A waterproof impact prediction system and device

Abstract

This invention provides a water hammer prediction system and device. The invention relates to the field of pipeline fluid transport safety technology. The water hammer prediction system is configured to: acquire static topology and physical attribute information and multi-dimensional real-time operating condition data of the fluid transport system, and pre-calibrate a dynamic simulation model based on the information and data; when a water hammer triggering event is detected, invoke the pre-calibrated dynamic simulation model to perform an initial prediction to generate an initial prediction result, and issue a pre-control command to the first-level protection device based on the initial prediction result; acquire the measured impact index at the first-level protection device and compare it with the initial prediction result to determine the prediction error; and, based on the prediction error, correct the dynamic simulation model in real time, and drive the corrected dynamic simulation model to perform cascade control of the next-level protection device.

Description

Technical Field

[0001] This invention relates to the field of pipeline fluid transport safety technology, and more specifically, to a water hammer prediction system and device. Background Technology

[0002] In modern industrial fields such as long-distance water pipelines, petrochemicals, cryogenic fluid transportation, and aerospace propulsion systems, the stable and safe transport of fluids within pipelines is crucial for ensuring the normal operation of the entire system. In these systems, water hammer, also known as water surge, is a common and highly hazardous transient hydraulic phenomenon. This phenomenon is usually caused by the rapid opening, closing, switching, or failure of pipeline components such as valves and pumps, leading to a drastic change in fluid velocity within the pipeline. Essentially, it is a transient pressure wave that forms and propagates rapidly within the pipeline. The extremely high pressure peaks generated by intense water hammer can cause destructive effects on the pipeline itself, valves, pumps, and even the entire supporting structure in an instant, potentially leading to major safety accidents such as pipeline rupture and equipment failure. Therefore, effective proactive protection against water hammer is a long-standing and critical technical requirement in this field.

[0003] To mitigate water hammer hazards, existing technologies typically employ two main approaches. The first involves installing mechanical protective devices, such as safety valves, slow-closing check valves, or pneumatic buffers at critical locations in pipelines. These devices operate passively, and their design parameters, such as the safety valve's opening pressure and the buffer's pre-charge pressure, are usually based on theoretical calculations and pre-settings for the worst-case operating conditions the system might encounter. However, the actual operating conditions of industrial systems are complex and variable; factors such as the temperature and viscosity of the fluid medium, the flow velocity in the pipeline, and the aging of the equipment are all dynamically changing. This leads to an inherent limitation of this approach: its protection logic is rigid and non-adaptive. Protection parameters that seem suitable under one condition may be completely inapplicable under another, failing to provide refined and differentiated responses to water hammer of varying intensities and forms.

[0004] The second type of approach is based on hydraulic theoretical models, such as using the method of characteristics to solve the water hammer equation or performing computational fluid dynamics (CFD) simulations to predict the water hammer process. While this type of approach has high theoretical accuracy, its calculation process is extremely complex, requiring precise system boundary conditions and substantial computational resources. This makes it difficult to meet the stringent requirements of real-time performance and rapid response in industrial settings. It is usually limited to offline analysis and verification during the design phase and lacks engineering application value for online, active closed-loop control.

[0005] The limitations of existing technologies, particularly the lack of precise dynamic adjustment, not only reduce protection efficiency but may even introduce new risks. For example, mechanical protection devices are designed based on fixed worst-case scenarios. When a relatively weak water hammer actually occurs, the device may over-protect. An excessively forceful bypass depressurization or an overly rigid buffer can itself generate severe secondary disturbances to the fluid system, potentially leading to negative pressure and cavitation in the main pipeline due to excessive depressurization. When the cavitation bubbles collapse, they generate a secondary impact with equally destructive force. Furthermore, if adjustable valves are used for damping dissipation, without precise predictive guidance, an excessive valve-closing damping action can itself become a new flow obstacle, thereby triggering a more violent new water hammer upstream of the valve. Therefore, water hammer protection is not about maximizing the intensity of the action; mismatched and excessive protection is itself a significant safety hazard. Therefore, existing technologies generally lack a technical means to sense the system's operating conditions in real time, accurately predict the intensity of impending water hammer, and make refined and graded dynamic adjustments to protection measures accordingly. This constitutes a key technical bottleneck that restricts the improvement of pipeline system safety and operational reliability. Summary of the Invention

[0006] This invention provides a water impact prediction system, the system being configured to:

[0007] Acquire static topology and physical property information and multidimensional real-time operating condition data of the fluid transport system, and pre-calibrate the dynamic simulation model based on the information and data;

[0008] When a water hammer triggering event is detected, the pre-calibrated dynamic simulation model is invoked to perform an initial prediction to generate an initial prediction result, and a pre-control command is issued to the first-level protection device based on the initial prediction result.

[0009] The measured impact index at the first-level protection device is obtained and compared with the initial prediction result to determine the prediction error; and the dynamic simulation model is corrected in real time according to the prediction error, and the corrected dynamic simulation model is driven to perform cascade control of the next-level protection device.

[0010] The multidimensional real-time operating data includes: the effective propagation velocity of pressure waves obtained through the fluid medium sensing subprocess. The current system operating mode is obtained through the operating mode identification sub-process. ; and the actual response time of key actuators obtained through the equipment health status assessment sub-process. .

[0011] The actual response time It is calculated using the following formula:

[0012]

[0013]

[0014] in, The current health index of critical actuators; This is the feature vector extracted from its vibration signal; For the preset weight vector, For bias terms; The standard response time calibrated for the key actuator.

[0015] The initial prediction results include the predicted peak pressure head. The system is further configured to be based on the Quantitative prediction of overpressure risk :

[0016]

[0017]

[0018] in, For the downstream terminal valve nodes in the initial prediction results in the future time... Pressure head time series; The preset maximum simulation duration; This refers to the maximum permissible working pressure head of the pipeline system.

[0019] The pre-control command includes a target pre-charge pressure for the intelligent buffer chamber deployed on the valve side. The adjustment command, the :

[0020]

[0021]

[0022] in, To match the current operating mode The corresponding control gain coefficient; To match the aforementioned operating mode Matching risk weights; and These are the base gain coefficient and the maximum gain coefficient, respectively. Pre-charge pressure for standard standby; The current fluid density; This is the acceleration due to gravity.

[0023] The prediction error The measured impact index and predicted shock index The deviation between them is determined; the real-time correction of the dynamic simulation model is achieved by updating a model correction factor. accomplish:

[0024]

[0025] in, This is a measured time series of pressure head; and These are the first time the measured pressure exceeds the safe pressure head and the second time it falls back to the safe pressure head. The moment; These are the factor values ​​before correction; To correct the gain.

