WO2007116590A1 - 運転制御方法,運転制御装置及び運転制御システム - Google Patents
運転制御方法,運転制御装置及び運転制御システム Download PDFInfo
- Publication number
- WO2007116590A1 WO2007116590A1 PCT/JP2007/050682 JP2007050682W WO2007116590A1 WO 2007116590 A1 WO2007116590 A1 WO 2007116590A1 JP 2007050682 W JP2007050682 W JP 2007050682W WO 2007116590 A1 WO2007116590 A1 WO 2007116590A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- control
- evaluation value
- model
- deviation
- amount
- Prior art date
Links
Classifications
-
- 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
- G05B13/042—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 in which a parameter or coefficient is automatically adjusted to optimise the performance
Definitions
- the present invention relates to a driving control apparatus to which unsupervised learning is applied, and a driving control method.
- Reinforcement learning is a framework of learning control that generates operation signals to the environment so that the measurement signal obtained from the environment becomes desirable by trial-and-error interaction with the environment such as the control target.
- Reinforcement Learning an evaluation value of scalar quantity calculated using a measurement signal that can also obtain environmental strength (in Reinforcement Learning, it is called a reward) is used as a clue, and an expectation of the evaluation value obtained from the present state to the future It has a learning function that generates an operation signal to the environment so that the value is maximized.
- there are algorithms such as Actor-Critic, Q-learning, and real-time Dynamic Programming.
- Dyna-architecture As a framework of reinforcement learning developed from the above-mentioned method, there is a framework called Dyna-architecture. This is a method of learning in advance what kind of operation signal should be generated for a model simulating a control target, and using this learning result to determine an operation signal to be applied to the control target. It also has a model adjustment function that reduces the error between the control target and the model.
- Patent Document 1 As a technology to which reinforcement learning is applied, the technology described in Patent Document 1 can be mentioned.
- a plurality of reinforcement learning modules which are a set of systems having a model and a learning function, are provided, and a responsible signal that takes a larger value as the prediction error between the model and the control object in each reinforcement learning module is smaller
- a responsible signal that takes a larger value as the prediction error between the model and the control object in each reinforcement learning module is smaller
- Patent Document 1 Japanese Patent Laid-Open No. 2000-35956
- the present invention provides an operation control device and an operation control method that can be operated without adversely affecting the operation state of a control target even when a deviation between a model and a real machine (model error) occurs. With the goal.
- an operation control method of a control device which derives and controls an operation amount that maximizes or minimizes an evaluation value based on a control deviation that is a deviation between a control amount to be controlled and its target value.
- the model has a model simulating the characteristics of the control target, and the operation quantity that maximizes or minimizes the evaluation value based on the control deviation of the model is calculated for the model, and the control target is calculated by the operation quantity.
- the evaluation value is calculated based on the control deviation in the case of control, and based on the evaluation value of the control deviation of the model and the deviation of the evaluation value of the control deviation of the controlled object, the manipulated variable of the current step and the next It is an operation control method characterized by determining a change width of an operation amount which is a difference from the operation amount determined in the step.
- the vehicle can be operated without adversely affecting the operating state of the controlled object.
- FIG. 1 is a diagram for explaining an example in which a control device 200 according to the present invention is applied to a control target 100.
- the operation signal generation unit 300 provided in the control device 200 generates an operation signal 201 to be applied to the control target. Further, the evaluation value calculation unit 500 calculates the evaluation value signal 203 using the measurement signal 202 from the control target. Operation signal generator 300 receives this evaluation value signal. Receive 203
- the operation signal generation unit 300 has a function of generating the operation signal 201 so that the total sum of the expected values of the evaluation value signal 203 up to the present state is also the maximum or the minimum. In the following, a case will be described where the operation signal generation unit 300 generates the operation signal 201 so that the sum of the expected values of the evaluation value signal 203 is maximized.
- An evaluation value calculation unit 500 generates an evaluation value signal 203 according to the deviation between the measurement signal 202 and its target value. For example, when the measurement signal 202 matches the target value, the evaluation value signal 203 is set to “1”, and when it does not match, it is set to “0”. Alternatively, the evaluation value signal 203 is set to be inversely proportional to the deviation between the measurement signal 202 and its target value. That is, as described in FIG. 5 described later, the evaluation value is closer to the target as the numerical value is larger as +30, and the target power is farther as the numerical value is smaller as 30. In this case, the evaluation value can be calculated by a plurality of methods. An example of evaluation value calculation is shown in FIG.
