Driving support device, driving support method, and program

Abstract

To present a plurality of optimal operations for recovering from an abnormality to normal.SOLUTION: A driving support device according to one aspect of the present disclosure comprises: an abnormality diagnosis unit that uses driving data of a control object to diagnose whether an abnormality or a symptom of the abnormality has occurred in the control object; an operation amount calculating unit that calculates a plurality of optimal operations for returning the control object to normal when the control object is diagnosed as having the abnormality or the symptom of the abnormality; and an output unit that outputs the plurality of optimal operations to a predetermined output destination.SELECTED DRAWING: Figure 5

Description

Technical Field

[0001] The present disclosure relates to a driving support device, a driving support method, and a program.

Background Art

[0002] In a plant such as a waste incineration plant, for example, an operator operates the air flow rate into the furnace and the like based on measurement values such as the combustion state of the furnace to operate the plant. As one of the technologies for supporting the operation of such a plant, when an abnormality or its sign is diagnosed or detected, considering the magnitude of the operation amount and the degree of deviation from the past normal operation data, a technique for presenting an optimal operation for returning the plant to normal is known (Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the conventional technology described in Patent Document 1, it is necessary to adjust with a weight parameter λ which of the two indicators, the magnitude of the operation amount and the degree of deviation from the past normal operation data, is emphasized. Therefore, depending on the value of the weight parameter λ, an impractical operation may be presented.

[0005] The present disclosure has been made in view of the above points, and an object thereof is to present a plurality of optimal operations for returning from an abnormality to normal.

Means for Solving the Problems

[0006] According to one aspect of the present disclosure, an operation support device includes an abnormality diagnosis unit that diagnoses whether an abnormality or a sign of the abnormality has occurred in a control target using operation data of the control target, an operation amount calculation unit that calculates a plurality of optimal operations for returning the control target to normal when it is diagnosed that an abnormality or a sign of the abnormality has occurred in the control target, and an output unit that outputs the plurality of optimal operations to a predetermined output destination.

Effect of the Invention

[0007] It is possible to present a plurality of optimal operations for returning from an abnormality to normal.

Brief Description of the Drawings

[0008]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Embodiments for Carrying Out the Invention

[0009] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. Hereinafter, in the case where the plant is controlled by the operator's operation, when an abnormality or its sign or the like occurs in the plant is diagnosed or detected, a plurality of optimal operations for returning the plant to normal are presented to the operator. The plant control system 1 will be described.

[0010] <Overall Configuration Example of Plant Control System 1> An overall configuration example of the plant control system 1 according to this embodiment will be described with reference to FIG. 1. FIG. 1 is a diagram showing an example of the overall configuration of the plant control system 1 according to this embodiment.

[0011] As shown in FIG. 1, the plant control system 1 according to this embodiment includes an operation support device 10, a control device 20, and a plant 30. Here, the operation support device 10 and the control device 20 are communicably connected via an arbitrary communication network, and similarly, the control device 20 and the plant 30 are communicably connected via an arbitrary communication network.

[0012] The operation support device 10 acquires the operation data of the plant 30 from the control device 20, and diagnoses or detects the presence or absence of an abnormality or its sign in the plant 30 from the operation data. Further, when the occurrence of an abnormality or its sign in the plant 30 is diagnosed or detected, the operation support device 10 presents a plurality of optimal operations for returning the plant 30 to normal to an operator or the like. Thereby, the operator can select a desired operation from among the plurality of optimal operations or determine an appropriate operation with reference to the plurality of optimal operations, so that the operation of the plant 30 by the operator is supported. Hereinafter, an abnormality or its sign, etc. will be collectively referred to simply as "abnormality, etc.", and the diagnosis or detection of the presence or absence of an abnormality, etc. will be collectively referred to simply as "diagnosing the presence or absence of an abnormality, etc.". Here, the operation data is measurement data obtained by measuring the state of the process executed by the plant 30 (for example, temperature, pressure, flow rate, gas concentration, etc.) by various sensors or the like, and is generally expressed as multivariate data (that is, a multi-dimensional vector).

[0013] The control device 20 collects operation data from the plant 30 and controls the plant 30 according to the operation set by the operator. Here, examples of the control device 20 include a PLC (programmable logic controller) or the like. Note that the operator may directly set an operation for the control device 20, or may set an operation for the control device 20 using a terminal or device that can communicate with the control device 20 via a communication network. In particular, the operator may set an operation for the control device 20 using the operation support device 10.

[0014] The plant 30 is equipment or facilities for executing various processes under the control of the control device 20. Specific examples of the plant 30 include, for example, a waste incineration plant, a petrochemical plant, a food plant, a steel plant, and the like.

[0015] Note that the overall configuration of the plant control system 1 shown in FIG. 1 is an example, and other configurations may be used. For example, the operation support device 10 and the control device 20 may be integrally configured, or a terminal or the like used by an operator may be included in the plant control system 1.

