Device, method and program

Abstract

SOLUTION: A device includes: an acquisition unit that acquires measurement data indicating a state of an object; a supply unit that supplies the measurement data acquired by the acquisition unit to a plurality of classification models classifying the measurement data as either normal or abnormal in response to input of the measurement data, that is the plurality of classification models each of which has been trained using training data for different periods including measurement data during a period in which the state of the object is normal; and a determination unit that determines whether the state of the object is normal or abnormal based on a plurality of classification results output from the plurality of classification models.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an apparatus, a method and a program. [Background technology]

[0002] Patent Document 1 states that "...the presence or absence of an abnormality in the equipment 10 may be detected using a threshold value" (paragraph 0055). [Prior art document] [Patent documents] [Patent Document 1] JP 2020-95425 A Summary of the Invention

[0003] In a first aspect of the present invention, there is provided an apparatus comprising: an acquisition unit that acquires measurement data indicating a state of an object; a supply unit that supplies the measurement data acquired by the acquisition unit to a plurality of classification models, each of which has been trained using learning data for different periods including measurement data during a period in which the state of the object is normal, and which classifies the measurement data as either normal or abnormal in response to input of the measurement data; and a judgment unit that judges the state of the object as either normal or abnormal based on a plurality of classification results output from the plurality of classification models.

[0004] The above device may further include a learning processing unit that generates, for each period during which the condition of the target is normal, a new classification model to be included in the multiple classification models by a learning process using learning data including the measurement data for that period acquired by the acquisition unit.

[0005] Any of the above devices may further include a selection unit that selects from two or more classification models that have been trained using learning data for different periods including measurement data during a period when the condition of the target is normal, the classification models classifying the measurement data as either normal or abnormal in response to input of the measurement data.

[0006] In the above device including a selection unit, the selection unit may select at least one classification model that has been trained using training data including measurement data from a most recent period as at least one of the plurality of classification models.

[0007] In any of the above devices including a selection unit, the selection unit may select at least one classification model designated by an operator from among the two or more classification models as at least one of the plurality of classification models.

[0008] In any of the above devices, the determination section may determine the state of the object as either normal or abnormal by taking a logical product of the plurality of classification results.

[0009] In any of the above devices, the determination section may determine the state of the object as either normal or abnormal by taking a majority vote of the plurality of classification results.

[0010] In the above-described device, the determination unit may assign a larger weight to a classification result of a classification model among the plurality of classification models that has been trained using training data including measurement data from a newer period, and determine the condition of the target as either normal or abnormal by taking a weighted majority vote of the plurality of classification results.

[0011] Any of the above devices may further include a determination unit that determines the severity of the abnormal state of the object based on the number of times that the state of the object is consecutively determined to be abnormal.

[0012] The above device may further include a setting unit that sets, as measurement data to be included in new learning data, either the measurement data for each period acquired by the acquisition unit and supplied from the supply unit to the multiple classification models, for a period in which the severity determined by the determination unit is lower than a standard severity, or the measurement data acquired by the acquisition unit and supplied from the supply unit to the multiple classification models, excluding each piece of measurement data that caused consecutive abnormality determinations until the severity determined by the determination unit became the standard severity, and each piece of measurement data that caused consecutive abnormality determinations following the measurement data.

[0013] In a second aspect of the present invention, there is provided a method comprising: an acquisition step of acquiring measurement data indicating a state of an object; a supply step of supplying the measurement data acquired in the acquisition step to a plurality of classification models, each of which has been trained using learning data for different periods including measurement data during a period in which the state of the object is normal, and which classifies the measurement data as either normal or abnormal in response to input of the measurement data; and a determination step of determining the state of the object as either normal or abnormal based on a plurality of classification results output from the plurality of classification models.

[0014] In a third aspect of the present invention, there is provided a program that causes a computer to function as an acquisition unit that acquires measurement data indicative of a state of an object, a supply unit that supplies the measurement data acquired by the acquisition unit to a plurality of classification models, each of which has been trained using learning data for different periods including measurement data during a period in which the state of the object is normal, and which classifies the measurement data as either normal or abnormal in response to input of the measurement data, and a judgment unit that judges the state of the object as either normal or abnormal based on a plurality of classification results output from the plurality of classification models.

[0015] The above summary of the invention does not list all of the necessary features of the present invention. Also, subcombinations of these features may also be inventions. [Brief description of the drawings]

[0016] [Figure 1] 1 shows a system 1 according to an embodiment. [Diagram 2] The operation of device 3 is shown. [Diagram 3] The relationship between the judgment result, the number of abnormality judgments x, and the severity is shown. [Figure 4] 22 illustrates an example computer 2200 in which aspects of the present invention may be embodied, in whole or in part. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0017] The present invention will be described below through embodiments of the invention, but the following embodiments do not limit the scope of the invention according to the claims. Furthermore, not all of the combinations of features described in the embodiments are necessarily essential to the solution of the invention.

[0018] <1. System 1> 1 shows a system 1 according to the present embodiment. The system 1 includes a facility 2 and an apparatus 3.

[0019] <1.1. Equipment 2> The equipment 2 may be an example of a target, and the state may be monitored by the device 3. The equipment 2 may be a facility or device equipped with one or more devices (not shown). For example, the equipment 2 may be a plant, or a composite device that combines multiple devices. Examples of plants include industrial plants such as chemical and bio plants, plants that manage and control wellheads and surrounding areas of gas fields and oil fields, plants that manage and control power generation such as hydroelectric power, thermal power, and nuclear power, plants that manage and control environmental power generation such as solar power and wind power, and plants that manage and control water and sewage systems, dams, etc. The devices installed in the equipment 2 are tools, machines, devices, etc., and may be so-called field devices. For example, the devices may be sensor devices such as pressure gauges, flow meters, and temperature sensors, valve devices such as flow control valves and on-off valves, or actuator devices such as fans and motors.