- the difference between the control amount and the target value is associated with the evaluation value, and the evaluation value can be generated with reference to this table.
- the evaluation value can be set and calculated as a function of the difference between the control amount and the target value.
- the operation signal 201 is generated by trial and error in the initial stage of learning. Thereafter, as the learning progresses, an operation signal 201 is generated such that the evaluation value signal 203 becomes larger.
- Such a learning algorithm can use, for example, an algorithm such as Actor-Critic or Q learning.
- the controller in FIG. 1 uses a framework called Dyna-architecture. It has a model unit 400 for simulating the control target 100, and the operation signal generation unit 300 learns in advance the method of generating the operation signal 201 for the model unit 400, It is a framework for generating the signal 201.
- the operation signal generation unit 300 has a function of generating the operation signal 204 input to the model unit 400 and receiving the measurement signal 205 and the evaluation value signal 206 from the model unit 400.
- the evaluation value signal 206 is calculated using the measurement signal 205 in the evaluation value calculation unit 510.
- the evaluation value calculation unit 510 has the same function as the evaluation value calculation unit 500.
- the operation signal generation unit 300 refers to the data stored in the operation signal generation parameter storage unit 600 to determine the operation signal 201 to be applied to the control target 100.
- FIG. 2 is a view for explaining an aspect of data stored in the operation signal generation parameter storage unit 600.
- the operation signal generation parameter storage unit 600 stores the name of the operation end provided in the control target 100, the change width per cycle of the operation amount, and data on the unit. There is.
- the operation end can increase or decrease the manipulated variable within the range of manipulated variable change.
- FIG. 2 describes the case where there are a plurality of operating terminals, the number of operating terminals may be one.
- the operation change width is described for each operation end in FIG. 2, a plurality of operation ends can be put together and the sum of the change widths of the operation ends can be limited.
- the limit value of the manipulated variable change width in FIG. 2 is determined by the manipulation signal generation parameter updating unit 700.
- the setting values necessary for the process of updating the nometer are input from an external input device 20 configured of a keyboard 30 and a mouse 40. These pieces of information are displayed on an image display device 10 such as a CRT. The operator of the control target 100 inputs the set value 214 using the image display device 10 and the external input device 20.
- FIG. 3 is an example of a screen displayed on the image display device 10. Through this screen, the operator can set the initial value, upper limit, lower limit, and update rate of the operation amount change width at the operation end.
- FIG. 4 is a diagram for explaining the processing in the operation signal generation parameter updating unit 700. The contents of each process in Fig. 4 will be explained below.
- process 710 it is determined whether the number of steps t is greater than 0. If it is 0 (if NO), process 720 is performed; if it is greater than 0 (if YES), process 740 is performed. carry out.
- the number of steps is the number of times the operation signal applied to the control target 100 has been changed, and is a value that increases from 0 to the initial value and by 1 each time the operation is performed.
- processing 720 the initial value set in FIG. 3 is acquired.
- processing 730 the initial value acquired in processing 720 is transmitted as data 209 to the operation signal generation parameter storage unit 600.
- process 740 the previous operation signal generation parameter stored in the operation signal generation parameter storage unit 600 is acquired as data 208.
- processing 750 the evaluation value signal 203 and the evaluation value signal 206 are acquired.
- r (t) is the value of the evaluation value signal 203
- r (t) is the evaluation value signal 206
- f (r (t), r (t)) is a function with r (t) and r (t) as variables.
- G (t + l) G (t) + f (r (t), r (t))
- G (t + 1) calculated using Equations 1 and 2 exceeds the upper limit set in FIG. 3, G (t + 1) is taken as the upper limit value set, and if smaller than the lower limit Let G (t + 1) be the lower limit value you set.
- G (t + 1) obtained by processing 770 is transmitted as data 209 to the operation signal generation parameter storage unit 600.
- the difference between the evaluation value signals 203 and 206 and the change amount of operation amount G (t + 1) ⁇ G (t) are stored as a table in correspondence with each other, and the operation amount may be determined with reference to this. .
- the change amount of the manipulated variable is calculated based on the difference between the evaluation value 206 based on the control deviation of the model and the evaluation value 203 based on the control deviation when the control target is controlled. It can be operated without adversely affecting the condition. In addition, it can be flexibly controlled according to the difference between the model and the actual machine.
