Nuclear power important parameter abnormity identification and alarm system

Abstract

The invention particularly relates to a nuclear power important parameter abnormity identification and alarm system. The system comprises a data warehouse and at least one group of parameter abnormity identification modules arranged in the data warehouse. And each parameter abnormity identification module receives the real-time operation data of the unit as input so as to carry out abnormity identification and alarm on specified nuclear power parameters. According to the invention, the operation state of the unit is monitored, and the abnormity of important parameters, equipment and systems can be identified timely and accurately.

Description

Technical Field

[0001] The present invention relates to the technical field of nuclear power safety protection, and in particular to a nuclear power important parameter abnormality recognition and alarm system. Background Art

[0002] During the normal operation of a nuclear power plant, unexpected protection actions such as load shedding, shutdown, and reactor shutdown may occur due to faulty operation or failure of process equipment, instrumentation and control system failure, power grid failure, fire, insufficient personnel skills, human error, etc. The current detection of abnormal conditions during the operation of the unit mainly relies on the alarm information of the DCS system. The corresponding alarm is triggered only when a single measurement value or a combination of multiple measurement values ​​reaches the corresponding threshold, and there is a lack of trend analysis; in addition, most of the typical abnormal events of the unit have different precursors before causing an accident, but the DCS system alarm cannot fully cover these events, nor can it prompt the operator in time according to the correlation fluctuations of parameters, so it is difficult to attract attention in the early stage of abnormal events.

[0003] Traditional monitoring methods based on single-variable control charts can no longer reflect changes in complex industrial processes. Although the big data of modern industrial systems brings rich information, nuclear power plant operators cannot empirically evaluate process operations based on massive amounts of data. Summary of the invention

[0004] The purpose of the present invention is to provide a nuclear power important parameter abnormality identification and early warning system to monitor the operating status of the unit and to promptly and accurately identify abnormalities of important parameters, equipment and systems.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] A nuclear power important parameter abnormality identification and alarm system comprises a data warehouse and at least one group of parameter abnormality identification modules arranged in the data warehouse; each parameter abnormality identification module receives the real-time operation data of the unit as input to perform abnormality identification and alarm of the specified nuclear power parameter.

[0007] In the present invention, each parameter anomaly identification module includes a first anomaly monitoring submodule based on knowledge drive, a second anomaly monitoring submodule based on data drive, and a comprehensive alarm submodule;

[0008] The first abnormality monitoring submodule builds an abnormality recognition model based on the abnormality recognition rules described in the preset abnormal event tree diagram; when the first abnormality monitoring submodule determines that the data characteristics of the corresponding parameters in the received real-time operation data of the unit meet the description in the abnormal event tree diagram, it determines that an abnormality exists;

[0009] The second abnormality monitoring submodule includes a normal state recognition subunit and an abnormal feature recognition subunit;

[0010] The normal state identification subunit uses the normal data in the historical operation data of the unit as training data and constructs a normal state reconstruction model based on artificial intelligence technology; the normal state reconstruction model predicts the value range of the corresponding parameters according to the real-time operation data of the unit input online; when the normal state identification subunit determines that the value range of the corresponding parameter in the received real-time operation data of the unit predicted by the normal state reconstruction model exceeds the normal range, it determines that there is an abnormality;

[0011] The abnormal feature recognition subunit uses the abnormal data in the historical operation data of the unit as training data and builds an abnormal feature recognition model based on artificial intelligence technology; when the abnormal feature recognition subunit determines that the received real-time operation data characteristics of the unit are consistent with the abnormal characteristics in the abnormal feature recognition model, it determines that an abnormality exists;

[0012] The comprehensive alarm submodule calculates the inputs of the first abnormal monitoring submodule, the normal state identification subunit and the abnormal feature identification subunit based on the preset weights, generates a corresponding comprehensive calculation result, and determines whether there is any abnormality in the current unit operation data based on the comprehensive calculation result. If there is an abnormality, an abnormal alarm is issued.

[0013] In the present invention, the normal state identification subunit includes an abnormal alarm discrimination threshold unit, a normal state reconstruction model, a residual calculation unit, and an alarm unit;

[0014] During normal operation, the normal state reconstruction model is called, and the corresponding parameter combination in the real-time operation data of the unit is input into the normal state reconstruction model, and the normal state reconstruction model outputs the corresponding reconstruction value; the residual calculation unit calculates N groups of residuals between the reconstruction value and the current value of each parameter in the parameter combination; the alarm unit determines that an abnormality exists when the residual exceeds the abnormal alarm judgment threshold.