[0020] The facility 2 may be provided with one or more sensors 20. Each sensor 20 may measure a physical quantity indicating the state of the facility 2. The measured physical quantity may be, for example, pressure, temperature, pH, speed, flow rate, etc., or may be the yield of a product produced by the facility 2 or the ratio of impurities mixed in the product. The sensors 20 may be different from each other, or at least a part of the two or more sensors 20 may be the same type. Each sensor 20 may supply a measured value of a physical quantity, etc. to the device 3. The communication between the sensor 20 and the device 3 may be performed using, for example, a wireless communication protocol of the International Society of Automation (ISA), such as ISA100, Highway Addressable Remote Transducer (HART) (registered trademark), BRAIN (registered trademark), FOUNDATION Fieldbus, PROFIBUS, etc.

[0021] <1.2.Device 3> The device 3 includes an acquisition unit 31 , a storage unit 32 , a supply unit 33 , a determination unit 34 , a decision unit 35 , an output unit 36 ​​, a setting unit 37 , a learning processing unit 38 , and a selection unit 39 .

[0022] <1.2.1. Acquisition part 31> The acquiring unit 31 acquires measurement data indicating the state of the equipment 2. The acquiring unit 31 may acquire measurement data including one type of measurement value measured by a single sensor 20, or may acquire measurement data including multiple types of measurement values ​​measured by multiple sensors 20 at the same (or approximately the same) timing. A timestamp indicating the timing of measurement by the sensor 20 or the timing of acquisition by the acquiring unit 31 may be added to the measurement data. The acquiring unit 31 may supply the acquired measurement data to the storage unit 32 to store it in a measurement data file 321 described later. The acquiring unit 31 may supply the acquired measurement data to the supply unit 33.

[0023] <1.2.2. Storage section 32> The storage unit 32 stores various information. For example, the storage unit 32 may store a measurement data file 321 and a plurality of classification models 322.

[0024] <1.2.2(1). Measurement data file 321> The measurement data file 321 stores the measurement data supplied from the acquisition unit 31. Note that for each piece of measurement data in the measurement data file 321, the timestamp of the measurement data may indicate to which of a plurality of periods that differ for each reference time width (one week as an example in this embodiment) the measurement data belongs. For example, a timestamp of January 1, 2023 may indicate that the measurement data belongs to the first week of 2023, and a timestamp of January 8, 2023 may indicate that the measurement data belongs to the second week of 2023.

[0025] <1.2.2(2).Classification model 322> Each classification model 322 classifies the measurement data as either normal or abnormal in response to the measurement data being input. Each classification model 322 may output a classification result binarized into a value indicating normality and a value indicating abnormality. Each classification model 322 may output the classification result to the determination unit 34.

[0026] Each classification model 322 may have been trained using learning data for each different period, including measurement data for a period in which the equipment 2 is in a normal state. The measurement data for a period in which the equipment 2 is in a normal state may be measurement data including measurement values ​​measured during the period in which the equipment 2 is in a normal state. The period in which the equipment 2 is in a normal state may be a period in which the equipment 2 is determined to be in a normal state by at least one of an operator or a determination unit 34 described later. The learning data for each period may include measurement data measured at two or more timings (measurement data in chronological order, for example). In this embodiment, as an example, the learning data for each period may include measurement data within a period of the same time width (also referred to as a reference time width). The reference time width may be one week, for example. The different periods may be periods in which at least one of the start and end periods is different, and in this embodiment, as an example, may be periods that do not include a common time point.

[0027] Each classification model 322 may classify the measurement data to be classified as normal if the measurement data to be classified satisfies a condition derived from the characteristics of the measurement data included in the learning data, and may classify the measurement data as abnormal if the measurement data does not satisfy the condition. The characteristics of the measurement data included in the learning data may be values ​​indicating the distribution of each type of measurement value, and may be at least one of the maximum value, minimum value, average value, median value, mode value, standard deviation, or variance, or may be values ​​indicating the waveform of the measurement value over time. The condition derived from the characteristics of the measurement data may be, for example, that the measurement value of the measurement data to be classified falls within a range from the minimum value to the maximum value, or that the measurement value falls within a range from (average value - 3σ) to (average value + 3σ) (where σ indicates standard deviation).

[0028] Each classification model 322 may be trained by unsupervised learning using only training data including measurement data previously classified as normal measurement data, and may be generated by the same training algorithm. In the present embodiment, as an example, each classification model 322 may be generated by a one-class SVM (Support Vector Machine) training algorithm. Each classification model 322 may be trained using different training data.

[0029] At least one of the classification models 322 stored in the storage unit 32 may be generated by a learning process by the learning processing unit 38 described later. At least one of the classification models 322 stored in the storage unit 32 may be a classification model 322 used for determination by the determination unit 34 (also referred to as a classification model 322 to be used). The classification model 322 to be used may be selected by a selection unit 39, the details of which will be described later.

[0030] <1.2.3. Supply section 33> The supplying unit 33 supplies the measurement data acquired by the acquiring unit 31 to one or more classification models 322. The supplying unit 33 may supply the measurement data to each classification model 322 to be used. As a result, a classification result may be output to the determining unit 34 from each classification model 322 to which the measurement data has been supplied.

[0031] <1.2.4. Judgment unit 34> The determination unit 34 determines whether the state of the equipment 2 is normal or abnormal based on the multiple classification results output from the multiple classification models 322. The determination unit 34 may obtain the classification results from each classification model 322 to be used and make the determination.

[0032] The determination unit 34 may determine the state of the facility 2 as either normal or abnormal by taking the logical product of a plurality of classification results. In this embodiment, as an example, the determination unit 34 may take the logical product of a plurality of classification results with the classification result that the measurement data is abnormal being true (True) and the classification result that the measurement data is normal being false (False). In other words, the state of the facility 2 may be determined as abnormal in response to the measurement data being determined to be abnormal in all classification results, and the state of the facility 2 may be determined as normal in response to the measurement data being determined to be normal in at least one classification result. The determination unit 34 may supply the determination result to the decision unit 35.

[0033] <1.2.5. Decision unit 35> The decision unit 35 determines the severity of the abnormal state of the facility 2 based on the number of consecutive times the state of the facility 2 has been determined to be abnormal (also referred to as the number of consecutive abnormal determination times). Being continuously determined to be abnormal may mean that the determination results of being abnormal continue without being sandwiched by a determination result of being normal.

