An accident diagnosis tracking and rapid prediction system for third-generation passive nuclear power plants

Abstract

The present invention relates to an accident diagnosis, tracking and rapid prediction system for a third-generation passive nuclear power plant, including an accident diagnosis module, an accident tracking module and an accident prediction module; the accident diagnosis module reads data signals, and determines the operating state of the power plant and generates an accident sequence according to the data signals; the accident tracking module uses the data signals and the accident sequence to correct the thermal hydraulic calculation engine contained therein, so that it can accurately track and predict the actual operating conditions of the thermal hydraulics; the accident prediction module uses the corrected thermal hydraulic calculation engine to predict the development of the accident in multiple times. The system of the present invention can complete the accident diagnosis, prediction and state tracking of nuclear power units by collecting only a small amount of signals, and quickly complete the accident diagnosis, prediction, state tracking of nuclear power units and perform high-time nuclear power unit state and accident prediction.

Description

Technical Field

[0001] The present invention belongs to the technical field of nuclear safety and nuclear emergency, and in particular relates to an accident diagnosis, tracking and rapid prediction system for a third-generation passive nuclear power plant. Background Art

[0002] Nuclear power plants are complex structures involving numerous devices. To ensure timely and comprehensive status monitoring, each device is equipped with numerous sensors for real-time monitoring. Therefore, once an anomaly occurs in a nuclear power plant, especially when conditions develop towards a severe accident, plant operators and emergency response technicians must identify the anomaly from a vast amount of monitoring data and alarm signals to make informed decisions. Even during a severe accident, which is often accompanied by power loss and extensive instrument damage or failure, the plant status monitoring information provided by the nuclear power plant is incomplete and inaccurate, further complicating the difficulty of making informed decisions. Misjudgment, improper operation, or other human risk can lead to a catastrophic escalation of the accident. Consequently, these situations can create significant work and psychological stress for the personnel involved. In the event of a severe accident at a nuclear power plant, it is crucial to diagnose the root cause and current operating status based on limited reliable operational data, analyze the accident's evolution, and identify effective mitigation measures to ultimately restore the plant to a controllable state. This presents a significant challenge for nuclear emergency response personnel. There is an urgent need for relevant support systems that can diagnose the root causes of accidents, track the operating status of the units, evaluate the effectiveness of different intervention measures, and provide strong technical support and auxiliary decision-making for emergency response to nuclear accidents.

[0003] Among existing systems, there are also some systems that diagnose, evaluate, and predict nuclear power plant accidents, such as a method and system for diagnosing and responding to severe accidents in nuclear power plants, published with the number CN104915768 A. This system proposes a method for diagnosing the root causes of severe accidents in nuclear power plants. The method mainly matches relevant characteristic parameters in a pre-established severe accident sequence sample library. Once the characteristic parameters match an accident or several accident sequences in the sample library, the possible probability of the accident sequence or multiple accident sequences is output as the diagnosis result, and the accident is used as the input of the prediction module to predict the development of the accident and the effectiveness of intervention measures.

[0004] However, the system has some serious deficiencies, specifically:

[0005] 1. Diagnostic results are heavily dependent on the sample library. Severe accidents are the result of the superposition of multiple accidents, which is related to the time of occurrence of the initial accident and the superimposed accidents. It is difficult for the sample library to enumerate all accident sequences.

[0006] 2. The root cause of an accident is determined by matching characteristic parameters with a sample library. The late-stage phenomena of a severe accident are generally consistent, with core heating and melting. This characteristic parameter matching approach can yield multiple root causes, making it difficult to provide an accurate diagnosis and providing limited support for emergency response.

[0007] 3. Using this root cause diagnosis as input to the predictive calculation module, it is difficult to accurately predict the progression of an accident. This is because the prediction of a severe accident depends not only on the accident sequence but also on the timing of multiple accidents. Therefore, predictions based solely on the accident sequence cannot accurately reflect the true evolution of nuclear power plant accident conditions.

