output 0-10V voltage signal or 4-20mA current signal can be directly used as the control input signal for downstream valve positioners or frequency converters.

[0026] The wireless communication module is responsible for the data link between the nodes and the central decision correction server. To adapt to communication requirements in different deployment environments, this module is preferably a dual-mode communication unit. One is a cellular communication module supporting LTE Cat-M1 or NB-IoT protocols, suitable for areas with cellular network coverage, providing wide-area, reliable data connectivity. The other is a low-power wide-area network module supporting the LoRaWAN protocol, operating in the unlicensed ISM band, suitable for mountainous areas or remote river sections with weak cellular signals. Nodes can aggregate data to a gateway node with cellular backhaul capabilities through ad-hoc or star networks, and then uniformly report to the server.

[0027] The central decision-making and correction server is the brain of the entire system, implemented through a complex software architecture. (See reference...) Figure 4 The server primarily comprises a data aggregation and alignment module, a command issuance and tracking module, a deviation identification and quantization engine, a hydrodynamic simulation engine, a model coupling interface module, and an emergency decision generation module. In a typical deployment, the server is a cluster of one or more high-performance physical servers, and the various functional modules are deployed using containerization technology (such as Docker) to achieve resource isolation and elastic scaling.

[0028] The underlying technology stack of the data aggregation and alignment module employs a high-throughput message queue middleware, such as Apache Kafka, as the entry point for data reported by all edge spatiotemporal reference nodes. The Kafka cluster reliably receives and persists massive amounts of state data packets. After passing through Kafka, the data is consumed in real-time by a stream processing application (such as one written using Apache Flink or Spark Streaming). This application is responsible for parsing the data packets and, using the high-precision global second-level timestamps and microsecond-level offsets carried within the packets, writing the data to a database specifically optimized for time-series data, such as InfluxDB or TimescaleDB. In this database, the timestamp is used as the primary key or core index, thus naturally achieving accurate time-series alignment of data from different nodes and of different types across the entire network.

[0029] The instruction issuance and tracking module maintains a relational database (such as PostgreSQL) to store detailed information on all issued flood control dispatch instructions, including the instruction's unique identifier, target node list, action parameters, issuance timestamp, expected execution time window, and current execution status (such as "issued," "received," "in execution," "completed," "execution deviation," etc.). This module provides a set of RESTful APIs for the emergency decision generation module to call, enabling secure and reliable generation and issuance of instructions.

[0030] The deviation identification and quantification engine is a crucial component in achieving closed-loop correction. Technically, this engine can be a Complex Event Processing (CEP) engine, such as Esper or a CEP library based on Apache Flink. It subscribes to real-time physical state data streams from the data aggregation and alignment module, as well as command status change event streams from the command issuance and tracking module. The engine internally predefines a series of association rules. For example, one rule could be defined as: "When a command to open 'Sluice Gate XX' changes to 'issued,' if, within a timeout threshold (e.g., 30 seconds) after the expected execution window deadline, the latest opening sensor value reported from the edge node corresponding to that sluice gate still does not reach 95% of the target opening value required by the command, then a 'partial execution' deviation event is triggered." Once the event is triggered, the engine immediately calculates the deviation (e.g., the difference between the target opening of 80% and the actual opening of 30%) and encapsulates information such as the deviation type, occurrence time, deviation amount, and associated command ID into a structured deviation event report, which is then published to a dedicated Kafka topic for downstream modules to consume.

[0031] The hydrodynamic simulation engine is the core computational component of the system. It integrates mature one-dimensional and two-dimensional hydrodynamic numerical model solvers. The one-dimensional model, based on Saint-Venant's equations, is suitable for simulating flood propagation processes in river networks and tributaries, offering high computational efficiency. The two-dimensional model, based on two-dimensional shallow water equations, employs the finite volume method on unstructured grids, enabling accurate simulation of flood inundation range and water depth distribution in complex terrains (such as floodplains). A key feature of this engine is that its model parameters, particularly the boundary conditions representing the operational status of hydraulic structures such as sluices and pumping stations (e.g., weir flow coefficient, gate opening height, pumping flow rate), are designed to be modified dynamically and in real-time via an external API.

[0032] The model coupling interface module acts as an adapter and translator between the deviation identification engine and the hydrodynamic simulation engine. It subscribes to deviation event reports. Upon receiving a deviation report, the module parses its content, identifying, for example, a deviation in the opening degree of "sluice gate XX". Then, it immediately calls the API provided by the hydrodynamic simulation engine to precisely update the boundary condition parameters representing the sluice gate in the model from the originally planned command values ​​(e.g., opening height of 8 meters) to the actual values ​​fed back by the sensors (e.g., opening height of 3 meters). After the parameter update is successful, the module also sends a "trigger recalculation" command to the simulation engine.