[0034] The decision unit 35 may determine a higher severity as the number of consecutive abnormal determination times is larger. For example, when the decision unit 35 sets the number of consecutive abnormal determination times as "x", the first reference number as "a" (where a is a natural number of 2 or more), and the second reference number as "b" (where b is a natural number with a < b), the severity may be determined to be at a low level in response to x < a, at a medium level (also referred to as an attention level) in response to a ≤ x < b, and at a high level (also referred to as a warning level) in response to b ≤ x. As an example, the first reference number a may be 5 times, and the second reference number b may be 10 times.

[0035] The decision unit 35 may determine the severity each time a determination result is supplied from the determination unit 34. The decision unit 35 may supply the determined severity and the determination result by the determination unit 34 together to the output unit 36 and the setting unit 37.

[0036] <1.2.6. Output unit 36> The output unit 36 ​​outputs the result of the determination made by the determination unit 34 and the result of the determination made by the determination unit 35. The output unit 36 ​​may cause a display device (not shown) to display the results of the determination and the determination.

[0037] <1.2.7. Setting section 37> The setting unit 37 sets the measurement data for a period in which the severity determined by the determining unit 35 is lower than the reference severity, among the measurement data for each period acquired by the acquiring unit 31 and supplied from the supplying unit 33 to the classification model 322 to be used, as the measurement data to be included in the new learning data (also referred to as measurement data for learning). Among the measurement data for each period supplied to the classification model 322 to be used, the measurement data for a period in which the severity determined by the determining unit 35 is lower than the reference severity may be the measurement data for each period in which the severity determined by the determining unit 35 is maintained lower than the reference severity. The severity being lower than the reference severity may mean that the equipment 2 is normal, or that the equipment 2 is abnormal but the severity of the abnormality is lower than the reference severity. In the present embodiment, the reference severity may be a high level, that is, a warning level, as an example.

[0038] In addition, in a period in which the severity is lower than the reference severity, even if the state of the equipment 2 is determined to be abnormal, the abnormal state occurs only temporarily and singularly, and therefore it is considered that the measurement data is classified as abnormal due to noise, drift, etc. in the measurement value even though the equipment 2 is essentially normal. The setting unit 37 may set the measurement data during such a period in which the equipment 2 is essentially normal as the measurement data for learning.

[0039] The setting unit 37 may add a label indicating that the measurement data is for learning (also referred to as a learning label) to the measurement data for each period in the measurement data file 321, the measurement data for each period having a severity lower than a reference severity. In the present embodiment, as an example, the setting unit 37 may determine whether the determined severity is lower than a high level each time the determination result of the severity is supplied from the determination unit 35, and add a learning label to the measurement data for the period in which the severity was maintained lower than the high level. In response to adding the learning label to the measurement data, the setting unit 37 may supply a notification to that effect to the learning processing unit 38.

[0040] <1.2.8. Learning processing unit 38> The learning processing unit 38 generates a new classification model 322 included in the multiple classification models 322 stored in the storage unit 32, for each period during which the state of the equipment 2 is normal, by a learning process using learning data including the measurement data for that period acquired by the acquisition unit 31. Generating a new classification model 322 for each period during which the state of the equipment 2 is normal may mean generating a new classification model 322 for each period during which the state of the equipment 2 is normal.

[0041] In response to receiving a notification from the setting unit 37 that the learning labels have been added to the measurement data, the learning processing unit 38 may generate a new classification model 322 using learning data including the measurement data of the most recent period among the measurement data to which the learning labels have been added in the measurement data file 321. Therefore, when learning labels are added to the measurement data of multiple consecutive periods, a new classification model 322 may be generated each time each period passes.

[0042] The learning processing unit 38 may generate the classification model 322 by unsupervised learning using only learning data including measurement data previously classified by adding learning labels, and in this embodiment, as an example, may generate the classification model 322 by a one-class SVM learning algorithm. In response to generating the classification model 322, the learning processing unit 38 may supply a notification indicating this to the selection unit 39.

[0043] <1.2.9. Selection section 39> The selection unit 39 selects the classification model 322 to be used from each classification model 322 stored in the storage unit 32. When two or more classification models 322 are stored in the storage unit 32, the selection unit 39 may select the classification models 322 to be used from the two or more classification models 322. When a reference number (four, for example) or more of classification models 322 are stored in the storage unit 32, the selection unit 39 may select the reference number of classification models 322 as the classification models 322 to be used.

[0044] The selection unit 39 may select at least one classification model 322 that has been trained using training data including measurement data from the most recent period as at least one of the classification models 322 to be used. The selection unit 39 may select each classification model 322 to be used in response to receiving a notification from the training processing unit 38 that a new classification model 322 has been generated. The selection unit 39 may supply identification information of the classification model 322 to be used to the supply unit 33, thereby causing the supply unit 33 to supply measurement data to the classification model 322 to be used.

[0045] According to the above-described device 3, the state of the equipment 2 is judged as normal or abnormal based on a plurality of classification results of normal or abnormal for the measurement data, which are output by supplying the measurement data to the classification model 322. Therefore, unlike the case where an index value indicating the degree of normality or abnormality of the measurement data is obtained and the index value is compared with a threshold value to make a judgment, it is possible to eliminate the trouble of the operator setting a threshold value by trial and error and to prevent the judgment result from differing depending on the size of the threshold value. In addition, since the classification model 322 trained by learning data for different periods including measurement data for a period when the equipment 2 is in a normal state is used, a classification result with high accuracy can be obtained compared to the case of using a single classification model trained by learning data including each of the measurement data for the multiple periods when the equipment 2 is in a normal state. That is, even if the equipment 2 is in a normal state, the trend of the measurement value may change, for example, when the measurement value drifts, when the state of the equipment 2 changes, or when the operation method of the equipment 2 is changed. Furthermore, even if the equipment 2 is in a normal state, the measurement value may contain various noises. In this way, when the trend of the measurement value changes or the measurement value contains various noises, if a single classification model trained by learning data including each of the measurement data for the multiple periods is used, even if the measurement data to be classified deviates from the normal range in each period, if the measurement data is included in the normal range as a whole for the multiple periods, the measurement data to be classified can be classified as normal measurement data. In contrast, if a plurality of classification models 322 trained by learning data for different periods are used, when the measurement data to be classified deviates from the normal range in each period, each classification model 322 can correctly classify the measurement data as abnormal measurement data. Then, by making a judgment using a plurality of such classification results, it is possible to obtain a highly accurate judgment result.

