Equipment status notification device, equipment status notification method, and program

Abstract

To provide a state notification device, a device state notification method, and a program capable of easily grasping and notifying the state of a device, among devices constituting a facility, which operates to repeat a first status and a second status and maintain one of the statuses for a fixed time.SOLUTION: An acquisition unit 41 of a device state notification device 21 acquires the first continuance time of a first status of each of a plurality of devices 19. A first extraction unit 44 extracts a device 19 equal to or lower in standard deviation of history first continuance time than a first threshold value as a first target device. A correlation data storage unit 49 stores first correlation data indicating a relationship between the history first continuance time and the abnormality probability of the device 19. An abnormality calculation unit 52 calculates an abnormality probability corresponding to newly acquired first continuance time on the basis of the first correlation data of the first target value. An image generation unit 53 causes a display unit 54 to generate and display an image indicating an abnormality probability.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present invention relates to a device state notification apparatus, a device state notification method, and a program.

Background Art

[0002] Many types of equipment, such as various power generation equipment including hydroelectric power generation equipment, and manufacturing equipment for petrochemical products, are composed of devices. These devices operate based on the operation of these devices, and the operating conditions can be changed. Therefore, when an abnormality occurs in a device, it affects the operation of the entire equipment.

[0003] Therefore, for example, Patent Document 1 discloses an abnormality diagnosis apparatus that performs abnormality diagnosis related to a mechanical system using a pattern recognition method. This abnormality diagnosis apparatus includes a diagnosis target data acquisition unit, a data set storage unit, a learning data extraction unit, a learning information creation unit, and a diagnosis unit. The diagnosis target data acquisition unit acquires diagnosis target data, which is a data set including parameter values of a plurality of items, from a mechanical system. The data set storage unit holds a plurality of data sets having parameter values of a plurality of items corresponding to the diagnosis target data. The learning data extraction unit extracts learning data, which is a data set used for creating learning information used in a pattern recognition method, from the plurality of data sets held in the data set storage unit based on a parameter value related to an external situation included in the diagnosis target data and extraction condition information, which is information for determining an extraction condition held in advance and in which an extraction method is specified using a specific parameter value included in the diagnosis target data. The learning information creation unit creates learning information from the learning data, and the diagnosis unit determines whether the diagnosis target data is abnormal based on the learning information. The parameter value that changes depending on the external situation is a value specified by an external device or the like or a value indicating information related to the external environment. Patent Document 1 describes the temperature around the mechanical system, the output setting value of a specific device, and the like.

Prior Art Documents

Patent Documents

[0004] [Patent Document 1] Japanese Patent Publication No. 2017-102826 [Overview of the Initiative] [Problems that the invention aims to solve]

[0005] As described above, the anomaly diagnosis device in Patent Document 1 requires that parameter values ​​that change depending on external conditions be identified in advance, and furthermore, that these parameter values ​​be acquired. However, in some cases, it is sufficient to diagnose anomalies in the equipment that constitutes the system or equipment, rather than diagnosing anomalies in the entire system or equipment. In addition, some equipment performs simple operations, such as repeatedly switching on and off, and maintaining an on state as the first status or an off state as the second status for a certain period of time. For equipment that operates in this manner, identifying parameter values ​​that change depending on external conditions and acquiring those parameter values ​​for each piece of equipment is not easy.

[0006] Therefore, the present invention aims to provide an equipment status notification device, an equipment status notification method, and a program that can easily grasp and notify the status of equipment that operates to repeatedly switch between a first status and a second status, and to maintain one of the states for a certain period of time, among the equipment that constitutes the equipment. [Means for solving the problem]

[0007] The device status notification device of the present invention comprises an acquisition unit, a first extraction unit, a correlation data storage unit, an anomaly calculation unit, and an image generation unit. The acquisition unit acquires status information indicating the state of a plurality of devices that are in either a first status or a second status, and also acquires the first duration of the first status. The first extraction unit extracts devices as first target devices in which the standard deviation of the history first duration, which is the first duration acquired in the past, is less than or equal to a preset first threshold. The correlation data storage unit stores first correlation data that is generated based on the history first duration and shows the relationship between the history first duration and the anomaly probability, which is the degree of anomaly of the device. When the first duration of a device extracted as a first target device is newly acquired by the acquisition unit, the anomaly calculation unit identifies the first correlation data generated for the first target device from the correlation data storage unit, and calculates the anomaly probability corresponding to the newly acquired first duration based on the identified first correlation data. The image generation unit generates an image showing the anomaly probability calculated by the anomaly calculation unit and displays it on the display unit.

[0008] The first correlation data is generated for each phase that separates the operating status of the equipment, which is composed of multiple devices, and it is preferable that the first extraction unit extracts the first target device for each phase.

[0009] The acquisition unit acquires the second duration of the second status, and the equipment status notification device preferably further includes a second extraction unit that extracts equipment as second target equipment for which the standard deviation of the historical second duration, which is the second duration acquired in the past, is less than or equal to a preset second threshold. The correlation data storage unit stores second correlation data that is generated based on the historical second duration and shows the relationship between the historical second duration and the abnormality probability, which is the degree of abnormality of the equipment. When the second duration of equipment extracted as second target equipment is newly acquired by the acquisition unit, the abnormality calculation unit identifies the second correlation data generated for the second target equipment from the correlation data storage unit, and calculates the abnormality probability corresponding to the newly acquired second duration based on the identified second correlation data.

[0010] The second correlation data is generated for each phase that separates the operating status of at least some of the equipment composed of multiple devices, and it is preferable that the second extraction unit extracts the second target equipment for each phase.