[0008] 4. Considering that third-generation nuclear power plants are equipped with passive dedicated safety facilities, the factors affecting the progress of accidents are more complex, which further increases the difficulty of accurately predicting the progress of accidents. Summary of the Invention

[0009] Based on the above-mentioned shortcomings and deficiencies in the prior art, one of the objects of the present invention is to at least solve one or more of the above-mentioned problems in the prior art. In other words, one of the objects of the present invention is to provide an accident diagnosis, tracking and rapid prediction system for a third-generation passive nuclear power plant that meets one or more of the above-mentioned needs.

[0010] In order to achieve the above-mentioned object of the invention, the present invention adopts the following technical solutions:

[0011] An accident diagnosis, tracking and rapid prediction system for a third-generation passive nuclear power plant, the system comprising an accident diagnosis module, an accident tracking module and an accident prediction module;

[0012] The accident diagnosis module includes a reading unit, a state recognition unit, and an accident recognition unit; the reading unit is used to read the data signals sent by the power plant equipment; the state recognition unit is used to identify the current power plant operating status based on the data signals; the accident recognition unit is used to identify whether there is an accident or accident sign, as well as the corresponding accident location and type based on the current power plant operating status. Several accidents or accident signs constitute an accident sequence;

[0013] The data tracking module includes a thermal hydraulic calculation engine and a correction unit. The thermal hydraulic calculation engine is used to calculate the simulated operation status of the thermal hydraulic according to the data signal and the accident sequence. The data tracking module compares the simulated operation status of the thermal hydraulic with the actual operation status of the thermal hydraulic, and makes the simulated operation status of the thermal hydraulic accurately track the actual operation status of the thermal hydraulic; the correction unit determines whether the current thermal hydraulic calculation engine needs to be corrected. If the simulated operation status of the thermal hydraulic deviates significantly from the actual operation status of the thermal hydraulic, the thermal hydraulic calculation engine is corrected to accurately track;

[0014] The accident prediction module is used to predict the development of accidents using the thermal hydraulic calculation engine.

[0015] As a preferred solution, the accident diagnosis module also includes a data correction unit for eliminating abnormal data from multiple data signals, averaging and merging normal data, and merging high and low pressure injection data.

[0016] As a preferred solution, the state recognition unit contains a recurrent neural network model, which can identify the current operating state of the power plant based on the input data signal.

[0017] As a preferred solution, the power plant operating states identified by the state identification unit include core flooding, partial core exposure, complete core exposure, core damage, pressure vessel failure and containment failure.

[0018] As a preferred solution, the accident identification unit identifies the accident or accident sign based on the data difference between two samplings.

[0019] As a preferred solution, the correction unit corrects the thermal hydraulic calculation engine by adjusting the boundary conditions of the thermal hydraulic calculation engine.

[0020] As a further preferred solution, the correction unit is used to correct the boundary conditions based on the deviation between the predicted operation of the thermal hydraulics and the actual operation of the thermal hydraulics, and use the corrected boundary conditions and the re-acquired data signals and accident sequence calculations to obtain the simulated operation of the thermal hydraulics, and determine whether the thermal hydraulic calculation engine needs to be corrected again until the simulated operation of the thermal hydraulics accurately tracks the predicted actual operation of the thermal hydraulics.

[0021] As a preferred solution, the system further includes an intervention evaluation module, which is used to obtain the intervention plan to be implemented, predict the effect of the intervention plan, and evaluate the effect.

[0022] As a further preferred solution, the prediction and evaluation of the intervention assessment module are based on severe accident management guidelines used in nuclear power plants.

[0023] As a further preferred solution, the intervention assessment module has multiple prediction channels, each prediction channel predicts the effect of an intervention solution.

