WO2021124861A1 - Dispositif de traitement d'informations, procédé de traitement d'informations et programme de traitement d'informations - Google Patents

Dispositif de traitement d'informations, procédé de traitement d'informations et programme de traitement d'informations Download PDF

Info

Publication number
WO2021124861A1
WO2021124861A1 PCT/JP2020/044600 JP2020044600W WO2021124861A1 WO 2021124861 A1 WO2021124861 A1 WO 2021124861A1 JP 2020044600 W JP2020044600 W JP 2020044600W WO 2021124861 A1 WO2021124861 A1 WO 2021124861A1
Authority
WO
WIPO (PCT)
Prior art keywords
prediction
control
model
unit
prediction model
Prior art date
Application number
PCT/JP2020/044600
Other languages
English (en)
Japanese (ja)
Inventor
通孝 古林
佐藤　拓朗
常平 山本
美香 西尾
央紗 阪口
Original Assignee
日立造船株式会社
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 日立造船株式会社 filed Critical 日立造船株式会社
Priority to CN202080088498.5A priority Critical patent/CN114868090A/zh
Publication of WO2021124861A1 publication Critical patent/WO2021124861A1/fr

Links

Images

Classifications

    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F23—COMBUSTION APPARATUS; COMBUSTION PROCESSES
    • F23G—CREMATION FURNACES; CONSUMING WASTE PRODUCTS BY COMBUSTION
    • F23G5/00—Incineration of waste; Incinerator constructions; Details, accessories or control therefor
    • F23G5/50—Control or safety arrangements
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators

Definitions

  • the present invention relates to an information processing device and the like used for updating a prediction model constructed by machine learning.
  • Patent Document 1 describes that the waste treatment plant is automatically controlled by outputting the amount of steam generated by the waste heat boiler after a predetermined time, which is predicted by the neural network model, to the control logic. Has been done.
  • Patent Document 1 When using a prediction model constructed by machine learning, including a neural network model, it is necessary to update the prediction model by re-learning or the like in order to maintain or improve the prediction accuracy.
  • a new model is constructed during the period of use of one model, and if the evaluation index of the new model is better than that of the model in use, the new model is applied. Is described.
  • the prediction model that predicts the appropriate control content is information indicating the control content to be performed when the device is in the operating state based on the input data indicating the operating state of the device itself or the facility in which the device is used. It is a prediction model to output.
  • the control based on the prediction result of the second prediction model is not performed, the control based on the prediction result of the first prediction model is performed, and the state of the plant is changed by the control. To do.
  • the prediction model cannot be appropriately updated at the end of the predetermined period.
  • One aspect of the present invention has been made in view of the above-mentioned problems, and an object of the present invention is to provide an information processing apparatus or the like capable of appropriately updating a prediction model for predicting control contents. ..
  • the information processing apparatus uses a plurality of prediction models for predicting the control contents to be performed on the controlled object, and a plurality of units constituting the first predetermined period.
  • the evaluation unit that evaluates the prediction result in each unit period when the control target is controlled by taking turns every period, and the above-mentioned first from a plurality of the above prediction models based on the evaluation result by the evaluation unit. It includes a selection unit for selecting a prediction model to be used in a second predetermined period after the predetermined period of 1.
  • the information processing method is an information processing method executed by one or a plurality of information processing devices in order to solve the above-mentioned problems, and is a control content to be performed on a controlled object.
  • the evaluation step for evaluating the prediction result in each unit period when the control target is controlled by alternately using a plurality of prediction models constituting the first predetermined period for each of a plurality of unit periods, and the above.
  • the selection step includes a selection step of selecting a prediction model to be used in the second predetermined period after the first predetermined period from the plurality of prediction models based on the evaluation result in the evaluation step.
  • FIG. 1 is a block diagram showing an example of a main configuration of the information processing device 1.
  • the information processing device 1 includes a control unit 10 that controls and controls each unit of the information processing device 1, and a storage unit 12 that stores various data used by the information processing device 1. .
  • the information processing device 1 includes an input unit 14 for receiving an input to the information processing device 1 and an output unit 16 for the information processing device 1 to output data.
  • the control unit 10 includes an update management unit 101, a prediction unit 102, a plant control unit 103, an evaluation unit 104, a selection unit 105, a teacher data generation unit 106, a learning unit 107, a manual control detection unit 108, and a reference information acquisition unit 109.
  • the state determination unit 110 is included.
  • the storage unit 12 stores the prediction result DB (database) 121, the prediction model DB 122, and the correct answer data determination information 123.
  • the teacher data generation unit 106 to the state determination unit 110 and the correct answer data determination information 123 will be described later.
  • the information processing device 1 is a device that realizes an automatic control system that predicts the control content to be performed on a control target using a prediction model and operates the control target according to the prediction result.
  • the above prediction model is a model constructed so that the control content to be performed on the controlled object can be predicted.
  • the control target is various devices used in the plant. That is, the information processing apparatus 1 controls various devices and the like included in the plant based on the prediction result of the prediction model, and automatic control of the plant is realized.
  • the control target is arbitrary and is not limited to this example.
  • the above prediction model may be constructed by a statistical method, may be constructed by machine learning, or may be a model in which they are combined.
  • the machine learning algorithm is arbitrary, and a prediction model such as a neural network may be used.
  • the update management unit 101 switches the prediction model used for automatic control of the controlled object every unit period (for example, one day). Further, the update management unit 101 updates the prediction model used for automatic control when a predetermined period including a plurality of unit periods (first predetermined period, for example, 60 days) has elapsed. Then, in the predetermined period after the update (second predetermined period, for example, 60 days), automatic control using the updated prediction model is performed.
  • the evaluation unit 104 and the selection unit 105 are involved in updating the prediction model.
  • the prediction unit 102 predicts the control content to be performed on the control target using the above-mentioned prediction model.
  • the above-mentioned prediction model includes various data indicating the control target and the operating state of the plant (for example, process data such as sensor detection values) and the control contents to be executed when the data is observed. This is a learning of the correspondence between. Therefore, the prediction unit 102 inputs the above-mentioned various data into the prediction model, and predicts the control content to be performed on the control target based on the output value of the prediction model.
  • the plant control unit 103 controls the operation of various devices used in the plant according to the prediction result of the prediction unit 102.
  • the plant control unit 103 may directly control the above-mentioned equipment, or may indirectly control the equipment by notifying the control device of the equipment of the control contents.
  • the evaluation unit 104 evaluates the prediction result in each unit period when the control target is controlled by alternately using the plurality of prediction models for each unit period.
  • the evaluation method may be any method that can compare the prediction accuracy of each prediction model in a unit period. Specific examples of the evaluation method will be described later.
  • the selection unit 105 selects a prediction model to be used by the prediction unit 102 thereafter from the plurality of prediction models based on the evaluation result by the evaluation unit 104. That is, the selection of the selection unit 105 updates the prediction model used by the prediction unit 102.
  • the information processing device 1 alternately uses a plurality of prediction models for predicting the control contents to be performed on the control target for each of a plurality of unit periods constituting the first predetermined period, and uses the control target.
  • the evaluation unit 104 for evaluating the prediction result in each unit period when controlled is provided.
  • the information processing device 1 is a selection unit that selects a prediction model to be used in the second predetermined period after the first predetermined period from the plurality of prediction models based on the evaluation result by the evaluation unit 104. It is equipped with 105.
  • the prediction results in each unit period when the control target is controlled by alternately using a plurality of prediction models for each unit period are evaluated, and based on the evaluation results, in the second predetermined period. Select the prediction model to use. Therefore, a plurality of prediction models can be evaluated under substantially equal conditions, which makes it possible to appropriately update the prediction model that predicts the control content.
  • the prediction result DB 121 is a database that stores the prediction result of the prediction unit 102 and various data related thereto. More specifically, in the prediction result DB 121, the input data input to each prediction model and the date and time of the prediction made based on the input data (or the acquisition date and time of the input data) are stored in association with each other. Has been done. Then, among the stored input data, the input data that is the basis of the prediction control is associated with the data indicating the content of the prediction control. Further, in the prediction result DB 121, the data indicating the content of the manual control manually performed by the operator with respect to the control target and the date and time when the manual control is performed are stored in association with each other.
  • the "input data on which the prediction control is based" refers to the input data input to the prediction model, which is the output data indicating that the prediction control should be performed, which is output from the prediction model. ..