[0016] <Example of Hardware Configuration of Operation Support Device 10> An example of the hardware configuration of the operation support device 10 according to the present embodiment will be described with reference to FIG. 2. FIG. 2 is a diagram showing an example of the hardware configuration of the operation support device 10 according to the present embodiment.

[0017] As shown in FIG. 2, the operation support device 10 according to the present embodiment includes an input device 101, a display device 102, an external I / F 103, a communication I / F 104, a RAM (Random Access Memory) 105, a ROM (Read Only Memory) 106, an auxiliary storage device 107, and a processor 108. These pieces of hardware are communicably connected to each other via a bus 109.

[0018] The input device 101 is, for example, a keyboard, a mouse, a touch panel, a physical button, or the like. The display device 102 is, for example, a display, a display panel, or the like. Note that the operation support device 10 may not have at least one of the input device 101 and the display device 102.

[0019] The external I / F 103 is an interface with an external device such as a recording medium 103a. Examples of the recording medium 103a include a CD (Compact Disc), a DVD (Digital Versatile Disk), an SD memory card (Secure Digital memory card), a USB (Universal Serial Bus) memory card, and the like.

[0020] The communication I / F 104 is an interface for connecting the driving assistance device 10 to a communication network. The RAM 105 is a volatile semiconductor memory (storage device) that temporarily holds programs and data. The ROM 106 is a non-volatile semiconductor memory (storage device) that can hold programs and data even when the power is turned off. The auxiliary storage device 107 is a non-volatile storage device such as, for example, an HDD (Hard Disk Drive), an SSD (Solid State Drive), or a flash memory. The processor 108 is various arithmetic devices such as, for example, a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit).

[0021] Note that the hardware configuration shown in FIG. 2 is an example, and the hardware configuration of the driving assistance device 10 is not limited to this. For example, the driving assistance device 10 may have a plurality of auxiliary storage devices 107 and a plurality of processors 108, may not have some of the illustrated hardware, or may have various hardware other than the illustrated hardware.

[0022] <Functional configuration example of the driving assistance device 10> An example of the functional configuration of the driving assistance device 10 according to the present embodiment will be described with reference to FIG. 3. FIG. 3 is a diagram showing an example of the functional configuration of the driving assistance device 10 according to the present embodiment.

[0023] As shown in FIG. 3, the driving assistance device 10 according to the present embodiment includes an input unit 201, an abnormality diagnosis unit 202, an operation amount calculation unit 203, and an output unit 204. Each of these units is realized, for example, by processing executed by one or more programs installed in the driving assistance device 10 on a processor 108 or the like. Further, the driving assistance device 10 according to the present embodiment has a database 205. The database 205 is realized, for example, by a storage area such as the auxiliary storage device 107. Note that the database 205 may be realized by a storage area such as a storage device that is communicably connected to the driving assistance device 10 via a communication network.

[0024] The input unit 201 acquires (inputs) operation data to be subjected to abnormality diagnosis (hereinafter also referred to as "data to be diagnosed") from the control device 20 at predetermined time intervals (for example, every control cycle of the plant 30). Hereinafter, the variable representing the operation data is denoted as x, and the data to be diagnosed is denoted as x (p) . Since the operation data is generally multivariate data, for example, if the total number of variables representing the state of the process (which is also called a state variable or the like) is M, and the state variables are x 1 , ···, x M , then x = (x 1 , ···, x M ). Also, among the state variables x 1 , ···, x M , there may be not only variables that can be operated by an operator or the like but also variables that cannot be operated (for example, when considering a waste incineration plant, the amount of waste input and the air flow rate can be operated, but the CO concentration etc. cannot be operated).

[0025] The abnormality diagnosis unit 202 diagnoses whether an abnormality or the like has occurred in the plant 30 by using the abnormality diagnosis model stored in the database 205 and the data to be diagnosed acquired by the input unit 201. Here, the abnormality diagnosis model is a model for diagnosing whether an abnormality or the like has occurred in the plant 30, and is created in advance, for example, by using machine learning techniques or the like with past operation data. In general, since an abnormality diagnosis model considering a time delay (that is, the time delay from when a certain operation is performed until the response thereto is measured) is created in the plant 30, hereinafter as well, the abnormality diagnosis model is assumed to be a machine learning model considering the time delay of the plant 30.

[0026] Also, hereinafter, the abnormality diagnosis model is represented by H(·), and when H(x (p) ) = +1, the plant 30 is normal (that is, no abnormality or the like has occurred), and when H(x (p)When =-1, it is assumed that the plant 30 is diagnosed as abnormal (i.e., an abnormality or the like has occurred). The abnormality diagnosis model H may be a linear model (i.e., a model that can linearly separate the region satisfying H(x)=+1 and the region satisfying H(x)=-1), or a non-linear model. Examples of linear models include linear regression, upper and lower limits, decision trees, and the like.