[0011] The present invention provides a device status notification method comprising an acquisition step, a first extraction step, a correlation data storage step, an anomaly calculation step, and an image generation step. The acquisition step acquires status information indicating the state of a plurality of devices that are in either a first status or a second status, and also acquires the first duration of the first status. The first extraction step extracts devices as first target devices in which the standard deviation of the history first duration, which is the first duration acquired in the past, is less than or equal to a preset first threshold. The correlation data storage step stores first correlation data generated based on the history first duration, which shows the relationship between the history first duration and the anomaly probability, which is the degree of abnormality of the device. The anomaly calculation step identifies the first correlation data generated for the first target device when the first duration of a device extracted as a first target device is newly acquired by the acquisition step, and determines the anomaly probability corresponding to the newly acquired first duration based on the identified first correlation data. The image generation step generates an image showing the anomaly probability determined in the anomaly calculation step and displays it on the display unit.

[0012] The equipment status notification program of the present invention causes a computer to execute the above-described acquisition step, first extraction step, correlation data storage step, error calculation step, and image generation step. [Effects of the Invention]

[0013] According to the present invention, the status of a device that operates to repeatedly switch between a first status and a second status, and to maintain one of these states for a certain period of time, can be easily grasped and notified. [Brief explanation of the drawing]

[0014] [Figure 1] This is a schematic diagram of a power plant. [Figure 2]This is a functional block diagram of the equipment status notification device. [Figure 3] This is an explanatory diagram of time-series data showing the first and second durations for each device. [Figure 4] This diagram illustrates the distribution of the first duration in the history. [Figure 5] This diagram illustrates the distribution of the second duration in the history. [Figure 6] This is an explanatory diagram showing an example of the memory configuration of the history memory unit. [Figure 7] This is an explanatory diagram of the process for generating correlation data and calculating anomaly probabilities. [Figure 8] This is an explanatory diagram illustrating an example of an image displayed on the display unit. [Modes for carrying out the invention]

[0015] The power plant 10 shown in Figure 1 is one embodiment of the present invention. The power plant 10 comprises, for example, a power generation system 11 that generates and transmits power, and an equipment status notification system (hereinafter simply referred to as the "notification system") 12 connected to the power generation system 11. The power generation system 11 comprises a power generation equipment 15, which is an example of equipment, and a controller 16 that comprehensively controls each part of the power generation equipment 15. The controller 16 performs sequence control of the power generation equipment 15. That is, the controller 16 causes the power generation equipment 15 to execute a sequence, and controls the operation of the power generation equipment 15, including operation and shutdown.

[0016] In this example, the power generation facility 15 is a hydroelectric power generation facility and consists of multiple devices such as a generator 18A, a water supply device 18B, a drainage device 18C, a turbine 18D, and a power transmission device 18E. The generator 18A is connected to the turbine 18D to generate electricity, and the power transmission device 18E sends the electricity generated by the generator 18A to the outside of the power plant 10. The water supply device 18B supplies cooling water to cool various parts of the generator 18A that become hot, such as the stator windings and various bearings. The drainage device 18C collects water (from leaks, etc.) from other devices such as the generator 18A and the turbine 18D and discharges it to the outside of the power generation facility 15.

[0017] The power generation facility 15 has various devices at each part within each of the above devices and at the connection parts between devices. For example, the generator 18A has devices such as an air supply damper and an exhaust damper, the water supply device 18B has devices such as a water supply pump, the drainage device 18C has devices such as a drainage pump, and the water turbine 18D has devices such as an oil level detector for detecting the oil level height of the lubricating oil in which the runner, guide vane, and bearings of the rotating shaft are immersed, and a temperature detector for detecting the temperature of the lubricating oil. The power transmission device 18E has devices such as a parallel breaker. Also, at the connection part between the generator 18A and the water supply device 18B, there are devices such as a valve and a flow meter for detecting the flow rate of cooling water, and at the connection part between the generator 18A and the drainage device 18C, there are devices such as a valve and a contamination detector for detecting the presence or absence of water contamination. Thus, the power generation facility 15 is equipped with many devices. However, in FIG. 1, in order to avoid complication of the drawing, among the power generation facility 15 and the device groups constituting each of the above devices, only the devices 19 for the generator 18A, the water supply device 18B, and the drainage device 18C are illustrated.

[0018] The controller 16 comprehensively controls each part of the power generation facility 15 to cause the power generation facility 15 to execute a sequence, etc., and controls the operation including the operation and stop of the generator 18A, which is a part of the power generation facility 15. The controller 16 may obtain the status information indicating the state of each device 19 of the power generation facility 15 from each device 19 and perform feedback control on the power generation facility 15 based on the obtained status information.

[0019] As status information, there is status information that is binarized information obtained by binarizing the state of device 19, and status information that is not binarized information. The status information is generated in association with the corresponding device 19. The binarized information indicates either the first status or the second status of device 19 in which two states (statuses), i.e., the first status and the second status, are switched. Examples of the binarized information include information indicating on and off, information indicating an open state and a closed state of opening and closing, etc. For example, information indicating on (during operation) and off (during stop) of an air supply damper provided at an air supply port and for adjusting the air volume, information indicating an open state and a closed state of a valve that opens and closes, information indicating an open state and a closed state of an exhaust damper provided at an exhaust port and that opens and closes, etc. The status information that is not binarized information is such that the possible values are continuous, such as the opening degree of a valve, the height of the oil level of the above-described lubricating oil, the temperature of the lubricating oil, etc., and indicates a value corresponding to the state. These status information are generated for each operation in which the state of device 19 is switched, or for each detection performed by the device, such as the detection of the height of the oil level or the temperature of the lubricating oil, and are output to controller 16. For example, when the drainage pump operates so that its state is switched from off to on, an operation signal indicating that the drainage pump has been switched to on is generated as status information and output to controller 16 Also, when the oil level detector detects the height of the oil level of the lubricating oil, height information indicating the detected oil level height is generated as status information and output to controller 16.