[0026] The cascaded control of the next-level protection device includes: the time series of pressure head after the valve predicted based on the modified dynamic simulation model. The control output of the variable damping valve is calculated using a proportional-integral-derivative control algorithm. And map it to the target opening degree of the valve. :

[0027]

[0028]

[0029] in, This represents the prediction error; These are the proportional, integral, and differential gain coefficients, respectively.

[0030] The cascaded control is a cyclic recursive process, and the system is further configured as follows:

[0031] The measured impact index of the second-level node is obtained at the next-level protective device. And determine the new prediction error. And based on the new prediction error, the model correction factor is updated a second time: in, These are the factor values ​​after the second correction; The correction gain is for the second-level node; the cyclic recursive process continues to execute until the water hammer event ends.

[0032] The system is further configured to perform post-event learning and model solidification after the water hammer event, using a batch gradient descent algorithm to adjust the pipe friction coefficient in the dynamic simulation model. Optimize:

[0033]

[0034]

[0035] in, Based on The average loss function calculated from the historical events; and These are the friction coefficient values ​​before and after optimization, respectively. It is the learning rate; It is the gradient of the total loss function with respect to the friction coefficient.

[0036] This invention provides a waterproofing impact prediction device, which includes:

[0037] Acquisition module: Acquires static topology and physical property information and multi-dimensional real-time operating condition data of the fluid transport system, and pre-calibrates the dynamic simulation model based on the information and data;

[0038] Triggering module: When a water hammer triggering event is detected, the pre-calibrated dynamic simulation model is invoked to perform an initial prediction to generate an initial prediction result, and a pre-control command is issued to the first-level protection device based on the initial prediction result;

[0039] The cyclic correction module acquires the measured impact index at the first-level protection device and compares it with the initial prediction result to determine the prediction error; and corrects the dynamic simulation model in real time based on the prediction error, and drives the corrected dynamic simulation model to perform cascade control on the next-level protection device.

[0040] Compared to existing technologies, this invention significantly improves the accuracy and foresight of water hammer prediction. Unlike traditional solutions that rely on fixed parameters or worst-case scenarios for design, this invention, before making any predictions, deeply integrates real-time, multi-dimensional operational data into the dynamic simulation model through a pre-calibration process. The system can not only correct key physical parameters such as pressure and wave velocity in real time based on changes in fluid medium temperature, but also automatically identify the system's current operating mode to match the corresponding risk level. More importantly, it can assess the health status of key actuators online and quantify performance degradation factors such as response delays caused by equipment aging into model parameters for compensation. This multi-dimensional model pre-calibration mechanism ensures that every initial prediction is based on the most realistic physical state of the pipeline system at present.

[0041] Secondly, this invention achieves refined adaptive control of the protective device, effectively avoiding secondary hazards caused by over- or under-protection. Based on high-precision initial prediction results, this invention can accurately quantify the impending overpressure risk and calculate control commands that perfectly match the risk level for different protective devices. For example, when controlling the intelligent buffer chamber, the system can dynamically adjust its control gain according to the risk weight of the current operating mode, thereby providing stronger buffer stiffness in high-risk modes and a gentler response in low-risk modes. This differentiated and appropriate protection strategy not only effectively suppresses the main water hammer peak but also avoids serious problems such as negative pressure cavitation or secondary impacts caused by excessive force in traditional solutions, significantly improving the operational stability of the entire system and the service life of the equipment.

[0042] More importantly, this invention constructs a multi-level closed-loop feedback correction and cascaded control mechanism that runs through the entire process of a water hammer event. The system does not rely solely on the initial prediction; instead, it compares the actual measurement with the prediction as the pressure wave propagates to each protection node, and uses the error between the two to correct the dynamic simulation model in real time. This cascaded recursive loop of prediction-execution-measurement-correction-repreneurial allows the system to continuously and dynamically optimize its prediction and control strategies for subsequent water hammer waves based on feedback from the real physical process. This mechanism endows the system with strong robustness and adaptability to uncertainties; even if there are minor deviations in the initial prediction, the system can quickly converge to the true state during the event's development, ensuring the final protective effect.

[0043] This invention, by introducing post-event learning and model solidification processes, endows the system with long-term self-learning and self-evolution capabilities. After each water hammer event, the system automatically archives full-cycle data and, through offline learning algorithms, optimizes and solidifies the baseline physical parameters (such as the pipe friction coefficient) in the model that slowly change due to factors such as pipe aging and scaling. This means that the system of this invention can not only adapt to changing operating conditions in the short term, but also adapt to the life cycle evolution of the pipeline system itself over a longer timescale, ensuring that its protective performance does not decline over time, but rather becomes increasingly accurate and reliable due to the accumulation of experience, achieving true full life cycle intelligent protection. Attached Figure Description

[0044] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1This is a flowchart illustrating a water impact forecasting method according to the present invention. Detailed Implementation

[0046] The embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0047] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. This application can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0048] It should be noted that various aspects of embodiments within the scope of the appended claims are described below. It will be apparent that the aspects described herein can be embodied in a wide variety of forms, and any particular structure and / or function described herein is merely illustrative. Based on this application, those skilled in the art will understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number and aspects set forth herein can be used to implement the device and / or practice the method. Additionally, this device and / or method can be implemented using structures and / or functionalities other than one or more of the aspects set forth herein.

[0049] Additionally, specific details are provided in the following description to facilitate a thorough understanding of the examples. However, those skilled in the art will understand that practice can be carried out without these specific details.

[0050] This invention provides a water hammer forecasting system, the system being configured as follows:

[0051] Acquire static topology and physical property information and multidimensional real-time operating condition data of the fluid transport system, and pre-calibrate the dynamic simulation model based on the information and data;

[0052] When a water hammer triggering event is detected, the pre-calibrated dynamic simulation model is invoked to perform an initial prediction to generate an initial prediction result, and a pre-control command is issued to the first-level protection device based on the initial prediction result.

[0053] The measured impact index at the first-level protection device is obtained and compared with the initial prediction result to determine the prediction error; and the dynamic simulation model is corrected in real time according to the prediction error, and the corrected dynamic simulation model is driven to perform cascade control of the next-level protection device.