- the operation amount change width is increased by setting the upper limit of the operation amount change width, the difference between the actual machine and the model has a large effect, so changing the operation amount quickly and operating the control target It can balance the negative impact on the changing state.
- the model parameter storage unit 800 stores parameters necessary to configure the model unit 400.
- the model unit 400 is a physical model
- the model parameter storage unit 800 stores physical constants necessary to construct the physical model. For example, when the control target 100 is a thermal power plant, values such as the heat transfer coefficient are stored.
- the model parameter updating unit 900 reads out the parameters 212 stored in the model parameter storage unit 800 so that the characteristics of the control object and the model match, corrects the parameters, and transmits the corrected parameter 213. , Update model parameters.
- the model parameters 211 can be transferred to the model unit 400 using the techniques described in Japanese Patent Application Laid-Open Nos. 10-214112 and 2001-154705. Set and update model parameters.
- FIG. 5 to 7 are diagrams for explaining problems that may occur when the conventional control device is applied to the control target 100.
- FIG. 5 to 7 are diagrams for explaining problems that may occur when the conventional control device is applied to the control target 100.
- FIG. 5 shows the relationship between the space of the operation amount and the obtained evaluation value.
- the value of the manipulated variable A is A
- the evaluation value is -30. Also, the evaluation value for A and B is +10.
- the action that maximizes the sum of the expected values of the evaluation values avoids the area where the evaluation value is negative as shown by the dotted line in FIG. 5, and becomes a positive action to the area that is positive.
- FIG. 6 is a diagram showing the change width of the movable operation amount by one action with an arrow.
- the change width of the operation amount is constant.
- the starting point force also has an evaluation value of +
- Fig. 7 is a diagram showing an example where the characteristics of the model and the controlled object are different.
- the condition of the manipulated variable for which the evaluation value is negative differs between the model and the control target.
- the evaluation value after one step is 30, which is not a desirable state.
- FIG. 8 to 10 explain the effects of applying the control device of the present invention to plant 100.
- FIG. 8 to 10 explain the effects of applying the control device of the present invention to plant 100.
- FIG. 8 to 10 the variation width of the manipulated variable is determined through the processing of FIG. 3 without making the manipulated variable variation width constant.
- the evaluation value after one step is -10. This is a better value than 30 obtained after one step in the conventional method.
- the operating state approximates to the initial state, so that the safety of the controlled object can be maintained.
- the control device 200 obtains information on the controlled object 100 and the model unit 400 having different characteristics! /! /.
- the model parameter update unit 900 stores the measured signal 202 from the control target 100 and the output signal 205 from the model unit 400 in the model parameter storage unit 800 so that the characteristics of the model unit 400 and the control target 100 match. Update the parameters that have been set. If the characteristics of the model and the controlled object are different, the operation signal 201 is returned to return to the initial state (Start in FIG. 8). As described above, when the difference between the evaluation values is larger than the predetermined value and the model is corrected, the difference between the model and the actual machine can be safely controlled along the model when the difference between the model and the actual machine is smaller than the predetermined value.
- FIG. 9 is a view for explaining the relationship between the space of the operation amount and the evaluation value obtained by the model after correction.
- the action that maximizes the sum of the expected values of the evaluation values is the direction toward a positive area, avoiding the area where the evaluation value is negative as shown by the dotted line in Fig. 9.
- This operation path differs between when the model before correction is used and when the model after correction is used.
- FIG. 10 shows a path when the control target 100 is controlled using the corrected operation path.
- FIG. 11 is a view for explaining the relationship between the number of steps and the amount of change in the amount of operation when the operation in FIG. 10 is performed.
- Equation 2 Since both the evaluation value of the model and the evaluation value from the control target are 0, the second item in Equation 2 is 0. Therefore, the change amount of the manipulated variable is increased by ⁇ per step.
- An operation signal 201 is displayed on the CRT 10 of FIG. It is also possible to display data such as the amount of change in operation amount, which is the data 210 stored in the operation signal generation parameter storage unit 600.
- the control amount 202 of the control target 100 can also be displayed.
- the CRT 10 can display the relationship between the space of the operation amount and the evaluation value shown in FIGS. 5 to 10 on the screen.