  • the prediction model DB 122 is a database that stores the prediction model used by the prediction unit 102. Although the details will be described later, the prediction model DB 122 stores a plurality of prediction models constructed in advance, and also stores the prediction model constructed by the learning unit 107.
  • FIG. 2 is a diagram showing an example of switching control and updating the prediction model to be used.
  • the update management unit 101 switches the prediction model used by the prediction unit 102 between the prediction model of the A group and the prediction model of the B group every day.
  • the plant control unit 103 controls the operation of various devices included in the plant according to the prediction result of the prediction unit 102.
  • the plant is a waste incinerator plant.
  • the prediction unit 102 makes a prediction using the prediction model of the group A, and the plant control unit 103 controls the operation of the waste incinerator plant according to this prediction result. Further, on the second day, the prediction unit 102 makes a prediction using the prediction model of the group B, and the plant control unit 103 controls the operation of the waste incinerator plant according to the prediction result.
  • the prediction model of group A includes two abnormality avoidance models A1 and A2 and four normal maintenance models A3 to A6.
  • the prediction model of group B includes two abnormality avoidance models B1 and B2 and four normal maintenance models B3 to B6. Since the prediction model of the group A and the prediction model of the group B are used alternately, the prediction targets of the A1 to A6 prediction models and the B1 to B6 prediction models are the same, respectively.
  • the abnormality avoidance model is a prediction model that predicts that the plant will not be in a normal operating state (for example, an abnormality will occur) and predicts the control contents to be performed in order to avoid it.
  • At least one abnormality avoidance model may be prepared for each type of abnormality to be avoided.
  • the abnormality avoidance models A1 and B1 predict the detection of "small amount of waste abnormality", which is the deterioration of the combustion state caused by the decrease in the amount of waste in the incinerator, and the control contents for avoiding it. It may be a predictive model.
  • abnormality avoidance models A2 and B2 are used to detect "flame-retardant waste abnormality", which is a deterioration of the combustion state caused by the introduction of low-calorie waste into the incinerator, and the control contents to avoid it. It may be a prediction model for prediction.
  • the normal maintenance model is a prediction model that predicts the control contents to be performed in order to maintain the normal operation of the waste incinerator.
  • the normal maintenance model may be prepared for each device controlled by the plant control unit 103 in the waste incinerator plant.
  • the normal maintenance models A5 and B5 may be used as predictive models for controlling a dust supply device that feeds dust into an incinerator.
  • the normal maintenance models A6 and B6 may be used as a prediction model for controlling the operation of the grate that conveys the waste in the incinerator.
  • individual prediction models may be prepared for each grate.
  • the "normal” condition and the “abnormal” condition may be set in advance.
  • a normal range is set for the output values of various sensors installed in various places in the plant, and it is defined as a "normal” state that the output values of all sensors are within the normal range, and at least a part of them.
  • the fact that the output value of the sensor is out of the normal range may be defined as "abnormal”.
  • the type of "abnormality" may be specified according to the type of the sensor whose output value is out of the normal range.
  • the evaluation unit 104 evaluates the prediction results for each of the plurality of abnormality avoidance models and the plurality of normal maintenance models. .. Then, the selection unit 105 selects the abnormality avoidance model and the normal maintenance model to be used from the next time onward based on the evaluation result by the evaluation unit 104. This makes it possible to properly update both the anomaly avoidance model and the normal maintenance model. Then, by using these appropriately updated prediction models, stable automatic control of the plant becomes possible.
  • the update management unit 101 updates the forecast model to be used.
  • the update management unit 101 updates the prediction model to be used from the A group and the B group to the C group and the D group.
  • Group C consists of a prediction model selected by selection unit 105 from groups A and B based on the evaluation results evaluated by evaluation unit 104 for 60 days of prediction results by prediction models of groups A and B.
  • the evaluation unit 104 evaluates the abnormality avoidance model by the abnormality avoidance rate, and evaluates the normal maintenance model by the normal state maintenance rate.
  • the evaluations of the first abnormality avoidance models A1 and B1 are 79% and 70%, respectively, and the abnormality avoidance model A1 is higher. Therefore, the selection unit 105 selects the abnormality avoidance model A1 as the first abnormality avoidance model in the C group.
  • the evaluations of the second abnormality avoidance models A2 and B2 are 75% and 78%, respectively, and the abnormality avoidance model B2 is higher. Therefore, the selection unit 105 selects the abnormality avoidance model B2 as the second abnormality avoidance model in the C group.
  • the selection unit 105 also selects the predicted model of the A group and the B group, which has a higher evaluation, in the same manner as described above. As a result, the C group consisting of the prediction models that were highly evaluated in the A group and the B group is constructed.
  • the D group consists of a prediction model newly constructed by the learning unit 107 based on the data collected during the 60 days when the prediction control by the prediction models of the A group and the B group was performed.
  • the D group is composed of the newly constructed prediction models D1 to D6.
  • machine learning only the above data collected in 60 days may be used, or the data collected in other periods may also be used.
  • machine learning may be performed using both the data used for constructing the prediction model of group A or group B and the above data collected in 60 days.
  • the operation control of the waste incinerator plant is performed by alternately using the C group prediction model and the D group prediction model every day for 60 days. .. Further, after the prediction control for 60 days is performed using the prediction models of the C group and the D group, a new prediction model is constructed and the prediction model to be used is updated in the same manner as described above.
  • the evaluation unit 104 also evaluates the prediction result when the prediction model of the C group selected by the selection unit 105 and the newly constructed prediction model of the D group are alternately used for control. Then, the selection unit 105 is used in a subsequent period (a third predetermined period after the second predetermined period in which the prediction models of the C group and the D group are used) based on the evaluation result by the evaluation unit 104. Select a predictive model.
  • the prediction model selected by the selection unit 105 was highly evaluated in the A group and the B group, and highly accurate prediction control can be expected even in the subsequent period. Further, the newly constructed prediction model reflects data that is not used for constructing the prediction model selected by the selection unit 105. Therefore, according to the above configuration, it is possible to select a prediction model that can be expected to perform highly accurate prediction control in the subsequent period while considering the newly acquired data.
  • the information processing apparatus 1 it is possible to stably control the operation of the waste incinerator while appropriately updating the prediction model to be used on a regular basis. Since the group after Group C is a prediction model group newly constructed at the time of renewal, the validity of the prediction should be verified by performing a simulation before using it for the actual operation of the waste incinerator plant. Is preferable.
  • the prediction using each prediction model may be performed even during the period when the result of the prediction is not used for control.
  • the prediction model of group B is not used for control on the first day.
  • the prediction unit 102 may also perform prediction using the prediction model of group B in parallel with the prediction using the prediction model of group A, and record the prediction result in, for example, the prediction result DB 121 or the like. .. Such records can be used for verification of the validity of prediction control using the prediction model of group A.
  • a plurality of devices used in the plant are included in the control target, and the plurality of devices are controlled according to the prediction result of the prediction model corresponding to each device.
  • the evaluation unit 104 may evaluate the prediction result of the prediction model corresponding to each of the plurality of devices.
  • the selection unit 105 selects a prediction model to be used in a predetermined period thereafter for each of the plurality of the above-mentioned devices based on the evaluation result by the evaluation unit 104.
  • the quality of the waste to be incinerated may change from day to day. For example, there are days when a lot of garbage that burns stably is contained, and there are days when a lot of garbage that burns unstable is contained.
  • the combustion stability of waste may change depending on the season. Therefore, the difference in the quality of the incinerated waste may appear as the difference in the evaluation of the prediction model.
  • the evaluation result by the evaluation unit 104 reflects the difference in the performance of the prediction model itself as it is. I can't say I'm doing it.
