JP7384182B2 - Abnormality diagnosis method, model update method, and abnormality diagnosis device - Google Patents

Description

The present disclosure relates to an abnormality diagnosis method, a model update method, and an abnormality diagnosis apparatus. The present disclosure particularly relates to an abnormality diagnosing method and an abnormality diagnosing apparatus that monitor the state of a manufacturing process or the like and estimate equipment where an abnormality has occurred and when the abnormality has occurred. Further, the present disclosure particularly relates to a model updating method for updating a model used in a manufacturing process or the like at a timing when an abnormality is estimated to occur.

BACKGROUND ART Conventionally, methods for diagnosing abnormal conditions in manufacturing processes and the like have been proposed. Conventional process operation monitoring, for example as in Patent Document 1 and Patent Document 2, involves constructing a predictive model and evaluating the error between the prediction by the predictive model and the current state as a degree of deviation.

Japanese Patent Application Publication No. 2010-049359 International Publication No. 2018/235807

Here, when detecting an abnormality in a manufacturing process or the like, the above-mentioned predictive model is created by learning the normal state, and an abnormality is diagnosed based on the degree of deviation. However, when the state of a manufacturing process or the like is normal, but there is a change in equipment (for example, maintenance and repair), a problem arises in which it is diagnosed that the state has deviated from the learned normal state. Although it is possible to immediately update the prediction model when there is a change in equipment, it is not practical to update the prediction model every time maintenance is performed on a daily basis, for example. Further, as a result of adjusting equipment, it may be difficult to determine which state is normal before or after the adjustment.

An object of the present disclosure, which was made to solve the above problems, is to provide an abnormality diagnosing method and an abnormality diagnosing device that can estimate with high precision the equipment in which the abnormality has occurred and the time in which the abnormality has occurred. Another object of the present disclosure is to provide a model update method that allows models used in manufacturing processes and the like to be updated at appropriate timing.

An abnormality diagnosis method according to an embodiment of the present disclosure includes:
a period setting step for setting multiple periods;
a data measurement step of measuring data regarding the operation of the equipment in each of the plurality of periods for each of the plurality of equipment that is similar to each other;
a distribution calculation step of calculating the distribution of the measured data for each combination of the plurality of facilities and the plurality of periods;
a statistical difference calculation step of calculating a statistical difference between a plurality of said distributions;
The method includes an abnormality estimation step of estimating the equipment where the abnormality has occurred and the time when the abnormality has occurred based on the statistical difference.

A model update method according to an embodiment of the present disclosure includes:
Obtain the presence or absence of an abnormality estimated by the abnormality diagnosis method described above, and when it is estimated that an abnormality has occurred, and the estimation contents do not indicate that it is within the normal range or that the data is abnormal, Update models used to operate multiple facilities.

An abnormality diagnosis device according to an embodiment of the present disclosure includes:
a period setting section for setting multiple periods;
a data measurement unit that measures data regarding the operation of the equipment in each of the plurality of periods for each of the plurality of equipment that are similar to each other;
a distribution calculation unit that calculates a distribution of the measured data for each combination of the plurality of facilities and the plurality of periods;
a statistical difference calculation unit that calculates a statistical difference between a plurality of the distributions;
The apparatus includes an abnormality estimating unit that estimates the equipment in which the abnormality has occurred and the time when the abnormality has occurred based on the statistical difference.

According to the present disclosure, it is possible to provide an abnormality diagnosing method and an abnormality diagnosing device that can estimate with high accuracy the equipment where the abnormality has occurred and the time when the abnormality has occurred. Further, according to the present disclosure, it is possible to provide a model updating method that can update a model used in a manufacturing process or the like at an appropriate timing.

FIG. 1 is a diagram illustrating a configuration example of an abnormality diagnosis device according to an embodiment. FIG. 2 is a diagram illustrating equipment for a manufacturing process targeted by the abnormality diagnosis device according to an embodiment. FIG. 3 is a flowchart illustrating an abnormality diagnosis method according to an embodiment. FIG. 4 is a diagram showing the distribution of data groups in Example 1. FIG. 5 is a diagram showing the distribution of data groups in Example 2.

DESCRIPTION OF THE PREFERRED EMBODIMENTS An abnormality diagnosis method, model update method, and abnormality diagnosis apparatus according to an embodiment of the present disclosure will be described below with reference to the drawings.

