阅读说明：本技术 一种蒸汽发生器故障诊断方法 (Steam generator fault diagnosis method ) 是由 王明军 王嘉诚 何少鹏 章静 田文喜 苏光辉 秋穗正 于 2021-07-30 设计创作，主要内容包括：本发明公开了一种蒸汽发生器故障诊断方法,包括以下步骤：通过核电厂蒸汽发生器现有传感器系统,采集蒸汽发生器运行数据；利用核电厂工业总线,实现蒸汽发生器运行数据的实时传输；经过信号处理单元,形成标准化数据；处理后的特征信号,分别传输至历史数据库与蒸汽发生器三维数字孪生模型系统(以下简称,三维数字孪生系统)压力、温度、速度等特征信号；通过三维数字孪生系统中物理引擎实现蒸汽发生器内部运行状况实时模拟；根据计算流体力学模拟与历史数据库,构建蒸汽发生器智能故诊断模型；结合模拟图像通过蒸汽发生器智能故诊断模型,实现蒸汽发生器故障诊断。(The invention discloses a steam generator fault diagnosis method, which comprises the following steps: collecting operating data of a steam generator through an existing sensor system of the steam generator of the nuclear power plant; the real-time transmission of the operation data of the steam generator is realized by utilizing the industrial bus of the nuclear power plant; forming standardized data through a signal processing unit; the processed characteristic signals are respectively transmitted to a historical database and a steam generator three-dimensional digital twin model system (hereinafter referred to as a three-dimensional digital twin system) to obtain characteristic signals of pressure, temperature, speed and the like; real-time simulation of the internal operating condition of the steam generator is realized through a physical engine in a three-dimensional digital twin system; constructing an intelligent failure diagnosis model of the steam generator according to the computational fluid dynamics simulation and historical database; and combining the simulation image to realize steam generator fault diagnosis through a steam generator intelligent diagnosis model.)

1. A steam generator fault diagnosis method characterized by: diagnosing faults in real time in the operating state of the steam generator; the fault diagnosis system consists of a nuclear power plant steam generator sensor system and a nuclear power plant industrial bus, and realizes real-time monitoring and data interaction of the operation state of a steam generator; performing feature extraction and feature standardization on the detection data by utilizing signal processing; simulating the flow field characteristics in the steam generator by using a physical engine according to the characteristic signals; constructing an intelligent failure diagnosis model of the steam generator according to the computational fluid dynamics simulation and historical database; the steam generator fault diagnosis is realized through an intelligent steam generator fault diagnosis model by combining the simulation image;

the method comprises the following steps:

step 1: the method comprises the following steps of collecting steam generator operation data by utilizing a steam generator sensor system of a nuclear power plant, and specifically comprising the following steps of:

step 1-1: marking the position of a sensing device of a steam generator of a nuclear power plant to realize the mutual correspondence of the sensor and the spatial information;

step 1-2: classifying the sensor groups in the step 1-1 into a primary steam generator measuring group, a secondary steam generator measuring group, a steam generator inlet and outlet pipeline group and a factory environment group;

step 1-3: collecting steam generator operation data with a time sequence characteristic by using a nuclear power plant steam generator sensor system;

step 2: storing the steam generator operation data with the time sequence characteristics acquired in the step 1 into a nuclear power plant data center by using a nuclear power plant industrial bus, so as to realize real-time transmission of the steam generator operation data;

and step 3: the method comprises the following steps of enabling operation data of a steam generator with a time sequence characteristic in a nuclear power plant data center to pass through a filter function, enabling the operation data of the steam generator to form a standardized data format, and specifically comprising the following steps:

step 3-1: extracting effective data from the steam generator operation data with the time sequence characteristics stored in the data center of the nuclear power plant in the step 2 by adopting a Kalman filtering method, a Gaussian filtering method, a median filtering method, an amplitude limiting value filtering method, a recursive average filtering method, a recursive median filtering method, a first-order lag filtering method or a weighted recursive average filtering method according to the signal characteristics;

step 3-2: enabling the effective data filtered in the step 3-1 to correspond to the sensor position information in the step 1-1 one by one, and realizing the association of sensor signals and position information;

step 3-3: performing characteristic coding classification on the information correlated in the step 3-2 according to the physical quantity pressure, temperature, amplitude, radiant quantity, flow rate and humidity measured by the steam generator of the nuclear power plant;