[0054] Example 1

[0055] This invention provides a standardized deployment framework and system architecture for a water hammer prediction system and device in typical industrial fluid transportation scenarios. This embodiment aims to illustrate a complete and feasible solution applicable to various pipeline systems where water hammer risks arise due to drastic changes in flow velocity, such as, but not limited to, long-distance raw water / refined oil pipelines, large-scale water conservancy projects, process fluid networks in chemical industrial parks, and aerospace ground refueling systems. In the initial stage of system deployment, static topology and physical attribute information of the pipeline system is first obtained through import or on-site mapping. This information originates from as-built design drawings, piping and instrumentation diagrams, or 3D BIM models. The controller will structure this information and store it in a database, specifically including: the overall layout of the pipeline system; the precise length and path coordinates of each main and branch pipe section; the nominal diameter, actual wall thickness, and detailed specifications of the pipe material for each pipe section (e.g., the elastic modulus and Poisson's ratio corresponding to the steel grade); and the precise spatial location and hydraulic characteristic parameters of all fixed components along the pipeline, such as elbows, tees, reducers, and major equipment (pumps, valves). The purpose of obtaining the above information is to ensure that the geometric boundaries and material properties of the physical model are completely consistent with the actual engineering. These basic parameters determine the calculation benchmark of core variables such as pressure wave propagation speed and pipeline characteristic impedance in the model, and are the fundamental prerequisite for achieving high-precision prediction.

[0056] Based on the static data modeling, this embodiment deploys active protection devices and a sensor network. The deployment principle of this invention is to construct a multi-layered, in-depth defense system that includes source suppression, mid-course dissipation, and end-point protection. An electrically controlled bypass loop is configured at the most significant potential source of water hammer in the system, i.e., at the downstream terminal rapid control valve, connected in parallel across both ends of the valve. The deployment of the electrically controlled bypass loop can weaken the initial intensity of water hammer by actively diverting the flow before the valve closes and the fluid momentum is completely converted into pressure energy. Simultaneously, to provide focused protection for critical equipment in the system, two intelligent buffer chambers are deployed at key pressure nodes in the pipeline system, i.e., at potential pressure wave antinodes. One chamber is located upstream of the downstream terminal valve, its function being to absorb and buffer the most intense initial pressure shock, protecting the valve itself; the other chamber is deployed downstream of the upstream main pump outlet, its core mission being to protect the expensive and critical pump unit equipment from damage caused by pressure waves reflected back from the end of the pipeline network. In addition, a set of variable damping valves is integrated in series at the midpoint of long-distance pipelines or at the optimal energy dissipation location determined by hydraulic analysis. As an active energy dissipation center, the variable damping valve can effectively intercept and weaken pressure waves propagating in the pipeline, preventing the impact energy from oscillating back and forth in the pipeline network without attenuation, thereby stabilizing the entire system.

[0057] To accurately capture the pressure transients during water hammer, pressure transmitters with high dynamic response characteristics are deployed at the inlet and outlet of the main pump, before and after the variable damping valve, before and after the terminal valve, and at the inlet of the two intelligent buffer chambers. Preferably, Rosemount 3051S series high-frequency pressure transmitters are used. To obtain the macroscopic state of the system operation, non-contact ultrasonic flow meters are configured on key pipe sections. Preferably, OPTISONIC 3400 series from Krohne is used to provide steady-state and transient flow reference data. Simultaneously, to achieve adaptive operation of the model, multiple industrial-grade resistance temperature sensors are arranged along the pipeline. The controller can query and correct key parameters such as the density and bulk modulus of the fluid from the physical property database based on the real-time temperature. Even small changes in fluid properties with temperature will significantly affect the propagation characteristics of pressure waves; fixed parameters cannot meet the requirements of refined control. In addition, to ensure the reliability of the protection system itself, the present invention integrates an equipment health status sensing subsystem. By installing industrial-grade accelerometers on each valve actuator and pump body, the vibration status is continuously monitored to assess whether there is any deterioration in its mechanical performance, thereby enabling the main controller to dynamically compensate for response delays or action deviations caused by equipment aging.

[0058] The entire system in this embodiment consists of a high-performance real-time industrial controller deployed in the upstream pumping station or central control room, preferably using a Siemens SIMATIC S7-1500 series PLC as the core for computation and decision-making. This controller was chosen because of its powerful processing capabilities and deterministic cycle time, ensuring that complex model simulations and control command calculations can be completed within the limited time window of pressure wave propagation. The system's signal transmission employs a hybrid network architecture that balances reliability and real-time performance. Devices and sensors located in the same area as the controller are wired via PROFIBUS-DP fieldbus technology, ensuring low latency and high stability in core area communication. For mid-point and end-point device nodes, communication is achieved through an industrial fiber optic ring network laid along the pipeline, using the PROFINET industrial Ethernet protocol. Fiber optic communication, due to its high bandwidth, long-distance transmission capability, and crucial electromagnetic interference resistance, ensures the integrity and purity of control commands and feedback signals in complex industrial environments.

[0059] Example 2

[0060] The following section will elaborate on the specific implementation method for online acquisition and analysis of multi-dimensional real-time operating conditions of the system. Online acquisition of multi-dimensional real-time operating conditions is the data foundation and decision-making prerequisite for all prediction and control processes. This embodiment will provide a detailed explanation of three parallel sub-processes: fluid medium sensing, operating mode recognition, and equipment health status assessment.

[0061] In the fluid medium sensing subprocess, two physical parameters need to be calculated based on real-time operating conditions for the subsequent water hammer simulation model: the actual density of the fluid and the effective propagation velocity of the pressure wave in the medium and pipe. Following the deployment in Example 1, the controller continuously acquires real-time temperature readings of the fluid within the pipe from multiple industrial-grade resistance temperature sensors arranged along the pipeline. And based on the nonlinear relationship between fluid density and bulk modulus and temperature, the actual density of the fluid is obtained. and bulk modulus .

[0062] The nonlinear functions of fluid density, bulk modulus, and temperature are pre-calculated and stored in the controller database during the system initialization phase based on the publicly available physical property data sheet (IAPWS-97) for the specific fluid medium (such as industrial water in this embodiment).