(Abnormality diagnosis device)
FIG. 1 is a diagram showing a configuration example of an abnormality diagnosis device 10 according to the present embodiment. The abnormality diagnosis device 10 includes a communication section 11, a storage section 12, a control section 13, and a display section 14. The control unit 13 includes a period setting unit 21, a data measurement unit 22, a distribution calculation unit 23, a statistical difference calculation unit 24, and an abnormality estimation unit 25. Details of each component of the abnormality diagnosis device 10 will be described later.

The abnormality diagnosis device 10 detects abnormalities in the manufacturing process. In this embodiment, the manufacturing process is a continuous casting process in the steel industry, but is not limited thereto. In the present embodiment, the abnormality diagnosis device 10 communicates with a host system including a process computer that manages the continuous casting process, and acquires necessary measurement values, operating parameters, etc. of equipment in the continuous casting process. As another example, the abnormality diagnosis device 10 may be part of a process computer or a host system.

FIG. 2 is a diagram showing a continuous casting machine, which is an example of equipment for a manufacturing process targeted by the abnormality diagnosis device 10. A continuous casting machine casts slabs. In continuous casting machines, there may be multiple strands branching from the mold. A process computer controls the rotational speed of the pinch rolls in the strand to draw out the molten steel. The abnormality diagnosis device 10 measures data related to the operation of each of a plurality of pieces of equipment that are similar to each other. In this embodiment, the abnormality diagnosis device 10 targets the first equipment 31 and the second equipment 32 for data measurement. Specifically, as shown in FIG. 2, the first equipment 31 is one strand, and the second equipment 32 is the other strand. Here, the number of mutually similar pieces of equipment to be measured by the abnormality diagnosis device 10 is not limited to two, and may be three or more.

The communication unit 11 has an interface function for wired or wireless communication. The communication unit 11 transmits and receives necessary data and signals to and from the process computer or host system. For example, the abnormality diagnosis device 10 transmits a signal that causes the communication unit 11 to measure the current value that drives the pinch rolls of the first equipment 31 and the second equipment 32, and receives the measured data by the communication unit 11. . The communication method performed by the communication unit 11 may be a wired communication standard or a wireless communication standard. For example, wireless communication standards may include cellular phone communication standards such as 3G, 4G, and 5G. Further, for example, the wireless communication standard may include IEEE802.11, Bluetooth (registered trademark), and the like. The communication unit 11 can support one or more of these communication standards.

The storage unit 12 may have a function as a memory that stores various information. The storage unit 12 may store, for example, programs executed by the control unit 13, data used in processes executed by the control unit 13, results of the processes, and the like. Furthermore, the storage unit 12 may function as a work memory for the control unit 13. The storage unit 12 can be configured by, for example, a semiconductor memory or the like, but is not limited thereto, and can be any storage device. For example, the storage unit 12 may be an internal memory of a processor used as the control unit 13, which will be described later, or may be a hard disk drive (HDD) accessible from the control unit 13.

In the present embodiment, the storage unit 12 stores intermediate data such as a plurality of time periods, data regarding equipment operation, data distribution and statistical differences between distributions, and results of magnitude determination of a plurality of statistical differences, which will be described later. A correspondence table between combinations and estimated abnormalities is stored.

The control unit 13 controls and manages each functional unit constituting the abnormality diagnosis device 10 and the entire abnormality diagnosis device 10. The control unit 13 includes at least one processor, such as a CPU (Central Processing Unit), to control and manage various functions. The control unit 13 may be composed of one processor or a plurality of processors. The processor constituting the control unit 13 reads a program from the storage unit 12 and executes it to function as a period setting unit 21, a data measurement unit 22, a distribution calculation unit 23, a statistical difference calculation unit 24, and an abnormality estimation unit 25. It may work.

The period setting unit 21 sets a plurality of periods. The plurality of periods is, for example, two periods, but may be three or more periods. In this embodiment, the period setting unit 21 sets a period P and a period Q. Period Q is a period after period P, but a portion thereof may overlap. The period P and the period Q may be set, for example, before and after an event related to the equipment, such as repair, replacement, maintenance, or start of actual operation, occurs in at least one of the first equipment 31 and the second equipment 32. Events related to equipment may include a certain period of blank space (non-operational period) for periodic management. Further, the period P and the period Q may not have the same length. For example, period P may be longer or shorter than period Q. Further, the period P may be a period during which a simulation of the equipment is operated, and the period Q may be a period during which the actual equipment is operated.