  • the difference in waste quality is less likely to appear as the difference in the evaluation of the prediction model. This enables appropriate updates based on the difference in performance of the prediction model itself.
  • FIG. 3 is a flowchart showing an example of a process of updating the prediction model to be used.
  • the update management unit 101 determines whether or not the prediction model to be used needs to be updated. Specifically, the update management unit 101 determines whether or not the usage period of the prediction model currently in use has reached a predetermined period. This predetermined period may be appropriately set, and may be 60 days as shown in FIG. 2, for example.
  • the update management unit 101 determines that renewal is required (YES in S1), and then the process proceeds to S2.
  • the update management unit 101 determines that the update is unnecessary (NO in S1) when the predetermined period has not been reached, and in this case, determines S1 again after the predetermined time.
  • the evaluation unit 104 evaluates the prediction result of the prediction model in the predetermined period. Specifically, since the prediction control is performed by alternately using a plurality of prediction models for each of a plurality of unit periods (one day in the example of FIG. 2) in the predetermined period, the evaluation unit 104 performs each unit period. Evaluate the prediction results in.
  • the selection unit 105 selects a prediction model to be used for prediction control from the next time onward from the plurality of prediction models based on the evaluation result of S2.
  • the selection unit 105 may select each of a plurality of types of prediction models, as in the example in which the C group is constructed from the prediction models selected from the A group and the B group in FIG.
  • the teacher data generation unit 106 generates teacher data to be used for constructing a new prediction model by using the data acquired in the predetermined period.
  • the method of generating teacher data will be described later.
  • the learning unit 107 builds a new prediction model by machine learning using the teacher data generated in S4.
  • the prediction model of group D in the example of FIG. 2 corresponds to the prediction model constructed in S5.
  • the update management unit 101 designates the prediction model selected in S3 and the prediction model constructed in S5 as prediction models to be used by the prediction unit 102 for prediction control from the next time onward. After S6, the process returns to S1.
  • the information processing apparatus 1 includes a teacher data generation unit 106, a learning unit 107, a manual control detection unit 108, a reference information acquisition unit 109, and a state determination unit 110. Further, the storage unit 12 stores the correct answer data determination information 123.
  • the teacher data generation unit 106 associates the input data input to the prediction model with the correct answer data indicating the content of the control to be executed, and uses it as the teacher data for constructing the prediction model.
  • the control to be executed should be executed in the state in order to keep the plant or equipment operating normally according to the state indicated by the input data (the operating state of the plant or the operating state of the device to be controlled). It is control.
  • the learning unit 107 builds a prediction model that predicts the control content to be performed on the controlled object by performing machine learning using the teacher data generated by the teacher data generation unit 106. Then, the learning unit 107 stores the constructed prediction model in the prediction model DB 122.
  • the information processing device 1 since the information processing device 1 includes the teacher data generation unit 106, the teacher data for constructing the prediction model is automatically generated from the input data input to the prediction model and used for the prediction. can do. Further, since the information processing device 1 includes the learning unit 107, it is possible to construct a prediction model that reflects the input data actually used for the prediction control.
  • the manual control detection unit 108 detects the manual control manually performed on the controlled object. Then, when the manual control detection unit 108 detects the manual control, the teacher data generation unit 106 responds to the detected manual control with respect to the input data input to the prediction model before the predetermined time of the detection.
  • the correct answer data may be associated with each other and used as teacher data for constructing the above prediction model.
  • Manual control is performed when the predicted control content of the prediction model is incorrect or insufficient, or when the prediction result that the control should be executed is not output even though the control should be executed. May be done. That is, it is possible to specify the control content of the "correct answer" that the prediction model should have predicted from the fact that the manual control was performed and the content of the manual control performed. Therefore, according to the above configuration, it is possible to automatically generate highly valid teacher data.
  • the reference information acquisition unit 109 controls the control target by inputting the input data input to the prediction model into another prediction model that predicts the control content to be performed on the control target. It is acquired as a prediction result for reference, which is not used in.
  • the teacher data generation unit 106 may be used as teacher data for constructing the prediction model by associating the input data with correct answer data according to the prediction result acquired by the reference information acquisition unit 109.
  • the prediction result of the prediction model used to control the control target may differ from the prediction result of other prediction models not used to control the control target. Further, in the control according to the prediction result of the prediction model, the control may not be performed at the timing when the control should be performed. Then, in such a case, it is possible that another prediction model could appropriately predict that the control should be performed. Therefore, according to the above configuration, it is possible to automatically generate highly valid teacher data in consideration of the prediction results of other prediction models.
  • the state determination unit 110 determines the operating state of the plant provided with the control target. This determination may be made based on the output values of various sensors provided in the plant. Examples of the operating state of the plant include normal and abnormal occurrence. Further, the state determination unit 110 stores the determination result in the prediction result DB 121 in association with the date and time of the determination.
  • the teacher data generation unit 106 corresponds to the input data input to the prediction model with correct answer data according to the state of the plant after the control of the control target based on the prediction result of the prediction model is performed. Attached as teacher data for constructing the above prediction model.
  • the operating state of the plant can be specified by referring to the prediction result DB 121.
  • control When control is performed according to the prediction result of the prediction model, it is understood that the control was appropriate if the operating state of the plant thereafter was normal, and that it was not appropriate if it was not normal. In this way, the operating state of the plant after control according to the prediction result of the prediction model indicates the "correct answer" of the control content. Therefore, according to the above configuration, it is possible to automatically generate highly valid teacher data.
  • Correct answer data determination information 123 is information for determining correct answer data based on various preset conditions.
  • the teacher data generation unit 106 determines the correct answer data using the correct answer data determination information 123.
  • the details of the correct answer data determination information 123 will be described below with reference to FIG.
  • the correct answer data determination information 123 may be, for example, table-type information as shown in FIG.
  • FIG. 4 is a diagram showing an example of correct answer data determination information 123. More specifically, FIG. 4 shows the correct answer data determination information 123A for determining the correct answer data in the teacher data of the abnormality avoidance model and the correct answer data determination information 123B for determining the correct answer data in the teacher data of the normal maintenance model. It shows that.
  • the correct answer data determination information 123A determines a method of determining correct answer data according to whether or not predictive control is performed based on the abnormality avoidance model (with predictive control / without predictive control) and the operating state of the plant at a predetermined time. It stipulates.
  • the predetermined time is the time after the time when the prediction control is performed, and is such that the effect of the prediction control is reflected in the operating state of the plant. For example, the predetermined time may be 10 minutes or 15 minutes immediately after the prediction control is performed.
  • “With predictive control” corresponds to the case where predictive control is performed by the target abnormality avoidance model.
  • the teacher data generation unit 106 associates the control content determined using the correct answer data determination information 123A with the input data on which the predictive control is based as correct answer data. Generate teacher data.
  • the input data on which the prediction control is based is the data input to the abnormality avoidance model at the time of the prediction.