[0020] When controller 16 acquires the status information, it identifies the status information that is binarized information from the acquired status information, and outputs the identified status information to the device status notification device (hereinafter referred to as the notification device) 21 of notification system

[0021] The controller 16 further outputs instruction information to the generator 18A during sequence control of the power generation equipment 15, instructing it to switch from the operating phase to the stop phase and from the stop phase to the operating phase. It is preferable that the controller 16 outputs phase information to the notification device 21, indicating that the generator 18A has switched phases from the operating phase to the stop phase, or from the stop phase to the operating phase, along with this instruction information to the generator 18A, and this is done in this example as well. The phases are set by dividing the operating status of the generator 18A. In this example, the operating status is divided into two phases, the operating phase and the stop phase, but the number of phases may be three or more. For example, the operating phase may be divided into an operating stable phase, in which power generation in the generator 18A is in a stable state, and an operating unstable phase, in which power generation is being performed but the generator is in an unstable state at the beginning (initial power generation) or at the end (end of power generation) when it switches to the stop phase, thus dividing the operating status into three phases. The controller 16 may also output to the notification device 21 certain status information, which is binarized information, from the status information acquired from the power generation equipment 15, treating it as phase information.

[0022] In this example, the status information, which is binarized information, is output from the power generation equipment 15 to the notification device 21 via the controller 16. However, it may also be output directly from each device of the power generation equipment 15 to the notification device 21 without going through the controller 16.

[0023] In this example, phase information is output from the controller 16 to the notification device 21, but the configuration is not limited to this. For example, when the controller 16 acquires status information, it may output the most recent past phase information to the notification device 21 in association with the status information.

[0024] In this example, the power plant 10 is equipped with a notification system 12 consisting of a notification device 21 and a plurality of terminals 22a to 22c, although the plurality of terminals 22a to 22c may be part of the notification device 21. That is, the power plant 10 may consist of a power generation system 11 and a notification device having terminals 22a to 22c. In the following description, when terminals 22a to 22c are not distinguished, they will be referred to as terminal 22.

[0025] The notification device 21 is for notifying the status of the equipment 19 of the power generation equipment 15 based on status information input from the power generation system 11. It is more preferable for the notification device 21 to notify the status of the equipment 19 based on both status information and phase information, and this is done in this example as well.

[0026] In this example, the notification device 21 is configured to communicate with each of the multiple terminals 22a to 22c, and the notification device 21 transmits the probability of an anomaly to the terminals 22, which then notify the terminals 22. Transmission from the notification device 21 to the terminals 22 occurs, for example, when the notification device 21 receives a transmission instruction from the terminals 22 to the notification device 21, and it does so in response to that acquisition. The terminals 22 in this example are, for example, personal computers (hereinafter referred to as PCs) equipped with a display unit such as a liquid crystal display (not shown), and examples of PCs include desktop PCs and mobile PCs (including tablet terminals, smartphones, etc.). Note that the number of terminals 22 equipped in the notification system 12 is not limited to the three in this example.

[0027] The notification device 21 notifies the abnormality probability, which is the degree of abnormality of a device 19 that is in either a first status or a second status, and whose duration in the first status is constant, that is, the probability that an abnormality is occurring in that device 19. For example, if there are multiple devices 19 that alternate between the first status and the second status, the notification device 21 identifies and extracts a device 19 from among these devices 19 whose duration in the first status is constant, calculates the abnormality probability of the extracted device 19, and notifies it. It is also preferable for the notification device 21 to further notify the abnormality probability of a device 19 whose duration in the second status is constant, and this is done in this example as well.

[0028] In Figure 2, the notification device 21 includes an input unit 40, an acquisition unit 41, a history storage unit 43, a first extraction unit 44 and a second extraction unit 45, a correlation data generation unit 48, a correlation data storage unit 49, an error calculation unit 52, an image generation unit 53, and a display unit 54. The first extraction unit 44 and the second extraction unit 45 constitute an extraction module 57, but the extraction module 57 is not necessarily required.

[0029] The input unit 40 is for inputting execution instructions to cause the correlation data generation unit 48, the first extraction unit 44 and the second extraction unit 45 of the extraction module 57 to perform predetermined processing. The input unit 40 is, for example, an input device such as a keyboard or mouse, and in this example it is part of the notification device 21, but it may also be connected to the notification device 21 as an external device of the notification device 21. Note that the processing conditions and timing of the processing of the first extraction unit 44, the second extraction unit 45 and the correlation data generation unit 48 may be set in the program, in which case the input unit 40 may not be necessary.

[0030] The acquisition unit 41 is connected to the controller 16 and acquires phase information and status information for each of the multiple devices 19 from the controller 16. As mentioned above, the controller 16 inputs status information for each of the multiple devices 19 that alternate between a first status and a second status, each time the status changes. Therefore, the acquisition unit 41 repeatedly acquires status information indicating the first status and status information indicating the second status for each device 19.

[0031] The acquisition unit 41 has a timer 41a that counts the duration of the status indicated in the status information. In response to the acquisition of status information indicating a first status from the controller 16, the acquisition unit 41 turns on the timer 41a. When the acquisition unit 41 acquires status information indicating a second status for the same device 19 from the controller 16, in response to this acquisition, it acquires the timer value counted by the timer 41a as the duration of the first status (hereinafter referred to as the first duration). In this way, when the acquisition unit 41 acquires status information indicating a second status after status information indicating a first status, it acquires the first duration for the device 19 indicated in the status information. The acquisition unit 41 stores the acquired first duration in the history storage unit 43 as a history first duration, which is a past first duration, associated with the device 19. In this example, the most recent past phase information is further associated and stored in the history storage unit 43.

