トヨタ自動車株式会社東京本社3F応接302

第１回MBD調査研究会運営委員会

2008年1月7日

大畠 明

トヨタ自動車株式会社

第1回運営委員会アジェンダ１．日時：2008年1月7日、13:00～17:00２．場所：トヨタ自動車株式会社，東京本社

3F 応接室３．議事：

(1) 主査挨拶(2) MBD調査研究会の設立状況（大畠）

-設立趣意書の確認-制御部門、部門協議会、理事会承認

(3) 部門協議会新体制について（大畠）-部門協議会規程改定-部門連携・活性化専門委員会規程-部門協議会技術委員会規程-部門協議会下の技術委員会について・ライフサイエンス・ES (Embedded System)

(4) ES技術委員会の説明(船橋）(5) ES技術委員会とMBD調査研究会との関係(6) MBD調査研究会委員登録の確認

(7) MBD調査研究会ホームページ,-掲載内容- JMAABホームページとのリンク

- その他(8) 運営基本方針

-委員募集について(各自募集、名簿公開可否確認）-活動計画について・研究会活動・OSへの参加・Conferenceなど・ES技術委員会との連携・先進パワトレ制御研究会・IFAC ＆ SICE Symposium on Theory and Technology

for Automotive Controlについて・その他の連携

-自動車技術会との連携について(9) その他

経緯

2007年 9月7日：制御部門運営委員会で設置承認(調査研究会規程により正式決定）

2007年11月9日：部門協議会で設置報告2007年12月7日：理事会にて設置報告

2007年11月1日：JMAAB Open Conferenceにて報告（参加者500名以上）

2007年12月1日：JMAAB Core Member会議で報告（参加会社自動車OEM, Tire1 ECUSuppliers15社、TMW, CSC）

JMAAB Core Member Meeting

Collaboration withSICE Research Committees

November 29th, 2007Alex Ohata

TOYOTA MOTR CORPORATION

Background 1. JMAAB is a voluntary user group of Matlab.

2. The JMAAB core members were limited to OEMs and tire 1 ECU suppliers. (Recently, the policy has been changed to ease the condition.)

3. The operation and administration may be getting troublesome when many members participate in JMAAB.

4. JMAAB isn’t a corporate organization and doesn’t have the chair.

5. The activities which need a payment and a contract are difficult.

Confliction1. JMAAB wants to deploy MBD widely. However, many

members may make the management difficult.

2. JMAAB thinks the tool chain is important. However, discussions with other tool suppliers may be restricted.

3. Supports from academic parties may be helpful but it is difficult to get them because of the confidentiality.

A different party could be necessary.

Purpose of Collaboration with Academic Parties

To deploy MBD technologies more widely

1. Collaboration with academic parties

2. To deploy MBD to more widely

3. Activities to need cost and contracts

Collaboration with SICE Research Committee on Model-Based Development for Embedded Control Systems

Why SICE?

1. SICE leads “the Trasdisciplinary Federation of Society and Technology” consisting of 47 academic societies.

2. The major academic party in control system design area.

3. The most internationalized party.(Official Language of Annual Conference is English.）（Strong collaboration with IFAC and IEEE.）

4. They understand Matlab very well.

5. Cybernet and almost JMAAB member companies are members supporting SICE.

Organization of Control Dept.

Operational Committee

Control Theory Committee

Control Engineering Committee

Research Committee on Advanced Powertrain Control

Research Committee on Model-Based Development for Embedded control systems

Control Dept.

The competition of control design with an V6 engine model written with Simulink provider from TMC.

MBD will be redefined from the academic point of view and make the collaboration network.

Purpose of MBD Research Committee

The purpose of this committee is to define :

1. MBD Process consisting of plant modeling, control design, implementation, verification & validation and calibration environments

2. the collaborations between academic parties and industries

3. The integration of advanced methodologies, including modeling, simulation technologies, Identification, control theories, verification, data management, project management and software engineering

4. How to apply it to the processes from planning to production and maintenance phases.

Term, Organization and ActivitiesTerm: January/2008 ～ December/2009Organization:

Chair Seiichi Shin (UEC)Co-chair Shigeru Oho(Hitachi Laboratory)Secretary Kazushi Nakano(UEC)

Alex Ohata (Toyota)Activity:To redefine MBD based on the one defined by JMAAB MBD Promotion WG

To get a big amount of budget from the government

To make the collaboration network with other academic parties.

Support from JMAAB

1. Announcement of MBD research committee’s activities within JMAAB

2. Linkage with each home page

3. JMAAB member’s presentation in the committee.

JMAAB members will be given preference to be a member of the MBD research committee.

Summary

1. SICE research committee will be established in the next January.

2. That is a voluntary activity to deploy MBD to suppliers widely.

3. The major activity is to make the collaboration network to realize MBD.

4. Matlab will be the central tool to construct the MBD tool chain.

5. “The 1st Symposium on Theory and Technology for Automotive Control Systems” will be held.