  • “No predictive control” corresponds to the case where predictive control is not performed by the target abnormality avoidance model.
  • the teacher data generation unit 106 provides correct answer data determination information 123A to the input data input to the target abnormality avoidance model at a time retroactively from the time of occurrence of the abnormality by a predetermined time.
  • Teacher data is generated by associating the control contents determined using the data with the correct answer data.
  • the correct answer data determination information 123A defines four states as the operating state of the plant: "abnormality occurrence”, “other abnormality occurrence”, “abnormality avoidance by predictive control”, and “normal regardless of predictive control”.
  • "abnormality occurrence” includes a state in which the occurrence could be avoided by predictive control in the end, but a state in which it took a predetermined time or more to avoid, and a state in which the abnormality was avoided by additional control such as manual control. Is done.
  • Abnormality occurrence is a state in which an abnormality predicted by the abnormality avoidance model has occurred.
  • another abnormality occurrence is a state in which an abnormality that the target abnormality avoidance model does not predict has occurred. For example, after performing predictive control based on the avoidance model of the small amount of waste, if the small amount of waste occurs, it will be “abnormality”, and if the flame-retardant waste abnormality occurs, “other abnormalities will occur”. ".
  • the method of determining the correct answer data for the combination of "predictive control” and "abnormality occurrence” is defined as "plus control amount”. Therefore, when the correct answer data determination information 123A is used, the teacher data generation unit 106 performs predictive control using the abnormality avoidance model, but in the case where an abnormality occurs, the teacher data in which the control amount of the correct answer data is increased is used. Generate. In this case, the content of the predictive control was correct, but the amount of control was insufficient, so it is possible that the abnormality was not avoided. Therefore, by increasing the amount of control of the correct answer data, highly relevant teacher data is generated. be able to.
  • control amount may be increased by a predetermined amount, or the control amount may be increased by multiplying the actual control amount by a predetermined magnification larger than 1. This also applies when the correct answer data determination information 123B, which will be described later, is used.
  • the teacher data generation unit 106 may determine how much the control amount is to be increased in consideration of the contents of the manual control and the additional control by another prediction model. For example, when the teacher data generation unit 106 manually controls the same control target as the target abnormality avoidance model, the correct answer of the control amount obtained by adding the control amount of the manual control to the control amount of the target abnormality avoidance model. Data may be generated. Further, when the teacher data generation unit 106 performs predictive control based on the prediction results of another prediction model for the same control target as the target abnormality avoidance model, the plant returns to the normal state in the subsequent predetermined period. On condition that, the same correct answer data as above may be generated.
  • the method of determining the correct answer data for the combination of "predictive control” and "other abnormality occurrence” is defined as "minus control amount”. Therefore, when the correct answer data determination information 123A is used, the teacher data generation unit 106 performs predictive control using the abnormality avoidance model, and then reduces the control amount of the correct answer data in the case where another abnormality occurs. Generate teacher data. In this case, since no abnormality of the prediction target has occurred, it is possible that the content of the prediction control was correct. However, it is possible that other anomalies have occurred due to predictive control. Therefore, by reducing the control amount of the correct answer data, it becomes possible to generate highly valid teacher data in which the control amount is such that neither the abnormality of the prediction target nor the other abnormality is generated as the correct answer data.
  • control amount may be reduced by a predetermined amount, or the control amount may be reduced by multiplying the actual control amount by a predetermined magnification smaller than 1. This also applies when the correct answer data determination information 123B, which will be described later, is used.
  • the teacher data generation unit 106 correctly corrects the control that the input data on which the prediction control is based is not performed in the case of "with prediction control" and "another abnormality occurs". It may be associated as data and used as teacher data.
  • abnormality avoidance by predictive control is a state in which anomalies are avoided as a result of predictive control using the target abnormality avoidance model.
  • correct answer data determination information 123A it is stipulated that the method of determining the correct answer data for the combination of "with predictive control” and “abnormal avoidance by predictive control” is "do not adjust the control amount”.
  • the teacher data generation unit 106 correctly answers the control amount in the predictive control as it is in the case where the abnormality is avoided as a result of the predictive control using the abnormality avoidance model. Generate teacher data as data. This makes it possible to generate highly valid teacher data. This is because it is considered that the content of the prediction control was appropriate in this case.
  • the correct answer data determination information 123A it is stipulated that the method of determining the correct answer data for the combination of "with predictive control” and "normal regardless of predictive control” is "not used as teacher data". Therefore, when the correct answer data determination information 123A is used, the teacher data generation unit 106 performs predictive control using the abnormality avoidance model and no abnormality occurs, but the abnormality does not occur even if the predictive control is not performed. No teacher data is generated for cases that are not considered to have occurred. This makes it possible to avoid generating unreliable teacher data. This is because it is considered that the predictive control did not need to be performed in this case.
  • the teacher data generation unit 106 may associate the input data on which the prediction control is based with the control that the prediction control is not performed as the correct answer data and use it as the teacher data.
  • the correct answer data determination information 123A it is stipulated that the method of determining the correct answer data for the combination of "no predictive control” and “abnormality occurrence” is "the control that can avoid abnormalities is the correct answer data”.
  • the teacher data generation unit 106 does not perform predictive control using the abnormality avoidance model, and in the case where an abnormality occurs, the control for avoiding the abnormality is used as the correct answer data. Generate teacher data. In this way, the teacher data generation unit 106 can also generate teacher data from the input data in the case where the prediction control is not performed.
  • control for avoiding the abnormality may be determined in consideration of manual control and control by another prediction model, or may be set according to the user's input operation via the input unit 14. For example, when the teacher data generation unit 106 has recovered from the abnormal state by manual control or control by another prediction model, the content of the control may be set as correct answer data.
  • the teacher data generation unit 106 does not perform predictive control using the abnormality avoidance model, and in the case where another abnormality occurs, the correct answer is the control for avoiding the abnormality.
  • Generate teacher data as data Note that this teacher data is not for the target abnormality avoidance model, but for other abnormality avoidance models for avoiding other abnormalities. In this way, the teacher data generation unit 106 can also generate teacher data for other anomaly avoidance models.
  • the teacher data generation unit 106 may determine the correct answer data according to a predetermined rule, and may determine the correct answer data without using the correct answer data determination information 123A.
  • the teacher data generation unit 106 determines the correct answer data based on the prediction result of the other abnormality avoidance model, for example, when the prediction result is different between the target abnormality avoidance model and the other abnormality avoidance model. May be good.
  • the other anomaly avoidance model is a prediction model that is not used for prediction control, but input the same input data as the target anomaly avoidance model and make a prediction in parallel with the target anomaly avoidance model. It is a prediction model. There may be two or more other anomaly avoidance models.
  • the reference information acquisition unit 109 acquires the prediction results of the other abnormality avoidance models from the prediction result DB 121.
  • the teacher data generation unit 106 determines the control amount, which is the prediction result of each abnormality avoidance model, when the control target is the same but the control amount is different between the target abnormality avoidance model and the other abnormality avoidance model.
  • the total value may be determined as the control amount of the correct answer data.