[0032] In this example, the notification device 21 further notifies the probability of an abnormality for equipment 19 whose second status duration (hereinafter referred to as the second duration) is constant. To this end, the acquisition unit 41, even when it acquires status information indicating the second status, acquires the second duration by counting the timer 41a and stores the second duration in the history storage unit 43 as a historical second duration, which is a past second duration, associated with the equipment and phase information. Therefore, if the probability of an abnormality for equipment 19 with a constant second duration is not to be notified, it is not necessary to acquire the second duration, and it is not necessary to store the historical second duration in the history storage unit 43. As described above, the acquisition unit 41 repeatedly acquires status information for each of the multiple equipment 19, and repeatedly acquires the historical first duration and historical second duration for each equipment 19 and stores them in the history storage unit 43.

[0033] The acquisition unit 41 outputs the acquired first duration to the anomaly calculation unit 52, associating it with the device 19 that indicated this first duration. In cases where an anomaly probability is reported for a device 19 with a constant second duration, as in this example, the acquisition unit 41 further outputs the acquired second duration to the anomaly calculation unit 52.

[0034] The history storage unit 43 stores the first history duration in association with the equipment and phase information. For a single piece of equipment 19 that alternates between a first status and a second status, the first history duration is acquired multiple times by the acquisition unit 41, so the history storage unit 43 stores multiple first history durations for each piece of equipment 19. The notification device 21 in this example also notifies the abnormality probability of equipment 19 where the duration of the second status is constant. Therefore, the history storage unit 43 similarly stores the second history duration in association with the equipment 19 and phase information. In this way, the history storage unit 43 stores multiple first history durations and multiple second history durations for each piece of equipment 19.

[0035] The first extraction unit 44 is for extracting a first target device that notifies an abnormal probability associated with a first duration. The first extraction unit 44 receives an instruction to execute a process based on an input operation at the input unit 40, and in response to this input, the first extraction unit 44 reads out multiple history first durations for each device 19 stored in the history storage unit 43 and calculates their standard deviation.

[0036] In this example, when generating the first correlation data described later for each phase, the first extraction unit 44 calculates the standard deviation for each phase in a single device 19. That is, it reads out multiple history first durations associated with the same device 19 and phase, and calculates the standard deviation. In this way, in this example, the standard deviations for the operating phase and the stopping phase are calculated for each device 19. If the first correlation data described later is not generated for each phase, it is not necessary to calculate the standard deviation of the operating phase and the standard deviation of the stopping phase separately.

[0037] The first extraction unit 44 compares the calculated standard deviation with a preset first threshold to determine whether the standard deviation is less than or equal to the first threshold. If it is less than or equal to the first threshold, it identifies the device 19 as a device 19 with a constant first duration and extracts it as the first target device. In this way, the first extraction unit 44 extracts devices 19 whose standard deviation of the first history duration is less than or equal to the first threshold as the first target devices. The first extraction unit 44 associates information indicating that the extracted device 19 is the first target device with the history storage unit 43 and stores it there. In cases where the standard deviation is calculated for each phase, as in this example, the phase is also associated with the device and stored in the history storage unit 43. The first threshold is not particularly limited as long as it is set according to the balance between the desired extraction accuracy and tolerance, etc., and in this example it is set to 500.

[0038] The second extraction unit 45 is for extracting a second target device that notifies an abnormal probability related to the second duration. Similar to the first extraction unit 44, the second extraction unit 45 receives an execution instruction for processing from the input unit 40, and in response to this input, the second extraction unit 45 reads multiple history second durations for each device 19 from the history storage unit 43 and calculates their standard deviation.

[0039] In this example, when generating the second correlation data described later for each phase, the second extraction unit 45 calculates the standard deviation for each phase in a single device 19, similar to the first extraction unit 44. That is, it reads out multiple history second durations associated with the same device 19 and phase, and calculates their standard deviations. In this way, in this example, the standard deviations for the operating phase and the stopping phase are calculated for each device 19. If the second correlation data described later is not generated for each phase, it is not necessary to calculate the standard deviations for the operating phase and the stopping phase separately.

[0040] The second extraction unit 45 compares the calculated standard deviation with a preset second threshold to determine whether the standard deviation is less than or equal to the second threshold. If it is less than or equal to the second threshold, the unit identifies the device 19 as a device 19 with a constant second duration and extracts it as a second target device. In this way, the second extraction unit 45 extracts devices 19 whose standard deviation of the history second duration is less than or equal to the second threshold as second target devices. The second extraction unit 45 associates information indicating that the extracted device 19 is a second target device with the history storage unit 43 and stores it there. In cases where the standard deviation is calculated for each phase, as in this example, the phase is also associated with the device and stored in the history storage unit 43. The second threshold is not particularly limited as long as it is set according to the balance between the desired extraction accuracy and tolerance, etc., and in this example it is set to 500. The first threshold and the second threshold may be the same as in this example, but they may also be different.

[0041] The correlation data generation unit 48 is for generating first correlation data that shows the relationship between the first duration of history of the device 19 extracted as the first target device and the abnormality probability, which is the degree of abnormality of the device 19. When the correlation data generation unit 48 receives an execution instruction for processing from the input unit 40, it responds to this input by reading the first duration of history of the device 19 extracted as the first target device from the history storage unit 43 and generating first correlation data that shows the relationship with the abnormality probability, which is the degree of abnormality of the device 19. The first correlation data is generated for each of the multiple devices 19.