[0015] In the present invention, the normal state identification subunit extracts parameter combinations describing all modes of the detection object from the historical operation data of the unit in the data warehouse as training data for the normal state reconstruction model.

[0016] In the present invention, each abnormal alarm discrimination threshold in the abnormal alarm discrimination threshold unit is obtained based on the normal state reconstruction model, including the following steps:

[0017] After the normal state reconstruction model is constructed, the normal operation historical data of the corresponding parameter combination is obtained from the data warehouse and input into the normal state reconstruction model for prediction, so as to predict the characteristic value based on the normal operation historical data;

[0018] The difference between each characteristic value and the corresponding parameter value in the parameter combination is obtained as each abnormal alarm discrimination threshold in the abnormal alarm discrimination threshold unit.

[0019] In the present invention, an alarm condition is preset in the alarm unit. When the abnormal alarm discrimination threshold and the residual satisfy the preset alarm condition, the normal state recognition subunit determines that there is a possibility of abnormality.

[0020] In the present invention, the preset alarm conditions include the following two groups:

[0021] (1) At least one set of residuals exceeds the corresponding abnormal alarm discrimination threshold;

[0022] (2) In a specified time window, the number of cases exceeding the abnormal alarm threshold exceeds the preset percentage;

[0023] (3) Specify a time window, and the situation where the abnormal alarm threshold is exceeded continuously must meet the preset length.

[0024] In the present invention, the abnormal feature recognition subunit extracts the abnormal features of the detection object from the historical operation data of the unit in the data warehouse as the training data of the abnormal feature recognition model, and the abnormal features of the detection object are the most relevant features when the detection object is abnormal;

[0025] During normal operation, the abnormal feature recognition model is called, and the corresponding parameter combination in the real-time operation data of the unit is input into the abnormal feature recognition model. The abnormal feature recognition model determines whether there is an abnormality based on the degree of abnormal feature matching. When the abnormal feature recognition subunit determines that the characteristics of the real-time operation data of the unit are consistent with the abnormal characteristics in the abnormal feature recognition model, it determines that there is an abnormality.

[0026] In the present invention, based on the unit operation experience, the experience of state abnormality judgment is converted into a configuration diagram of computer operation to construct an abnormality recognition model, which includes the following steps:

[0027] (1) For different detection objects, the abnormal event tree diagram is drawn using the VISIO diagram based on the operation experience. The abnormal event tree diagram includes multiple branches at different levels, and each branch corresponds to an abnormal type;

[0028] (2) Using knowledge modeling tools, develop a knowledge-driven anomaly detection configuration diagram based on the abnormal event tree diagram verified by unit operation experience;

[0029] (3) Construct the corresponding anomaly recognition model based on the anomaly detection configuration diagram;

[0030] During normal operation, the knowledge-driven anomaly detection configuration diagram is deployed to the data warehouse, and then the real-time operation data of the unit is input into the corresponding anomaly recognition model for processing. The first anomaly monitoring submodule judges the data features of the corresponding parameters in the real-time operation data of the unit based on the anomaly recognition rules described in the abnormal event tree diagram, and inputs the judgment results into the comprehensive alarm submodule for processing.

[0031] In the present invention, the comprehensive alarm submodule includes a weight adjustment unit; the weight adjustment unit presets different weights for different inputs to generate corresponding comprehensive calculation results;

[0032] In the weight adjustment unit, the weights inputted by the first abnormality monitoring submodule, the normal state identification submodule and the abnormality feature identification submodule are adjusted according to the actual operation status of the unit.

[0033] In the present invention, the nuclear power important parameter abnormality identification and alarm system also includes an online update module; the online update module uses updated unit historical operation data according to user feedback to implement the update of the data-driven normal state model and abnormal feature identification model.