[0046] In addition, since the status of the equipment 2 is determined as normal or abnormal by taking the logical product of multiple classification results from multiple classification models 322, a more accurate determination result can be obtained compared to when a determination is made by taking a logical sum.

[0047] Furthermore, for each period during which the equipment 2 is in a normal state, a new classification model 322 is generated by a learning process using learning data including the measurement data for that period. Therefore, classification models 322 corresponding to changes in the trend of the measurement data can be generated one after another and used for judgment, so that highly accurate judgment results can be obtained.

[0048] Furthermore, the multiple classification models 322 to be used are selected from two or more classification models 322 stored in the storage unit 32. Therefore, an appropriate classification model 322 can be appropriately selected and used for the determination.

[0049] In addition, one classification model 322 that has been trained using learning data including measurement data from the most recent period is selected as at least one of the classification models 322 to be used, so that a classification model 322 corresponding to the most recent state of the equipment 2 can be used for judgment.

[0050] Furthermore, since the severity of the abnormal state of the equipment 2 is determined based on the number of consecutive times that the state of the equipment 2 is determined to be abnormal, the severity of the abnormal state can be obtained.

[0051] Furthermore, of the measurement data for each period supplied to the multiple classification models 322, the measurement data for a period in which the severity is lower than the reference severity is set as the measurement data to be included in the new learning data. Therefore, since the measurement data for a period in which the equipment 2 is essentially normal can be set as the measurement data for learning, by generating the classification model 322 using new learning data including such measurement data, it is possible to obtain a classification model 322 that can correctly classify essentially normal measurement data as normal.

[0052] <2. Operation of Device 3> 2 shows the operation of the device 3. The device 3 performs the processes of steps S1 to S55 to assist in monitoring the facility 2. Note that, as an example, in this embodiment, a description will be given on the assumption that the classification model 322 is not stored in the storage unit 32 at the start of this operation.

[0053] In step S1, the acquisition unit 31 acquires measurement data for a period during which the state of the equipment 2 is normal (for example, a period of a reference time width). In this step, the period during which the state of the equipment 2 is normal may be a period during which the operator has determined that the state of the equipment 2 is normal. For example, if the operator has determined that the state of the equipment 2 is normal over a period of the reference time width, the period may be the period, or if the operator has determined that the state of the equipment 2 is normal over a period longer than the reference time width, the period may be a period of the reference time width included in the period. The acquired measurement data may include measurement values ​​for a period during which the operator has determined that the state of the equipment 2 is normal. The acquired measurement data may be for learning the classification model 322, and a learning label may be added in advance. The acquisition unit 31 may acquire measurement values ​​from one or more sensors 20 to obtain measurement data. The acquisition unit 31 may perform so-called preprocessing on the acquired measurement data. As an example, the acquisition unit 31 may calculate a moving average of the measurement values ​​from the most recent multiple measurement data acquired, and use this as the measurement value of the acquired measurement data, or may remove outliers detected by statistical analysis using past measurement data from the measurement values ​​of the acquired measurement data.

[0054] In step S3, the learning processing unit 38 generates a new classification model 322 by a learning process using the learning data including the measurement data acquired in step S1. The learning processing unit 38 may store the generated new classification model 322 in the storage unit 32.

[0055] In step S5, the selection unit 39 selects the classification model 322 generated in step S3 as the classification model 322 to be used. As a result, measurement data may be supplied to the selected classification model 322 in step S15 described later.

[0056] In step S11, the supply unit 33 starts measuring the reference time width, which may start a new period.

[0057] In step S13, the acquisition unit 31 acquires measurement data indicating the state of the equipment 2. The acquisition unit 31 may acquire measurement values ​​from one or more sensors 20 as the measurement data. The acquisition unit 31 may perform preprocessing on the acquired measurement data in the same manner as in step S1 described above.

[0058] In step S15, the supply unit 33 supplies the measurement data acquired in step S13 to at least one classification model 322. As a result, each classification model 322 to which the measurement data is supplied may output a classification result in which the measurement data is classified as normal or abnormal.

[0059] The supplying unit 33 may supply the measurement data to each classification model 322 to be used that is selected by the selecting unit 39. As an example, when only a single classification model 322 is selected to be used in the above-mentioned step S5, the supplying unit 33 may supply the measurement data to the single classification model 322. When multiple classification models 322 are selected to be used in the below-described step S55, the supplying unit 33 may supply the measurement data to each selected classification model 322. When step S15 is performed for the second or subsequent times, the process of step S15 may be executed at a reference interval (for example, 30 minutes).

[0060] In step S17, the determination unit 34 obtains the classification results of the measurement data from each classification model 322 to be used.

[0061] In step S19, the determination unit 34 determines whether the state of the equipment 2 is normal or abnormal, based on each of the classification results acquired in step S17.

[0062] When only a single classification result is obtained from the single classification model 322 in step S17 described above, the determination unit 34 may determine the state of the facility 2 as normal or abnormal based on the single classification result. As an example, the determination unit 34 may determine the state of the facility 2 as normal based on the classification result that the measurement data is normal, and may determine the state of the facility 2 as abnormal based on the classification result that the measurement data is abnormal.

[0063] When a plurality of classification results are obtained from the plurality of classification models 322 in step S17 described above, the determination unit 34 may determine the state of the facility 2 as either normal or abnormal based on the plurality of classification results. The determination unit 34 may take the logical product of the plurality of classification results and determine the state of the facility 2 as either normal or abnormal. When it is determined in step S19 that the facility 2 is normal (step S19; normal), the process may proceed to step S41. When it is determined in step S19 that the facility 2 is abnormal (step S19; abnormal), the process may proceed to step S21.