The data measurement unit 22 measures data regarding the operation of the equipment for each of the plurality of mutually similar equipment in each of the plurality of periods. In the present embodiment, the data measuring unit 22 measures data regarding the operation of the equipment in the period P and the period Q for each of the first equipment 31 and the second equipment 32 via the process computer. The data regarding the operation of the equipment is not particularly limited as long as it is data indicating the operation status of the equipment. For example, the data regarding the operation of the equipment may be a current value for driving a pinch roll, a rotation speed, etc.

The data measuring unit 22 measures a plurality of pieces of data for each combination of a plurality of facilities and a plurality of periods. Although the number of data is not limited, a large amount of data is required to obtain a probability density function representing the characteristics of the above combinations. In this embodiment, the data measurement unit 22 measures data groups P ₁ , P ₂ , Q ₁ and Q ₂ shown in Table 1. The data measurement unit 22 may accumulate measured data in the storage unit 12 for each combination of a plurality of facilities and a plurality of periods. That is, the data groups P ₁ , P ₂ , Q ₁ and Q ₂ may be stored in the storage unit 12.

Here, at least one of the data groups P ₁ and P ₂ may be a simulation value or an empirically determined ideal state (a calculated value that is not an actual measurement value). Further, the data group of the first equipment 31 (at least one of the data groups P ₁ and Q ₁ ) or the data group of the second equipment 32 (at least one of the data groups P ₂ and Q ₂ ) is a simulation value or an ideal value. It may be a state (a calculated value that is not an actual value).

The distribution calculation unit 23 calculates the distribution of measured data for each combination of a plurality of facilities and a plurality of periods. The distribution calculation unit 23 may read the stored data groups P ₁ , P ₂ , Q ₁ and Q ₂ from the storage unit 12 . The distribution calculation unit 23 calculates a statistical data distribution including at least a probability density function for each of the data groups P ₁ , P ₂ , Q ₁ and Q ₂ . The distribution calculation unit 23 may calculate the average value, standard deviation, etc. for each of the data groups P ₁ , P ₂ , Q ₁ and Q ₂ in order to obtain a probability density function. The distribution calculation unit 23 may store the calculated data distribution in the storage unit 12 as intermediate data.

The statistical difference calculation unit 24 calculates statistical differences between a plurality of distributions. The statistical difference calculation unit 24 may read the distribution data of the data groups P ₁ , P ₂ , Q ₁ and Q ₂ stored from the storage unit 12 . In this embodiment, the statistical difference calculation unit 24 uses the formulas shown in Table 2 to calculate four statistical differences for the data groups P ₁ , P ₂ , Q ₁ and Q ₂ . The meanings of the first to fourth statistical differences are shown in Table 2.

The above D _JS indicates the calculation of JS information amount (Jensen-Shannon divergence). However, the statistical difference calculation unit 24 may use a histogram intersection, a Kullback-Leibler divergence, an L1 norm, an L2 norm, or the like instead of the JS information amount. In other words, the statistical difference calculation performed by the statistical difference calculation unit 24 is not limited to a specific calculation, but may be any calculation that can measure the difference between distributions.

Here, regarding the data groups P ₁ , P ₂ , Q ₁ and Q ₂ , an example of calculating the statistical difference will be shown in the case where they can be treated as normal distributions. The data distribution may be other than normal distribution. In the case of normal distribution, the probability density function can be expressed as p(x) in the following equation (1), and can be obtained by calculating the mean μ and variance σ ² .

In this embodiment, the statistical difference calculation unit 24 uses a symmetric JS information amount as an index for measuring the difference in probability distribution. The JS information amount is calculated based on the KL information amount. The KL information amount is expressed by D _KL in the following equation (2). The JS information amount is expressed by the following equation (3).

The statistical difference calculation unit 24 may store the calculated statistical difference between the distributions in the storage unit 12 as intermediate data.

The abnormality estimating unit 25 calculates a statistical index from a plurality of statistical differences, and estimates whether an abnormality has occurred based on the statistical index. In this embodiment, the plurality of statistical differences are the first to fourth statistical differences described above. The abnormality estimating unit 25 may read the first to fourth statistical difference data stored from the storage unit 12. In this embodiment, the statistical index is the average value of a plurality of statistical differences, and is represented by S in equation (4).

The abnormality estimating unit 25 estimates that there is an abnormality when the statistical index exceeds a threshold (hereinafter referred to as a statistical index threshold). The statistical index threshold is set such that it can be detected that at least one of the first to fourth statistical differences is large. For example, the statistical index threshold may be set in advance based on first to fourth statistical differences in the same manufacturing process measured in a state where no abnormality occurs.