[0063] The propagation speed of pressure waves depends not only on the fluid itself but also on the elasticity of the pipe wall. Therefore, it is necessary to calculate the effective bulk modulus. Effective bulk modulus It combines the combined effects of fluid and piping:

[0064]

[0065] in, The nominal diameter of the pipe. This is the actual wall thickness of the pipe. This refers to the elastic modulus of the pipe material. These three parameters are all static physical parameters, stored in the controller database during system initialization. The controller acquires the real-time temperature... Then, through bulk modulus Combined static parameters Calculate The controller calculates the effective propagation speed of the pressure wave under the current operating conditions. :

[0066]

[0067] In the operation mode recognition sub-process, the system needs to be able to automatically identify the current macroscopic operating state so that matching safety thresholds and intervention strategies can be invoked in subsequent control. This embodiment uses a decision tree logic model based on real-time monitoring data to achieve this function. The controller analyzes the flow data from the flow meter. The set of control commands it issues to the main pump and valves The system also includes feedback signals from valve position sensors. Four main operating modes are preset. Conventional conveying mode ( Emergency stop mode ), pipeline filling mode ( ), and standby maintenance mode ( ).

[0068] The operation mode recognition logic is as follows: if traffic is detected... Within a preset high-flow working range [ Stable fluctuations, and the absolute value of their rate of change over time. Less than a tiny threshold Meanwhile, instruction set If there are no high-risk instructions, then the current mode is determined. For conventional conveying mode .

[0069] If the controller issues a shutdown command to the main pump or a rapid shut-off command to the terminal valve with a shut-off time less than the preset safety time, the system will immediately switch to the current mode. Switch to emergency stop mode

[0070] If traffic is detected It starts from zero and gradually increases, while remaining below the minimum normal workload. Then the determination mode For pipeline filling mode .

[0071] If traffic If the value is almost zero and there are no operating commands from the main power equipment (such as the main pump) in the system, then the determination mode is... Standby maintenance mode .

[0072] In the equipment health status assessment sub-process, the system needs to quantify the performance degradation of key active components, such as valve electric actuators, due to long-term operation, and convert this degradation level into a correction factor for the simulation model. This embodiment uses a regression model based on vibration signal feature analysis to achieve this. According to the deployment in Embodiment 1, the controller collects a high-frequency vibration time-domain signal from an industrial-grade accelerometer mounted on the valve actuator's housing each time the actuator operates. The controller first processes the signal. Perform a Fast Fourier Transform to obtain its frequency domain representation. And extract a set of feature vectors that can characterize its operating state from it. .

[0073] Among them, the root mean square (RMS) value reflects the intensity of vibration energy, the kurtosis reflects the amount of impact component, and the peak factor (CF) reflects the presence of extreme impact. The controller uses this feature vector... Input a pre-trained multiple linear regression model to calculate the current health index of the actuator. Its value ranges from [0, 1], where 1 represents a completely new state and 0 represents complete failure. The regression model is expressed as:

[0074]

[0075] in, This is the weight vector for each vibration characteristic. These parameters, acting as bias terms, are obtained through offline training using vibration data from the entire lifecycle of the device. Health Index The significance lies in correcting the standard technical parameters of the actuator, such as its actual response time. The standard response time can be specified by the manufacturer. and health index Joint decision:

[0076]

[0077] This formula shows that, with the health index As the value decreases, the actual response time will increase accordingly. The controller will calculate... Update the dynamic simulation model, replacing the original idealized parameters. In this way, the present invention ensures that the equipment model on which its prediction and control decisions are based can accurately reflect the current performance status of the equipment, thereby significantly improving the real-world reliability of the entire forecasting and protection system.

[0078] Example 3

[0079] After obtaining the multidimensional real-time operating parameters of the system, this invention needs to pre-calibrate the dynamic simulation model used for water hammer prediction.

[0080] Based on the static topology information of the pipeline system obtained in Example 1, the controller performs spatial discretization processing on the main pipeline. The controller then calculates the total length of the pipeline. Divided into There are three equal-length calculation pipe segments, each with a length of [length missing]. The time step in this invention is a variable that is dynamically adjusted based on real-time operating conditions. The controller calls the effective propagation velocity of the current pressure wave calculated in Example 2. Set the time step for this simulation:

[0081]

[0082] The simulation time step setting ensures that the feature lines of the computational mesh fall precisely on the mesh nodes, fundamentally guaranteeing the accuracy of subsequent calculations. Total number of pipe segments. Then based on the total length of the pipeline and spatial step size Decision, that is Through spatial and temporal discretization, the pipeline system is transformed into a system composed of... A one-dimensional computing grid consisting of computing nodes.

[0083] The controller establishes a set of core hydraulic transient calculation equations for each internal computing node, namely, positive and negative characteristic line compatibility equations. For any internal computing node... ( ), in the next moment ( (Time) pressure head and traffic The following two linear equations must be satisfied simultaneously:

[0084]

[0085]

[0086] in, and These are known coefficients calculated from the hydraulic parameters of adjacent nodes at the previous (known) time step. and This is a key coefficient characterizing the hydraulic properties of the pipeline. In this embodiment, the coefficient... and For uniform pipes, these values ​​are equal and are collectively referred to as the pipe characteristic impedance coefficient. The calculation formula is as follows:

[0087]

[0088] Here This refers to the effective propagation velocity of the pressure wave calculated in real time in Example 2. It is the acceleration due to gravity. The cross-sectional area of ​​the pipe is given by the nominal pipe diameter defined in Example 1. Calculation yields ( Similarly, the coefficient of the friction term related to pipe friction. It also needs to be calculated:

[0089]

[0090] in, This represents the Darcy-Weisbach friction coefficient of the pipe, a value derived from the pipe material and wall roughness defined in Example 1. (Known coefficient) and The expression depends on the neighboring nodes at the previous time step. and Status:

[0091]

[0092]

[0093] This represents the pressure head at the adjacent computing node i-1 upstream of the current computing node i at the previous time (a known time). This represents the pressure head at the adjacent computing node i+1 downstream of the current computing node i at the previous time (a known time).

[0094] This represents the flow at the adjacent computing node i-1 upstream of the current computing node i at the previous time (a known time). This represents the flow at the adjacent computing node i+1 downstream of the current computing node i at the previous time (a known time).

[0095] During the pre-calibration phase, the controller will use the static parameters obtained in Example 1 ( ) and the dynamic parameters obtained from real-time analysis in Example 2 ( (to be integrated)

[0096] Specifically, the controller pre-calculates the core constants required by the simulation model under the current operating conditions, including the characteristic impedance coefficient. and friction factor At the same time, the controller will identify the current operating mode. Loaded as a state label for the model, this label will be used in subsequent predictions to select the boundary condition handling model that matches the pattern (e.g., emergency shutdown mode). This corresponds to a boundary model for a fast valve closure. Furthermore, the controller will use the actual response times of each key actuator evaluated in Example 2. Loaded into the model, it serves as the dynamic parameter for the corresponding boundary condition model (such as valves and pumps).