[0027] When the abnormality diagnosis unit 202 diagnoses that the plant 30 is abnormal, the operation amount calculation unit 203 calculates a plurality of optimal operation amounts for returning the plant 30 from the abnormal state to the normal state. That is, the operation amount calculation unit 203 calculates S optimal operation amounts a (p) +a (s) satisfying H(x (s) +a (s) )=+1 (s = 1, ···, S). At this time, the operation amount calculation unit 203 uses the multi-objective optimization technique to calculate the Pareto optimal solution that minimizes two indexes, namely, the magnitude of the operation amount and the degree of deviation from the past normal operation data, as the S optimal operation amounts a (s) (s = 1, ···, S). Here, S is a predetermined integer of 2 or more. Each operation amount a (s) is also multi-variable data. When the operation data is represented as x=(x 1 , ···, x M ), a (s) =(a 1 (s) , ···, a M (s) ) is represented.

[0028] The output unit 204 outputs the S optimal operation amounts a (s) (s = 1, ···, S) calculated by the operation amount calculation unit 203 to a predetermined output destination and presents them to the operator of the plant 30 or the like. As a result, the operation amount selected from among the S optimal operation amounts a (s) (s = 1, ···, S) or the operation amount determined with reference to the S optimal operation amounts a (s) (s = 1, ···, S) is set in the control device 20, and the plant 30 is controlled according to the operation amount. The S optimal operation amounts a (s)As the output destination of (s = 1, ···, S), for example, it may be the display device 102, or it may be a terminal or device that is communicably connected via a communication network, etc.

[0029] The database 205 stores various data (for example, an abnormality diagnosis model, past operation data (especially, past operation data when the plant 30 is normal), etc.).

[0030] <Operation support process> The operation support process according to the present embodiment will be described with reference to FIG. 4. FIG. 4 is a flowchart showing an example of the operation support process according to the present embodiment. Note that steps S101 to S105 in FIG. 4 are repeatedly executed at predetermined time intervals (for example, for each control cycle of the plant 30).

[0031] Step S101: The input unit 201 acquires the diagnostic target data x (p) from the control device 20.

[0032] Step S102: Next, the abnormality diagnosis unit 202 uses the abnormality diagnosis model stored in the database 205 and the diagnostic target data x (p) acquired in step S101 above to diagnose whether an abnormality or the like has occurred in the plant 30. That is, the abnormality diagnosis unit 202 calculates whether H(x (p) ) is either +1 or -1, and diagnoses the occurrence of an abnormality or the like in the plant 30 based on that value.

[0033] Step S103: Next, the abnormality diagnosis unit 202 determines whether it is diagnosed in step S102 above that an abnormality or the like has occurred in the plant 30. If it is diagnosed that an abnormality or the like has occurred in the plant 30, the process proceeds to step S104; otherwise, the process of this repetition ends.

[0034] Step S104: When it is diagnosed in step S103 above that an abnormality or the like has occurred in the plant 30, the operation amount calculation unit 203 calculates a plurality of optimal operation amounts for returning the plant 30 from the abnormal state to the normal state. That is, the operation amount calculation unit 203 uses the multi-objective optimization technique to minimize two indicators, namely, the magnitude of the operation amount and the degree of deviation from the past normal operation data, while satisfying H(x (p) +a (s) ) = +1, and calculates the Pareto optimal solutions of S optimal operation amounts a (s) (s = 1, ···, S). The calculation method of the S optimal operation amounts a (s) (s = 1, ···, S) will be described later.

[0035] Step S105: The output unit 204 outputs the S optimal operation amounts a (s) (s = 1, ···, S) calculated in step S104 above to a predetermined output destination and presents them to the operator or the like of the plant 30. As a result, the operator or the like can, for example, set the operation amount selected from among the S optimal operation amounts a (s) (s = 1, ···, S) or the operation amount determined with reference to the S optimal operation amounts a (s) (s = 1, ···, S) in the control device 20, and the plant 30 can be controlled according to this operation amount.

[0036] <Calculation method of a plurality of optimal operation amounts> Hereinafter, in step S104 above, using the multi-objective optimization technique, the method of calculating S optimal operation amounts a (p) +a (s) (s = 1, ···, S) that minimize two indicators, namely, the magnitude of the operation amount and the degree of deviation from the past normal operation data, and satisfy H(x (s) (s = 1, ···, S) will be described. Note that S is an integer of 2 or more determined in advance.

[0037] Let d(·) be a distance function (for example, Euclidean distance, Manhattan distance, Mahalanobis distance, etc.). Also, let X be the set of past operation data of the plant 30 during normal operation, that is, X = {x (n) |H(x (n)) = +1, n = 1, ···, N = |X|}. Let q k (x (p) +a (s) |X) be the outlier degree of x when X is given. (p) +a (s) That is, it is the outlier degree of

[0038] As the outlier degree q k The local outlier factor (LOF) can be used. The local outlier factor q k (x|X) when X is given can be calculated as follows.