[0064] In step S21, the determination unit 35 increments the number x of consecutive abnormal determination times. When step S21 is performed for the first time, the number of consecutive abnormal determination times may be incremented from the initial value of 0 to 1.

[0065] In step S23, the determination unit 35 determines whether the number x of consecutive abnormal determination times is less than the first reference number a, that is, whether x < a. When it is determined that x < a (step S23; Yes), the process may proceed to step S25. When it is determined that x is not less than a, that is, a ≤ x (step S23; No), the process may proceed to step S27.

[0066] In step S25, the determination unit 35 determines that the severity of the abnormal state of the facility 2 is at a low level. After the process of step S25 is completed, the process may proceed to step S45 described later.

[0067] In step S27, the determination unit 35 determines whether the continuous number of abnormal determination times x is equal to or greater than the first reference number of times a and less than the second reference number of times b, that is, whether a ≤ x < b. If it is determined that a ≤ x < b (step S27; Yes), the process may proceed to step S29. If it is determined that a ≤ x < b does not hold, that is, b ≤ x (step S27; No), the process may proceed to step S31.

[0068] In step S29, the determination unit 35 determines that the severity level of the abnormal state of the facility 2 is at a medium level, that is, at the caution level. After the process of step S29 is completed, the process may proceed to step S45 described later.

[0069] In step S31, the determination unit 35 determines that the severity level of the abnormal state of the facility 2 is at a high level, that is, at the warning level. After the process of step S31 is completed, the process may proceed to step S45 described later.

[0070] In step S41, the determination unit 35 resets the continuous number of abnormal determination times x to 0. In step S43, the determination unit 35 determines that the severity level of the abnormal state of the facility 2 is 0.

[0071] In step S45, the output unit 36 outputs the result of the determination by the determination unit 34 in step S19 and the result of the determination by the determination unit 35 in step S25, S29, S31, or S43. The output unit 36 may display the determination and decision results on a display device (not shown).

[0072] In step S47, the supply unit 33 determines whether the measurement of the reference time width has been completed, that is, whether the period of the reference time width (one week in this embodiment as an example) has elapsed since step S11 described above. If it is determined that the period of the reference time width has elapsed (step S47; Yes), the process may proceed to step S49. If it is determined that the period of the reference time width has not elapsed (step S47; No), the process may proceed to step S13.

[0073] In step S49, the setting unit 37 judges whether or not the seriousness has been maintained below the reference seriousness (high level in this embodiment as an example) during the most recent period. If it is judged that the seriousness has not been maintained below the high level (step S49; No), the process may proceed to step S11 described above. If it is judged that the seriousness has been maintained below the high level (step S49; Yes), the process may proceed to step S51.

[0074] In step S51, the setting unit 37 sets the measurement data of the most recent period as the measurement data to be included in the new learning data. In the present embodiment, as an example, the setting unit 37 may set the measurement data of the period during which the determination that the state of the equipment 2 is normal is maintained in step S19 as the measurement data to be included in the new learning data. In addition, the setting unit 37 may set the measurement data of the period during which the seriousness is determined to be low level, medium level, or 0 in steps S25, S29, and S43 and thus the seriousness is maintained below the high level as the measurement data to be included in the new learning data. The setting unit 37 may add a learning label to the measurement data of the most recent period stored in the measurement data file 321.

[0075] In step S53, the learning processing unit 38 performs a learning process using learning data including measurement data from the most recent period to which the learning labels have been added, to generate a new classification model 322. The learning processing unit 38 may add the generated new classification model 322 to the storage unit 32 for storage.

[0076] In step S55, the selection unit 39 selects a classification model 322 to be used from the multiple classification models 322 stored in the storage unit 32. As a result, the measurement data may be supplied to each of the selected classification models 322 to be used in step S15 described above.

[0077] When the storage unit 32 stores only the classification models 322 equal to or less than the reference number, the selection unit 39 may select each classification model 322 as the classification model 322 to be used. When the storage unit 32 stores classification models 322 greater than the reference number, the selection unit 39 may select the reference number of classification models 322 from these classification models 322 as the classification model 322 to be used. For example, the selection unit 39 may select at least one classification model 322 that is the most recent one generated in step S53 from among the classification models 322 in the storage unit 32 as the classification model 322 to be used. As a result, the measurement data may be classified by all the classification models 322 until the number of the classification models 322 in the storage unit 32 reaches the reference number, and after the number of the classification models 322 in the storage unit 32 exceeds the reference number, the measurement data may be classified by the classification models 322 of the reference number to be selected. The selection unit 39 may sequentially delete the classification models 322 that have been previously generated according to the capacity of the storage unit 32. When the process of step S55 is completed, the process may proceed to step S11 described above.

[0078] <3. Example of operation> Figure 3 shows the relationship between the result of the judgment, the number of consecutive abnormality judgments x, and the severity of the abnormality. In this figure, the horizontal axis indicates time, and the dashed lines in the vertical direction indicate the division of the period.

[0079] As shown in the figure, the number of consecutive abnormality determinations x is incremented according to the consecutive determination results that the state of the equipment 2 is abnormal, and is reset according to the determination result that the state of the equipment 2 is normal. Furthermore, when the number of consecutive abnormality determinations x is 0, the severity is 0, when the number of consecutive abnormality determinations x is 1 to 4, the severity is low, when the number of consecutive abnormality determinations x is 5 to 9, the severity is medium (i.e., caution level), and when the number of consecutive abnormality determinations x is 10 abnormalities, the severity is high (i.e., warning level).