[0034] In the present invention, the online update module includes a manual discrimination unit, a normal state reconstruction model online update unit, and an abnormal feature recognition model offline update unit;

[0035] The manual judgment unit judges the abnormal alarm status issued by the comprehensive alarm submodule;

[0036] If it is judged as a false alarm, it means that there may be new normal data, and the new normal data does not exist in the historical operation data of the unit, then it will be transferred to the online update unit of the normal state reconstruction model for processing;

[0037] The normal state reconstruction model online updating unit adds new normal data to the training data of the normal state reconstruction model to obtain updated training data; subsequently, the normal state reconstruction model is retrained based on the updated training data to obtain an updated normal state reconstruction model;

[0038] And based on retraining or incremental learning, an updated normal state reconstruction model is obtained;

[0039] If it is judged as a missed report, it means that there are features that should be identified by the existing abnormal feature recognition model, but are not actually identified, and then it is transferred to the abnormal feature recognition model offline update unit for processing;

[0040] The abnormal feature recognition model offline update unit extracts the missed features and establishes an abnormal feature library, and subsequently retrains the abnormal feature recognition model based on the abnormal data in the abnormal feature library to obtain an updated abnormal feature recognition model.

[0041] Beneficial technical effects of the present invention:

[0042] The nuclear power important parameter abnormality identification and alarm system of the present invention has a group of parameter abnormality identification modules for each type of important parameter. When the unit is operating normally, each parameter abnormality identification module is used to continuously and automatically monitor the important parameters, and early warning of abnormalities in equipment, functions and parameters can be given. When a single important parameter deviates from normal operation or an abnormality occurs in the status of important systems or equipment, the corresponding staff is reminded, thereby reducing the workload of the operating personnel and improving the stability level of the unit. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 The diagram is a structural diagram of an embodiment of a nuclear power important parameter abnormality identification and alarm system of the present invention. DETAILED DESCRIPTION

[0044] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field to which this application belongs; the terms used in the specification of the application are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned figure descriptions are intended to cover non-exclusive inclusions.

[0045] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0046] The technical solution of the present invention is clearly and completely described below in conjunction with the accompanying drawings and specific embodiments.

[0047] See also Figure 1 This embodiment provides a nuclear power important parameter abnormality identification and alarm system, including a data warehouse and at least one group of parameter abnormality identification modules arranged in the data warehouse; each parameter abnormality identification module receives the real-time operation data of the unit as input to perform abnormality identification and alarm of specified nuclear power parameters.

[0048] In this embodiment, each parameter anomaly identification module includes a first anomaly monitoring submodule based on knowledge drive, a second anomaly monitoring submodule based on data drive, and a comprehensive alarm submodule;

[0049] The first abnormality monitoring submodule builds an abnormality recognition model based on the abnormality recognition rules described in the abnormal event tree diagram summarized and sorted out from the preset unit operation experience; when the first abnormality monitoring submodule determines that the data characteristics of the corresponding parameters in the received unit real-time operation data meet the description in the abnormal event tree diagram, it determines that there is an abnormality;

[0050] The second abnormality monitoring submodule includes a normal state recognition subunit and an abnormal feature recognition subunit;

[0051] The normal state identification subunit uses the normal data in the historical operation data of the unit as training data and constructs a normal state reconstruction model based on artificial intelligence technology; the normal state reconstruction model predicts the value range of the corresponding parameters according to the real-time operation data of the unit input online; when the normal state identification subunit determines that the value range of the corresponding parameter in the received real-time operation data of the unit predicted by the normal state reconstruction model exceeds the normal range, it determines that there is an abnormality;

[0052] The abnormal feature recognition subunit uses the abnormal data in the historical operation data of the unit as training data and builds an abnormal feature recognition model based on artificial intelligence technology; when the abnormal feature recognition subunit determines that the received real-time operation data characteristics of the unit are consistent with the abnormal characteristics in the abnormal feature recognition model, it determines that an abnormality exists;

[0053] The comprehensive alarm submodule calculates the inputs of the first abnormal monitoring submodule, the normal state identification subunit and the abnormal feature identification subunit based on the preset weights, generates a corresponding comprehensive calculation result, and determines whether there is any abnormality in the current unit operation data based on the comprehensive calculation result. If there is an abnormality, an abnormal alarm is issued.

[0054] Under stable operation, many parameters of nuclear power plants are stable. However, due to their stability, they are easily overlooked in operation monitoring and lack continuous dynamic monitoring, which results in abnormal changes in these parameters not being discovered and identified in time, and only being noticed after reaching the alarm threshold. The nuclear power important parameter abnormality identification and alarm system of the present invention adopts a coexistence of two methods: a knowledge-driven first abnormality monitoring submodule and a data-driven second abnormality monitoring submodule to identify abnormalities of nuclear power important parameters.