[0097] Example 4

[0098] After completing pre-calibration, the system transitions from monitoring to proactive forecasting. This process begins with real-time identification of potential water hammer risks, and immediately after risk identification, it invokes the pre-calibrated dynamic simulation model to simulate an impending water hammer event.

[0099] First, this embodiment describes the mechanism for identifying water hammer trigger events. Unlike passively waiting for pressure to rise before responding, this invention employs an active triggering mechanism based on command prediction. The controller continuously monitors control commands issued by itself or its superior scheduling system to critical terminal actuators (especially downstream terminal fast-control valves). The controller has an internal risk command evaluation logic; when a received command meets preset risk conditions, it is determined to be a water hammer trigger event. In this embodiment, the mechanism targets terminal valves (i.e., computing nodes). The risk criterion is the valve closing time required by the instruction. The relationship with the hydraulic characteristics of the pipeline. When the valve closing time is less than the time required for a pressure wave to travel one round trip within the pipeline, the maximum intensity of water hammer will occur. Therefore, the controller will use this as a criterion:

[0100]

[0101] in, It is the planned closing time parsed from the valve control command. The total length of the pipeline. This is the calculated effective propagation speed of the pressure wave. Once the controller detects a command sent to the terminal valve, it will extract [the value / speed]. Parameters, and the real-time calculated pipeline timescale. The comparison is performed. If the above inequality holds, the controller marks this event as a high-risk water hammer trigger event and immediately initiates the initial prediction process.

[0102] In the initial prediction process, the controller invokes the pre-calibrated dynamic simulation model from Example 3 to simulate the entire water hammer process triggered by the event. This simulation process spans from the current moment in the time dimension. Start with time step Step by step, until a preset maximum simulation duration is reached. This duration needs to cover the process of the water hammer wave making several round trips within the pipe. Within each time step, the controller updates all data in the pipe. Hydraulic state (pressure head) of each computation node and traffic For internal nodes () The state at the next moment is solved by simultaneously solving the positive and negative characteristic line phase equations defined in Example 3. For the boundary nodes at both ends of the pipeline ( and If so, then specific boundary condition equations need to be applied.

[0103] For the upstream boundary ( (i.e., the main pump outlet), assuming the main pump maintains a constant outlet pressure head during this event, then its boundary conditions are: ,in Given the known pump outlet design head, substitute it into the negative characteristic line compatibility equation for that node. The flow rate at the next moment can be directly calculated. .

[0104] For the downstream boundary ( The hydraulic behavior of a valve (i.e., a terminal valve) is determined by both the orifice outflow equation and the positive characteristic line compatibility equation. At any given moment, the flow rate through the valve is... With the pressure head before the valve satisfy:

[0105]

[0106] in, This refers to the flow area when the valve is fully open. For flow coefficient, This is an instantaneous relative opening function describing the valve closing process, with its value varying from 1 (fully open) to 0 (fully closed). The specific form of this function is determined by the closing time in the trigger command. It is determined by the preset shutdown curve.

[0107] For linear shutdown: ,in It is worth noting that the controller here will use the actual valve actuator response time evaluated in Example 2. To correct the time parameters This simulates the valve operation process that realistically considers equipment aging. The valve equation is then compared with the positive characteristic linear equation of this node. By combining the equations, the valve node at the next time step can be solved. and .

[0108] The controller, at each time step The inner loop performs the solution for all internal and boundary nodes until the maximum simulation time is reached. The output of the entire simulation process is a two-dimensional data matrix describing future time. Inside, every calculation node along the pipeline. Pressure head and traffic The complete process of change over time.

[0109] Example 5

[0110] After completing the initial water hammer prediction, the controller generates and executes pre-control commands based on the prediction results in the first-level protection area, namely the protection device (including the electrically controlled bypass circuit and the intelligent buffer chamber) on the downstream terminal valve side.

[0111] The controller outputs a two-dimensional data matrix, which is the predicted pressure head. and traffic Key feature extraction and risk assessment are performed. The controller focuses on analyzing the downstream terminal valve node (i.e., the computational node) that serves as the source of water hammer. Pressure head time series By iterating through this time series, the controller determines the predicted peak pressure head. :

[0112]

[0113] The controller will use this predicted peak value The maximum allowable working pressure head of the piping system is pre-stored in the database. By making comparisons, the predicted overpressure risk of this water hammer event can be quantified. :

[0114]

[0115] This prediction indicates an overpressure risk. It is the fundamental basis for calculating all control commands. If This indicates that the valve operation will not cause an overpressure risk, the system does not need to activate active protection, and the process terminates. If Then the controller immediately begins generating pre-control commands for the first-level protection device.

[0116] For the electrically controlled bypass circuit, the controller needs to determine its optimal activation sequence, activation amplitude, and activation duration. In this embodiment, its activation sequence... Set to close when the main valve begins to close ( A small pre-opening time margin prior to the start. Start the bypass valve to ensure that the bypass pressure relief path is established before the main pipeline flow rate begins to change. Its opening extent is the target opening degree of the bypass valve. With the predicted overpressure risk Related.

[0117] The controller calculates a target bypass flow. This flow rate is proportional to the risk of overpressure and aims to remove enough hydraulic energy to suppress pressure peaks near a safe level.

[0118]

[0119] in, It is a preset bypass control proportional gain coefficient. The controller adjusts the target flow rate based on the hydraulic characteristic curve of the bypass valve itself. The required valve physical opening for reverse calculation Its operating time Then, through the predicted pressure time series To determine, the duration of coverage begins when the pressure head first exceeds [a certain value]. To eventually fall back to The following is the entire time window, plus an additional safety margin.

[0120] Finally, the controller generates a complete pre-control command sequence for the electronic bypass circuit, which includes the precise start time, target opening degree, and duration of hold.

[0121] For the intelligent buffer chamber on the valve side, the controller's task is to regulate the pre-charge gas pressure inside. This ensures that its stiffness matches the upcoming impact energy, achieving optimal buffering and energy absorption. The controller is based on the predicted overpressure risk. To calculate the target precharge pressure :

[0122]

[0123] in, It is the standard standby pre-charge pressure of the buffer chamber under non-risk conditions. It is the perceived current fluid density. Both are gravitational acceleration and are used to measure overpressure head. Converted to pressure units (Pa). An improvement in this embodiment also lies in the control gain coefficient of the buffer chamber. It is not a fixed value, but rather a value that varies with the current system operating mode. Related functions .