[0039]

Equation

[0040]

Equation

[0041]

Equation

[0042] At this time, in this calculation method, S operation amounts a (p) +a) that satisfy H(x k (x (p) +a|X) and minimize d(a) and q (s) (s = 1, ···, S) are calculated by multi-objective optimization technology. That is, Min a (d(a), q k (x (p) +a|X)) s.t. H(x (p)By solving the multi-objective optimization problem of +a)=+1, S Pareto optimal solutions are calculated as the optimal operation amount a (s) (s = 1, ···, S). Here, a is a variable representing the operation amount.

[0043] In this way, considering two indicators, namely the magnitude of the operation amount and the degree of deviation from the past normal operation data, a plurality of optimal operation amounts can be calculated. By considering the magnitude of the operation amount, it becomes possible to return from an abnormal state to a normal state with a small operation amount based on the current state of the plant 30. On the other hand, by considering the degree of deviation from the past normal operation data, it becomes possible to return from an abnormal state to a normal state with an operation amount that is highly feasible in view of past normal operations.

[0044] As an example, when S = 3, M = 2, R = {x=(x 1 ,x 2 )|H(x)=+1}, H(·) is a linear model, and the set E of operation data 1 ⊂X has a high density, and the set E of operation data 2 ⊂X has a low density, a calculation example of the optimal operation amount a (s) (s = 1, 2, 3) is shown in Fig. 5. In the example shown in Fig. 5, the operation amount a (1) such that the point P (p) = x (1) + a (1) , the operation amount a (2) such that the point P (p) = x (2) + a (2) , and the operation amount a (3) such that the point P (p) = x (3) + a (3) are calculated as the optimal operation amounts. Hereinafter, when a (s) is taken as the operation amount that is the Pareto optimal solution, the point P (s) = x (p) + a (s) will also be referred to as the Pareto optimal operating point.

[0045] The Pareto optimal operating points P (1) , P (2) , P (3)The relationship is shown in Fig. 6. As shown in Fig. 6, the Pareto-optimal operating points P (1) , P (2) , P (3) are all points on the Pareto curve. However, the Pareto-optimal operating point P (1) has the highest density in the neighborhood (that is, the outlier degree q k (x (p) +a (1) |X) is the smallest), while the magnitude of the operation amount a (1) (that is, d(a (1) )) is the largest. The Pareto-optimal operating point P (2) has the smallest density in the neighborhood, while the magnitude of the operation amount a (2) is the smallest. The Pareto-optimal operating point P (3) has a medium level both in terms of the density in the neighborhood and the magnitude of the operation amount a (3) . Hereinafter, the two-dimensional space with density and distance as axes will be referred to as the distance-density space.

[0046] Accordingly, the operator can select a desired operation from among the optimal operation amounts a (s) (s = 1, 2, 3) according to how much importance is attached to either the magnitude of the operation amount or the degree of deviation from the normal operation data in the past, or can determine an appropriate operation with reference to the optimal operation amounts a (s) (s = 1, 2, 3).

[0047] Note that in this calculation method, the range of values that the operation amount a can take is not particularly limited. However, for example, the range of values that the operation amount a can take may be limited to grid points.

[0048] ≪Calculation process of the optimal operation amount a (s) (s = 1, ···, S)≫ Hereinafter, an example of the process of calculating the optimal operation amount a (s) (s = 1, ···, S) will be described with reference to Fig. 7. Fig. 7 is a flowchart showing an example of the optimal operation calculation process according to this embodiment.

[0049] Step S201: The operation amount calculation unit 203 sets an arbitrary point in the distance-density space as a reference point. The reference point is a point that serves as a reference when calculating the hypervolume described later. As the reference point, for example, a point randomly selected in the distance-density space may be used. However, it is preferable that the reference point is not on the Pareto curve, has a certain degree of low density, and has a certain degree of large distance.

[0050] Step S202: Next, the operation amount calculation unit 203 generates a set A = {a (s) | s = 1, ···, S} composed of S search points. At this time, the operation amount calculation unit 203, for example, randomly selects S points that satisfy H(x 1 , ···, x M + a (p) + a (s) ) = +1 within the space spanned by the state variables x (s) (hereinafter referred to as the state variable space) to generate each search point a

[0051] . Hereinafter, A will be referred to as the search point set.

[0052] Step S203: Next, the operation amount calculation unit 203 initializes the ranked extraction point set Z = φ. The ranked extraction point set Z is a set composed of ranked extraction points (described later). Also, the rank is an evaluation index for evaluating the goodness of the Pareto optimal solution. Hereinafter, the rank is an integer of 1 or more, and it is assumed that the smaller the value, the better the Pareto optimal solution.