[0042] In this example, the probability of anomaly in equipment 19 is determined for each phase, so first correlation data is generated for each phase. Therefore, when the correlation data generation unit 48 reads the first history duration from the history storage unit 43, it reads multiple first history durations that are associated with the same phase and generates first correlation data. Thus, in this example, the correlation data generation unit 48 generates first correlation data for each of the operating and stopping phases for a single piece of equipment 19. If the probability of anomaly is determined without considering the phases, it is not necessary to generate first correlation data for each phase.

[0043] In this example, the anomaly probability of a device 19 with a constant second duration is also reported. Therefore, the correlation data generation unit 48 is also used to generate second correlation data showing the relationship between the history second duration of a device 19 extracted as a second target device and the anomaly probability, which is the degree of anomaly of the device 19. When the correlation data generation unit 48 receives an execution instruction for processing from the input unit 40, it responds to this input by reading the history second duration of a device 19 extracted as a second target device from the history storage unit 43 and generating second correlation data showing the relationship with the anomaly probability, which is the degree of anomaly of the device 19. The second correlation data is generated for each device 19 of the multiple devices 19. In this way, the correlation data generation unit 48 generates first correlation data based on multiple history first durations of a device 19 extracted as a first target device, and second correlation data based on multiple history second durations of a device 19 extracted as a second target device, for each device 19. The correlation data generation unit 48 associates the generated first correlation data with the corresponding device 19, and similarly associates the generated second correlation data with the corresponding device 19, and stores them in the correlation data storage unit 49. Details of the processing of the correlation data generation unit 48 will be described later using another diagram.

[0044] The correlation data storage unit 49 stores the first target device 19, its phase, and the first correlation data of the device 19 in association with each other. In this example, the correlation data storage unit 49 further stores the second target device 19, its phase, and the second correlation data of the device 19 in association with each other. If the probability of anomalies is not calculated for each phase, association with the phase is not necessary.

[0045] The anomaly calculation unit 52 receives information from the acquisition unit 41, including a first duration, the device 19 that represents this first duration, and the phase. In response to this input, the anomaly calculation unit 52 reads the correlation data storage unit 49 and identifies the first correlation data associated with the device 19 and the phase. If identified, it calculates the anomaly probability corresponding to the first duration input from the acquisition unit 41 based on the first correlation data and outputs it to the image generation unit 53. If not identified, it terminates processing for the device 19 and the first duration.

[0046] In this example, the anomaly calculation unit 52 receives the second duration and the equipment 19 and phase associated with this second duration as information from the acquisition unit 41. In response to this input, the anomaly calculation unit 52 reads the correlation data storage unit 49 and identifies the second correlation data associated with the equipment 19 and the phase. If identified, it calculates the anomaly probability corresponding to the second duration input from the acquisition unit 41 based on the second correlation data and outputs it to the image generation unit 53. If not identified, it terminates processing for the equipment 19 and the second duration.

[0047] The error calculation unit 52 may perform processing in response to input from the acquisition unit 41 as described above, but is not limited to this configuration. For example, it may perform processing in response to an input from the input unit 40 indicating the start of processing, input the start timing of processing through an input operation at the input unit 40 and perform processing based on that start timing, or perform processing at the start timing set in the program. Details of the processing of the error calculation unit 52 will be described later using separate diagrams.

[0048] The image generation unit 53 generates an image representing the anomaly probability obtained by the anomaly calculation unit 52, and outputs image data showing that image to the display unit 54, thereby displaying the image on the display unit 54. The display unit 54 is a well-known display device, such as a liquid crystal display.

[0049] The extraction process performed by the first extraction unit 44 and the second extraction unit 45 will be explained with reference to Figures 3 to 5. In the example shown in Figure 3, the horizontal axis is time, the first status is "on" and "open", and the second status is "off" and "closed", and the time-series data for each of the devices 19 (see Figures 1 and 2), from devices A to E, are for the operating phase. For device A, the duration of the "on" state as multiple acquired history first durations ta is roughly equal to each other, and the same is true for devices B and C. However, for devices D and E, there is a large variation in the multiple acquired history first durations ta. Regarding the second duration tb, for device C, the multiple acquired second durations tb are roughly equal to each other, and the same is true for device E. However, for devices A, B, and D, the multiple acquired second durations tb differ significantly from each other. Thus, the history first duration ta and history second duration tb, which represent the time-series data, differ for each device 19.

[0050] For device A, where the ON durations are roughly equal and the OFF durations vary greatly, the distributions of the first duration ta and second duration tb are as follows. That is, the historical first duration ta shows a narrow distribution as shown in Figure 4, while the historical second duration tb shows a wide distribution as shown in Figure 5. The first extraction unit 44 (see Figure 2) calculates the standard deviation of the multiple historical first durations ta shown in Figure 4. Here, it is assumed that the standard deviation is less than or equal to the first threshold. In this case, the first extraction unit 44 extracts device A as the first target device in the operation phase. The second extraction unit 45 (see Figure 2) also calculates the standard deviation of the multiple historical second durations tb shown in Figure 5. Here, it is assumed that this standard deviation is greater than the second threshold. In this case, the second extraction unit 45 does not extract device A as the second target device in the operation phase. The first extraction unit 44 and the second extraction unit 45 perform the same extraction process in the operation phase for devices B to E.