[0024] Compared with the prior art, the present invention has the following beneficial effects:

[0025] The system of the present invention can complete the accident diagnosis, prediction and status tracking of nuclear power units by collecting only a small amount of signals;

[0026] Able to quickly complete nuclear power unit accident diagnosis, prediction and status tracking;

[0027] Able to predict the status and accidents of nuclear power units at high times;

[0028] Ability to simultaneously predict the effects of multiple different intervention programs. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 is a structural block diagram of an intervention assessment module according to an embodiment of the present invention;

[0030] Figure 2 The present invention is a flowchart of an accident diagnosis, tracking and rapid prediction system for a third-generation passive nuclear power plant according to an embodiment of the present invention. DETAILED DESCRIPTION

[0031] To more clearly illustrate the embodiments of the present invention, specific embodiments of the present invention will be described below with reference to the accompanying drawings. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings and other embodiments can be obtained based on these drawings without inventive efforts.

[0032] Example:

[0033] The present invention is a third-generation passive nuclear power plant accident diagnosis tracking and rapid prediction system, the structure of which is as follows: Figure 1 As shown, it includes an accident diagnosis module, an accident tracking module and an accident prediction module.

[0034] The accident diagnosis module includes a reading unit, a state recognition unit, and an accident recognition unit. The reading unit can read data signals from various devices or detection points in the power plant and send the data signals to the state recognition unit. The state recognition unit identifies the current operating status of each device in the power plant based on the input data signal, and further identifies the overall operating status of the power plant. The data signal and the operating status of the power plant are also sent to the accident recognition unit. The accident recognition unit performs two samples based on the operating status of the power plant and the data signal. Based on the data difference between the two samples, it identifies whether an accident has occurred or whether there are signs of an accident, and then further determines the accident type and location corresponding to the accident or accident sign. A sequence consisting of a series of accidents or accident signs is regarded as an accident sequence.

[0035] The state recognition unit incorporates a recurrent neural network model, which takes data signals from various devices or detection points as input. After identification, the recurrent neural network model outputs the operating status of each device and the overall operating status of the power plant. Specifically, the state recognition unit can identify the current power plant operating state and determine whether it is in one of the following accident states: core flooding, partially exposed core, completely exposed core, damaged core, pressure vessel failure, or containment failure.

[0036] The accident identification unit is capable of identifying the signs and types of accidents, identifying events that occurred between two data signal sampling periods, evaluating the mass and energy change rates of the steam generator, main system, and containment, assessing the LOCA area and location (in the case of a breach sequence), checking whether the containment has failed, and generating event sequences and boundary conditions for the thermal-hydraulic calculation engine.

[0037] To improve data accuracy, the accident diagnosis module also includes a data correction unit, which checks the consistency of multiple data signals, removes abnormal data from these signals, averages and merges normal data, and also combines high- and low-pressure safety injection data. Multiple data signals refer to the possibility that multiple data sources may output data signals for the same device or test point, thereby ensuring more accurate data signals.

[0038] In order to facilitate the analysis and use of the data signal, the data correction unit is also used to convert the data signal into a signal unit that matches the recognition program.

[0039] The data tracking module includes a thermal-hydraulic calculation engine and a correction unit. The accident sequence and data signals are input into the thermal-hydraulic calculation engine. After a series of thermal-hydraulic calculations under boundary conditions, the module outputs a predicted thermal-hydraulic operating status for the power plant. The correction unit compares the predicted thermal-hydraulic operating status with the actual thermal-hydraulic operating status at the next moment to determine whether the thermal-hydraulic calculation engine needs correction. If the predicted thermal-hydraulic operating status deviates significantly from the actual operating status, the thermal-hydraulic calculation engine is deemed to require correction. The thermal-hydraulic calculation engine is then corrected and the predicted and actual operating status are re-compared until the predicted operating status accurately tracks the actual operating status at the next moment. Specifically, corrections to the thermal-hydraulic calculation engine generally involve modifying the boundary conditions used. The thermal-hydraulic calculation engine can utilize the internationally widely used severe accident analysis software MAAP, and a physical model is built for the characteristics of third-generation advanced passive nuclear power plants.

[0040] The accident prediction module is used to use the thermal-hydraulic calculation engine to predict the development of accidents. Since the thermal-hydraulic calculation engine adjusted by the data tracking module is now able to accurately track and predict the actual operation of the thermal-hydraulic power plant, the accident prediction module can use the thermal-hydraulic calculation engine's ability to calculate in multiple times to predict the power plant status and accident development for a longer period of time in the future.