[0055] Among them, the data-driven second abnormal monitoring submodule is further divided into a normal state identification subunit and an abnormal feature identification subunit. When working, the real-time operation data of the unit is input into the normal state identification subunit, the abnormal feature identification subunit and the first abnormal monitoring submodule respectively. Each module makes a judgment on whether it is abnormal according to the preset model or abnormal judgment condition, and inputs its own judgment into the comprehensive alarm submodule. The comprehensive alarm submodule makes a judgment according to the preset judgment logic, and finally gives an identification result of whether it is abnormal, thereby realizing continuous automatic monitoring of important parameters during normal operation of the unit, and issuing an early warning when important parameters deviate from normal operation and abnormalities occur, so as to promptly remind the operator to pay attention to the parameter change trend in time.

[0056] In this embodiment, the normal state identification subunit includes an abnormal alarm discrimination threshold unit, a normal state reconstruction model, a residual calculation unit, and an alarm unit;

[0057] During normal operation, the normal state reconstruction model is called, and the corresponding parameter combination in the real-time operation data of the unit is input into the normal state reconstruction model, and the normal state reconstruction model outputs the corresponding reconstruction value; the residual calculation unit calculates N groups of residuals between the reconstruction value and the current value of each parameter in the parameter combination; the alarm unit determines that an abnormality exists when the residual exceeds the abnormal alarm judgment threshold.

[0058] The normal state reconstruction model is based on the historical time series data of the unit, and the corresponding normal samples of the historical operation data of the unit are established through data analysis technology. The corresponding evaluation model is established based on the normal samples of the historical operation data of the unit. For different detection objects, corresponding normal state reconstruction models are constructed respectively; when constructing the normal state reconstruction model, the most relevant parameter combination with the detection object is sorted out as the training data of the normal state reconstruction model.

[0059] In order to identify various abnormalities of the detection object as much as possible, when training the normal state reconstruction model, a parameter combination that can describe all modes of the detection object is selected as training data. In this embodiment, the normal state recognition subunit extracts a parameter combination that describes all modes of the detection object from the historical operation data of the unit in the data warehouse as the training data of the normal state reconstruction model;

[0060] All modes include the following combinations: (1) Identification and early warning of abnormalities in important parameters of the primary circuit, secondary circuit, turbine, peripheral, electrical, three wastes and ventilation and fire protection systems when the unit is operating at 2%, 8%, 17%, 30%, 50%, 80% and 100% power levels; (2) Identification and early warning of abnormalities in important parameters during the unit's 2MW / min power reduction and 0.5MW / min power increase; (3) Identification and early warning of abnormalities in core parameters during the unit's non-power operation up and down periods.

[0061] In this embodiment, each abnormal alarm discrimination threshold in the abnormal alarm discrimination threshold unit is obtained based on the normal state reconstruction model, including the following steps:

[0062] After the normal state reconstruction model is constructed, the normal operation history data of the corresponding parameter combination is obtained from the data warehouse and input into the normal state reconstruction model for prediction, so as to predict the characteristic value based on the normal operation history data; the characteristic value includes the value, the value range and the change rate;

[0063] The difference between each characteristic value and the corresponding parameter value in the parameter combination is obtained as each abnormal alarm discrimination threshold in the abnormal alarm discrimination threshold unit.

[0064] In this embodiment, an alarm condition is preset in the alarm unit. When the abnormal alarm discrimination threshold and the residual satisfy the preset alarm condition, the normal state recognition subunit determines that there is a possibility of abnormality.

[0065] In this embodiment, the preset alarm conditions include the following two groups:

[0066] (1) At least one set of residuals exceeds the corresponding abnormal alarm discrimination threshold;

[0067] (2) In a specified time window, the number of cases exceeding the abnormal alarm threshold exceeds the preset percentage;

[0068] (3) Specify a time window, and the situation where the abnormal alarm threshold is exceeded continuously must meet the preset length.

[0069] The preset alarm conditions are set in this way to avoid false alarms as much as possible. Due to data jitter, business knowledge is introduced into the alarm unit to reduce the false alarm rate. For example, based on business knowledge, it can be confirmed that there is no jitter data accounting for more than 1 / 3 of the data within 1 minute; the alarm unit only determines that an alarm is needed when it determines that more than 1 / 3 of the data within one minute has a residual that exceeds the abnormal alarm judgment threshold. In addition, users can modify the abnormal alarm judgment threshold based on experience, and establish fixed upper and lower value thresholds and change rate thresholds for each parameter through the experience of operating personnel, so as to realize automatic monitoring of operating parameters, change from people looking for problems to problems looking for people, assist operators in discovering unit abnormalities, and improve work efficiency and unit safety level.