Here, the statistical index may be calculated using a formula different from the above formula (4). For example, for each of the first to fourth statistical difference terms, a formula with a weighting coefficient may be used.

Furthermore, the abnormality estimating unit 25 estimates the equipment where the abnormality occurred and the time when the abnormality occurred based on statistical differences. When the abnormality estimating unit 25 estimates that an abnormality has occurred based on the statistical index, it may estimate the equipment where the abnormality has occurred and the time when the abnormality has occurred. In other words, when the abnormality estimating unit 25 estimates that no abnormality has occurred based on the statistical index, it does not need to estimate the equipment where the abnormality has occurred and the time when the abnormality has occurred. The abnormality estimating unit 25 compares a plurality of statistical differences (first to fourth statistical differences) with respective threshold values to determine the magnitude. At this time, the threshold values may be the same value (0.5 as an example) or different threshold values may be used for the first to fourth statistical differences. Then, the abnormality estimating unit 25 uses the combination of the results of the magnitude determination to estimate the equipment where the abnormality has occurred and the time when the abnormality has occurred. The abnormality estimation unit 25 may use the correspondence table shown in Table 3 in this estimation. The correspondence table is data that associates combinations of results of determining the magnitude of a plurality of statistical differences with estimated abnormalities.

The abnormality estimating unit 25 estimates that an abnormality has occurred in the first equipment 31 during the period P when Case 6 applies, for example. In case 6, since the first statistical difference (D _JS (P ₁ | | Q ₁ )) is large, it is considered that an abnormality occurred in the first equipment 31 or the second equipment 32 during period P. It will be done. Here, while the third statistical difference (D _JS (P ₁ | | P ₂ )) is large, the second statistical difference (D _JS (P ₂ | | Q ₂ )) and the fourth Since the statistical difference (D _JS (Q ₁ | | Q ₂ )) is small, it is estimated that the abnormality occurred in the first equipment 31, not the second equipment 32. In addition, in cases corresponding to Case 1, Case 14, Case 15, and Case 16, the abnormality estimating unit 25 determines that the estimation that an abnormality has occurred based on the statistical index is incorrect, and requests re-measurement of the data. can be executed. The abnormality estimating unit 25 can output the estimation result including the presence or absence of an abnormality based on the statistical index and the details of the estimated abnormality shown in Table 3 to the display unit 14 for display.

The display unit 14 displays estimation results regarding the occurrence of an abnormality under the control of the control unit 13. The display unit 14 may be a display device such as a liquid crystal display or an organic electroluminescence panel.

(Model update method)
When the operator of the manufacturing process confirms the occurrence of an abnormality and the details of the abnormality displayed on the display unit 14, the operator of the manufacturing process may change the operating conditions according to the details of the abnormality. Generally, in a manufacturing process, equipment is modeled to determine optimal operating conditions using simulation. When equipment needs to be repaired, it may be difficult to decide whether or not to update such a model. Further, as a result of adjusting equipment, it may be difficult to determine which state is normal before or after the adjustment.

As described above, the abnormality diagnosis device 10 calculates the statistical difference between the distributions of data acquired for each combination of multiple pieces of equipment and multiple periods that are similar to each other, and calculates the statistical difference between the distributions in Table 3. As shown in Figure 2, it is possible to estimate with high accuracy the equipment where the abnormality occurred and the time when the abnormality occurred. Therefore, an operator of the manufacturing process can check the occurrence of an abnormality and the details of the abnormality displayed on the display unit 14 and update the model at an appropriate timing. Here, such model updating processing may be automatically executed by, for example, a process computer. That is, a device such as a process computer that manages a model of equipment in a manufacturing process or the like obtains the presence or absence of an abnormality estimated by the abnormality diagnosis device 10, and when it is estimated that an abnormality has occurred, performs operation control. A model update method may be performed to update a model used in a model. However, as shown in Table 3, the estimation contents ("estimated abnormality" in Table 3) may indicate that the data is estimated to be within the normal range or that the data is estimated to be abnormal. . A device such as a process computer may update the model of the equipment when it is estimated that an abnormality has occurred and the estimation does not indicate that it is within a normal range or that the data is abnormal. In addition, in the model updating method, a device such as a process computer determines which state is normal before and after equipment adjustment, etc., based on the content of the abnormality estimated by the abnormality diagnosis device 10, and correctly determines which state is normal. The model can be updated.