[0053] Step S204: Next, the operation amount calculation unit 203 extracts the Pareto optimal solution from the search point set A as an extraction point. The operation amount calculation unit 203 may extract the extraction point, for example, according to the following procedures 1-1 to 1-3. (s) For each a (s) ∈ A, the operation amount calculation unit 203 calculates d(a k ) and q (p) (x (s) + a

[0054] Step 1-2: The operation amount calculation unit 203 determines whether there exists an a (s) ∈A that satisfies any one of the following Conditions 1 to 3 for each a (s') ∈A (where s≠s').

[0055] Condition 1) d(a (s) ) > d(a (s') ), and q k (x (p) +a (s) |X) > q k (x (p) +a (s') |X) Condition 2) d(a (s) ) > d(a (s') ), and q k (x (p) +a (s) |X) = q k (x (p) +a (s') |X) Condition 3) d(a (s) ) = d(a (s') ), and q k (x (p) +a (s) |X) > q k (x (p) +a (s') |X) Step 1-3: The operation amount calculation unit 203 extracts these a (s') ∈A that are determined not to exist in a (s) ∈A in Step 1-2 above as Pareto optimal solutions and uses these a (s) ∈A as extraction points.

[0056] Hereinafter, the set of extraction points extracted in Step S204 above will be denoted as B⊆A and referred to as the extraction point set.

[0057] Step S205: Next, the operation amount calculation unit 203 ranks the extraction point a (s) ∈B extracted in Step S204 above. The operation amount calculation unit 203 may rank the extraction point a (s) ∈B, for example, according to the following Steps 2-1 to 2-4.

[0058] Step 2-1: The operation amount calculation unit 203 uses d(a (s) ) and q (s) (x k +a (p) |X) for each extraction point a (s) ∈B and the reference point set in the above step S201 to calculate the hypervolume HV. The hypervolume refers to the area or (hyper)volume of a plane or (hyper)solid formed by each extraction point a (s) and the reference point. For details of the hypervolume, for example, refer to Reference 1 etc. In this embodiment, the hypervolume HV is defined as the area of a plane represented by the sum of planes with the extraction point a (s) and the reference point as diagonal vertices in the distance-density space.

[0059] As an example, the hypervolume HV when the extraction points extracted in the above step S204 are B = {a (1) , a (2) , a (3)}, and the operating point corresponding to the extraction point a (s) is Q (s) =x (p) +a (s) is shown in FIG. 8. As shown in FIG. 8, the hypervolume HV is the area of a plane calculated by the point (d(a (s) ), q k (x (p) +a (s) |X)) and the reference point.

[0060] Step 2-2: The operation amount calculation unit 203 uses d(a (s) ) and q (s') (x (s) +a (s') |X) for each extraction point a k ∈B and the reference point set in the above step S201 to calculate the hypervolume HV (p) (s') (s) for each extraction point a (s) ∈B\{a (s)}. That is, the operation amount calculation unit 203 calculates the hypervolume HV for each extraction point a (s)The hypervolume HV when removed from the extraction point set B (s) is calculated.

[0061] Step 2-3: The operation amount calculation unit 203 calculates, for each extraction point a (s) ∈ B, HV - HV (s) as the contribution degree C (s) to the hypervolume.

[0062] As an example, an example of the contribution degree C (2) to the hypervolume of a ∈ B is shown in FIG. 9. As shown in FIG. 9, the contribution degree C (2) to the hypervolume is calculated by C (2) = HV - HV (2) . (2) is calculated by.

[0063] Step 2-4: The operation amount calculation unit 203 ranks the extraction points a (s) ∈ B in descending order of the contribution degree C (s) to the hypervolume, using the number of elements |Z| of the ranked extraction point set Z and the contribution degree C (s) to the hypervolume. That is, the operation amount calculation unit 203 ranks the extraction points a (s) ∈ B in descending order of the contribution degree C (s) to the hypervolume, starting from |Z| + 1 in order.

[0064] Step S206: Next, the operation amount calculation unit 203 adds the ranked extraction points a (s) ∈ B ranked in step S205 above to the ranked extraction point set Z. If the rank of the extraction point a (s) ∈ B is r (s) , the ranked extraction point can also be expressed as (a (s) , r (s) ), but hereinafter, it is assumed that the ranked extraction point set Z is an ordered set in which the order between extraction points is defined by the rank. That is, the ranked extraction point set Z is the extraction point a (s)Based on a set of elements, assume that an order is defined by rank between any two elements. When comparing the ranks of two extraction points, the extraction point with the smaller rank value is called the "extraction point with a higher rank", and the extraction point with the larger rank value is called the "extraction point with a lower rank".

[0065] Step S207: Next, the operation amount calculation unit 203 determines whether all the search points in the search point set A have been extracted as extraction points. That is, when treating Z as just a set, the operation amount calculation unit 203 determines whether A\Z = φ.