[0041] To provide technical support for nuclear emergency response and minimize accident consequences, an intervention assessment module has been added to the aforementioned system. This module is used to identify intervention plans, predict their effectiveness, and evaluate them. As an implementation, the intervention measures assessment module is based on the Severe Accident Management Guidelines (SAMG) for nuclear power plants. According to the SAMG, the positive and negative evaluations of intervention measures, as well as long-term concerns, are coupled with the development prediction results of the accident prediction module to achieve dynamic evaluation of intervention measures.

[0042] In order to improve the evaluation effect and quickly find the optimal solution, four prediction channels are set in the intervention evaluation module. Each prediction channel predicts the effect of an intervention plan, and realizes the comparative evaluation of the accident evolution under different intervention plans and assumptions. The structural block diagram of the intervention evaluation module is shown in the figure below. Figure 1 shown.

[0043] The above system works as follows:

[0044] The flowchart of accident diagnosis tracking and rapid prediction is as follows Figure 2 As shown, coarse data (i.e., a relatively small amount of data information) is first collected from various devices or collection points in the power plant. In this embodiment, after the coarse data is collected, a data correction unit checks whether the data signal is within a reasonable range. It then removes abnormal data from multiple data signals, averages and merges normal data, identifies damaged steam generator circuits, merges high- and low-pressure safety injection data, and converts the data into signal units that facilitate identification and processing.

[0045] The corrected data signal is input into the state recognition module to identify the current operating status of the power plant and determine whether there is an accident in the power plant as a whole, especially core flooding, partial core exposure, complete core exposure, core damage, pressure vessel failure, containment failure, etc.

[0046] The accident identification unit further determines whether there are any accidents or signs of accidents, as well as the corresponding accident locations and types based on the current power plant operating status and data signals. By evaluating the mass-energy change rates of the steam generator, main system and containment, and checking whether the containment has failed, the event sequence and boundary conditions of the thermal-hydraulic calculation engine are set.

[0047] The accident tracking module establishes and runs the thermal hydraulic tracking process, inputs the data signal into the thermal hydraulic calculation engine, performs calculations based on the set event sequence and boundary conditions, and outputs the predicted operation status of the thermal hydraulics. The correction unit compares the predicted operation status of the thermal hydraulics with the actual operation status of the thermal hydraulics, and determines whether the thermal hydraulic calculation engine needs to be corrected based on the size of the deviation between the two. If correction is required, the boundary conditions are modified accordingly based on the deviation value. If the deviation value is too large, the thermal hydraulic calculation engine is directly initialized. After correcting or initializing the thermal hydraulic calculation engine, new data signals are obtained again, and the deviation value of the next predicted operation status is compared with the actual operation status. If a large deviation value still exists, correction is made again until the predicted operation status can accurately track the actual operation status at the next moment.

[0048] The accident prediction module uses a modified thermal hydraulic calculation engine, combined with rough data and accident sequences to make long-term multiple-time predictions of future accident developments.

[0049] The intervention evaluation module obtains the planned intervention plan for the accident, and at the same time predicts the effects of multiple intervention plans and evaluates the effects to obtain the optimal plan through comparison.

[0050] Based on the above-described system and method, this embodiment can track the status of nuclear power units and diagnose the cause of accidents based on a small number of signals from the nuclear power unit (120 for the AP1000 reactor type). The prediction speed is relatively fast, and accident status tracking and diagnosis can be completed within 300 seconds after receiving the data, maintaining an accuracy rate of over 95%. Furthermore, the above-described system and method provide comprehensive diagnostic information for accidents, including accident type (LOCA, SBO, MSLB, etc.), accident severity, and the area where the accident occurred. The accident prediction fast-time multiplier is also high, capable of achieving an overall fast-time effect of over 40 times. Furthermore, the effects of four different intervention plans can be simultaneously predicted. Each intervention plan can simultaneously provide accident progression for five different hypothetical models: best estimate, conservative estimate, optimistic estimate, and an intermediate estimate between conservative and best estimate, or an intermediate estimate between optimistic and best estimate. This means that up to 20 possible evolution scenarios for an accident can be provided at a time.