When the model is a machine learning model, a device such as a process computer may update the model by, for example, performing machine learning on data in a period determined to be non-abnormal by the abnormality diagnosis device 10 as new learning data. . If the model is a mathematical model, a device such as a process computer may update the model by, for example, adjusting numerical values in the mathematical formula to match data in a period determined to be non-abnormal by the abnormality diagnosis device 10. . Here, when it is estimated to be within the normal range, as in case 8 and case 12 of Table 3, for example, a device such as a process computer does not need to update the model. In this way, by determining whether to update the model based on the content of the estimated abnormality, it is possible to avoid a situation where the predictive model is updated every time maintenance is performed on a daily basis.

(Abnormality diagnosis method)
FIG. 3 is a flowchart illustrating the abnormality diagnosis method according to this embodiment. The abnormality diagnosis device 10 described above executes processing according to the flowchart in FIG.

The abnormality diagnosis device 10 sets a plurality of periods (step S1). Here, step S1 corresponds to a period setting step executed by the period setting section 21.

The abnormality diagnosis device 10 measures data related to the operation of the equipment in each of the plurality of periods for each of the plurality of pieces of equipment that are similar to each other (step S2). Here, step S2 corresponds to a data measurement step executed by the data measurement section 22.

The abnormality diagnosis device 10 calculates the distribution of measured data for each combination of a plurality of facilities and a plurality of periods (step S3). Here, step S3 corresponds to a distribution calculation step executed by the distribution calculation unit 23.

The abnormality diagnosis device 10 calculates statistical differences between the plurality of distributions (step S4). Here, step S4 corresponds to a statistical difference calculation step executed by the statistical difference calculation unit 24.

The abnormality diagnosis device 10 calculates a statistical index from a plurality of statistical differences (step S5).

The abnormality diagnosis device 10 estimates whether an abnormality has occurred based on the calculated statistical index. When the abnormality diagnosis device 10 estimates that no abnormality has occurred (No in step S6), it ends the series of processes.

When the abnormality diagnosis device 10 estimates that an abnormality has occurred (Yes in step S6), the abnormality diagnosis device 10 compares the plurality of statistical differences with respective threshold values to perform magnitude determination (step S7).

The abnormality diagnosis device 10 estimates the equipment where the abnormality occurred and the time when the abnormality occurred, using the combination of the results of the size determination (step S8). Here, steps S5 to S8 correspond to an abnormality estimation step executed by the abnormality estimation unit 25.

The abnormality diagnosis device 10 presents the estimation result (step S9). The abnormality diagnosis device 10 displays, for example, the content of the estimated abnormality on the display unit 14, and ends the series of processing.

Table 4 shows the distribution (mean and standard deviation) of the data for Example 1. Table 5 shows the distribution (mean and standard deviation) of the data for Example 2. FIG. 4 is a diagram showing the distribution (histogram) of the data group of Example 1. Further, FIG. 5 is a diagram showing the distribution (histogram) of the data group of Example 2.

The data groups P ₁ , P ₂ , Q ₁ and Q ₂ of Example 1 and Example 2 assume normal distribution. Here, the data is the current value of the drive motor for the pinch roll, which is part of the equipment of the continuous casting machine. As mentioned above, in a continuous casting machine, the equipment after the mold is branched, and a plurality of similar equipments exist.

The continuous casting machine in this embodiment supplies molten steel from one tundish to two molds to simultaneously produce two slabs. Once the molten steel has solidified to the inside, it is cut with a cutter to produce a slab. One of the slab manufacturing lines manufactured from two molds is a strand corresponding to the first equipment 31 described above. The other side of the slab production line is a strand corresponding to the second equipment 32 described above. Pinch rolls on these strands are used to pull out the slab from the mold while performing secondary cooling, and are driven by an electric motor or the like.

Period P and period Q are each about one week. Adjustments to both equipment were performed between period P and period Q. The number of data groups P ₁ , P ₂ , Q ₁ and Q ₂ is approximately 1400 each. The statistical index threshold was set at 2.3. In addition, the threshold value used for size determination is 0.3 for the first statistical difference and the second statistical difference, and 0.5 for the third statistical difference and the fourth statistical difference. It was done. Table 6 shows the results of the abnormality estimation carried out by the abnormality diagnosis device 10 for Example 1 and Example 2.

The abnormality diagnosis device 10 estimated that an abnormality occurred in both Example 1 and Example 2, and that the first equipment 31 was abnormal in the period Q. This matched the distribution shown in the histograms of FIGS. 4 and 5, and good results were obtained. Moreover, the statistical index shows a higher value in Example 2 than in Example 1. This also agrees with the distribution shown in the histograms of FIGS. 4 and 5, indicating that the abnormality diagnosis apparatus 10 can estimate abnormalities well.