step 3-4: according to the following table, the first column from the left is a sensor code column initial letter P, namely detected physical quantity type pressure P, temperature T, amplitude A, radiation quantity R, flow speed F and environment humidity H, and AABB is a measuring point and a specific sensor unit respectively; the second column is a sensor space position, S is a primary measurement group P of the steam generator, a secondary measurement group S of the steam generator, an inlet and outlet pipeline group T of the steam generator and a factory environment group S), and AABBCC is a Cartesian coordinate system space coordinate absolute or a specified origin relative height; the third column is detection start time YY \ MM \ DD; the fourth column is a specific measurement value to form a standard signal in a standard format conforming to data analysis;

sensor encoding Spatial position of sensor Time of detection Amount of detection P-AABB S-AABBCC YY\MM\DD Value of

And 4, step 4: respectively transmitting the standard signals in the step 3-4 to a historical database and a three-dimensional digital twinning system, and specifically comprising the following steps;

step 4-1: storing the standard signals in the step 3-4 to a local operation data center of the nuclear power plant, and drawing a detection curve;

step 4-2: according to the requirements of the analog physical quantity of the three-dimensional digital twin system, signals related to pressure, temperature, speed and radiation characteristics are transmitted to the three-dimensional digital twin system;

and 5: the characteristic signals in the step 4-2 are subjected to real-time simulation of the internal operation condition of the steam generator through a physical engine in a three-dimensional digital twin system, so that the three-dimensional operation state and environment simulation of the steam generator are realized, and a visual effect diagram is obtained;

step 6: according to the computational fluid dynamics simulation result and the characteristic signal of the historical database, the intelligent fault diagnosis of the steam generator is realized, and the method comprises the following specific steps:

step 6-1: respectively simulating the steam generator internal flow field under the safe operation condition and the accident condition of the steam generator by utilizing a computational fluid mechanics program to form a pressure, speed, temperature and radiation cloud chart; meanwhile, simulating the operation environment of a steam generator plant, realizing numerical simulation under the working conditions of steam generator failure and safe operation, and forming a radiation and temperature cloud chart;

step 6-2: correlating the simulated cloud picture in the step 6-1 according to the spatial position of the sensor in the step 3-4 to realize that the numerical simulated cloud picture corresponds to the position of the acquired signal and marking;

step 6-3: taking the numerical simulation cloud picture marked in the step 6-2 as an intelligent fault diagnosis learning sample of the steam generator;

step 6-4: forming an intelligent diagnosis model of the steam generator by utilizing a supervised deep learning Convolutional Neural Network (CNN);

and 7: and (3) the visualized effect diagram in the step (5) is used for realizing the fault diagnosis of the steam generator through the intelligent fault diagnosis model of the steam generator in the step (6), and the specific steps are as follows:

step 7-1: taking the visual effect diagram in the step 5 as input data, and utilizing a 6-4 intelligent fault diagnosis model of the steam generator to classify fault types and predict service life, so as to finally realize the full three-dimensional fault diagnosis and service life prediction of the steam generator;

step 7-2: and (4) updating the intelligent fault diagnosis unit model of the steam generator irregularly according to the intelligent fault diagnosis result of the steam generator in the step 7-1, and repeating the step 6 and the step 7.

Technical Field

The invention belongs to the technical field of nuclear reactor thermal hydraulic calculation, and particularly relates to a steam generator fault diagnosis method based on a numerical simulation and digital twinning system.

Background

The steam generator of the pressurized water reactor is a heat exchange device for generating steam required by a steam turbine, in the nuclear reactor, heat generated by nuclear fission is taken out by a coolant, and is transferred to a two-loop working medium through the steam generator, so that the two-loop working medium generates steam with certain temperature, certain pressure and certain dryness. Steam generators are not only primary but also secondary devices, and are important energy exchange devices in nuclear power plants.

According to the operation experience of nuclear power plants, the steam generator has the problems of vibration, abrasion, fatigue, corrosion and the like during the operation process, so that the performance of heat transfer pipes and other components is degraded and the normal use is influenced. The performance degradation of the steam generator relates to numerous disciplines such as thermal hydraulic power, mechanics, materials and the like, if the safe operation of the steam generator is difficult to ensure from a single aspect and a single dimensionality, the operation fault of the steam generator can be more intuitively and timely predicted by means of a deep learning algorithm, computational fluid mechanics and a digital twinning technology.