  • the predictive control performed based on the target abnormality avoidance model is a control that increases the operating speed of a certain device by two steps
  • another abnormality avoidance model reduces the operating speed of the device by one step. It is assumed that the prediction result that control should be performed is output. In such a case, the teacher data generation unit 106 may set the control for increasing the operating speed of the device by one step as the correct answer data.
  • the teacher data generation unit 106 may evaluate the validity of the prediction results of the target anomaly avoidance model and other anomaly avoidance models, and the evaluation result evaluated as the most appropriate may be the correct answer data, or may be valid. Each prediction result whose sex evaluation result is equal to or greater than the threshold value may be used as correct answer data.
  • the method of evaluating the validity of the prediction result is not particularly limited. For example, if the input data input to the anomaly avoidance model is within the distribution range of the teacher data used for learning the anomaly avoidance model, it may be evaluated as valid, and if it is outside the range, it may be evaluated as invalid. Further, when the abnormality avoidance model outputs a numerical value indicating the accuracy of the prediction as the prediction result, it may be determined that the prediction result having a large numerical value is appropriate.
  • the validity of the prediction result of the target abnormality avoidance model may be judged based on the operating state of the plant after the prediction control based on the prediction result. That is, it may be determined that the prediction result is valid if the operating state of the plant after the prediction control based on the prediction result is stable, and that it is not valid if it is unstable.
  • the validity of the prediction result may be evaluated by the teacher data generation unit 106, or a block different from the teacher data generation unit 106 may be provided and the block may be used.
  • the correct answer data determination information 123B for the normal maintenance model determines the correct answer data according to whether or not the predictive control is performed based on the normal maintenance model (with / without predictive control) and the additional control at a predetermined time.
  • the method is specified.
  • the predetermined time is the same as the correct answer data determination information 123A described above. Further, it is the same as the case of the above-mentioned correct answer data determination information 123A that the input data associated with the correct answer data is different in each case of “with predictive control” and “without predictive control”.
  • the additional control is at least one of control performed based on the prediction result of a prediction model other than the target normal maintenance model and manual control performed manually.
  • five additional controls “same control”, “reverse control”, “other control”, “no additional control (normal maintenance)”, and “no additional control (normal maintenance cannot be performed)" are specified. doing.
  • Standard control is the same control as the predictive control by the target normal maintenance model.
  • the same control is a control in which the control target and the control direction are the same
  • the "reverse control” is a control in which the control target is the same but the control direction is opposite.
  • the control to increase the dust supply speed of the dust supply device that supplies waste to the incinerator by one step and the control to increase the dust supply speed by two steps are the same, but the control to decrease the dust supply speed is the above two controls. Is the opposite control.
  • the method of determining the correct answer data for the combination of "with predictive control” and “same control” is defined as "plus control amount”. Therefore, when the correct answer data determination information 123B is used, the teacher data generation unit 106 controls the correct answer data in the case where the same control as the control is performed after the predictive control using the normal maintenance model is performed. Generate teacher data with increased. In this case, the content of the predictive control was correct, but the amount of control was insufficient, so it is considered that additional control was performed. Therefore, by increasing the amount of control of the correct answer data, highly relevant teacher data is generated. be able to.
  • the method of determining the correct answer data for the combination of "with predictive control” and "reverse control” is defined as "minus control amount”. Therefore, when the correct answer data determination information 123B is used, the teacher data generation unit 106 determines the correct answer data in the case where the predictive control using the normal maintenance model is performed and then the control opposite to the control is performed. Generate teacher data with reduced control. In this case, the content of the predictive control was correct, but the amount of control was too large, and it is considered that the reverse control was performed. Therefore, by reducing the amount of control of the correct answer data, highly valid teacher data is generated. be able to.
  • the teacher data generation unit 106 may use the reverse control as the correct answer data. Further, the teacher data generation unit 106 may use the control amount in the correct answer data as the difference between the control amount of the predictive control and the control amount of the additional control.
  • “other control” is control for a control target different from the predictive control by the normal maintenance model of the target.
  • the predictive control based on the normal maintenance model of the target is the control that increases the dust supply speed
  • the control that changes the operating speed of the grate is another control.
  • the correct answer data determination information 123B the method of determining the correct answer data for the combination of "with predictive control” and “other control” is defined as "adjusted according to other control contents". The adjustment method for each of the other control contents will be determined separately.
  • the teacher data generation unit 106 adjusts the correct answer data and obtains the teacher data in the case where the predictive control using the normal maintenance model is performed and then other control is performed. Generate. For example, if the other control is a control that brings about the same effect as the prediction control by the target normal maintenance model (for example, increasing the amount of steam generated from the incinerator), the teacher data generation unit 106 may perform the correct answer data. Teacher data with an increased amount of control may be generated. Further, for example, when the other control is a control that has the opposite effect to the predictive control by the normal maintenance model of the target, the teacher data generation unit 106 may generate the teacher data in which the control amount of the correct answer data is reduced. Good.
  • the input data that is the basis of the prediction control by the target normal maintenance model may be associated with the control that the prediction control is not performed as the correct answer data and used as the teacher data. Further, since it is possible that an abnormality has occurred due to the control amount being too large, the teacher data generation unit 106 may generate correct answer data in which the control amount is reduced compared to the predictive control by the normal maintenance model of the target. Good.
  • the teacher data generation unit 106 performs predictive control using the normal maintenance model, and in the case where the normal state is maintained, the control amount in the predictive control is used as it is. Generate teacher data as correct answer data. This makes it possible to generate highly valid teacher data. This is because it is considered that the content of the prediction control was appropriate in this case. In addition to this case, if it can be judged that the predictive control using the normal maintenance model is appropriate, the teacher data may be generated in the same manner.
  • the method of determining the correct answer data in the case of "without additional control (cannot be maintained normally)" is not “teacher data” in both “with predictive control” and “without predictive control”. It is stipulated.
  • the teacher data generation unit 106 does not generate teacher data in the case where the additional control is not performed and the normal state cannot be maintained. This makes it possible to avoid generating unreliable teacher data. This is because in this case, it is difficult to automatically determine what kind of control should be performed to maintain the normal state.
  • the teacher data generation unit 106 also associates the control content input by the user via the input unit 14 with the input data on which the prediction control is based as correct answer data and uses it as teacher data. Good.
  • the method of determining the correct answer data for the combination of "no predictive control” and “other control” is defined as "apply the content of additional control”. Therefore, when the correct answer data determination information 123B is used, the teacher data generation unit 106 sets the content of the additional control as the correct answer data in the case where the predictive control using the normal maintenance model is not performed and the additional control is performed. Generate teacher data. In this case, it is considered that the normal maintenance model should have predicted the control performed as the additional control.
  • the teacher data generation unit 106 may determine the correct answer data based on other prediction results performed at the same timing as or at the same timing as the normal maintenance model of the target, instead of the additional control. For example, the teacher data generation unit 106 may determine the correct answer data based on the prediction result of the other prediction model when the prediction is also performed in the other prediction model in parallel with the normal maintenance model of the target.