[0124] In order to identify discrete operating modes (such as...) (etc.) are quantified into risk levels that can be used for continuous calculation. This embodiment provides each operating mode. A dimensionless operational mode risk weight is defined. Its value ranges from [0, 1]. This weight value is used during the system commissioning phase to calibrate the potential hazard level of the pipeline system according to different modes and is stored in the controller. In this embodiment, the risk weight of each mode... Set as:

[0125] Emergency stop mode ( The first option typically involves the fastest valve closure and carries the highest risk; therefore, its weight is set as follows: .

[0126] Conventional conveying mode ( The valve operation risk under these conditions is moderate, and its weight can be set as follows: .

[0127] Pipeline filling mode ( The lower the flow rate, the lower the risk; its weight can be set as follows: .

[0128] Standby maintenance mode ( There is no current or risk of water hammer below, and its weight is... .

[0129] The controller obtains the current operating mode Then, its corresponding risk weight will be matched immediately. Subsequently, the controller calculates the control gain coefficient ultimately used in the voltage regulation formula using the following linear interpolation model. :

[0130]

[0131] in, This is the base gain coefficient, corresponding to the buffer response strength at the lowest risk level; This represents the maximum gain coefficient, corresponding to the strongest buffer response strength at the highest risk level. and As calibration parameters for the system, they are pre-set during the deployment phase based on the physical properties of the buffer chamber and the safety requirements of the pipeline.

[0132] The controller determines the risk weight based on the current operating mode. By precisely interpolating between the base and maximum gain coefficients, a control gain that perfectly matches the current risk level is calculated. The beneficial effect of this approach is that it is effective when the system is at high risk. In mode, Take the maximum value, such that It also approaches its maximum value, resulting in a higher calculated target pre-charge pressure, thus allowing the buffer chamber to withstand the most intense impacts with greater stiffness; while in the lower-risk conventional operating mode, With a moderate value, the response of the buffer chamber is more gentle, which can effectively absorb the impact and avoid the secondary high-frequency oscillation caused by excessive protection (excessive stiffness of the buffer), which would cause new disturbances to the system.

[0133] Finally, after completing the calculation of the above command parameters, the controller will send the generated command sequence (including the action timing command for the electronic bypass circuit and the pressure regulation command for the intelligent buffer chamber) to the local execution unit of each protection device located on the downstream terminal valve side through the communication network architecture, before the water hammer event actually occurs, thereby completing the pre-deployment of the first level of protection.

[0134] Example 6

[0135] This embodiment specifically illustrates that after the first water hammer wave actually reaches the protection area on the downstream terminal valve side, the system performs actual measurement of the first-level node, model correction, and prediction of the next-level node. This process is crucial for achieving the transition from open-loop prediction to closed-loop adaptive control, ensuring that the entire protection system can dynamically optimize its subsequent control behavior based on feedback from the actual physical process.

[0136] In Example 5, the controller has issued pre-control commands to the intelligent buffer chamber and the electrically controlled bypass circuit on the valve side. When the water hammer wave is actually generated and reaches the area due to the closing action of the terminal valve, the pressure transmitter deployed at the inlet of the intelligent buffer chamber on the valve side begins to collect pressure data in real time at a high sampling rate. The controller processes the collected pressure time series and calculates the measured impact index, which can characterize the intensity of this impact. In this embodiment, the index is defined as the actual pressure head during the first pressure pulse cycle. Exceeding safe pressure head Partial integral over time:

[0137]

[0138] in, This is the first time the measured pressure has exceeded [a certain level]. At that moment, This is the first time the pressure has fallen back to The following moments. The measured impact index. Compared to a single peak pressure, it can more comprehensively reflect the magnitude and duration of the impact energy, providing a more robust basis for model correction.

[0139] The controller will measure the above-mentioned impact index Compared with the predicted shock index calculated based on the uncorrected model in the initial prediction of Example 4 Comparison. Predicting the impact index. The calculation method is the same as the measured value, except that its data source is the predicted pressure time series. The deviation between the two This reflects the prediction error of the initial model:

[0140]

[0141] This deviation The existence of these uncertainties is attributed to complex factors that are difficult to model accurately, such as undefined local resistance losses along the pipeline and the buffering effect of trace amounts of undissolved gas in the fluid. To compensate for these uncertainties, this invention introduces a model correction factor. This factor serves as the coefficient for the defined friction resistance term. The dynamic multiplier is used to correct the energy dissipation characteristics of the entire model online. In this embodiment, the correction factor... The update employs a recursive correction algorithm based on prediction error feedback:

[0142]

[0143] in, This is the new correction factor obtained after this revision. This is the factor value before correction (its default value is 1 in the initial prediction). This is a preset correction gain, which determines the magnitude by which the system adjusts based on a single error. If the prediction is too strong ( The system will automatically reduce the dissipation term, making the model more aggressive; if the prediction is weak ( If the dissipation term is increased, the model becomes more conservative, thus enabling the model to adapt quickly to real physical processes.

[0144] After calculating the new model correction factor The controller then immediately uses this factor to update the dynamic simulation model online. Specifically, this factor is used to update the friction factor of all calculated pipe segments in the model. Updated to Subsequently, the controller immediately initiates a corrected secondary prediction. The initial condition for this prediction is no longer the system's steady state, but rather the moment the first-stage node completes its measurement, serving as the new starting point and inheriting the true or best-estimated hydraulic state of all nodes along the pipeline at that moment. The controller drives the corrected simulation model to continue calculations into the future, with the core objective of accurately predicting the hydraulic state of the pressure wave, weakened by the first-stage protection, as it propagates to the next-stage protection node—the variable damping valve in the middle of the pipeline—specifically, its arrival time and impact intensity.

[0145] Example 7

[0146] This embodiment takes a variable damping valve deployed in the middle of a pipeline as an example to illustrate how it receives and executes the predicted command corrected by the previous stage, and performs actual measurement and further correction. At the same time, this embodiment also illustrates that the closed-loop logic of prediction-execution-actual measurement-correction is used to iteratively extrapolate to subsequent protection nodes and reflected waves, thereby achieving adaptive management of the entire water hammer event process.