[0070] The abnormal feature recognition model is based on the historical time series data of the unit, and the corresponding abnormal samples of the historical operation data of the unit are established through data analysis technology. The corresponding evaluation model is established based on the abnormal samples of the historical operation data of the unit. For different detection objects, corresponding abnormal feature recognition models are constructed respectively; when constructing the abnormal feature recognition model, the most relevant parameter combination with the detection object is sorted out as the training data of the abnormal feature recognition model.

[0071] Depending on the detection object, the corresponding abnormal feature settings are also different. In this embodiment, the abnormal feature recognition subunit extracts the abnormal features of the detection object from the historical operation data of the unit in the data warehouse as the training data of the abnormal feature recognition model. The abnormal features of the detection object are the most relevant features when the detection object is abnormal, such as a value, a range of values ​​or a rate of change;

[0072] During normal operation, the abnormal feature recognition model is called, and the corresponding parameter combination in the real-time operation data of the unit is input into the abnormal feature recognition model. The abnormal feature recognition model determines whether there is an abnormality based on the degree of abnormal feature matching. When the abnormal feature recognition subunit determines that the characteristics of the real-time operation data of the unit are consistent with the abnormal characteristics in the abnormal feature recognition model, it determines that there is an abnormality.

[0073] The abnormality recognition model is constructed based on knowledge and experience; for different detection objects, corresponding abnormality feature recognition models are constructed respectively. In this embodiment, based on the unit operation experience, the experience of state abnormality judgment is converted into a computer operation configuration diagram to construct the abnormality recognition model, including the following steps:

[0074] (1) For different detection objects, the abnormal event tree diagram is drawn using the VISIO diagram based on the operation experience. The abnormal event tree diagram includes multiple branches at different levels, and each branch corresponds to an abnormal type;

[0075] (2) Using knowledge modeling tools, develop a knowledge-driven anomaly detection configuration diagram based on the abnormal event tree diagram verified by unit operation experience;

[0076] (3) Construct the corresponding anomaly recognition model based on the anomaly detection configuration diagram;

[0077] During normal operation, the knowledge-driven anomaly detection configuration diagram is deployed to the data warehouse, and then the real-time operation data of the unit is input into the corresponding anomaly recognition model for processing. The first anomaly monitoring submodule judges the data features of the corresponding parameters in the real-time operation data of the unit based on the anomaly recognition rules described in the abnormal event tree diagram, and inputs the judgment results into the comprehensive alarm submodule for processing.

[0078] In this embodiment, the comprehensive alarm submodule includes a weight adjustment unit; the weight adjustment unit presets different weights for different inputs to generate corresponding comprehensive calculation results;

[0079] In the weight adjustment unit, the weights of the inputs of the first abnormal monitoring submodule, the normal state identification subunit and the abnormal feature identification subunit are adjusted according to the actual operation of the unit. For example, in the actual operation of the unit, if it is found that the abnormal feature identification model is more suitable for the abnormal identification of a certain type of parameter, the normal state reconstruction model is second, and the abnormal identification model is even less suitable, then for this type of parameter, in the comprehensive alarm submodule, a higher weight is set for the abnormal feature identification model.

[0080] In this embodiment, the detection object is an important parameter of nuclear power.

[0081] The above models are detection models built based on the historical operation data of the unit, and the accuracy of the model is constrained by the data. Since the historical operation data of the unit fails to cover all normal or abnormal conditions, the abnormality detection model developed based on the historical operation data of the unit will have false alarms or missed alarms when encountering new normal or abnormal conditions. For example, as the performance of equipment such as fans and pumps decreases, or the working efficiency of components such as heat exchangers decreases, differences will appear between the features in real-time operation data and historical operation data. For example, if the electric power decreases, the trend of the change in the heater drain flow is different from that in the historical operation data. If the previous model is used for detection, the false alarm rate will increase, so an online update process is added; historical operation data is used for verification, and adjustments are made based on new data.

[0082] In this embodiment, the nuclear power important parameter abnormality identification and alarm system also includes an online update module; the online update module uses updated unit historical operation data based on user feedback to quickly implement the update of the data-driven normal state model and abnormal feature identification model.