[0066] If it is not determined in step S207 that all the search points in the search point set A have been extracted as extraction points (NO in step S207), the operation amount calculation unit 203 proceeds to the process of step S208. On the other hand, if it is determined in step S207 that all the search points in the search point set A have been extracted as extraction points (YES in step S207), the operation amount calculation unit 203 proceeds to the process of step S209.

[0067] Step S208: Next, the operation amount calculation unit 203 excludes the extraction points included in the extraction point set B from the search point set A and returns to step S204 above. That is, the operation amount calculation unit 203 sets A←A\B and returns to step S204 above. As a result, after excluding the extraction points included in the extraction point set B, the processes after step S204 above are executed again using the search point set A. Therefore, until all the search points in the search point set A are extracted as extraction points, the processes of step S204 to step S206 above are repeatedly executed, and all the search points included in the search point set A will be ranked.

[0068] Step S209: Next, the operation amount calculation unit 203 updates all or some of the ranked extraction points a (s) ∈Z included in the ranked extraction point set Z. The operation amount calculation unit 203 updates the ranked extraction point a, for example, by Differential Evolution described in Reference 1 (s)∈Z may be updated, or the ranked extraction point a may be obtained by metaheuristics technology (s) ∈Z may be updated. Alternatively, for example, the ranked extraction point a may be updated by one or more of the following update methods 1 to 3 (s) ∈Z may be updated.

[0069] Update method 1) For the ranked extraction point a ∈ Z whose rank value is less than a predetermined threshold (that is, the extraction point a ∈ Z with a high rank (s) ∈Z), leave it as it is without updating. On the other hand, for the ranked extraction point a ∈ Z whose rank value is greater than or equal to the threshold (that is, the extraction point a ∈ Z with a low rank (s) ∈Z), translate it by a predetermined amount in the direction of the extraction point a ∈ Z with the highest rank (s) ∈Z). (s) ∈Z), translate it parallel by a predetermined amount in the direction of the extraction point a ∈ Z with the highest rank (s) ∈Z).

[0070] Update method 2) For each ranked extraction point a (s) ∈Z, translate it randomly by a predetermined amount, and then recalculate the rank of the translated ranked extraction point a' (s) ∈Z. When recalculating the rank, use Z ∪ {a' (s)}\{a (s)} to calculate the hypervolume HV and the contribution degree of each ranked extraction point to the hypervolume, and then rank each ranked extraction point in descending order of the contribution degree to the hypervolume. If the rank of the translated ranked extraction point a' (s) ∈Z is higher than the rank of the ranked extraction point a (s) ∈Z before translation, then adopt the translation and update the ranked extraction point a (s) ∈Z. On the other hand, if the rank of the translated ranked extraction point a' (s) ∈Z is not higher than the rank of the ranked extraction point a (s) ∈Z before translation, then revert the translation and do not update the ranked extraction point a (s) ∈Z.

[0071] However, instead of rank, for example, when the contribution degree to the hypervolume becomes high, translation may be adopted, and if not, it may be reverted to the original translation.

[0072] Update method 3) Ranked extraction point a where the rank value is less than a predetermined threshold (s) For a ∈ Z, it is updated by the above update method 2. On the other hand, for a ranked extraction point a ∈ Z where the rank value is greater than or equal to the threshold, (s) it is updated by the above update method 1.

[0073] By the process of step S209 above, an updated set Z of ranked extraction points a is obtained, where at least some of the ranked extraction points a (s) have been updated.

[0074] Step S210: Next, the operation amount calculation unit 203 determines whether the number of updates in step S209 above has reached a predetermined upper limit.

[0075] If it is determined in step S210 above that the number of updates has not reached the upper limit (NO in step S210), the operation amount calculation unit 203 proceeds to the process of step S211. On the other hand, if it is determined in step S210 above that the number of updates has reached the upper limit (YES in step S210), the operation amount calculation unit 203 proceeds to the process of step S212.

[0076] Step S211: The operation amount calculation unit 203 deletes the rank from the updated set Z of ranked extraction points obtained in step S209 above to obtain a new search point set A, and returns to step S203 above. That is, the operation amount calculation unit 203 deletes the rank from the updated set Z of ranked extraction points, sets A←Z, and returns to step S203 above. As a result, using the new search point set A, the processes after step S203 above are executed again. Therefore, until the number of updates reaches the upper limit, the processes of steps S203 to S209 above are repeatedly executed.