[0051] When the extraction process is performed in this manner, the history storage unit 43 (see Figure 2) stores the equipment 19 (see Figures 1 and 2), the phase, the first history duration ta, the second history duration tb, and whether it is a first target equipment or a second target equipment, as shown in Figure 6. For example, in the operation phase, equipment A shows the first history duration ta1, ta2, ta3, ... and is indicated as a "○" in Figure 6 as being a first target equipment. In the operation phase, equipment A shows the second history duration tb1, tb2, tb3, ... and since there is no "○" in Figure 6 indicating that it is a second target equipment, it is not extracted as a second target equipment and is not applicable.

[0052] The processing of the correlation data generation unit 48 (see Figure 2) will be explained with reference to Figure 7. Since the generation processes for the first correlation data and the second correlation data are the same, the generation process for the first correlation data will be explained here, and the explanation for the generation process for the second correlation data will be omitted. The correlation data generation unit 48 identifies the equipment 19 (see Figures 1 and 2) extracted as the first target equipment and the first history duration for a specific phase of that equipment 19, and calculates a reference value based on these first history durations. In this way, the correlation data generation unit 48 calculates a reference value for each of the equipment 19 based on the first history duration stored in the history storage unit 43 for a specific phase. The calculation method for the reference value is not limited, as long as it is a value statistically derived from the first history duration, for example. The reference value is a value associated with the abnormality probability, which indicates the probability that the state of the equipment 19 is abnormal.

[0053] The correlation data generation unit 48 is preferably a learning calculation unit that learns the acquired history first duration and calculates a reference value. In this example, the learning calculation unit is also used in the correlation data generation unit 48. In this example, the correlation data generation unit 48 creates a distribution of history first durations using a non-parametric method from the history first durations from which outliers have been removed, and calculates a reference value from the statistical information of the data. It is not necessary to remove outliers, but it is preferable to remove them because it is possible to report the probability of an abnormality occurring in the device 19 with greater accuracy. In removing outliers, the One-Class SVM is used twice to remove outliers. It is preferable to remove excessively large outliers of the history first duration with the first application of the One-Class SVM, and remove small, hard-to-distinguish outliers with the second application of the One-Class SVM.

[0054] The first history duration ta, with outliers removed, is used as training data, and the distribution of the first history duration ta is determined from the first history duration ta using kernel density estimation, a nonparametric density estimation method (see Figure 7(A)). This obtained distribution is the distribution of the first history duration under normal operation conditions. From this distribution, elapsed time α1, which is likely to be frequently observed in normal first history durations, and elapsed time α2, which deviates slightly from α1, are determined. For example, α1 is set as the first history duration to which 65% of the distribution of first history durations belong, and α2 is preferably obtained by adding a value obtained by multiplying the standard deviation s of the distribution of first history durations by a predetermined coefficient to α1. In this example, it is obtained by adding 120 × (standard deviation s of the distribution of first history durations) to α1. α1 can be any duration to which a predetermined proportion of the distribution of first history durations belongs, and is not limited to the 65% mentioned above as in this example, but setting it to at least 50% is more preferable in order to ensure the validity of the obtained anomaly probability.

[0055] In this example, from the durations α1 and α2 obtained from the distribution of the first historical duration described above, and the standard deviation s of the distribution, a first reference value VA (see Figures 7(A) and (B)) and a second reference value VB (see Figures 7(A) and (B)), which is associated with an anomaly probability higher than the first reference value VA, are calculated using the following equations (1) and (2) as reference values ​​for generating an anomaly probability model. Then, the anomaly probability y1 taken for the first reference value VA is set to, for example, 2%, and the anomaly probability y2 taken for the second reference value VB is set to, for example, 80%. The anomaly probabilities y1 taken for the first reference value VA and y2 taken for the second reference value are not limited to this example. The anomaly probabilities taken for each reference value may be set appropriately based on the performance of the probability model to be generated. For example, if you want to calculate probabilities with a margin of error around the reference value, you should set the anomaly probability lower, and if you want to calculate probabilities strictly around the reference value, you should set the anomaly probability higher. VA = α1···(1) VB = {(α2 - α1) / s} + α1...(2)

[0056] The correlation data storage unit 49 associates the reference value, the device 19, the phase, the first target device, and the second target device, and stores them as reference value information.

[0057] The anomaly calculation unit 52 identifies the same equipment and phase as the equipment 19 and phase input from the acquisition unit 41 along with the first duration from the correlation data storage unit 49, determines whether it is associated with that equipment 19 and phase and whether it corresponds to the first target equipment, and if so, identifies its reference value. The anomaly calculation unit 52 calculates the degree of difference between the acquired first duration and the reference value identified from the correlation data storage unit 49 and outputs it to the image generation unit 53.

[0058] The degree of difference is determined using the following method. First, constants A and B are calculated by associating the first reference value VA and the anomaly probability y1 taken with the first reference value VA, and the second reference value VB and the anomaly probability y2 taken with the second reference value VB, respectively, with the sigmoid function in equation (3) below. Then, using the calculated constants A and B, the anomaly probability y (where 0.0 ≤ y ≤ 1.0) is calculated from the anomaly value (operation timing time) x for the anomaly probability model PM generated from the sigmoid function in equation (3) below. In other words, a probability model is generated by determining the constants A and B. y = 1 / (1 + e Ax+B ) ····(3)

[0059] The obtained probability model is graphed (see Figure 7(B)), and the first duration Ta of the acquired device 19 in that phase is taken as the x-value in equation (3) above on the horizontal axis of this graph (Figure 7(B)) (see Figure 7(C)). The anomaly probability Pn of the acquired device 19 can then be represented on the vertical axis as y in equation (3) above. In Figure 7(C), the acquired first duration Ta is shown as Tn. The expressed anomaly probability Pn represents the difference from the anomaly probability at the baseline, and this anomaly probability can be used as the degree of difference. In this example, the anomaly probability is shown as a percentage (%), but it may also be expressed using values ​​other than percentages, such as setting the lower limit to 0 (zero) and the upper limit to 1.