[0051] It should be noted that the above embodiments are only detailed descriptions of the preferred embodiments and principles of the present invention. For ordinary technicians in this field, there will be changes in the specific implementation methods based on the ideas provided by the present invention, and these changes should also be regarded as the scope of protection of the present invention.

Claims

1. An accident diagnosis, tracking and rapid prediction system for a third-generation passive nuclear power plant, characterized in that: The system includes an accident diagnosis module, an accident tracking module and an accident prediction module; The accident diagnosis module includes a reading unit, a state recognition unit, and an accident recognition unit; the reading unit is used to read the data signal sent by the power plant equipment; the state recognition unit is used to identify the current power plant operation state according to the data signal; the accident recognition unit is used to identify whether there is an accident or accident sign, and the corresponding accident location and type according to the current power plant operation state, and a number of accidents or accident signs constitute an accident sequence; The accident tracking module comprises a thermal hydraulic calculation engine and a correction unit. The thermal hydraulic calculation engine is used to calculate the simulation operation of the thermal hydraulic according to the data signal and the accident sequence. The accident tracking module compares the simulation operation of the thermal hydraulic with the actual operation of the thermal hydraulic, and makes the simulation operation of the thermal hydraulic accurately track the actual operation of the thermal hydraulic; the correction unit determines whether the current thermal hydraulic calculation engine needs to be corrected. If the simulation operation of the thermal hydraulic deviates greatly from the actual operation of the thermal hydraulic, the thermal hydraulic calculation engine is corrected to accurately track; The data sources of the accident identification unit include the data signal obtained by the reading unit and the power plant operating state obtained by the state identification unit. The accident identification unit performs two samplings in succession according to the power plant operating state and the data signal, and identifies whether an accident has occurred or whether there are signs of an accident according to the data difference between the two samples, and further determines the accident type and location corresponding to the accident or the sign of the accident. The accident prediction module is used to use the thermal-hydraulic calculation engine and predict the development of the accident based on the power plant operating state. The state identification unit contains a recurrent neural network model, and the recurrent neural network model can identify the current power plant operating state according to the input data signal. The correction unit corrects the thermal-hydraulic calculation engine by adjusting the boundary conditions of the thermal-hydraulic calculation engine. The correction unit is used to correct the boundary conditions according to the deviation between the simulated operation of the thermal-hydraulic and the actual operation of the thermal-hydraulic, and uses the corrected boundary conditions and the re-acquired data signal and the accident sequence to calculate the simulated operation of the thermal-hydraulic, and determines whether the thermal-hydraulic calculation engine needs to be corrected again until the simulated operation of the thermal-hydraulic accurately tracks the actual operation of the thermal-hydraulic.

2. The accident diagnosis, tracking and rapid prediction system for a third-generation passive nuclear power plant according to claim 1, characterized in that: The accident diagnosis module also includes a data correction unit, which is used to eliminate abnormal data in multiple data signals, average and merge normal data, and merge high and low pressure injection data.

3. The accident diagnosis, tracking and rapid prediction system for a third-generation passive nuclear power plant according to claim 1, characterized in that: The current power plant operation status identified by the status identification unit includes core flooding, core partial exposure, core complete exposure, core damage, pressure vessel failure and containment failure.

4. The accident diagnosis, tracking and rapid prediction system for a third-generation passive nuclear power plant according to claim 1, characterized in that: The system further comprises an intervention assessment module, which is used to obtain an intervention plan to be implemented, predict the effect of the intervention plan and evaluate the effect.

5. The accident diagnosis, tracking and rapid prediction system for a third-generation passive nuclear power plant as claimed in claim 4, characterized in that: The prediction and evaluation of the intervention assessment module are based on severe accident management guidelines used in nuclear power plants.

6. The accident diagnosis, tracking and rapid prediction system for a third-generation passive nuclear power plant according to claim 4, characterized in that: The intervention assessment module has a plurality of prediction channels, each of which predicts the effect of an intervention plan.

Images