A method for using artificial intelligence to provide actions to take in response to abnormal conditions in a plant.

Abstract

The present invention relates to a method for providing measures to be taken for an abnormal state of a plant by utilizing artificial intelligence, and includes the steps of: preparing material on measures to be analyzed for the abnormal state of the plant; analyzing the material on measures to be analyzed to grasp predicted information on measures to be taken for the abnormal state; confirming measures for a current operating state based on the predicted information on measures; and notifying an operator of the measures.

Description

Technical Field

[0001] The present invention relates to a method for providing measures for abnormal states of a plant using artificial intelligence.

Background Art

[0002] Power plants (Plants) and factories (Factories) have various states / situations, and in order for power plant operators or factory managers to prepare for specific states / situations, they possess procedure manuals or response plans (manuals) for each situation.

[0003] In the prior art, the operator or factory manager has to monitor the state of the power plant / factory, and there may be a situation where such monitoring time continues, or monitoring of the state / situation that occurred during a change of workers such as shift work is overlooked, or there is no time for prediction or preparation for measures / operations in a specific state / situation and one has to rush.

[0004] If it is possible to provide guidance through measure / operation prediction in such a state / situation, the operation burden of the operator or the management burden of the manager can be reduced.

Summary of the Invention

Problems to be Solved by the Invention

[0005] Therefore, an object of the present invention is to provide a method for providing measures for abnormal states of a plant using artificial intelligence.

Means for Solving the Problems

[0006] The object of the present invention described above is achieved by providing a method for providing countermeasures for abnormal conditions in a plant using artificial intelligence, which includes the steps of: preparing data on countermeasures to be analyzed for abnormal conditions in a plant; analyzing the data on countermeasures to be analyzed to obtain predicted information on countermeasures for the abnormal conditions; confirming countermeasures for the current operating conditions based on the predicted information on countermeasures; and notifying the operator of the countermeasures.

[0007] The data on the measures to be analyzed can be obtained from at least one of the following: (1) existing operational data, and (2) hypothetical data.

[0008] The aforementioned hypothetical data can be obtained using at least one of the accident scenarios and simulations.

[0009] In the step of analyzing the data on the measures to be taken as described above, it is possible to identify the abnormal condition operating variables related to the abnormal condition.

[0010] In the step of preparing the data on the measures to be analyzed using the aforementioned existing data, the aforementioned abnormal operating conditions variables can be taken into consideration.

[0011] The step may further include verifying the procedures and manuals using the aforementioned abnormal operating variables.

[0012] The abnormal state may include at least one of the following: an abnormal state, an emergency state, an alarm state, and a specific state of the plant. [Effects of the Invention]

[0013] According to the present invention, a method is provided for providing countermeasures for abnormal conditions in a plant using artificial intelligence. [Brief explanation of the drawing]

[0014] [Figure 1]This invention illustrates a system that uses artificial intelligence according to one embodiment of the present invention to provide measures to be taken in response to abnormal conditions in a plant. [Figure 2] This is a sequence diagram illustrating a method for providing countermeasures for abnormal conditions in a plant using artificial intelligence according to one embodiment of the present invention. [Modes for carrying out the invention]

[0015] In this invention, "abnormal state" can be any one of the following: alarm state, alert state, abnormal state, emergency state, and specific state in a plant.

[0016] In this invention, "measures to be taken" includes "operational matters."

[0017] In this invention, "plant" includes power plants, and in particular may be nuclear power plants.

[0018] In the following explanation, a power plant is used as an example of a plant, and an alarm state or abnormal state is used as an example of an abnormal state, but the present invention is not limited thereto.

[0019] In this invention, artificial intelligence is used to determine the state of a plant, making it easier for operators or managers to take action in preparation for specific states / situations. Similar to the navigation system that guides you when driving a car, when you confront a specific state / situation, the next action is predicted or guided, simplifying choices, reducing decision-making time, making decisions easier, and lowering the burden on the operator.

[0020] This invention presents a method for deriving key actions / operational items based on their contribution, using artificial intelligence technology and neural network theory.

[0021] Figure 1 shows a system that uses artificial intelligence according to one embodiment of the present invention to provide countermeasures for abnormal conditions in a plant.

[0022] The system 1 according to the present invention includes an input unit 10, an analysis unit 20, and an output unit 30.

[0023] The operation of the system 1 will be described in detail with reference to FIG. 2.

[0024] FIG. 2 is a sequence diagram showing a method for providing measures against abnormal states of a plant using artificial intelligence according to an embodiment of the present invention.

[0025] First, prepare measure items data for analysis (S100).

[0026] The measure items data for analysis is obtained from (1) existing operation data and (2) virtual data.

[0027] The virtual data is obtained through procedures, manuals, scenarios, simulators, and the like.

[0028] Examples of generating virtual data are as follows.

[0029] In the case of a power plant, scenarios may include emergency, abnormal, and alarm situation occurrence scenarios. Examples of measures for specific environmental situations, such as when a specific pump fails or when a turbine stops, can be cited as examples of scenarios. Simulate the normal state and include any failure or turbine stop situation in the scenario to derive the abnormal situation as a virtual state. The reason for the need for the virtual state is to generate an operation history for emergency, abnormal, and alarm judgments, and it is dangerous to create an abnormal situation in an actual power plant for such judgments.