[0033] The emergency decision generation module embodies the system's intelligence. This module receives continuously updated flood evolution predictions from the hydrodynamic simulation engine. Internally, it contains a decision rule engine or an offline-trained reinforcement learning model. The module compares the latest prediction (e.g., the highest inundation level in a protected area within the next 3 hours) with preset flood control thresholds. If the predicted water level evolution is found to exceed the safety threshold, the module will automatically or semi-automatically generate new response instructions. For example, based on the rule in the expert knowledge base: "IF Predicted water level > Warning water level - 0.2 meters AND Time < 2 hours THEN Activate backup pumping station, set flow rate to 50 cubic meters per second," a new pumping station activation instruction is generated and issued through the instruction issuance and tracking module. Example 2

[0034] This embodiment discloses a method for simulating river flood evolution and providing flood control emergency decision support, referring to... Figures 6 to 8 .

[0035] The first step is to construct a distributed spatiotemporal reference network covering the entire flood control area, possessing microsecond-level time synchronization capabilities. In engineering practice, this step corresponds to the deployment and initialization of the aforementioned edge spatiotemporal reference nodes. After deployment, all nodes are powered on and initialized. (Refer to...) Figure 5The central processing unit (CPU) inside the node begins executing the firmware. First, it initializes the high-precision time synchronization module, waiting for it to successfully lock onto the GNSS satellite signal and begin stably outputting the PPS signal. Simultaneously, the MCU internally configures a 32-bit or 64-bit hardware timer / counter, allowing it to run freely at a frequency of 1MHz (i.e., 1 microsecond resolution) without any frequency division. The MCU also configures the GPIO pin connected to the PPS signal as an external interrupt input, setting it to the highest priority. When the rising edge of the first PPS signal triggers an interrupt through this pin, the interrupt service routine (ISR) is executed. In this ISR, the MCU parses the RMC (Recommended Minimum Specific GNSS data) statement corresponding to the PPS signal from the GNSS module's NMEA 0183 data stream via the UART interface, extracting the current global second-level timestamp (YYYY-MM-DD HH:MM:SS, UTC). The ISR immediately binds this global second-level timestamp with the current hardware timer count (which should be close to zero) as a base time tuple and stores it in SRAM, then instantly resets the hardware timer. From then on, whenever a PPS signal rises, the ISR repeats this process: recording the global second-level timestamp associated with that PPS signal and resetting the microsecond-level hardware timer. Through this mechanism, at any given time, when a node needs to timestamp a local event, it only needs to read the most recently stored global second-level timestamp and add the accumulated count value of its local microsecond-level timer since that PPS event. Thus, the timestamps of all nodes in the network can be traced back to the same UTC time base source, and the synchronization error between them is mainly determined by the jitter of the PPS signal itself and the MCU interrupt delay, which can be stably controlled within the range of 100 nanoseconds to 1 microsecond, laying a solid foundation for the spatiotemporal consistency analysis of all subsequent data.

[0036] The second step involves connecting the status feedback interfaces and execution control interfaces of various flood control facilities to the data acquisition interface module and actuator control interface module of the corresponding edge spatiotemporal reference node, respectively. For example, the angular displacement sensor of a sluice gate (which linearly converts the gate opening degree into a 4-20mA current signal) is connected to the node's ADC input channel; the auxiliary contact of the contactor controlling the sluice gate winch motor is connected to the node's GPIO input port; and the node's relay output terminal is connected to the contactor's coil control circuit. After completing the physical connection, a data acquisition task is configured for each sensor in the node firmware. This task can be periodic (e.g., acquiring water level once per second) or event-triggered (e.g., acquiring data immediately when the gate's operating state changes). The moment the acquisition action is triggered and completed (e.g., when an ADC conversion completion interrupt occurs), the firmware program immediately reads the current high-precision timestamp and encapsulates this timestamp and the acquired status data (e.g., the converted gate opening percentage) into a structured status data packet. After encapsulation, this data packet is sent to the transmission queue of the wireless communication module and reported to the central decision correction server in real time.

[0037] The third step involves the central decision-making correction server issuing flood control dispatch instructions with precise time constraints to the target edge spatiotemporal reference nodes. When the server's internal hydrodynamic simulation engine predicts, based on upstream water conditions, that a floodgate needs to be opened to reduce the flood peak, the emergency decision generation module constructs a structured dispatch instruction. This instruction includes not only the target object (such as a node), the action to be performed (such as "open the gate"), and related parameters (such as an opening degree of 80%), but also an instruction execution time window, which explicitly defines the expected start time stamp and the required completion time stamp. The server issues this instruction via a wireless communication network. Upon receiving the instruction, the target node immediately records its local high-precision timestamp and performs CRC verification and instruction parsing. If the instruction is valid and is a control instruction, the node sends control signals to the connected physical actuators through its actuator control interface module, initiating the actual physical action.