[0077] Step S212: The operation amount calculation unit 203 uses the ranked extraction point a included in the updated ranked extraction point set Z obtained in step S209 above (s) as the final Pareto optimal solution, that is, the optimal operation amount a (s) . As a result, it is considered that S optimal operation amounts a (s) (s = 1, ···, S) have been calculated

[0078] . Note that in step S105 of FIG. 4, the output unit 204 outputs S optimal operation amounts a (s) (s = 1, ···, S) to a predetermined output destination, but this is not the only case. For example, S'(<S) optimal operation amounts a (s) (s = 1, ···, S') may be output to a predetermined output destination in descending order of rank. Also, the output unit 204 may output not only the optimal operation amounts a (s) (s = 1, ···, S), but also their ranks to a predetermined output destination. By outputting the ranks in addition to the optimal operation amounts a (s) (s = 1, ···, S) to a predetermined output destination, an operator or the like can select and determine the operation amount to be set in the control device 20 considering the rank

[0079] <Example> Hereinafter, an example of this embodiment will be described

[0080] In this example, it is assumed that the plant 30 is a waste incineration plant, and the case where S = 3 optimal operation amounts a (s) (s = 1, 2, 3) are presented to an operator or the like will be described

[0081] A schematic diagram of a waste incineration plant is shown in FIG. 10. As shown in FIG. 10, in a waste incineration plant, waste and air are fed into a combustion furnace, and the heat generated by the combustion is converted into steam, and steam and exhaust gases such as carbon monoxide (CO) are output. Generally, since steam is used for power generation and the like, it is required to increase the amount of steam generated and stabilize it. On the other hand, in order to increase the amount of steam generated, it is necessary to increase the amount of waste input and the air flow rate. However, this may cause incomplete combustion, and as a result, the CO concentration may increase. Therefore, it is necessary to appropriately operate the amount of waste input and the air flow rate. The amount of waste input is operated by the operating speed of a facility called a feeder, and the air flow rate is operated by the opening and closing angle of a valve or the like. Also, the steam flow rate and the CO concentration are generally measured not only for the current value but also for the rate of change thereof.

[0082] Therefore, the state variables of the waste incineration plant are as follows.

[0083] x 1 : Steam flow rate (current value) x 2 : Steam flow rate (speed) x 3 : CO concentration (current value) x 4 : CO concentration (speed) x 5 : Air flow rate x 6 : Feeder speed That is, the operation data is x = (x 1 , x 2 , x 3 , x 4 , x 5 , x 6 ). Among the above state variables, those that can be operated by an operator or the like are the air flow rate x 5 and the feeder speed x 6 . Therefore, the manipulated variable is a = (0, 0, 0, 0, a 5 , a 6 ). Here, a 5 is the manipulated variable for the air flow rate, and a 6 is the manipulated variable for the feeder speed.

[0084] Thus, for example, when using the Euclidean distance as d(·), Min a ((a 5 2 +a 6 2 ) 1 / 2 ,q k (x (p) +a|X)) s.t. H(x (p) +a)=+1, by solving the multi-objective optimization problem, three Pareto optimal solutions can be calculated as the optimal operation amounts a (s) (s = 1, 2, 3).

[0085] As an example, assume that the Pareto optimal operating points P (2) =x (p) +a (2) (s = 1, 2, 3) shown in Fig. 11 are obtained. The optimal operation amount a (1) has the smallest magnitude of the operation amount while the density in the vicinity is the smallest. The optimal operation amount a (2) has a medium magnitude of the operation amount and a medium density in the vicinity. The optimal operation amount a (3) has the largest magnitude of the operation amount but the highest density in the vicinity.

[0086] At this time, an example of the steam flow rate and CO concentration when controlling the waste incineration plant with each operation amount a (s) (s = 1, 2, 3) is shown in Fig. 12. As shown in Fig. 12(a), when operating with the operation amount a (1) , since the operation is to slightly increase the air input amount and keep the waste input amount unchanged, combustion is slightly promoted and the steam flow rate using the heat from combustion also increases slightly. On the other hand, the CO concentration in the exhaust gas does not increase. Also, as shown in Fig. 12(b), when operating with the operation amount a (2) , since the operation is to slightly increase the air input amount and also increase the waste input amount, the fuel required for combustion increases and the steam flow rate increases. On the other hand, since the amount of air is insufficient, the CO concentration due to incomplete combustion increases. Furthermore, as shown in Fig. 12(c), when operating with the operation amount a (3)When the operation is performed, the amount of air input and the amount of waste input will both increase significantly, resulting in an increase in the fuel required for combustion and an increase in the steam flow rate. On the other hand, the CO concentration does not increase, but the amount of operation to change from the current operating state becomes large.

[0087] In this way, three optimal operation amounts a (s) (s = 1, 2, 3) that are difficult to distinguish as superior or inferior to each other are presented to the operator or the like. Therefore, the operator can select a desired operation amount from these operation amounts a (s) (s = 1, 2, 3), or determine an appropriate operation amount with reference to these operation amounts a (s) (s = 1, 2, 3), and it becomes possible to perform a desired operation according to the situation.