[0083] In this embodiment, the online update module includes a manual discrimination unit, a normal state reconstruction model online update unit, and an abnormal feature recognition model offline update unit;

[0084] The manual judgment unit judges the abnormal alarm status issued by the comprehensive alarm submodule;

[0085] If it is judged as a false alarm, it means that there may be new normal data, and the new normal data does not exist in the historical operation data of the unit, then it will be transferred to the online update unit of the normal state reconstruction model for processing;

[0086] The normal state reconstruction model online updating unit adds new normal data to the training data of the normal state reconstruction model to obtain updated training data; subsequently, the normal state reconstruction model is retrained based on the updated training data to obtain an updated normal state reconstruction model;

[0087] And based on retraining or incremental learning, an updated normal state reconstruction model is obtained;

[0088] If it is judged as a missed report, it means that there are features that should be identified by the existing abnormal feature recognition model, but are not actually identified, and then it is transferred to the abnormal feature recognition model offline update unit for processing;

[0089] The abnormal feature recognition model offline update unit extracts the missed features and establishes an abnormal feature library, and subsequently retrains the abnormal feature recognition model based on the abnormal data in the abnormal feature library to obtain an updated abnormal feature recognition model.

[0090] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A nuclear power important parameter abnormality identification and alarm system, characterized in that: It includes a data warehouse and at least one group of parameter anomaly identification modules arranged in the data warehouse; each parameter anomaly identification module receives the real-time operation data of the unit as input to perform anomaly identification and alarm of the specified nuclear power parameters.

2. The nuclear power important parameter abnormality identification and alarm system according to claim 1 is characterized in that: Each parameter anomaly recognition module includes a first anomaly monitoring submodule based on knowledge drive, a second anomaly monitoring submodule based on data drive, and a comprehensive alarm submodule; The first abnormality monitoring submodule builds an abnormality recognition model based on the abnormality recognition rules described in the preset abnormal event tree diagram; when the first abnormality monitoring submodule determines that the data characteristics of the corresponding parameters in the received real-time operation data of the unit meet the description in the abnormal event tree diagram, it determines that an abnormality exists; The second abnormality monitoring submodule includes a normal state recognition subunit and an abnormal feature recognition subunit; The normal state identification subunit uses the normal data in the historical operation data of the unit as training data and builds a normal state reconstruction model based on artificial intelligence technology; the normal state reconstruction model predicts the value range of the corresponding parameters based on the real-time operation data of the unit input online; The normal state identification subunit determines that an abnormality exists when it determines that the value range predicted by the normal state reconstruction model based on the corresponding parameter in the received real-time operation data of the unit exceeds the normal range; The abnormal feature recognition subunit uses the abnormal data in the historical operation data of the unit as training data and builds an abnormal feature recognition model based on artificial intelligence technology; When the abnormal feature recognition subunit determines that the received real-time operation data characteristics of the unit are consistent with the abnormal features in the abnormal feature recognition model, it determines that an abnormality exists; The comprehensive alarm submodule calculates the inputs of the first abnormal monitoring submodule, the normal state identification subunit and the abnormal feature identification subunit based on the preset weights, generates a corresponding comprehensive calculation result, and determines whether there is any abnormality in the current unit operation data based on the comprehensive calculation result. If there is an abnormality, an abnormal alarm is issued.

3. The nuclear power important parameter abnormality identification and alarm system according to claim 2 is characterized in that: The normal state identification subunit extracts parameter combinations describing all modes of the detection object from the historical operation data of the unit in the data warehouse as training data for the normal state reconstruction model.

4. The nuclear power important parameter abnormality identification and alarm system according to claim 2 is characterized in that: The normal state recognition subunit includes an abnormal alarm discrimination threshold unit, a normal state reconstruction model, a residual calculation unit, and an alarm unit; During normal operation, the normal state reconstruction model is called, and the corresponding parameter combination in the real-time operation data of the unit is input into the normal state reconstruction model, and the normal state reconstruction model outputs the corresponding reconstruction value; the residual calculation unit calculates N groups of residuals between the reconstruction value and the current value of each parameter in the parameter combination; When the residual exceeds the abnormal alarm discrimination threshold, the alarm unit determines that an abnormality exists.