As described above, the abnormality diagnosis device 10 and the abnormality diagnosis method according to the present embodiment perform statistical analysis between the distributions of data acquired for each combination of a plurality of mutually similar equipment and a plurality of periods. By calculating the difference, it is possible to estimate with high precision the equipment where the abnormality occurred and the time when the abnormality occurred. Further, the model updating method according to the present embodiment can update a model used in a manufacturing process or the like at an appropriate timing based on the occurrence of an abnormality and the content of the abnormality estimated by this abnormality diagnosis method.

Although the embodiments of the present disclosure have been described based on the drawings and examples, it should be noted that those skilled in the art can easily make various changes or modifications based on the present disclosure. It should therefore be noted that these variations or modifications are included within the scope of this disclosure. For example, the functions included in each component or each step can be rearranged to avoid logical contradictions, and multiple components or steps can be combined or divided into one. It is. Embodiments according to the present disclosure can also be realized as a program executed by a processor included in the device or a storage medium recording the program. It is to be understood that these are also encompassed within the scope of the present disclosure.

In the above embodiment, the abnormality diagnosis method, model update method, and abnormality diagnosis apparatus 10 have been described assuming that the plurality of pieces of equipment that are similar to each other are strands of a continuous casting machine. However, the above-described abnormality diagnosis method, model update method, and abnormality diagnosis apparatus 10 are not limited to this example, and can be applied to a plurality of similar equipment used in manufacturing processes and the like. Similar equipment may be, for example, manufacturing machines arranged in parallel to perform the same process, such as heating, cooling, molding, etc.

10 Abnormality diagnosis device 11 Communication unit 12 Storage unit 13 Control unit 14 Display unit 21 Period setting unit 22 Data measurement unit 23 Distribution calculation unit 24 Statistical difference calculation unit 25 Abnormality estimation unit 31 First equipment 32 Second equipment

Claims (7)

a period setting step for setting multiple periods;
a data measurement step of measuring data regarding the operation of the equipment in each of the plurality of periods for each of the plurality of equipment that is similar to each other;
a distribution calculation step of calculating the distribution of the measured data for each combination of the plurality of facilities and the plurality of periods;
a statistical difference calculation step of calculating a statistical difference between a plurality of said distributions;
an abnormality estimation step of estimating the equipment in which the abnormality has occurred and the time in which the abnormality has occurred, using data in which a combination of the results of determining the magnitude of the plurality of statistical differences is associated with the estimated abnormality; including abnormality diagnosis methods. The abnormality estimation step compares the plurality of statistical differences with respective thresholds to determine the magnitude, and uses a combination of the results of the magnitude determination to determine the equipment in which the abnormality has occurred and the time period in which the abnormality occurred. The abnormality diagnosis method according to claim 1, wherein the abnormality diagnosis method estimates. 3. The abnormality diagnosis method according to claim 1, wherein the abnormality estimation step includes calculating a statistical index from the plurality of statistical differences and estimating whether or not an abnormality has occurred based on the statistical index. 4. The abnormality diagnosis method according to claim 3, wherein the abnormality estimation step estimates the equipment where the abnormality has occurred and the time when the abnormality has occurred when it is estimated that the abnormality has occurred based on the statistical index. The abnormality diagnosis method according to claim 3 or 4, wherein the statistical index is an average value of a plurality of the statistical differences. The presence or absence of an abnormality estimated by the abnormality diagnosis method according to any one of claims 3 to 5 is acquired, and even if it is estimated that an abnormality has occurred and the estimation content is within a normal range, the data is abnormal. A model updating method that predicts the states of the plurality of facilities and updates a model used for deriving the degree of deviation or optimizing the operation of the facilities, when the conditions of the plurality of facilities are not indicated. . a period setting section for setting multiple periods;
a data measurement unit that measures data regarding the operation of the equipment in each of the plurality of periods for each of the plurality of equipment that are similar to each other;
a distribution calculation unit that calculates a distribution of the measured data for each combination of the plurality of facilities and the plurality of periods;
a statistical difference calculation unit that calculates a statistical difference between a plurality of the distributions;
an anomaly estimating unit that estimates the equipment in which the anomaly has occurred and the time at which the anomaly occurred, using data in which a combination of the results of determining the magnitude of the plurality of statistical differences is associated with the estimated anomaly; Equipped with an abnormality diagnosis device.

Images