[0147] In the final step of Example 6, the controller, based on the measured data from the first-level nodes, drives the corrected dynamic simulation model to generate a calculation of the pressure wave reaching the middle section of the pipe (based on the calculated nodes). The secondary prediction of the hydraulic state at the center (point 1) is performed. This prediction includes the precise time when the pressure wave arrives at that point, as well as the predicted pressure and flow time series. The variable damping valve's task is to dissipate the energy of the shock wave by actively changing the valve opening to create throttling losses. In this embodiment, the controller uses a proportional-integral-derivative (PID) control algorithm to generate a dynamic target opening command for the variable damping valve based on the secondary prediction result. The setpoint of this PID controller... For the safety pressure head of the pipeline system The process feedback quantity is the predicted time series of the downstream pressure head. The controller calculates the prediction error. :

[0148]

[0149] The controller calculates the control output using a PID algorithm. This output is proportional to the damping strength required to counteract the predicted overpressure:

[0150]

[0151] in, These are the preset proportional, integral, and derivative gain coefficients, respectively. They control the output quantity. After normalization, it is mapped to a variable damping valve. Target opening at any time The controller will contain the target opening time series over the entire period from before the arrival of the pressure wave to after its departure. As a pre-control command, it is sent to the actuator of the intermediate variable damping valve through the communication network to guide it to perform precise dynamic throttling action.

[0152] When the water hammer wave, weakened by the first stage of protection, actually reaches and passes through the variable damping valve, high-frequency pressure transmitters deployed on both sides of the valve measure the actual pressure head across the valve in real time. Based on the measured pressure difference and real-time flow rate, the controller calculates the actual energy dissipated by the variable damping valve during this event and quantifies it as the measured impact index of the second-stage node. The controller compares this measured value with the predicted impact index calculated based on the secondary prediction results. By comparison, a new prediction error is obtained. This error reflects potential biases in the model's predictions of the mid-section pipe characteristics even after the initial correction. The controller utilizes this new error to adjust the model's correction factor. Second update:

[0153]

[0154] in, It is the factor value corrected by the first-level node. This is the latest value after this revision. This is a correction gain for the second-level nodes, and its value can be compared with that of the first-level nodes. The coefficients are different to accommodate corrections needed at different locations. The friction term coefficients in the model... It was then updated to .

[0155] After completing the field measurements and model correction of the variable damping valve, the controller immediately drives the simulation model, which has been corrected twice for higher fidelity, to perform a third prediction. The goal of this prediction is to accurately predict the state of the pressure wave as it continues to propagate upwards after the energy dissipation of the mid-stage damping valve and reaches the upstream third-level protection node, the pump-side intelligent buffer chamber. The prediction results will be used to guide the buffer chamber to perform more precise pre-charge pressure regulation. The cascaded control chain, which receives prediction commands from the previous level, executes them, performs field measurements at this level, corrects the model, and predicts and guides the next level, will be executed sequentially at each protection node as the pressure wave propagates. When the pressure wave propagates to the upstream boundary of the pipeline (such as the main pump) and is reflected, the cascaded control chain will start in reverse. That is, the measured data of the pump-side buffer chamber will serve as the first feedback for correcting the reflected wave model, guiding the mid-stage damping valve to respond to the reflected wave. The measured data of the mid-stage damping valve will then guide the devices in the downstream terminal valve area. This iterative process will continue until the pressure fluctuations at all nodes in the system are suppressed below the safety threshold, thereby achieving comprehensive, closed-loop, and adaptive control over the entire lifecycle of water hammer events (including multiple reflections and oscillations).

[0156] Example 8

[0157] Once the controller determines that the pressure fluctuations within the piping system have subsided, signifying the end of a complete water hammer event, the controller packages and structures the entire lifecycle data from triggering to termination of this event into an event log. This log is stored on the controller's non-volatile storage medium or uploaded to the database of the higher-level monitoring system, serving as a basis for subsequent learning. Each event log entry contains the following key information:

[0158] Initial operating condition vector: records the system state before the event is triggered, including steady-state flow. Fluid temperature and the initial effective wave velocity calculated in real time. and initial fluid density wait.

[0159] Trigger event parameters: Record the specific operational parameters that trigger water hammer, such as the planned closing time of the terminal valve. And the curve type is turned off.

[0160] Predicted and measured data pairs: Record the predicted and measured impact indices of each protection node during the event, such as the predicted value of the first-level node. and measured values The second-level node and wait.

[0161] Real-time correction factor sequence: recorded during the event, model correction factors All iterative update values, i.e. .

[0162] The goal of post-hoc learning is to optimize baseline physical parameters in the dynamic simulation model that have uncertainties or change slowly over time, thereby minimizing the initial prediction error of future events. In this embodiment, the preferred optimization objective is the pipe friction coefficient defined in Example 3. Because the roughness of the pipe's inner wall changes over time (e.g., due to scaling, corrosion), Treating it as a learnable parameter is crucial for maintaining the long-term accuracy of the model. To achieve this, the controller defines a loss function. Used to quantify a specific event In this context, the accuracy of the initial prediction is considered. The preferred loss function is the squared error between the initially predicted impact index and the measured impact index.

[0163]

[0164] in, It is an event The measured impact index of the first-level node in the middle. It is explicitly stated that the initial predicted impact index is related to the friction coefficient. The function changes It will affect the coefficient of the friction term in the model. This, in turn, changes the entire simulation result.

[0165] The controller is configured to preferentially select events during system idle periods or when a certain number of accumulated event logs are stored, provided that specific conditions are met. The offline learning process is automatically initiated at certain times. This process uses a batch gradient descent algorithm to update the friction coefficient. The controller first reads the most recent value from the database. Count the event logs and calculate the average loss for these events. :

[0166]

[0167] The goal of the gradient descent algorithm is to find the gradient descent method that makes the gradient descent algorithm more efficient. Minimized Value. Its iterative update rule is:

[0168]

[0169] in, It is the friction coefficient value currently stored in the system. It is the new value after one iteration of optimization. It is a preset learning rate used to control the step size of each update. It is the total loss function with respect to the friction coefficient The gradient of the function. Because the function... The output is a complex simulation model, and its analytical derivative is difficult to obtain. Therefore, in this embodiment, the gradient is approximated using numerical methods. The controller will repeatedly execute this update rule for multiple iterations until... Value convergence or loss function It drops below the preset threshold.

[0170] When the gradient descent algorithm converges and obtains an optimal friction coefficient The controller will then store this new value in the system static parameter database defined in Example 1. In the event of a new water hammer event in the future, the pre-calibration process of Example 3 will use a friction coefficient value that is closer to the physical reality. To calculate the coefficient of the friction term. This will directly improve the accuracy of initial predictions, thereby making the generated initial pre-control commands more precise and reducing the adjustment range required in the real-time correction stage. Through this closed loop of post-learning and model solidification, the system proposed in this invention possesses the self-evolutionary ability to learn from experience and continuously adapt to long-term changes in the pipeline system.