5. The nuclear power important parameter abnormality identification and alarm system according to claim 4 is characterized in that: Each abnormal alarm discrimination threshold in the abnormal alarm discrimination threshold unit is obtained based on the normal state reconstruction model, including the following steps: After the normal state reconstruction model is constructed, the normal operation historical data of the corresponding parameter combination is obtained from the data warehouse and input into the normal state reconstruction model for prediction, so as to predict the characteristic value based on the normal operation historical data; The difference between each characteristic value and the corresponding parameter value in the parameter combination is obtained as each abnormal alarm discrimination threshold in the abnormal alarm discrimination threshold unit.

6. The nuclear power important parameter abnormality identification and alarm system according to claim 4 is characterized in that: The alarm unit presets an alarm condition. When the abnormal alarm discrimination threshold and the residual satisfy the preset alarm condition, the normal state recognition subunit determines that there is a possibility of abnormality.

7. The nuclear power important parameter abnormality identification and alarm system according to claim 6 is characterized in that: The preset alarm conditions include the following two groups: (1) At least one set of residuals exceeds the corresponding abnormal alarm discrimination threshold; (2) In a specified time window, the number of cases exceeding the abnormal alarm threshold exceeds the preset percentage; (3) Specify a time window, and the situation where the abnormal alarm threshold is exceeded continuously must meet the preset length.

8. The nuclear power important parameter abnormality identification and alarm system according to claim 2 is characterized in that: The abnormal feature recognition subunit extracts the abnormal features of the detection object from the historical operation data of the unit in the data warehouse as the training data of the abnormal feature recognition model; During normal operation, the abnormal feature recognition model is called, and the corresponding parameter combination in the real-time operation data of the unit is input into the abnormal feature recognition model. The abnormal feature recognition model determines whether there is an abnormality based on the degree of abnormal feature matching. When the abnormal feature recognition subunit determines that the characteristics of the real-time operation data of the unit are consistent with the abnormal characteristics in the abnormal feature recognition model, it determines that there is an abnormality.

9. The nuclear power important parameter abnormality identification and alarm system according to claim 2 is characterized in that: According to the unit operation experience, the experience of abnormal state judgment is converted into a computer operation configuration diagram to construct an abnormality recognition model, which includes the following steps: (1) For different detection objects, the abnormal event tree diagram is drawn using the VISIO diagram based on the operation experience. The abnormal event tree diagram includes multiple branches at different levels, and each branch corresponds to an abnormal type; (2) Using knowledge modeling tools, develop a knowledge-driven anomaly detection configuration diagram based on the abnormal event tree diagram verified by unit operation experience; (3) Construct the corresponding anomaly recognition model based on the anomaly detection configuration diagram.

10. The nuclear power important parameter abnormality identification and alarm system according to claim 2 is characterized in that: The comprehensive alarm submodule includes a weight adjustment unit; the weight adjustment unit presets different weights for different inputs to generate corresponding comprehensive calculation results; In the weight adjustment unit, the weights inputted by the first abnormality monitoring submodule, the normal state identification submodule and the abnormality feature identification submodule are adjusted according to the actual operation status of the unit.

11. The nuclear power important parameter abnormality identification and alarm system according to claim 2 is characterized in that: The nuclear power important parameter abnormality identification and alarm system also includes an online update module; the online update module uses updated unit historical operation data according to user feedback to achieve the update of the data-driven normal state model and abnormal feature identification model.

12. The nuclear power important parameter abnormality identification and alarm system according to claim 11 is characterized in that: The online update module includes a manual judgment unit, a normal state reconstruction model online update unit, and an abnormal feature recognition model offline update unit; the manual judgment unit judges the abnormal alarm state issued by the comprehensive alarm submodule; If it is judged as a false alarm, it means that there may be new normal data, and the new normal data does not exist in the historical operation data of the unit, then it will be transferred to the online update unit of the normal state reconstruction model for processing; The normal state reconstruction model online updating unit adds new normal data to the training data of the normal state reconstruction model, retrains the normal state reconstruction model based on the updated training data, and obtains an updated normal state reconstruction model; If it is judged as a missed report, it means that there are features that should be identified by the existing abnormal feature recognition model, but are not actually identified, and then it is transferred to the abnormal feature recognition model offline update unit for processing; The abnormal feature recognition model offline updating unit extracts the missed features and establishes an abnormal feature library, retrains the abnormal feature recognition model based on the abnormal data in the abnormal feature library, and obtains an updated abnormal feature recognition model.

Images