[0080] <4. Modifications> In the above embodiment, the selection unit 39 has been described as selecting at least one classification model 322 that has been trained using training data including measurement data from the latest period as at least one of the classification models 322 to be used. In addition to this, or instead of this, at least one classification model 322 designated by the operator may be selected as at least one of the classification models 322 to be used from among the two or more classification models 322 stored in the storage unit 32. In this way, an arbitrary classification model 322 can be selected by the operator and used for the judgment. Therefore, when the equipment 2 is currently normal, the classification model 322 that has determined that the current measurement data is normal can be selected and used for the judgment. In addition, a classification model that has previously shown high classification accuracy in a test using normal measurement data and abnormal measurement data can be selected and used for the judgment. In response to the classification model 322 being designated by the operator, the selection unit 39 may fix the classification model 322 as the classification model 322 to be used, or may automatically select the classification model 322 in the processing of step S55.

[0081] In addition, the determination unit 34 has been described as determining whether the state of the equipment 2 is normal or abnormal by taking the logical product of the classification results from the multiple classification models 322 to be used. However, the determination unit 34 may determine whether the state of the equipment 2 is normal or abnormal by taking a majority vote of the multiple classification results. In addition, the determination unit 34 may apply a larger weight to the classification result of the classification model 322 that has been trained using learning data including measurement data from a newer period among the multiple classification models 322 to be used, and determine whether the state of the equipment 2 is normal or abnormal by taking a weighted majority vote of the multiple classification results. Applying a weight to the classification result and taking a weighted majority vote may mean increasing the number of each classification result by the weight and taking a majority vote. As an example, the classification result by the classification model 322 trained using the learning data from the most recent period is set as the first classification result, the classification result by the classification model 322 trained using the learning data from the second most recent period is set as the second classification result, and the classification result by the classification model 322 trained using the learning data from the third most recent period is set as the third classification result. Moreover, the weights for the first to third classification results are w1 to w3 (where w1>w2>w3), respectively. In this case, the judgment unit 34 may take a majority vote to determine the first classification result as the classification result for the number of items (1×w1), the second classification result as the classification result for the number of items (1×w2), and the third classification result as the classification result for the number of items (1×w3). As a result, the newer classification model 322, that is, the classification model 322 according to the state of the equipment 2 that is closer, has a greater influence on the judgment result, so that it is possible to obtain a judgment result according to the transition of the state of the equipment 2.

[0082] Further, the judgment unit 34 has been described as making a judgment based on a plurality of classification results output from a plurality of classification models 322 generated by a common learning algorithm, but the judgment may be made based on a classification result output from a classification model generated by another known learning algorithm, in addition to the plurality of classification results output from the plurality of classification models 322.

[0083] Also, although it has been described that the learning processing unit 38 generates the classification model 322 in response to receiving a notification from the setting unit 37 that a learning label has been added to the measurement data, the classification model 322 may be generated according to a predetermined schedule, or may be generated in response to an instruction from an operator. The learning processing unit 38 may generate the classification model 322 by using unused learning data for the learning process, among the learning data for each period in which the state of the equipment 2 was normal.

[0084] Further, the device 3 has been described as having the storage unit 32, the determination unit 35, the output unit 36, the setting unit 37, the learning processing unit 38, and the selection unit 39, but any of these may not be included. If the device 3 does not have the storage unit 32, the measurement data file 321 and the classification model 322 may be stored in a storage device externally connected to the device 3. If the device 3 does not have the determination unit 35, the output unit 36 ​​may output the determination result by the determination unit 34 without outputting the determination result by the determination unit 35, and the setting unit 37 may set the measurement data during the period in which the equipment 2 was determined to be normal by the determination unit 34 as the measurement data to be included in the new learning data. If the device 3 does not have the output unit 36, the determination unit 34 and the determination unit 35 may store the results of the determination and the judgment in association with the corresponding measurement data in the measurement data file 321. If the device 3 does not have the setting unit 37, the learning processing unit 38 may generate the classification model 322 using measurement data within a period selected by an operator, among the measurement data for each period. If the device 3 does not have the learning processing unit 38, the determination unit 34 may make a determination using only the classification models 322 stored in the storage unit 32 by default. If the device 3 does not have the selection unit 39, the determination unit 34 may make a determination using each classification model 322 in the storage unit 32.

[0085] Also, although each learning data is described as including measurement data within a period of the same reference time width, it may include measurement data within periods of different time widths. In this case, the operator may arbitrarily set the start and end of each period. Also, in the process of step S1 described above, the acquisition unit 31 may acquire measurement data for a period of any time width in which the state of the equipment 2 is normal, and as an example, may acquire measurement data for a period of any time width in which the state of the equipment 2 is determined to be normal by the operator.

[0086] In addition, it has been explained that the setting unit 37 sets the measurement data for the period during which the severity determined by the determination unit 35 was maintained lower than the standard severity (for example, a high level) as measurement data for learning, that is, excludes the period during which the severity became equal to or higher than the standard severity, and sets the measurement data for the remaining period as measurement data for learning, but the measurement data for learning may be set in other manners. For example, when the number of consecutive abnormality determinations is equal to or greater than the second reference number "b" (i.e., the reference number for determining the severity as high level), the setting unit 37 may exclude the measurement data that caused the "b"th abnormality determination or later from the measurement data acquired by the acquisition unit 31 and supplied from the supply unit 33 to the classification models 322, and set the remaining measurement data as the measurement data for learning, or may exclude each measurement data that caused consecutive abnormality determinations until the severity determined by the determination unit 35 became high level (i.e., each measurement data that caused the first abnormality determination to the "b"th abnormality determination) and each measurement data that caused consecutive abnormality determinations following the measurement data (i.e., each measurement data that caused consecutive abnormality determinations after the "b+1"th abnormality determination) from the measurement data acquired by the acquisition unit 31 and supplied from the supply unit 33 to the classification models 322, as the measurement data for learning. In these cases, the measurement data for learning may include measurement data within periods of different time widths.

[0087] Moreover, the target of monitoring by the device 3 has been described as the facility 2, but it may be a device (an actuator device, for example) provided in the facility 2.

[0088] 2, no classification model 322 is stored in the storage unit 32 at the start of the operation, but one or more classification models 322 may be stored in advance. In this case, the operation may start from step S11 in a state in which each classification model 322 in the storage unit 32 is pre-selected by the selection unit 39 as the classification model 322 to be used. In addition, in step S55, the selection unit 39 may select the classification model 322 generated by the learning processing unit 38, or may select the classification model 322 pre-stored in the storage unit 32.