[0060] The notification device 21 is composed of a computer, and by having the computer execute a predetermined program, the computer functions as described above. The program is a device status notification program, and causes the computer to execute an acquisition step, a first extraction step, a storage step, a correlation data storage step, an abnormality calculation step, and an image generation step. The acquisition step acquires status information indicating the state of multiple devices 19 that are in either a first status or a second status, and also acquires the first duration of the first status. The first extraction step extracts devices 19 as first target devices in which the standard deviation of the history first duration, which is the first duration acquired in the past, is less than or equal to a preset first threshold. The correlation data storage step stores first correlation data that is generated based on the history first duration and shows the relationship between the history first duration and the abnormality probability, which is the degree of abnormality of the device. The anomaly calculation step, when the first duration of the device 19 extracted as the first target device is newly acquired by the acquisition unit 41, identifies the first correlation data generated for the first target device from the correlation data storage unit, and calculates the anomaly probability corresponding to the newly acquired first duration based on the identified first correlation data. The image generation step generates an image showing the anomaly probability calculated by the anomaly calculation unit and displays it on the display unit.

[0061] The operation of the above configuration will now be explained. The power generation system 11 of the power plant 10 repeatedly starts and stops the power generation equipment 15 under the control of the controller 16. The power generation equipment 15 starts up by executing a sequence of operations by the controller 16 and generates and transmits power during the operation phase. Similarly, the controller 16 executes a sequence of operations to enter the shutdown phase, stopping power generation and transmission. The controller 16 outputs status information, including phase information and the status of the equipment 19 operating in the power generation equipment 15, to the acquisition unit 41 of the notification device 21.

[0062] When the acquisition unit 41 acquires status information for a specific piece of equipment 19 from the power generation equipment 15 via the controller 16, if the status information indicates a first status, the acquisition unit 41 acquires the first duration of the piece of equipment 19 when it subsequently acquires status information indicating a second status for that piece of equipment 19. The acquisition unit 41 then stores the first duration as the first history duration in the history storage unit 43, associating it with the piece of equipment 19 and the phase information. If the acquired status information indicates a second status, the acquisition unit 41 similarly acquires the second duration of the piece of equipment 19 when it subsequently acquires status information indicating a first status for that piece of equipment 19. The acquisition unit 41 then stores the second duration as the second history duration in the history storage unit 43, associating it with the piece of equipment 19 and the phase information. The acquisition unit 41 also outputs the acquired first duration and second duration to the error calculation unit 52.

[0063] The history storage unit 43 stores multiple first history durations and multiple second history durations for each device 19, associated with phase information.

[0064] The first extraction unit 44, in response to an execution instruction input from the input unit 40, reads multiple first durations for each device 19 from the history storage unit 43, calculates the standard deviation, and extracts the devices 19 whose standard deviation is less than or equal to the first threshold as the first target devices. The history storage unit 43 then stores the information and phase associated with the device 19 as the first target device. Similarly, the second extraction unit 45, in response to an execution instruction input from the input unit 40, reads multiple second durations for each device 19 from the history storage unit 43, calculates the standard deviation, and extracts the devices 19 whose standard deviation is less than or equal to the second threshold as the second target devices. The history storage unit 43 then stores the information and phase associated with the device 19 as the second target device. Since the first and second target devices are selected based on the first and second thresholds with an acceptable range for the standard deviation, even if a device 19 shows some discrepancy in the first and second durations, it is considered that the first and second statuses continued for a certain duration. As a result, omissions in the devices 19 from which the first and second correlation data are generated are suppressed, and the occurrence of devices 19 from which the anomaly probability cannot be calculated is prevented.

[0065] The correlation data generation unit 48, in response to the input of an execution instruction from the input unit 40, generates first correlation data for each of the devices 19 extracted as the first target devices and stores it in the correlation data storage unit 49. The first correlation data generated consists of two sets: first correlation data for the operation phase and first correlation data for the shutdown phase. Similarly, the correlation data generation unit 48, in response to the input of an execution instruction from the input unit 40, generates second correlation data for each of the devices 19 extracted as the second target devices, showing the relationship between the second history duration and the abnormality probability of the device 19, and stores it in the correlation data storage unit 49. The second correlation data generated consists of two sets: second correlation data for the operation phase and second correlation data for the shutdown phase. As described above, the first and second correlation data are generated for each phase that divides the operating status of the equipment. Therefore, even for devices 19 whose indicated state changes due to different operations depending on the phase, correlation data corresponding to the phase is generated, resulting in a higher accuracy in the abnormality probability. Furthermore, the information acquired to generate the first and second correlation data only needs to be the first and second durations, and the information acquired from the device 19 to determine each of these durations only needs to be status information indicating the timing of the switch between the first and second statuses. In this way, it is easy to understand the status of the device 19, including the work of collecting information related to notification.

[0066] When the first duration of the device 19 is acquired by the acquisition unit 41, the anomaly calculation unit 52 identifies the first correlation data generated for the device 19 in the correlation data storage unit 49, calculates the anomaly probability corresponding to the input first duration based on the identified first correlation data, and outputs it to the image generation unit 53. Similarly, when the second duration of the device 19 is acquired by the acquisition unit 41, the anomaly calculation unit 52 identifies the second correlation data generated for the device 19 in the correlation data storage unit 49, calculates the anomaly probability corresponding to the input second duration based on the identified second correlation data, and outputs it to the image generation unit 53. Since the first and second durations are time information, they are highly accurate, and therefore the first and second correlation data are also generated with high accuracy. The anomaly calculation unit 52 uses such first and second correlation data to calculate the anomaly probability, so the state of the device 19 is accurately understood.