- FIG. 14 shows an example when the relationship between the space of the operation amount and the evaluation value is displayed on the screen.
- the control device 100 sets the operation amounts of the plurality of operations to be applied to the control target to the plurality of axes respectively, displays the start point and the reach point of each operation applied to the control target, and reaches the operation one step before Connect the point and the start point of the operation of the next step, create the image information to be displayed, and display it on CRT10. This makes it possible to easily grasp the amount of change in each operation in comparison with the overall operation.
- an arrival point is also displayed by an arrow.
- control device 100 has model 400 simulating the characteristic of the control object, and calculates evaluation value based on the control deviation when the model is controlled to the target, and evaluation value calculation unit 510 of the model and the like. And an evaluation value calculation unit 500 of the control object that calculates the evaluation value based on the control deviation when the control object is controlled, and the evaluation value of the model when performing each operation and the evaluation value from the control object The difference is calculated, and display data to be displayed corresponding to the display of each operation is created and transmitted to the CRT 10. In this way, by displaying the difference between the evaluation value of the model and the evaluation value from the control target when each operation is performed according to each operation display, no operation is performed. can do.
- the change width of the operation amount is reduced immediately after the start of the operation, and the operation method learned for the model is controlled. Check if the target is also valid. After that, it turns out that the operation method learned for the model whose characteristics of the control object and model are close is effective for the control object, and then the change width of the operation amount gradually increases.
- FIG. 1 is a view for explaining an example in which a control device of the present invention is applied to a control target.
- FIG. 2 is a diagram for explaining an aspect of data stored in an operation signal generation parameter storage unit.
- FIG. 3 is a diagram for explaining a screen displayed on the image display device.
- FIG. 4 is a diagram for explaining processing of an operation signal generation parameter updating unit.
- FIG. 5 is a diagram for explaining the characteristics of a model.
- FIG. 6 is a diagram for explaining an arrival point for each step.
- FIG. 7 is a diagram for explaining the difference between a control target and a model characteristic.
- FIG. 8 is a view for explaining the operation method of the present invention.
- FIG. 9 is a diagram for explaining the characteristics of the model after correction.
- FIG. 11 is a diagram for explaining the relationship between the number of steps and the amount of change in operation amount.
- FIG. 12 This is an example of evaluation value calculation.
- FIG. 13 An example of a table for determining the amount of operation.
- FIG. 14 Example of displaying the relationship between the space of the operation amount and the evaluation value on the screen.
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Feedback Control In General (AREA)
Abstract
Description
Claims
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US12/279,350 US8155763B2 (en) | 2006-03-31 | 2007-01-18 | Operation control method, operation control device, and operation control system |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2006-096373 | 2006-03-31 | ||