[0171] This invention provides a waterproofing impact prediction device, which includes:

[0172] Acquisition module: Acquires static topology and physical property information and multi-dimensional real-time operating condition data of the fluid transport system, and pre-calibrates the dynamic simulation model based on the information and data;

[0173] Triggering module: When a water hammer triggering event is detected, the pre-calibrated dynamic simulation model is invoked to perform an initial prediction to generate an initial prediction result, and a pre-control command is issued to the first-level protection device based on the initial prediction result;

[0174] The cyclic correction module acquires the measured impact index at the first-level protection device and compares it with the initial prediction result to determine the prediction error; and corrects the dynamic simulation model in real time based on the prediction error, and drives the corrected dynamic simulation model to perform cascade control on the next-level protection device.

[0175] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0176] In this specification, the same or similar parts between the various embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the descriptions of the embodiments described later are relatively simple, and relevant parts can be referred to the descriptions of the foregoing embodiments.

[0177] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A waterproof impact prediction system, characterized in that, The system is configured as follows: Acquire static topology and physical property information and multidimensional real-time operating condition data of the fluid transport system, and pre-calibrate the dynamic simulation model based on the information and data; When a water hammer triggering event is detected, the pre-calibrated dynamic simulation model is invoked to perform an initial prediction to generate an initial prediction result, and a pre-control command is issued to the first-level protection device based on the initial prediction result. The measured impact index at the first-level protection device is obtained and compared with the initial prediction result to determine the prediction error; and the dynamic simulation model is corrected in real time according to the prediction error, and the corrected dynamic simulation model is driven to perform cascade control on the next level protection device. The prediction error is caused by It is stated that the measured impact index is derived from... It indicates that the predicted impact index is from express; The prediction error The measured impact index and predicted shock index Determine the deviation between them and predict the impact index. The calculation method is the same as the measured value, and its data source is the predicted pressure time series. ; Real-time correction of the dynamic simulation model is achieved by updating a model correction factor. accomplish: ； ； ； in, This is a measured time series of pressure head; and These are the first time the measured pressure exceeds the safe pressure head and the second time it falls back to the safe pressure head. The moment; These are the factor values ​​before correction; To correct the gain.

2. The waterproof impact prediction system according to claim 1, characterized in that, The multidimensional real-time operating data includes: the effective propagation velocity of pressure waves obtained through the fluid medium sensing subprocess. The current system operating mode is obtained through the operating mode identification sub-process. ; and the actual response time of key actuators obtained through the equipment health status assessment sub-process. .

3. A waterproof impact prediction system according to claim 2, characterized in that, The actual response time It is calculated using the following formula: in, The current health index of critical actuators; This is the feature vector extracted from its vibration signal; For the preset weight vector, For bias terms; The standard response time calibrated for the key actuator.

4. A waterproof impact prediction system according to claim 1, characterized in that, The initial prediction results include the predicted peak pressure head. The system is further configured to, according to Quantitative prediction of overpressure risk : in, The preset maximum simulation duration, More specifically, the downstream terminal valve nodes in the initial prediction results are in the future time... Pressure head time series; More specifically, this refers to the maximum permissible working pressure head of the pipeline system.

5. A waterproof impact prediction system according to claim 4, characterized in that, The pre-control command includes a target pre-charge pressure for the intelligent buffer chamber deployed on the valve side. The adjustment command, the : in, To match the current operating mode The corresponding control gain coefficient; To match the aforementioned operating mode Matching risk weights; and These are the base gain coefficient and the maximum gain coefficient, respectively. Pre-charge pressure for standard standby; The current fluid density; This is the acceleration due to gravity.

6. A waterproof impact prediction system according to claim 1, characterized in that, The cascaded control of the next-level protection device includes: the time series of pressure head after the valve predicted based on the modified dynamic simulation model. The control output of the variable damping valve is calculated using a proportional-integral-derivative control algorithm. And map it to the target valve opening. : in, This represents the prediction error; These are the proportional, integral, and differential gain coefficients, respectively.

7. A waterproof impact prediction system according to claim 6, characterized in that, The cascaded control is a cyclic recursive process, and the system is further configured as follows: The measured impact index of the second-level node is obtained at the next-level protective device. And determine the new prediction error. ; and based on the new prediction error The model correction factor is updated a second time: in, These are the factor values ​​after the second correction; The correction gain is for the second-level node; the cyclic recursive process continues to execute until the water hammer event ends.

8. A waterproof impact prediction system according to claim 1, characterized in that, The system is further configured to perform post-event learning and model solidification after the water hammer event, using a batch gradient descent algorithm to adjust the pipe friction coefficient in the dynamic simulation model. Optimize: in, Based on The average loss function calculated from the historical events; and These are the friction coefficient values ​​before and after optimization, respectively. It is the learning rate; It is the gradient of the total loss function with respect to the friction coefficient. It is an event The measured impact index of the first-level node in the middle. This indicates that the initial predicted impact index is related to the friction coefficient. The function.

9. A waterproof impact prediction device, characterized in that, The device includes: Acquisition module: Acquires static topology and physical property information and multi-dimensional real-time operating condition data of the fluid transport system, and pre-calibrates the dynamic simulation model based on the information and data; Triggering module: When a water hammer triggering event is detected, the pre-calibrated dynamic simulation model is invoked to perform an initial prediction to generate an initial prediction result, and a pre-control command is issued to the first-level protection device based on the initial prediction result; The cyclic correction module acquires the measured impact index at the first-level protection device and compares it with the initial prediction result to determine the prediction error; and corrects the dynamic simulation model in real time based on the prediction error, and drives the corrected dynamic simulation model to perform cascade control on the next-level protection device. The prediction error is caused by It is stated that the measured impact index is derived from... It indicates that the predicted impact index is from express; The prediction error The measured impact index and predicted shock index Determine the deviation between them and predict the impact index. The calculation method is the same as the measured value, and its data source is the predicted pressure time series. Real-time correction of the dynamic simulation model is achieved by updating a model correction factor. accomplish: ； ； ； in, This is a measured time series of pressure head; and These are the first time the measured pressure exceeds the safe pressure head and the second time it falls back to the safe pressure head. The moment; These are the factor values ​​before correction; To correct the gain.

Images