[0067] When the image generation unit 53 receives an abnormality probability from the abnormality calculation unit 52, it generates an image representing that abnormality probability and outputs it to the display unit 54. The display unit 54 displays an image based on the input image data. This notifies the abnormality probability. For example, as shown in Figure 8, an image G1 is generated that includes the abnormality probability Pn, which is the difference in the first duration Ta during a predetermined phase of the equipment 19, in this case the operation phase, and is displayed on the display unit 54. This notifies the abnormality probability Pn during the operation phase of the equipment 19, which shows the acquired first duration Tn. In the example shown in Figure 8, the acquired first duration Tn is plotted on the graph of the abnormality probability model PM shown in Figure 7, and an image G1 displaying the calculated abnormality probability Pn together with the graph PM is displayed on the display unit 54. However, the image to be displayed is not limited to this example. For example, only the numerical value of the calculated abnormality probability Pn may be displayed as image G1, or the abnormality probability Pn may be indicated by the brightness or color of the color in addition to the numerical value, or instead. As described above, the status of the device 19, which operates by repeatedly switching between the first and second statuses and maintaining one of the states for a certain period of time, can be easily grasped and reported.

[0068] In this embodiment, a power generation facility 15 is used as the facility equipped with the device 19. However, the facility is not limited to this, and other facilities may be used as long as they are equipped with a device that can repeatedly switch between a first status and a second status. [Explanation of symbols]

[0069] 15 Power generation equipment 19 Equipment 21. Notification device 41 Acquisition Department 43 History Storage Unit 44 1st extraction part 45 Second extraction part 48 Correlation Data Generation Unit 49. Correlation Data Storage Unit 52 Abnormal operation section 53 Image Generation Unit 54 Display section

Claims

1. An acquisition unit acquires status information indicating the state of multiple devices that are in either a first status or a second status, and also acquires a first duration of the first status. A first extraction unit extracts as a first target device devices devices in which the standard deviation of the historical first duration, which is the first duration acquired in the past, is less than or equal to a preset first threshold, A correlation data storage unit that stores first correlation data generated based on the first history duration, which shows the relationship between the first history duration and the abnormality probability, which is the degree of abnormality of the equipment, When the first duration of a device extracted as the first target device is newly acquired by the acquisition unit, the abnormality calculation unit identifies the first correlation data generated for the first target device from the correlation data storage unit, and calculates the abnormality probability corresponding to the newly acquired first duration based on the identified first correlation data, An image generation unit generates an image showing the abnormal probability determined by the abnormal calculation unit and displays it on the display unit. A device that provides notification of the status of equipment.

2. The first correlation data is generated for each phase that separates the operating status of the equipment composed of the multiple devices, The device status notification device according to claim 1, wherein the first extraction unit extracts the first target device for each phase.

3. The acquisition unit acquires the second duration of the second status, The system further includes a second extraction unit that extracts devices as second target devices whose standard deviation of the historical second duration, which is the second duration acquired in the past, is less than or equal to a preset second threshold, The correlation data storage unit stores second correlation data that is generated based on the second history duration and shows the relationship between the second history duration and the abnormality probability, which is the degree of abnormality of the equipment. The device status notification device according to claim 1 or 2, wherein the abnormality calculation unit identifies the second correlation data generated for the second target device from the correlation data storage unit when the second duration of the device extracted as the second target device is newly acquired by the acquisition unit, and determines the abnormality probability corresponding to the newly acquired second duration based on the identified second correlation data.

4. The second correlation data is generated for each phase that separates the operating status of at least a portion of the equipment composed of the multiple devices, The device status notification device according to claim 3, wherein the second extraction unit extracts the second target device for each phase.

5. A step of acquiring status information indicating the state of multiple devices that are in either a first status or a second status, and acquiring a first duration of the first status, A first extraction step in which devices whose standard deviation of the historical first duration, which is the first duration acquired in the past, is less than or equal to a predetermined first threshold are extracted as first target devices, A correlation data storage step involves storing first correlation data generated based on the first history duration, which shows the relationship between the first history duration and the abnormality probability, which is the degree of abnormality of the equipment. When the first duration of the device extracted as the first target device is newly acquired by the acquisition step, the first correlation data generated for the first target device is identified, and based on the identified first correlation data, the anomaly calculation step is to determine the anomaly probability corresponding to the newly acquired first duration, An image generation step which generates an image showing the abnormal probability obtained in the abnormal calculation step and displays it on the display unit. A method for notifying the status of equipment having the following characteristics.

6. A step of acquiring status information indicating the state of multiple devices that are in either a first status or a second status, and acquiring a first duration of the first status, A first extraction step in which devices whose standard deviation of the historical first duration, which is the first duration acquired in the past, is less than or equal to a predetermined first threshold are extracted as first target devices, A correlation data storage step involves storing first correlation data generated based on the first history duration, which shows the relationship between the first history duration and the abnormality probability, which is the degree of abnormality of the equipment. When the first duration of the device extracted as the first target device is newly acquired by the acquisition step, the first correlation data generated for the first target device is identified, and based on the identified first correlation data, the anomaly calculation step is to determine the anomaly probability corresponding to the newly acquired first duration, An image generation step which generates an image showing the abnormal probability obtained in the abnormal calculation step and displays it on the display unit. A device status notification program that causes a computer to execute a command.

Images