[0038] The fourth step involves the central decision-making correction server performing real-time aggregation and timing alignment of the received network-wide status data packets, and comparing them with the issued scheduling instructions to accurately identify and quantify instruction execution deviations. For example... Figure 8As shown, the server's data aggregation and alignment module continuously receives status data packets from various nodes. Utilizing the high-precision timestamps within these packets, data from different sources is precisely mapped onto a unified global timeline. Simultaneously, the deviation identification and quantification engine continues to operate. For the aforementioned floodgate opening command, the engine establishes a temporary monitoring window. It continuously monitors the data stream from node 0x1A2B3C4D with an event type of "gate opening feedback." The engine compares the actual opening time series with the expected state. The expected state is: before the set timestamp, the opening value should reach or exceed 80%. If, after this timestamp, the engine receives the latest opening data for the floodgate, with a timestamp after the set timestamp but an opening value of only 35.2%, the engine immediately determines that a composite deviation event has occurred: "delayed execution" and "partial execution." The engine quantifies this deviation: the difference between the two timestamps, resulting in an opening deviation of 44.8%. This information is encapsulated into a deviation event report and pushed to subsequent processing stages.

[0039] The final step, and the core of the closed-loop control of this invention, is to trigger the hydrodynamic simulation engine to dynamically correct the model and perform rolling re-simulation based on the identified and quantified instruction execution deviation events. For example... Figure 7As shown, when the model coupling interface module receives the aforementioned deviation event report, it immediately "translates" this abstract event into a specific adjustment to the mathematical description of the simulation model. Specifically, it calls the simulation engine's API to forcibly update the gate opening height parameter in the weir flow formula representing the spillway in the model from the originally planned command value (corresponding to 80% opening) to the actual value fed back (corresponding to 35.2% opening). This immediate update of the model parameters will immediately trigger a simulation recalculation. To ensure computational efficiency and response speed, the system adopts a tiered triggering mechanism. If the deviation occurs at a small tributary control gate with a limited impact range, the system may only initiate a rapid one-dimensional hydrodynamic recalculation for a local downstream river section. However, if the deviation occurs at a critical spillway on the main stream, as in this example, the system will immediately interrupt the current global simulation process, using the global hydrological field (water level and velocity distribution in each computational grid) at the time of the deviation as the new initial condition, and the corrected model parameters as the new boundary condition, to initiate a high-precision two-dimensional hydrodynamic model recalculation covering the entire downstream risk area. After recalculation, a completely new set of flood evolution predictions that more closely approximates physical reality (such as updated peak arrival time, highest inundation level, and inundation range) is generated and pushed to the emergency decision generation module in real time. Based on the new predictions, this module reassesses the effectiveness of the current flood control plan. If it is found that the original plan is no longer able to address new risks caused by execution deviations (for example, a protected area reaching the warning level two hours earlier due to inadequate flood discharge), the module will automatically generate new, adaptive emergency instructions, such as issuing an order to open nearby backup flood diversion gates ahead of schedule. These newly generated instructions also carry precise time constraints and are issued through the aforementioned process, thus forming a complete closed-loop control loop of continuous iteration and dynamic optimization, encompassing "simulation-decision-execution-perception-feedback-correction-resimulation." Example 3

[0040] Scenario setting: A 30-kilometer section of the downstream main stream of a certain river basin is selected for simulation. At the upstream entrance of this section, there is a key flood discharge gate A (monitored by node 01), 15 kilometers downstream, there is an important protected area, the water level of which is monitored by node 02, and at the downstream outlet, there is flood diversion gate B (monitored by node 03).

[0041] Initial conditions: The upstream reservoir will pre-discharge floodwater according to the weather forecast, which will cause the upstream flow of floodgate A to increase sharply at time T0.