  • the other predictive model may be a normal maintenance model, an anomaly avoidance model, or both.
  • the reference information acquisition unit 109 acquires the prediction results of the other prediction models from the prediction result DB 121.
  • the teacher data generation unit 106 may determine the correct answer data based on the evaluation result for each prediction result, as in the above-mentioned "other example of the correct answer data determination method for the abnormality avoidance model". ..
  • FIG. 5 is a flowchart showing an example of processing for generating teacher data. This process corresponds to S4 in FIG.
  • the processes S11 to S17 of FIG. 5 are performed for each abnormality avoidance model.
  • the teacher data generation unit 106 extracts from the prediction result DB 121 the prediction control executed according to the output of the abnormality avoidance model in a predetermined period, and the input data on which the prediction control is based.
  • the input data extracted here is associated with the correct answer data in the case of "predictive control" in the correct answer data determination information 123A of FIG.
  • the manual control detection unit 108 extracts from the prediction result DB 121 the manual control performed at a predetermined time after the prediction control for each of the prediction controls extracted in S11.
  • the extraction result of S12 is used to determine the correct answer data in the case of "abnormality occurrence" in the correct answer data determination information 123A of FIG.
  • the reference information acquisition unit 109 extracts the prediction results of other prediction models at a predetermined time after the prediction control for each of the prediction controls extracted in S11 from the prediction result DB 121.
  • the prediction results of other prediction models may include prediction results used for control and prediction results not used for control. Further, the other prediction model may be an abnormality avoidance model, a normal maintenance model, or both of them.
  • the extraction result of S13 is used for determining the correct answer data in the cases of "abnormality occurrence" and "other abnormal occurrence" in the correct answer data determination information 123A of FIG.
  • the teacher data generation unit 106 extracts the determination result of the operating state of the plant at a predetermined time after the prediction control for each of the prediction controls extracted in S11. It is assumed that the operating state of the plant is determined by the state determination unit 110 and the determination result is stored in the prediction result DB 121.
  • the state determination unit 110 determines whether an abnormality has occurred in the plant or a normal state. Then, when the state determination unit 110 determines that the state is not normal, it may determine the type of abnormality and store the determination result in the prediction result DB 121. As a result, the teacher data generation unit 106 can identify whether the operating state of the plant corresponds to "abnormality occurrence" or "other abnormality occurrence” by referring to the prediction result DB 121.
  • the teacher data generation unit 106 determines that the operating state of the plant is "abnormal avoidance by predictive control” and "normal regardless of predictive control” based on the extraction results of S12 and S13 for the "normal” state. It is possible to specify which one is applicable.
  • the teacher data generation unit 106 extracts input data satisfying a predetermined condition from the prediction result DB 121.
  • the difference between the input data extracted in S11 and the input data extracted in S15 is that the former is executing predictive control based on the input data, while the latter is not executing predictive control. It is in.
  • the teacher data is subjected to the following processing. It is said that.
  • the above-mentioned predetermined condition may be a condition in which the teacher data generation unit 106 can extract input data capable of determining the corresponding correct answer data. For example, when the state in which the abnormality has occurred is included in the state extracted in S14, the teacher data generation unit 106 inputs the input data input to the abnormality avoidance model at a time when a predetermined time is traced back from the occurrence of the abnormality. It may be extracted.
  • the teacher data generation unit 106 uses the correct answer data determination information 123A for each of the input data extracted in S11 and the input data extracted in S15 according to the extraction result and the detection result of S12 to S14. Determine the correct answer data.
  • the method for determining the correct answer data is as described above.
  • the teacher data generation unit 106 generates teacher data by associating the input data extracted in S11 with the correct answer data determined in S16 with each of the input data extracted in S15.
  • the processing of FIG. 5 is completed, and after that, an abnormality avoidance model is constructed using the generated teacher data (S5 of FIG. 3).
  • the teacher data generation unit 106 extracts from the prediction result DB 121 the prediction control executed according to the output of the normal maintenance model in a predetermined period, and the input data on which the prediction control is based.
  • the input data extracted here is associated with the correct answer data in the case of "predictive control" in the correct answer data determination information 123B of FIG.
  • the manual control detection unit 108 extracts from the prediction result DB 121 the manual control performed at a predetermined time after the prediction control for each of the prediction controls extracted in S11. Further, in S13, the reference information acquisition unit 109 extracts the prediction results of other prediction models at a predetermined time after the prediction control for each of the prediction controls extracted in S11 from the prediction result DB 121.
  • the teacher data generation unit 106 determines "same control”, “reverse control”, “other control”, and "additional control” for the additional control at a predetermined time after the prediction control in the correct answer data determination information 123B of FIG. Specify which of "None" is applicable.
  • the teacher data generation unit 106 extracts the detection result of the operating state of the plant at a predetermined time after the prediction control for each of the prediction controls extracted in S11.
  • the teacher data generation unit 106 determines "no additional control (normal maintenance)” or “no additional control (normal)” in the case of "no additional control” depending on the extraction result of S14. ”(Cannot be maintained)” is specified.
  • the teacher data generation unit 106 extracts input data satisfying a predetermined condition from the prediction result DB 121.
  • the difference between the input data extracted in S11 and the input data extracted in S15 is that the former is executing predictive control based on the input data, while the latter is not executing predictive control. It is in.
  • the input data extracted in S15 is used as teacher data by the following processing if it corresponds to the combination of "no predictive control" and "other control" in the correct answer data determination information 123B of FIG.
  • the input data extracted in S15 may be any input data that allows the teacher data generation unit 106 to determine the corresponding correct answer data. For example, when the operation state detected in S14 includes a state in which an abnormality has occurred, the teacher data generation unit 106 inputs the input to the normal maintenance model when a predetermined time is traced back from the occurrence of the abnormality. Data may be extracted.
  • the teacher data generation unit 106 uses the correct answer data determination information 123B for each of the input data extracted in S11 and the input data extracted in S15 according to the extraction result and the detection result of S12 to S14. Determine the correct answer data.
  • the method for determining the correct answer data is as described above. Further, since the processing of S17 is the same as that at the time of generating the teacher data for the abnormality avoidance model described above, the description will not be repeated.
  • the control block of the information processing device 1 (particularly each part included in the control unit 10) may be realized by a logic circuit (hardware) formed in an integrated circuit (IC chip) or the like, or may be realized by software. Good.
  • the information processing device 1 includes a computer that executes instructions of a program (information processing program) that is software that realizes each function.
  • the computer includes, for example, one or more processors and a computer-readable recording medium that stores the program. Then, in the computer, the processor reads the program from the recording medium and executes it, thereby achieving the object of the present invention.
  • the processor for example, a CPU (Central Processing Unit) can be used.
  • the recording medium in addition to a “non-temporary tangible medium” such as a ROM (Read Only Memory), a tape, a disk, a card, a semiconductor memory, a programmable logic circuit, or the like can be used.
  • a RAM RandomAccessMemory
  • the program may be supplied to the computer via an arbitrary transmission medium (communication network, broadcast wave, etc.) capable of transmitting the program.
  • a transmission medium communication network, broadcast wave, etc.
  • one aspect of the present invention can also be realized in the form of a data signal embedded in a carrier wave, in which the above program is embodied by electronic transmission.