Disclosure of Invention

The invention aims to provide a steam generator fault diagnosis method which can be realized by using a computational fluid dynamics means, an existing nuclear power plant detection system and a three-dimensional digital twin system.

In order to achieve the purpose, the invention adopts the following technical scheme:

a steam generator fault diagnosis method characterized by: diagnosing faults in real time in the operating state of the steam generator; the fault diagnosis system consists of a nuclear power plant steam generator sensor system and a nuclear power plant industrial bus, and realizes real-time monitoring and data interaction of the operation state of a steam generator; performing feature extraction and feature standardization on the detection data by utilizing signal processing; simulating the flow field characteristics in the steam generator according to the characteristic signals (related to pressure, temperature and speed) by using a physical engine; constructing an intelligent failure diagnosis model of the steam generator according to the computational fluid dynamics simulation and historical database; the steam generator fault diagnosis is realized through an intelligent steam generator fault diagnosis model by combining the simulation image;

the method comprises the following steps:

step 1: the method comprises the following steps of collecting steam generator operation data by utilizing a steam generator sensor system of a nuclear power plant, and specifically comprising the following steps of:

step 1-1: according to the steam plant structure model of the nuclear power plant, drawing a three-dimensional geographic information model by using 3D MAX and GIS software, and marking the position of a sensing device of a steam generator of the nuclear power plant on the three-dimensional model to realize the mutual correspondence of the sensor and spatial information;

step 1-2: classifying the sensor groups in the step 1-1 into a primary steam generator measuring group, a secondary steam generator measuring group, a steam generator inlet and outlet pipeline group and a factory environment group;

step 1-3: collecting steam generator operation data with a time sequence characteristic by using a nuclear power plant steam generator sensor system;

step 2: storing the steam generator operation data with the time sequence characteristics acquired in the step 1 into a nuclear power plant data center by utilizing a nuclear power plant industrial BUS (BUS) system, so as to realize real-time transmission of the steam generator operation data;

and step 3: the method comprises the following steps of enabling operation data of a steam generator with a time sequence characteristic in a nuclear power plant data center to pass through a filter function, enabling the operation data of the steam generator to form a standardized data format, and specifically comprising the following steps:

step 3-1: extracting effective data from the steam generator operation data with the time sequence characteristics stored in the data center of the nuclear power plant in the step 2 by adopting a Kalman filtering method, a Gaussian filtering method, a median filtering method, an amplitude limiting value filtering method, a recursive average filtering method, a recursive median filtering method, a first-order lag filtering method or a weighted recursive average filtering method according to the signal characteristics;

step 3-2: enabling the effective data filtered in the step 3-1 to correspond to the sensor position information in the step 1-1 one by one, and realizing the association of sensor signals and position information;

step 3-3: performing characteristic coding classification on the information correlated in the step 3-2 according to the physical quantity pressure, temperature, amplitude, radiant quantity, flow rate and humidity measured by the steam generator of the nuclear power plant;

step 3-4: according to the following table, the first column from the left is a sensor code column initial letter P, namely detected physical quantity type pressure P, temperature T, amplitude A, radiation quantity R, flow speed F and environment humidity H, and AABB is a measuring point and a specific sensor unit respectively; the second column is a sensor space position, S is a primary measurement group P of the steam generator, a secondary measurement group S of the steam generator, an inlet and outlet pipeline group T of the steam generator and a factory environment group S), and AABBCC is a Cartesian coordinate system space coordinate absolute or a specified origin relative height; the third column is detection start time YY \ MM \ DD; the fourth column is a specific measurement value to form a standard signal in a standard format conforming to data analysis;

sensor encoding	Spatial position of sensor	Time of detection	Amount of detection
				P-AABB	S-AABBCC	YY\MM\DD	Value of