[0030] The virtual data may be used when there is no existing measure items data or when it is insufficient, and may not be used when the existing operation data is sufficient.

[0031] Also, the virtual data enables securing measure items data for analysis against abnormal states that are difficult to secure with existing measure items data.

[0032] The prepared data on measures to be analyzed is input into system 1 via input unit 10.

[0033] Next, we analyze the data on the measures to be analyzed and prepare information predicting the measures to be taken (S200).

[0034] The analysis of the data on measures to be analyzed is performed by the analysis unit 20. The analysis can be performed through artificial intelligence, particularly neural network learning.

[0035] The analysis is conducted via neural network learning, and the abnormal operating variables obtained from neural network learning are used again to prepare data for the measures to be taken for further analysis.

[0036] The abnormal operating conditions variables can be explained in more detail as follows:

[0037] In the event of an alarm, abnormal conditions may occur, such as pressure exceeding a specific value, temperature exceeding a specific value, or pump failure. Thus, a variable inventory is selected from among the main equipment (not all equipment) that is associated with specific equipment or devices, and these variables are used to determine abnormal / emergency / alarm conditions through neural network learning, taking into account the analysis and application methods. While it is also possible to make a judgment by looking at the overall situation of the power plant or factory, it is often more accurate to make a judgment by looking at specific variables, and a variable inventory (abnormal condition operating variables) is necessary.

[0038] Abnormal condition operating variables are used to inversely calculate the contribution of abnormal condition detection (inverse factor calculation) and to verify the procedures or manuals. The procedures or manuals improved through verification are used to generate hypothetical documents.

[0039] In other words, we use neural network theory to calculate the degree of contribution of each action to the decision-making process, and derive predictive information about the actions to be taken based on that contribution.

[0040] Abnormal driving variables can become variables (factors), and can also be values ​​that confirm specific situations. "Inverse factor calculation" can also be used to confirm factors through artificial intelligence. It is also possible to confirm other factors linked to specific factors through explainable artificial intelligence. In this way, inverse factor calculation becomes possible by utilizing explainable artificial intelligence.

[0041] Subsequently, the necessary actions for the current operating status are confirmed using the predicted action information (S300).

[0042] Examples of situations requiring corrective action include the following: A specific piece of equipment must not overheat; if it does, the appropriate action is to either shut it down or take appropriate measures to lower its temperature. In this case, the necessary part is to monitor the temperature of the specific equipment. If the temperature of the equipment is critical, neural network learning associated with its temperature must be completed, and the actions or operation of such equipment must be designed to be predictable.

[0043] In other words, it involves proposing appropriate measures for the current operating conditions.

[0044] Finally, the driver is notified of the derived measures (S400).

[0045] Notifications are made using the output unit 30 and can be made in various ways, such as by sound and on-screen display.

[0046] In other embodiments, notifications may be output to various entities, and the measures to be taken in response to abnormal conditions may also be notified.

[0047] According to the present invention, a plant can be operated or managed without a manual, and by learning the history and actions taken in specific states / situations of the plant in the past, useful actions can be predicted at a specific point in time.

[0048] When there is no historical record of a specific state / condition in the plant's past, procedures or control plans can be used, and logical judgments can be employed to provide useful actions / measures.

[0049] Therefore, according to the present invention, the burden on operators or managers and the time required for decision-making are reduced, and the selection of appropriate measures becomes easier.

[0050] The embodiments described above are illustrative examples for explaining the present invention, and the present invention is not limited thereto. Since a person with ordinary skill in the art to which the present invention pertains could implement the present invention in various ways, the scope of technical protection of the present invention should be determined by the appended claims.

Claims

[Claim 1] In a method of providing countermeasures for abnormal conditions in a plant using artificial intelligence, The steps include preparing documents containing records of measures taken to prepare for abnormal conditions in the plant, and The steps include: analyzing the data on the measures to be analyzed via the artificial intelligence to obtain information predicting measures to be taken in response to the abnormal state; The steps include: obtaining the measures to be taken for the current operating state based on the aforementioned predicted information on measures; The steps include notifying the driver of the aforementioned measures, Includes, The aforementioned data on measures to be analyzed is obtained from (1) existing data on measures recorded in the past, and (2) hypothetical data recorded in the hypothetical, The aforementioned hypothetical data was obtained by utilizing accident scenarios and simulations. The accident scenarios were derived by first simulating the normal state of the plant, and then simulating the abnormal state as a hypothetical state. In the step of analyzing the data on the measures to be analyzed, The abnormal condition operation variables, which are variables of a specific piece of equipment among the equipment constituting the plant, are obtained, and the abnormal condition is acquired. In the step of preparing the data on measures to be analyzed using the existing data on measures, Utilizing the aforementioned abnormal operating conditions variables, Using the aforementioned abnormal operating variables, the procedures and manuals used to generate the aforementioned virtual data are improved. The abnormal state is a method that includes at least one of the following: an abnormal state of the plant; an alarm state in which a specific piece of equipment among the equipment constituting the plant has failed or stopped; and a specific state in which the value of a specific piece of equipment among the equipment constituting the plant is a specific value.

Images