[0089] 2 has been described as storing the new classification model 322 generated by the learning processing unit 38 in the storage unit 32, a new classification model 322 generated in an external device by a learning process similar to that of the learning processing unit 38 may be stored in the storage unit 32. In this case, in response to the classification models 322 being stored in the storage unit 32, the selection unit 39 may select one of the classification models 322 as the classification model 322 to be used.

[0090] Various embodiments of the present invention may also be described with reference to flow charts and block diagrams, where the blocks may represent (1) stages of a process in which operations are performed or (2) sections of an apparatus responsible for performing the operations. Particular stages and sections may be implemented by dedicated circuitry, programmable circuitry provided with computer readable instructions stored on a computer readable medium, and / or a processor provided with computer readable instructions stored on a computer readable medium. Dedicated circuitry may include digital and / or analog hardware circuitry and may include integrated circuits (ICs) and / or discrete circuits. Programmable circuitry may include reconfigurable hardware circuitry including logical AND, logical OR, logical XOR, logical NAND, logical NOR, and other logical operations, memory elements such as flip-flops, registers, field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), and the like.

[0091] A computer-readable medium may include any tangible device capable of storing instructions that are executed by a suitable device, such that the computer-readable medium having instructions stored thereon comprises an article of manufacture that includes instructions that can be executed to create means for performing the operations specified in the flowcharts or block diagrams. Examples of computer-readable media may include electronic storage media, magnetic storage media, optical storage media, electromagnetic storage media, semiconductor storage media, and the like. More specific examples of computer-readable media may include floppy disks, diskettes, hard disks, random access memories (RAMs), read-only memories (ROMs), erasable programmable read-only memories (EPROMs or flash memories), electrically erasable programmable read-only memories (EEPROMs), static random access memories (SRAMs), compact disk read-only memories (CD-ROMs), digital versatile disks (DVDs), Blu-ray (RTM) disks, memory sticks, integrated circuit cards, and the like.

[0092] The computer readable instructions may include either assembler instructions, Instruction Set Architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk®, JAVA®, C++, etc., and conventional procedural programming languages ​​such as the “C” programming language or similar programming languages.

[0093] Computer readable instructions may be provided to a processor or programmable circuitry of a general purpose computer, special purpose computer, or other programmable data processing apparatus, either locally or over a local area network (LAN), a wide area network (WAN), such as the Internet, etc., to execute the computer readable instructions to create means for performing the operations specified in the flowcharts or block diagrams. Examples of processors include computer processors, processing units, microprocessors, digital signal processors, controllers, microcontrollers, etc.

[0094] 4 illustrates an example of a computer 2200 in which aspects of the present invention may be embodied in whole or in part. Programs installed on the computer 2200 may cause the computer 2200 to function as or perform operations associated with an apparatus or one or more sections of the apparatus according to an embodiment of the present invention, and / or to perform a process or steps of the process according to an embodiment of the present invention. Such programs may be executed by the CPU 2212 to cause the computer 2200 to perform specific operations associated with some or all of the blocks of the flowcharts and block diagrams described herein.

[0095] The computer 2200 according to this embodiment includes a CPU 2212, a RAM 2214, a graphic controller 2216, and a display device 2218, which are connected to each other by a host controller 2210. The computer 2200 also includes input / output units such as a communication interface 2222, a hard disk drive 2224, a DVD-ROM drive 2226, and an IC card drive, which are connected to the host controller 2210 via an input / output controller 2220. The computer also includes legacy input / output units such as a ROM 2230 and a keyboard 2242, which are connected to the input / output controller 2220 via an input / output chip 2240.

[0096] The CPU 2212 operates according to a program stored in the ROM 2230 and the RAM 2214, thereby controlling each unit. The graphic controller 2216 retrieves image data generated by the CPU 2212 into a frame buffer or the like provided in the RAM 2214 or into the graphic controller 2216 itself, and causes the image data to be displayed on the display device 2218.

[0097] The communication interface 2222 communicates with other electronic devices via a network. The hard disk drive 2224 stores programs and data used by the CPU 2212 in the computer 2200. The DVD-ROM drive 2226 reads programs or data from the DVD-ROM 2201 and provides the programs or data to the hard disk drive 2224 via the RAM 2214. The IC card drive reads programs and data from an IC card and / or writes programs and data to an IC card.

[0098] The ROM 2230 stores therein a boot program or the like executed by the computer 2200 upon activation, and / or a program that depends on the hardware of the computer 2200. The input / output chip 2240 may also connect various input / output units to the input / output controller 2220 via a parallel port, a serial port, a keyboard port, a mouse port, and the like.

[0099] The programs are provided by a computer-readable medium such as a DVD-ROM 2201 or an IC card. The programs are read from the computer-readable medium, installed in the hard disk drive 2224, the RAM 2214, or the ROM 2230, which are also examples of computer-readable media, and executed by the CPU 2212. Information processing described in these programs is read by the computer 2200, and brings about cooperation between the programs and the various types of hardware resources described above. An apparatus or method may be constructed by realizing information manipulation or processing according to the use of the computer 2200.

[0100] For example, when communication is performed between the computer 2200 and an external device, the CPU 2212 may execute a communication program loaded in the RAM 2214 and instruct the communication interface 2222 to perform communication processing based on the processing described in the communication program. Under the control of the CPU 2212, the communication interface 2222 reads transmission data stored in a transmission buffer processing area provided in the RAM 2214, the hard disk drive 2224, the DVD-ROM 2201, or a recording medium such as an IC card, and transmits the read transmission data to the network, or writes reception data received from the network to a reception buffer processing area or the like provided on the recording medium.

[0101] Furthermore, the CPU 2212 may cause all or a necessary portion of a file or database stored in an external recording medium such as the hard disk drive 2224, the DVD-ROM drive 2226 (DVD-ROM 2201), an IC card, etc. to be read into the RAM 2214, and perform various types of processing on the data on the RAM 2214. The CPU 2212 then writes back the processed data to the external recording medium.