And 4, step 4: respectively transmitting the standard signals in the step 3-4 to a historical database and a three-dimensional digital twinning system, and specifically comprising the following steps;

step 4-1: storing the standard signals in the step 3-4 to a local operation data center of the nuclear power plant, and drawing a detection curve;

step 4-2: according to the requirements of the analog physical quantity of the three-dimensional digital twin system, signals related to pressure, temperature, speed and radiation characteristics are transmitted to the three-dimensional digital twin system;

and 5: the characteristic signals in the step 4-2 are subjected to real-time simulation of the internal operation condition of the steam generator through a Unity 3D physical engine in a three-dimensional digital twin system, so that the three-dimensional operation state and environment simulation of the steam generator is realized, and a visual effect diagram is obtained;

step 6: according to the simulation results of a computational fluid dynamics program Fluent, Star CCM + or OpenFoam and characteristic signals of a historical database, the intelligent fault diagnosis of the steam generator is realized, and the method comprises the following specific steps:

step 6-1: respectively simulating the steam generator internal flow field under the safe operation condition and the accident condition of the steam generator by utilizing a computational fluid mechanics program to form a pressure, speed, temperature and radiation cloud chart; meanwhile, simulating the operation environment of a steam generator plant, realizing numerical simulation under the working conditions of steam generator failure and safe operation, and forming a radiation and temperature cloud chart;

step 6-2: correlating the simulated cloud picture in the step 6-1 according to the spatial position of the sensor in the step 3-4 to realize that the numerical simulated cloud picture corresponds to the position of the acquired signal and marking;

step 6-3: taking the numerical simulation cloud picture marked in the step 6-2 as an intelligent fault diagnosis learning sample of the steam generator;

step 6-4: forming an intelligent diagnosis model of the steam generator by utilizing a supervised deep learning Convolutional Neural Network (CNN);

and 7: and (3) the visualized effect diagram in the step (5) is used for realizing the fault diagnosis of the steam generator through the intelligent fault diagnosis model of the steam generator in the step (6), and the specific steps are as follows:

step 7-1: taking the visual effect diagram in the step 5 as input data, and utilizing a 6-4 intelligent fault diagnosis model of the steam generator to classify fault types and predict service life, so as to finally realize the full three-dimensional fault diagnosis and service life prediction of the steam generator;

step 7-2: and (4) updating the intelligent fault diagnosis unit model of the steam generator irregularly according to the intelligent fault diagnosis result of the steam generator in the step 7-1, and repeating the step 6 and the step 7.

The method adopts the combination of computational fluid dynamics software and a deep learning algorithm, and realizes an intelligent fault diagnosis algorithm according to a fluid dynamics model; meanwhile, by utilizing a digital twinning system, the fault diagnosis of the steam generator under real-time visual simulation is finally realized. Compared with the prior art, the invention has the following advantages:

1. by utilizing the mature data acquisition scheme of the nuclear power plant, the realizability and the implementation safety of the method are improved

2. A computational fluid mechanics calculation model is used as a deep learning algorithm training parameter, so that the fault types are enriched, and the reliability and accuracy of fault simulation are improved;

3. and fault diagnosis in a man-machine interaction mode is realized through a three-dimensional digital twin system.

Drawings

Fig. 1 is a schematic view of a steam generator.

FIG. 2 is a flow chart of the present invention.

Detailed Description

The following takes a typical steam generator testing scheme as an example in conjunction with the flow chart shown in fig. 2. For further details of the present invention, another exemplary steam generator configuration is shown in FIG. 1.

A steam generator fault diagnosis method based on a numerical simulation and digital twinning system comprises the following steps:

step 1: the method comprises the following steps of collecting steam generator operation data by utilizing a steam generator sensor system of a nuclear power plant, and specifically comprising the following steps of:

step 1-1: according to the steam plant structure model of the nuclear power plant, drawing a three-dimensional geographic information model by using 3D MAX and GIS software, and marking the position of a sensing device of a steam generator of the nuclear power plant on the three-dimensional model to realize the mutual correspondence of the sensor and spatial information;

step 1-2: classifying the sensor groups in the step 1-1 into a primary steam generator measuring group, a secondary steam generator measuring group, a steam generator inlet and outlet pipeline group and a factory environment group;

step 1-3: collecting steam generator operation data with a time sequence characteristic by using a nuclear power plant steam generator sensor system;

step 2: storing the steam generator operation data with the time sequence characteristics acquired in the step 1 into a nuclear power plant data center by utilizing a nuclear power plant industrial BUS (BUS) system, so as to realize real-time transmission of the steam generator operation data;