[0042] Traditional open-loop system (comparative example): In the first 30 minutes (T0), the traditional decision support system of the flood control command center, based on hydrodynamic model calculations, formulated a scheduling plan: at time T0, an instruction was issued to floodgate A, requiring it to open to 100% within 10 minutes (i.e., before T0+600 seconds) to cope with the flood peak. The system model, based on the assumption of perfect execution of this instruction, predicted that the water level in the midstream protected area would reach its peak at T0+3 hours, but would still be 0.5 meters below the safe level. However, in the physical world, floodgate A, due to mechanical failure, jammed at 40% opening and could not be opened further. Because the traditional system lacked high-frequency, real-time status feedback and closed-loop correction mechanisms, its internal simulation evolution was completely disconnected from physical reality. It wasn't until T0+2 hours, when on-site patrol personnel in the protected area reported via walkie-talkie that the water level had exceeded the warning line, that the command center realized the problem. At this point, an emergency order was issued to open floodgate B downstream for compensation, but the optimal time had been missed, resulting in the flooding of some low-lying areas within the protected area.

[0043] Embodiments of the present invention: During the T0-30 minute period, the central decision correction server of this invention makes the same decision and issues an instruction to node 01 via wireless network: Target A floodgate, open to 100%, with an expected completion time window of [T0, T0+600 seconds].

[0044] At time T0+180, the opening of floodgate A reached 40% and then stopped increasing. Node 01 continuously reports status data packets with microsecond-level timestamps indicating "opening degree = 40%".

[0045] At T0+600 seconds, the instruction execution window closes. The server's deviation identification and quantification engine detects that the actual opening degree of floodgate A (40%) is far from the instruction target (100%), immediately classifying it as a "serious partial execution" deviation event and generating a deviation report.

[0046] At T0+600.1 seconds, the model coupling interface module receives the report and immediately forces the boundary conditions of the No. A spillway gate in the hydrodynamic simulation engine to be changed from 100% opening to 40% opening.

[0047] At time T0+600.2, the corrected model is triggered to undergo a global recalculation.

[0048] The recalculation was completed at T0+605 seconds. The new forecast results show that, without intervention, the water level in the protected area will exceed the safe level at T0+2.5 hours.

[0049] At T0+606 seconds, the emergency decision generation module automatically generated a new compensatory instruction based on the new forecast results: Target B flood diversion gate, immediately open to 80%, with an expected completion time window of [NOW, NOW+900 seconds]. This instruction was immediately issued to node 03.

[0050] Flood diversion gate B was successfully opened as instructed. Subsequent rolling simulations and real-time feedback data showed that, due to timely and precise compensation measures, the highest water level in the midstream protected area was successfully controlled 0.2 meters below the safe level, avoiding any flooding losses.

[0051] As can be seen from the above embodiments and data comparisons, the present invention, by constructing a perception-execution-feedback-correction closed loop based on a unified high-precision time benchmark, can capture the uncertainty of flood control command execution in real time and accurately, and dynamically couple it into the iterative evolution of the hydrodynamic model, thereby greatly improving the accuracy of flood evolution prediction, gaining valuable time windows for emergency decision-making, and significantly enhancing the scientific nature, initiative, and timeliness of flood control work.

[0052] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A river flood evolution simulation and flood control emergency decision support system, characterized in that, It includes multiple edge spatiotemporal reference nodes and a central decision correction server; The edge spatiotemporal reference nodes are distributed and deployed at preset locations in the flood control area. Each edge spatiotemporal reference node is configured to synchronize with a unified time reference and is responsible for collecting the actual status information of the physical facilities connected to it, generating status data with high-precision timestamps, and executing control commands from the outside. The central decision correction server communicates bidirectionally with the multiple edge spatiotemporal reference nodes. The central decision correction server internally includes: an instruction issuance and tracking module, used to construct and issue flood control dispatch instructions carrying expected execution time windows to the target edge spatiotemporal reference nodes; a data aggregation and alignment module, used to receive and perform time-series alignment processing on the status data reported by the edge spatiotemporal reference nodes based on the high-precision timestamps; a hydrodynamic simulation engine, which integrates a hydrodynamic numerical model for simulating the evolution of river floods, and the boundary conditions or initial conditions of the numerical model are configured to be dynamically modified externally; a deviation identification and quantification engine, used to identify and quantify instruction execution deviations by comparing the flood control dispatch instructions with the time-series aligned status data, and generate deviation events; a model coupling interface module, used to, upon detecting the deviation event, convert the quantified instruction execution deviations into specific numerical updates to the boundary conditions or initial conditions of the hydrodynamic simulation engine, and trigger the hydrodynamic simulation engine to re-simulate; and an emergency decision generation module, used to receive the updated prediction results generated by the hydrodynamic simulation engine's re-simulation, and generate new flood control dispatch instructions accordingly.