JP2006096373A JP4952025B2 (ja) | 2006-03-31 | 2006-03-31 | 運転制御方法,運転制御装置及び運転制御システム |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2007116590A1 true WO2007116590A1 (ja) | 2007-10-18 |
Family
ID=38580892
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2007/050682 WO2007116590A1 (ja) | 2006-03-31 | 2007-01-18 | 運転制御方法,運転制御装置及び運転制御システム |
Country Status (4)
Country | Link |
---|---|
US (1) | US8155763B2 (ja) |
JP (1) | JP4952025B2 (ja) |
CN (1) | CN101390024A (ja) |
WO (1) | WO2007116590A1 (ja) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2018105412A1 (ja) * | 2016-12-07 | 2018-06-14 | ソニー株式会社 | 情報処理装置および方法、並びにプログラム |
Families Citing this family (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP4427074B2 (ja) | 2007-06-07 | 2010-03-03 | 株式会社日立製作所 | プラントの制御装置 |
US8135653B2 (en) | 2007-11-20 | 2012-03-13 | Hitachi, Ltd. | Power plant control device which uses a model, a learning signal, a correction signal, and a manipulation signal |
JP4876057B2 (ja) * | 2007-11-20 | 2012-02-15 | 株式会社日立製作所 | プラントの制御装置、及び火力発電プラントの制御装置 |
JP4627553B2 (ja) * | 2008-03-28 | 2011-02-09 | 株式会社日立製作所 | プラントの制御装置および火力発電プラントの制御装置 |
CN105319966B (zh) * | 2014-07-30 | 2017-10-20 | 南京南瑞继保电气有限公司 | 一种避免系统间通讯恢复后冗余系统同时退出值班的方法 |
JP6497367B2 (ja) * | 2016-08-31 | 2019-04-10 | 横河電機株式会社 | プラント制御装置、プラント制御方法、プラント制御プログラム及び記録媒体 |
JP6742222B2 (ja) * | 2016-11-14 | 2020-08-19 | 株式会社日立製作所 | 運転支援装置及びプログラム |
JP6662310B2 (ja) * | 2017-01-11 | 2020-03-11 | 横河電機株式会社 | データ処理装置、データ処理方法及びプログラム |
DE112017007028B4 (de) * | 2017-02-09 | 2023-06-15 | Mitsubishi Electric Corporation | Positionskontrollvorrichtung und Positionskontrollverfahren |
CN111433694A (zh) * | 2017-11-29 | 2020-07-17 | 三菱日立电力系统株式会社 | 运转条件评价装置、运转条件评价方法及发电设备的控制系统 |
JP6970078B2 (ja) | 2018-11-28 | 2021-11-24 | 株式会社東芝 | ロボット動作計画装置、ロボットシステム、および方法 |
JP7206874B2 (ja) * | 2018-12-10 | 2023-01-18 | 富士電機株式会社 | 制御装置、制御方法及びプログラム |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH04256102A (ja) * | 1991-02-08 | 1992-09-10 | Toshiba Corp | モデル予測制御装置 |
JPH0695706A (ja) * | 1992-09-11 | 1994-04-08 | Toshiba Corp | モデル予測制御装置 |
JPH0883104A (ja) * | 1994-09-12 | 1996-03-26 | Toshiba Corp | プラント制御装置 |
JPH09128011A (ja) * | 1995-10-31 | 1997-05-16 | Fuji Electric Co Ltd | モデル予測制御方法 |
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 |
JP2003005804A (ja) * | 2001-06-19 | 2003-01-08 | Honda Motor Co Ltd | プラントの制御装置 |
JP2004094939A (ja) * | 2002-08-09 | 2004-03-25 | Omron Corp | モデル構造、制御装置、温度調節器および熱処理装置 |
Family Cites Families (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5166873A (en) * | 1989-09-13 | 1992-11-24 | Yokogawa Electric Corporation | Process control device |
US5930136A (en) * | 1990-06-04 | 1999-07-27 | Hitachi, Ltd. | Control device for controlling a controlled apparatus, and a control method therefor |
EP0524317A4 (en) * | 1991-02-08 | 1995-02-15 | Tokyo Shibaura Electric Co | Model forecasting controller |
US5448681A (en) * | 1992-03-27 | 1995-09-05 | National Semiconductor Corporation | Intelligent controller with neural network and reinforcement learning |
JPH10214112A (ja) | 1997-01-29 | 1998-08-11 | Hitachi Ltd | 火力プラントシミュレータ |
JP3086206B2 (ja) | 1998-07-17 | 2000-09-11 | 科学技術振興事業団 | エージェント学習装置 |
JP2001154705A (ja) | 1999-11-25 | 2001-06-08 | Hitachi Ltd | 火力プラントの運転制御方法及び運転制御装置 |
JP3742966B2 (ja) * | 2000-02-23 | 2006-02-08 | 株式会社日立製作所 | 制御装置 |
US7050864B2 (en) * | 2001-04-20 | 2006-05-23 | Honda Giken Kogyo Kabushiki Kaisha | Control system for a plant using identified model parameters |
JP3989775B2 (ja) * | 2002-06-05 | 2007-10-10 | 株式会社日立製作所 | 制御モデルを用いた制御系のプリセット制御方法および制御装置 |
ITMI20021231A1 (it) * | 2002-06-06 | 2003-12-09 | Nuovo Pignone Spa | Sistema di controllo e regolazione della temperatura di fiamma per turbine a gas monoalbero |