[0088] <Other calculation methods for multiple optimal operation amounts> In the above embodiment, S optimal operation amounts a (s) (s = 1, ···, S) were calculated by the multi-objective optimization technique. However, for example, if there are S adjusted weight parameters λ 1 , ···, λ S existing, these weight parameters λ 1 , ···, λ S can be used to calculate S optimal operation amounts a (s) (s = 1, ···, S) by a method similar to the method described in Patent Document 1. That is, for s = 1, ···, S, Min a d(a) + λ s q k (x (p) + a|X) s.t. H(x (p) + a) = +1, and the optimal operation amount a (s) can be calculated. Also, these S optimal operation amounts a (s) (s = 1, ···, S) can be evaluated by the contribution degree to the hypervolume, ranked, and then presented to the operator or the like.

[0089] <Summary> As described above, when an abnormality or the like occurs in the plant 30, the operation support device 10 according to the present embodiment can present a plurality of optimal operation amounts to an operator or the like for returning the plant 30 from the abnormality to the normal state. Therefore, the operator or the like selects a desired operation from among the plurality of optimal operations, or determines an appropriate operation with reference to the plurality of optimal operations, and then actually sets the operation in the control device 20, and it becomes possible to control the plant 30 to return to the normal state.

[0090] In the above embodiment, the S optimal operation amounts a (s) (s = 1, ···, S) are presented to an operator or the like. In addition to this, for example, the operation order may also be presented to the operator or the like after determining the operation order by a method similar to the method described in Patent Document 1.

[0091] The present invention is not limited to the above specifically disclosed embodiments, and various modifications, changes, combinations with known technologies, etc. are possible without departing from the description of the claims.

[0092] [References] Reference 1: Daichi Ueyama, Kenichi Tamura, Keiichiro Yasuda, Extension of Differential Evolution to Multi-objective Optimization Problems Using Local Descent Direction Vectors, Transactions of the Institute of Electrical Engineers of Japan, C, Vol. 132 No. 8 pp. 1356-1361 (2012)

Explanation of Reference Signs

[0093] 1 Plant control system 10 Operation support device 20 Control device 30 Plant 101 Input device 102 Display device 103 External I / F 103a Recording medium 104 Communication I / F 105 RAM 106 ROM 107 Auxiliary storage device 108 Processor 109 Bus 201 Input Unit 202 Abnormality Diagnosis Unit 203 Operation Amount Calculation Unit 204 Output Unit 205 Database

Claims

1. An abnormality diagnosis unit that diagnoses whether an abnormality or a sign of the abnormality has occurred in the control target using the operation data of the control target; An operation amount calculation unit that calculates a plurality of optimal operations for returning the control target to normal when it is diagnosed that an abnormality or a sign of the abnormality has occurred in the control target; An output unit that outputs the plurality of optimal operations to a predetermined output destination; An operation support device having the above.

2. The operation amount calculation unit: Calculates a plurality of operations such that the control target becomes normal and the magnitude of the operation amount and the degree of deviation from the past operation data when the control target is normal are small as the plurality of optimal operations. The operation support device according to claim 1.

3. The operation amount calculation unit: Calculates a Pareto optimal solution that minimizes the magnitude of the operation amount and the degree of deviation using a multi-objective optimization technique as the plurality of optimal operations. The operation support device according to claim 2.

4. The operation amount calculation unit: Ranks the Pareto optimal solutions using the contribution degree to the hypervolume, Calculates the ranked Pareto optimal solutions as the plurality of optimal operations. The operation support device according to claim 3.

5. The output unit: Outputs a predetermined number of optimal operations to the output destination in descending order of the ranks among the plurality of optimal operations. The operation support device according to claim 4.

6. The output unit: Outputs the plurality of optimal operations and a plurality of ranks corresponding to each of the plurality of optimal operations to the output destination. The operation support device according to claim 4.

7. The output unit: Outputs the plurality of optimal operations to a display device included in the operation support device or a terminal used by an operator of the control target. The operation support device according to claim 1.

8. The abnormality diagnosis unit: Diagnoses whether an abnormality or a sign of the abnormality has occurred in the control target using a machine learning model considering the time delay with respect to the operation of the control target and the operation data. The operation support device according to any one of claims 1 to 7.

9. An abnormality diagnosis procedure for diagnosing whether an abnormality or a sign of the abnormality has occurred in the control target using the operation data of the control target; An operation amount calculation procedure for calculating a plurality of optimal operations for returning the control target to normal when it is diagnosed that an abnormality or a sign of the abnormality has occurred in the control target; An output procedure for outputting the plurality of optimal operations to a predetermined output destination; A driving assistance method executed by a computer.

10. An abnormality diagnosis procedure for diagnosing whether an abnormality or a sign of the abnormality has occurred in the control target using the driving data of the control target, An operation amount calculation procedure for calculating a plurality of optimal operations for returning the control target to normal when it is diagnosed that an abnormality or a sign of the abnormality has occurred in the control target, An output procedure for outputting the plurality of optimal operations to a predetermined output destination, A program for causing a computer to execute the above.

Images