Landscapes

  • Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Mechanical Engineering (AREA)
  • General Engineering & Computer Science (AREA)
  • Feedback Control In General (AREA)
  • Incineration Of Waste (AREA)
  • Testing And Monitoring For Control Systems (AREA)

Abstract

La présente invention a pour objectif de réaliser la mise à jour appropriée d'un modèle de prédiction qui prédit un contenu de commande. L'invention concerne un dispositif de traitement d'informations (1) qui comprend : une unité d'évaluation (104) qui évalue des résultats de prédiction pour chaque période lorsqu'une cible de commande est commandée en utilisant alternativement une pluralité de modèles de prédiction qui prédisent un contenu de commande dans chaque période d'une pluralité de périodes ; et une unité de sélection (105) qui sélectionne un modèle de prédiction à utiliser par la suite parmi la pluralité de modèles de prédiction sur la base des résultats d'évaluation.
PCT/JP2020/044600 2019-12-19 2020-12-01 Dispositif de traitement d'informations, procédé de traitement d'informations et programme de traitement d'informations WO2021124861A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202080088498.5A CN114868090A (zh) 2019-12-19 2020-12-01 信息处理装置、信息处理方法、以及信息处理程序

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2019229429A JP7125383B2 (ja) 2019-12-19 2019-12-19 情報処理装置、情報処理方法、および情報処理プログラム
JP2019-229429 2019-12-19

Publications (1)

Publication Number Publication Date
WO2021124861A1 true WO2021124861A1 (fr) 2021-06-24

Family

ID=76431862

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2020/044600 WO2021124861A1 (fr) 2019-12-19 2020-12-01 Dispositif de traitement d'informations, procédé de traitement d'informations et programme de traitement d'informations

Country Status (3)

Country Link
JP (1) JP7125383B2 (fr)
CN (1) CN114868090A (fr)
WO (1) WO2021124861A1 (fr)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH05265513A (ja) * 1992-03-19 1993-10-15 Hitachi Ltd 排水ポンプ所のポンプ運転台数制御方法および装置
JPH0713611A (ja) * 1993-06-24 1995-01-17 Hitachi Ltd プロセスモデル評価装置およびプロセスモデル評価方法
JPH1074188A (ja) * 1996-05-23 1998-03-17 Hitachi Ltd データ学習装置およびプラント制御装置
JP2019101495A (ja) * 2017-11-28 2019-06-24 横河電機株式会社 診断装置、診断方法、プログラム、および記録媒体

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2008102720A1 (fr) 2007-02-19 2008-08-28 Kaneka Corporation Procédé de fabrication de 3-aminopipéridine optiquement active ou d'un sel de celle-ci

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH05265513A (ja) * 1992-03-19 1993-10-15 Hitachi Ltd 排水ポンプ所のポンプ運転台数制御方法および装置
JPH0713611A (ja) * 1993-06-24 1995-01-17 Hitachi Ltd プロセスモデル評価装置およびプロセスモデル評価方法
JPH1074188A (ja) * 1996-05-23 1998-03-17 Hitachi Ltd データ学習装置およびプラント制御装置
JP2019101495A (ja) * 2017-11-28 2019-06-24 横河電機株式会社 診断装置、診断方法、プログラム、および記録媒体

Also Published As

Publication number Publication date
JP7125383B2 (ja) 2022-08-24
JP2021096782A (ja) 2021-06-24
CN114868090A (zh) 2022-08-05

Similar Documents

Publication Publication Date Title
US20230350355A1 (en) Heuristic method of automated and learning control, and building automation systems thereof
TWI797418B (zh) 用於校正離子注入半導體製造工具中的部件故障的方法及系統
CN107871190A (zh) 一种业务指标监控方法及装置
CN109255517B (zh) 风控策略的生成方法、装置、服务器及可读存储介质
WO2020242553A1 (fr) Prédiction de panne utilisant une identification de capteur basée sur un gradient
CN109343489B (zh) 烟草制丝工艺参数自愈控制方法及系统
US11892819B2 (en) Control device, control system, control method, and computer-readable storage medium
JP2004178492A (ja) 強化学習法を用いたプラントシミュレーション方法
JP6492555B2 (ja) 異常診断方法、異常診断装置及び異常診断プログラム
CN111754024A (zh) 一种基于回归分析的电力行业时间序列预测方法和装置
JP6790154B2 (ja) 協調型学習システム及び監視システム
CN111222553B (zh) 机器学习模型的训练数据处理方法、装置和计算机设备
CN104076809A (zh) 数据处理装置以及数据处理方法
CN114647923A (zh) 用于监测资产剩余使用寿命的装置、方法和计算机程序产品
JP2022015250A (ja) 放流水質予測装置および放流水質予測方法
CN113868953A (zh) 工业系统中多机组运行优化方法、装置、系统及存储介质
JPWO2021199160A5 (ja) 情報処理装置、情報処理方法、プログラム、情報処理システム
WO2021124861A1 (fr) Dispositif de traitement d'informations, procédé de traitement d'informations et programme de traitement d'informations
WO2021210353A1 (fr) Système de prédiction de défaillance
JP6625183B1 (ja) リモデル支援装置、および、リモデル支援方法
US20140052425A1 (en) Method and apparatus for evaluating a model of an industrial plant process
JP7531177B2 (ja) 放流水質予測装置および放流水質予測方法
CN111932042B (zh) 自适应排程方法、设备及计算机存储介质
CN110955286B (zh) 一种禽蛋监测方法及装置
CN117871771B (zh) 一种基于大数据的燃气能源监测方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20901436

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20901436

Country of ref document: EP

Kind code of ref document: A1