2. The system according to claim 1, characterized in that, The hardware structure of the edge spatiotemporal reference node includes: a central processing unit, serving as the control core of the node; a high-precision time synchronization module, used to receive external standard time signals and generate hardware synchronization signals strictly aligned with the standard time signals; a data acquisition interface module, used to connect various types of hydrological sensors and status feedback devices to collect the actual status information of the physical facilities; an actuator control interface module, used to output control signals to external physical actuators; and a wireless communication module, used to realize bidirectional data exchange with the central decision correction server.

3. The system according to claim 2, characterized in that, The high-precision time synchronization module is a receiver module that supports receiving signals from the Global Navigation Satellite System and has a second pulse signal output function. The output pin of the second pulse signal is directly connected to the external interrupt request pin of the central processing unit through hardware circuitry. The rising edge of the second pulse signal is used to trigger the central processing unit to execute the highest priority interrupt service routine to achieve time synchronization.

4. The system according to claim 3, characterized in that, The method for establishing a high-precision time scale at the edge spatiotemporal reference node is as follows: the central processing unit is equipped with a hardware timer; when the external interrupt request pin detects the rising edge of the second pulse signal, the central processing unit clears and resets the hardware timer in the interrupt service routine; the precise timestamp of any local event is composed of the global second-level timestamp recorded by the central processing unit and the count value of the hardware timer since the clearing.

5. The system according to claim 1, characterized in that, The data aggregation and alignment module of the central decision correction server uses the high-precision timestamp carried in the status data as the primary key to achieve accurate temporal alignment of heterogeneous data from different edge spatiotemporal reference nodes.

6. A method for simulating river flood evolution and providing flood control emergency decision support, applicable to the system described in any one of claims 1-5, characterized in that, Includes the following steps: A distributed spatiotemporal reference network is constructed, which consists of multiple edge spatiotemporal reference nodes deployed at preset locations in the flood control area, and a high-precision time scale traceable to a unified time reference is established for all edge spatiotemporal reference nodes in the network. The central decision correction server issues flood control dispatch instructions with precise time constraints to the target edge spatiotemporal reference nodes. The precise time constraints define the expected execution time window of the instructions. After receiving the flood control dispatch instruction, the target edge spatiotemporal reference node drives the connected physical execution mechanism to perform corresponding actions, and uses the high-precision time scale to collect data on the actual execution status of the physical execution mechanism, generate a status data packet carrying a high-precision timestamp, and report the status data packet to the central decision correction server in real time. The central decision correction server aggregates and aligns the received status data packets in real time based on the high-precision timestamp, and compares them with the issued flood control dispatch instructions to identify and quantify the execution deviation between the actual execution status and the expected execution status of the instructions, and generates a structured deviation event report. Based on the identified and quantified execution deviation, the central decision correction server dynamically corrects the mathematical model parameters of its internal hydrodynamic simulation engine, which are the boundary conditions or initial conditions used by the hydrodynamic simulation engine to describe the state of the physical actuator. The hydrodynamic simulation engine is triggered to perform a rolling re-simulation based on the corrected mathematical model parameters, generating an updated set of flood evolution prediction results, and generating new flood control emergency decisions based on the updated flood evolution prediction results.

7. The method according to claim 6, characterized in that, The flood control dispatch instructions issued by the central decision correction server include a unique instruction identifier, a target node identifier, an action code, an action parameter set, an instruction issuance timestamp, an expected execution window start timestamp, and an expected execution window end timestamp.

8. The method according to claim 6, characterized in that, The status data packet reported by the edge spatiotemporal reference node includes a unique node identifier, event type encoding, status data payload, global second-level timestamp, and microsecond-level offset, wherein the global second-level timestamp and the microsecond-level offset together constitute the high-precision timestamp.

9. The method according to claim 6, characterized in that, The specific process of the deviation identification and quantification engine identifying and quantifying the execution deviation is as follows: For each issued flood control dispatch instruction, continuously monitor the status data stream reported by the target edge spatiotemporal reference node; after the expected execution window deadline of the instruction, compare the timestamp and data value of the latest received status data with the target status required by the instruction in time series. If the actual status does not reach the target status, it is determined as a deviation event, and the difference between the actual status and the target status in the time dimension or numerical dimension is calculated as the deviation quantification result.

10. The method according to claim 6, characterized in that, The process of the central decision correction server dynamically correcting the model and rolling resimulation based on execution deviation also includes a graded triggering mechanism: according to the degree of impact of execution deviation events on the global hydrological evolution, they are divided into different levels; if the deviation level is low, the hydrodynamic model of the local river section is recalculated. If the deviation level is high, the current global simulation process will be interrupted, and the global hydrological field at the time of the deviation will be used as the new initial conditions, the corrected model parameters will be used as the new boundary conditions, and the hydrodynamic model covering the entire downstream risk area will be recalculated.

Images