[0102] Various types of information, such as various types of programs, data, tables, and databases, may be stored in the recording medium and undergo information processing. The CPU 2212 may perform various types of processing on the data read from the RAM 2214, including various types of operations, information processing, conditional judgment, conditional branching, unconditional branching, information search / replacement, etc., as described throughout this disclosure and specified by the instruction sequence of the program, and write back the results to the RAM 2214. The CPU 2212 may also search for information in a file, database, etc. in the recording medium. For example, when a plurality of entries each having an attribute value of a first attribute associated with an attribute value of a second attribute are stored in the recording medium, the CPU 2212 may search for an entry that matches a condition, in which an attribute value of the first attribute is specified, from among the plurality of entries, read the attribute value of the second attribute stored in the entry, and thereby obtain the attribute value of the second attribute associated with the first attribute that satisfies a predetermined condition.

[0103] The above-described programs or software modules may be stored in a computer-readable medium on the computer 2200 or in the vicinity of the computer 2200. Also, a recording medium such as a hard disk or a RAM provided in a server system connected to a dedicated communication network or the Internet can be used as a computer-readable medium, thereby providing the programs to the computer 2200 via the network.

[0104] Although the present invention has been described above using the embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments. It is clear to those skilled in the art that various modifications and improvements can be made to the above embodiments. It is clear from the description of the claims that such modifications and improvements can also be included in the technical scope of the present invention.

[0105] It should be noted that the order of execution of each process, such as operations, procedures, steps, and stages, in the devices, systems, programs, and methods shown in the claims, specifications, and drawings is not specifically stated as "before," "prior to," etc., and may be realized in any order unless the output of a previous process is used in a later process. Even if the operational flow in the claims, specifications, and drawings is explained using "first," "next," etc. for convenience, it does not mean that it is essential to perform the process in this order. [Explanation of symbols]

[0106] 1 System 2 Equipment 3 equipment 20 Sensors 31 Acquisition Department 32 Storage section 33 Supply section 34 Judgment section 35 Decision Section 36 Output section 37 Setting section 38 Learning processing unit 39 Selection section 321 Measurement data file 322 Classification Model 2200 Computer 2201 DVD-ROM 2210 Host Controller 2212 CPU 2214 RAM 2216 Graphics Controller 2218 Display Device 2220 Input / Output Controller 2222 Communication Interface 2224 Hard Disk Drive 2226 DVD-ROM drive 2230 ROM 2240 Input / Output Chip 2242 Keyboard

Claims

1. An acquisition unit that acquires measurement data indicating a state of the target; a supply unit that supplies the measurement data acquired by the acquisition unit to a plurality of classification models each having been trained using training data for a mutually different period, including measurement data during a period in which the state of the target is normal, and that classifies the measurement data as either normal or abnormal in response to input of the measurement data; a determination unit that determines whether a state of the object is normal or abnormal based on a plurality of classification results output from the plurality of classification models; An apparatus comprising:

2. 2. The device according to claim 1, further comprising a learning processing unit that generates, for each period in which the condition of the target is normal, a new classification model to be included in the plurality of classification models by a learning process using learning data including the measurement data for that period acquired by the acquisition unit.

3. 2. The device according to claim 1, further comprising a selection unit that selects the plurality of classification models from two or more classification models that have been trained using training data for different periods including measurement data during a period in which the condition of the target is normal, and that classify the measurement data as either normal or abnormal in response to input of the measurement data.

4. The apparatus according to claim 3 , wherein the selection unit selects at least one classification model that has been trained using training data including measurement data from a most recent period as at least one of the plurality of classification models.

5. The apparatus according to claim 3 , wherein the selection unit selects at least one classification model designated by an operator from among the two or more classification models as at least one of the plurality of classification models.

6. The device according to claim 1 , wherein the determination unit determines the state of the object as either normal or abnormal by taking a logical product of the plurality of classification results.

7. The device according to claim 1 , wherein the determination unit determines the state of the object as either normal or abnormal by taking a majority vote of the plurality of classification results.

8. 8. The device according to claim 7, wherein the determination unit assigns a larger weight to a classification result of a classification model that has been trained using training data including measurement data from a newer period among the plurality of classification models, and determines the state of the target as either normal or abnormal by taking a weighted majority vote of the plurality of classification results.

9. The apparatus of claim 1 , further comprising a determination unit that determines a severity of the abnormal condition of the object based on a number of consecutive times the condition of the object is determined to be abnormal.

10. 10. The device according to claim 9, further comprising a setting unit that sets, as measurement data to be included in new learning data, one of the following: among the measurement data for each period acquired by the acquisition unit and supplied from the supply unit to the plurality of classification models, measurement data for a period in which the severity determined by the determination unit is lower than a reference severity; and among the measurement data acquired by the acquisition unit and supplied from the supply unit to the plurality of classification models, measurement data excluding each measurement data that caused consecutive abnormality determinations until the severity determined by the determination unit became the reference severity, and each measurement data that caused consecutive abnormality determinations following the said measurement data.

11. an acquisition stage for acquiring measurement data indicative of a state of the object; a supply step of supplying the measurement data acquired in the acquisition step to a plurality of classification models each having been trained using training data for a mutually different period including measurement data during a period in which the state of the subject is normal, the classification models classifying the measurement data as either normal or abnormal in response to input of the measurement data; A determination step of determining whether the state of the object is normal or abnormal based on a plurality of classification results output from the plurality of classification models; A method for providing the above.

12. Computer, An acquisition unit that acquires measurement data indicating a state of the target; a supply unit that supplies the measurement data acquired by the acquisition unit to a plurality of classification models each having been trained using training data for a mutually different period, including measurement data during a period in which the state of the target is normal, and that classifies the measurement data as either normal or abnormal in response to input of the measurement data; a determination unit that determines whether the state of the object is normal or abnormal based on a plurality of classification results output from the plurality of classification models; A program that functions as a

Images