EP1650672A3 (en) * | 2004-10-21 | 2006-08-23 | HONDA MOTOR CO., Ltd. | A neural network element with reinforcement/attenuation learning |
US7395251B2 (en) * | 2005-07-01 | 2008-07-01 | International Business Machines Corporation | Neural networks for prediction and control |
-
2006
- 2006-03-31 JP JP2006096373A patent/JP4952025B2/ja not_active Expired - Fee Related
-
2007
- 2007-01-18 US US12/279,350 patent/US8155763B2/en not_active Expired - Fee Related
- 2007-01-18 CN CNA2007800061107A patent/CN101390024A/zh active Pending
- 2007-01-18 WO PCT/JP2007/050682 patent/WO2007116590A1/ja active Application Filing
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH04256102A (ja) * | 1991-02-08 | 1992-09-10 | Toshiba Corp | モデル予測制御装置 |
JPH0695706A (ja) * | 1992-09-11 | 1994-04-08 | Toshiba Corp | モデル予測制御装置 |
JPH0883104A (ja) * | 1994-09-12 | 1996-03-26 | Toshiba Corp | プラント制御装置 |
JPH09128011A (ja) * | 1995-10-31 | 1997-05-16 | Fuji Electric Co Ltd | モデル予測制御方法 |
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 |
JP2003005804A (ja) * | 2001-06-19 | 2003-01-08 | Honda Motor Co Ltd | プラントの制御装置 |
JP2004094939A (ja) * | 2002-08-09 | 2004-03-25 | Omron Corp | モデル構造、制御装置、温度調節器および熱処理装置 |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2018105412A1 (ja) * | 2016-12-07 | 2018-06-14 | ソニー株式会社 | 情報処理装置および方法、並びにプログラム |
US10795326B2 (en) | 2016-12-07 | 2020-10-06 | Sony Corporation | Information processing apparatus, and method |
Also Published As
Publication number | Publication date |
---|---|
CN101390024A (zh) | 2009-03-18 |
JP2007272498A (ja) | 2007-10-18 |
US20090012632A1 (en) | 2009-01-08 |
US8155763B2 (en) | 2012-04-10 |
JP4952025B2 (ja) | 2012-06-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2007116590A1 (ja) | 運転制御方法,運転制御装置及び運転制御システム | |
JP4698656B2 (ja) | 制御システムおよび制御支援装置 | |
CN105313120A (zh) | 机器人程序修正系统 | |
CN108994818B (zh) | 控制装置以及机器学习装置 | |
JP4811997B2 (ja) | 状態推定装置、状態推定システム及びコンピュータプログラム | |
WO2020183864A1 (ja) | 自動操縦ロボットを制御する操作推論学習モデルの学習システム及び学習方法 | |
CN109849025A (zh) | 振动抑制装置 | |
EP0540168B1 (en) | Simulator using a neural network | |
JP6908144B1 (ja) | 自動操縦ロボットの制御装置及び制御方法 | |
JPH06266449A (ja) | 調整入力利用のカスケード炉温制御装置 | |
US11137728B2 (en) | Processing device, control parameter determination method, and non-transitory recording medium storing a control parameter determination program | |
KR102382047B1 (ko) | Pso를 이용한 모터 제어기의 자동 학습 튜닝 시스템 | |
DeBoon et al. | Multi-objective gain optimizer for a multi-input active disturbance rejection controller: Application to series elastic actuators | |
US20200379452A1 (en) | Plant operating condition setting support system, learning device, and operating condition setting support device | |
Kostadinov et al. | Online weight-adaptive nonlinear model predictive control | |
CN109507879B (zh) | 控制装置以及机器学习装置 | |
JP4546332B2 (ja) | 運転支援装置および運転支援方法 | |
CN116047888A (zh) | 一种基于bp神经网络pid的自平衡车的控制方法 | |
JP3214270B2 (ja) | 電動機の速度制御装置並びに速度及び位置制御装置 | |
WO2020162200A1 (ja) | 制御装置および制御プログラム | |
CN114761182B (zh) | 机器人控制系统和用于训练所述机器人控制系统的方法 | |
JP3040901B2 (ja) | ニューラルネットワークによる制御方法および内蔵制御装置 | |
CN101943889A (zh) | 使电气传动系统调节器自动启动和/或运行的方法及装置 | |
EP3835894A1 (en) | Method and apparatus for configuring a machine controller | |
CN115298529A (zh) | 被测体自动驾驶装置、被测体测试系统、指令车速生成程序及被测体自动驾驶方法 |
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: 07706986 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 6948/DELNP/2008 Country of ref document: IN |
|
WWE | Wipo information: entry into national phase |
Ref document number: 12279350 Country of ref document: US |
|
WWE | Wipo information: entry into national phase |
Ref document number: 200780006110.7 Country of ref document: CN |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 07706986 Country of ref document: EP Kind code of ref document: A1 |