and step 3: the method comprises the following steps of enabling operation data of a steam generator with a time sequence characteristic in a nuclear power plant data center to pass through a filter function, enabling the operation data of the steam generator to form a standardized data format, and specifically comprising the following steps:

step 3-1: extracting effective data from the steam generator operation data with the time sequence characteristics stored in the data center of the nuclear power plant in the step 2 by adopting a Kalman filtering method, a Gaussian filtering method, a median filtering method, an amplitude limiting value filtering method, a recursive average filtering method, a recursive median filtering method, a first-order lag filtering method or a weighted recursive average filtering method according to the signal characteristics;

step 3-2: enabling the effective data filtered in the step 3-1 to correspond to the sensor position information in the step 1-1 one by one, and realizing the association of sensor signals and position information;

step 3-3: performing characteristic coding classification on the information correlated in the step 3-2 according to the physical quantity pressure, temperature, amplitude, radiant quantity, flow rate and humidity measured by the steam generator of the nuclear power plant;

step 3-4: according to the following table, the first column from the left is a sensor code column initial letter P, namely detected physical quantity type pressure P, temperature T, amplitude A, radiation quantity R, flow speed F and environment humidity H, and AABB is a measuring point and a specific sensor unit respectively; the second column is a sensor space position, S is a primary measurement group P of the steam generator, a secondary measurement group S of the steam generator, an inlet and outlet pipeline group T of the steam generator and a factory environment group S), and AABBCC is a Cartesian coordinate system space coordinate absolute or a specified origin relative height; the third column is detection start time YY \ MM \ DD; the fourth column is a specific measurement value to form a standard signal in a standard format conforming to data analysis;

sensor encoding	Spatial position of sensor	Time of detection	Amount of detection
				P-AABB	S-AABBCC	YY\MM\DD	Value of

And 4, step 4: respectively transmitting the standard signals in the step 3-4 to a historical database and a three-dimensional digital twinning system, and specifically comprising the following steps;

step 4-1: storing the standard signals in the step 3-4 to a local operation data center of the nuclear power plant, and drawing a detection curve;

step 4-2: according to the requirements of the analog physical quantity of the three-dimensional digital twin system, signals related to pressure, temperature, speed and radiation characteristics are transmitted to the three-dimensional digital twin system;

and 5: the characteristic signals in the step 4-2 are subjected to real-time simulation of the internal operation condition of the steam generator through a Unity 3D physical engine in a three-dimensional digital twin system, so that the three-dimensional operation state and environment simulation of the steam generator is realized, and a visual effect diagram is obtained;

step 6: according to the simulation results of a computational fluid dynamics program Fluent, Star CCM + or OpenFoam and characteristic signals of a historical database, the intelligent fault diagnosis of the steam generator is realized, and the method comprises the following specific steps:

step 6-1: respectively simulating the steam generator internal flow field under the safe operation condition and the accident condition of the steam generator by utilizing a computational fluid mechanics program to form a pressure, speed, temperature and radiation cloud chart; meanwhile, simulating the operation environment of a steam generator plant, realizing numerical simulation under the working conditions of steam generator failure and safe operation, and forming a radiation and temperature cloud chart;

step 6-2: correlating the simulated cloud picture in the step 6-1 according to the spatial position of the sensor in the step 3-4 to realize that the numerical simulated cloud picture corresponds to the position of the acquired signal and marking;

step 6-3: the numerical simulation cloud picture marked in the step 6-2 is used as an intelligent fault diagnosis learning sample of the steam generator,

step 6-4: forming an intelligent diagnosis model of the steam generator by utilizing a supervised deep learning Convolutional Neural Network (CNN);

and 7: and (3) the visualized effect diagram in the step (5) is used for realizing the fault diagnosis of the steam generator through the intelligent fault diagnosis model of the steam generator in the step (6), and the specific steps are as follows:

step 7-1: taking the visual effect diagram in the step 5 as input data, and utilizing a 6-4 intelligent fault diagnosis model of the steam generator to classify fault types and predict service life, so as to finally realize the full three-dimensional fault diagnosis and service life prediction of the steam generator;

step 7-2: and (4) updating the intelligent fault diagnosis unit model of the steam generator irregularly according to the intelligent fault diagnosis result of the steam generator in the step 7-1, and repeating the step 6 and the step 7.