先進ﾊﾟﾜｰﾄﾚｰﾝ制御調査研究会

１．TMC提供モデルによるエンジン始動問題（6チャレンジャーが07年制御理論ｼﾝﾎﾟｼﾞｭｰﾑで報告）（IFAC2008にて、OS）

２．2008年度延長（トルク制御問題を新設定）

３．エンジンモデルと始動問題に関する出版を計画中

４．IFACとの共催ｼﾝﾎﾟｼﾞｭｰﾑを計画中1st Symposium on Theory and Technology

for Powertrain Control

SICE Annual Conference 2007 in KagawaSICE Annual Conference 2007 in Kagawa

Benchmark Problemfor Automotive Engine Control

September 19th, 2007

Akira OHATA (Toyota Motor Corporation)Junichi KAKO (Toyota Motor Corporation)Tielong SHEN (Sophia University)Kazuhisa ITOH (Tottori University)

The SICE Technical Committee on Advanced Powertrain Control

Contents

1. Motivation and purpose

2. Provided engine model

3. Benchmark problem

4. Overview of intermediate results

5. Summary

Contents

1. Motivation and purpose

2. Provided engine model

3. Benchmark problem

4. Overview of intermediate results

5. Summary

MotivationsAn advanced control hasn’t been continuously used even if it is put into the market once

Possible reasons

Academic side problems Industry side problems

difficult to catchrequirements from industryno engine test facility

difficult to know the advantagesfew mature controlengineers

Purpose

SICE Technical Committee on Advanced Powertrain Controlwas established in June, 2006.

( It provides a V6 engine model and a control problem. )Problem: Start the engine model!

1. To identify the gaps between academic and industrial sides

2. To find the way removing gaps if necessary

3. To improve the communication between both at least

Contents

1. Motivation and purpose

2. Provided engine model

3. Benchmark problem

4. Overview of intermediate results

5. Summary

Targeted Engine

1pc_L

6Conn10

5Conn8

4Conn7

3Conn5

2Conn3

1Conn1

cyl5

cyl3

cyl1

Mux

Atomspher

Atomspher

Atomspher

1pc_L

6Conn10

5Conn8

4Conn7

3Conn5

2Conn3

1Conn1

cyl5

cyl3

cyl1

Mux

Atomspher

Atomspher

Atomspher

V6 spark ignition engine model

4

mt

3

st

2

pr

1

rps

1

Startor

pcRConn1

Conn4

Conn7

Conn3

Conn6

Conn9

right_bank

pc_LConn1

Conn5

Conn8

Conn3

Conn7

Conn10

left_bank

darea TH_fld

U_st TH_fl

Throttle_valve

st pr

Pres_sensor

Piston_Crankshaft

V_state

IN_ch

IN_port1

IN_port3

IN_port5

IN_port2

IN_port4

IN_port6

Intake_chamber

[pc_R]

[pc_L]

Atomspher

fl flow

Air_flow_sensor

1

darea

Atmosphere

Top Layer of the Engine Model

mt

st_rad

Pa_pm

rps

rpm

kPa_pm

mt_mg/s

st_deg

monitor

Spark advance

Injected fuel

Throttle angle

Torque

mt

st

pr

rps

engine

rpm

Crank angle

Air flow rate

ECU_dth

ECU_fi

ECU_sa

controller

STOP

ECU_rps torqe

Starter Scope

>

An Example of Simulation Results

0 1 2 3 4 5 6 7 8 9 100

500100015002000

0 1 2 3 4 5 6 7 8 9 1002468

Engi

ne sp

eed

(rpm

)C

ylin

der p

ress

ure

(MPa

)

reference

Air fuel ratio = stoichiometric, Spark Advance = constantAir fuel ratio = stoichiometric, Spark Advance = constant

Time (s)

Contents

1. Motivation and purpose

2. Provided engine model

3. Benchmark problem

4. Overview of intermediate results

5. Summary

Requirements for the Control Design

The engine speed must be confined in this area!

rpm50±

Requirements for steady state performance （evaluation）1. Closed loop must be stable. OK2. Regulate mean rpm at 650rpm OK

Requirements for transient performance1. Control the engine speed at within 1.5s NG2. The overshoot must be sufficiently suppressed. NG3. Hunting must not occur. OK

Requirements for robustness1. Initial crank angle 2. friction torque 3. fuel evaporation characteristic

rpm50650

0 2 4 6 8 100

500

1000

1500

2000

0 2 4 6 8 100

500

1000

1500

2000

Engi

ne sp

eed

(rpm

)

Time (s)

±

Key Feature of the Engine Behavior

massFuelmassAir

AF = limitlowerLL : limitupperUL :

The rpm increases.

Fire&

Motor inactive

Misfire&

Motor inactive

Misfire&

Motor active

400≥rpm

Key ON

Fire&

Motor active

250=rpm

ULAFLL ≤≤

ULAFLL ≤≤

AFULLLAF

<<

AFULLLAF

<<

The rpm decreases.

Control Strategy

0 2 4 6 8 100

500

1000

1500

2000

Regulate rpm with throttle and spark advance

Robust start with fuel injection

Suppress the over shoot with throttle and spark advance

Engi

ne sp

eed

(rpm

)

time (s)

Complexity of the Control Design1. Change of characteristic

Firing : stable but strongly nonlinearMisfiring : unstable and slightly nonlinear

2. Redundancy of manipulations Throttle angle, fuel injection, spark advance

3. Time variant control strategy

4. Combination of feedback and feedforward

5. Hybrid control design

Contents

1. Motivation and purpose

2. Provided engine model

3. Benchmark problem

4. Overview of intermediate results

5. Summary

Challengers

1. Currently, 9 challengers mainly from universities are grappling with the benchmark problem.

2. 6 challengers reported their intermediate results in the 36th SICE Symposium on Control Theory in Sapporo on September 7th, 2007.

Approaches1. The model blocks of the intake and the fuel are used in

the control without the simplification.

2. Deriving a simple model by the physical consideration and the control is designed with advanced control theory.

(similar to the existing approach in the industry)

3. Simulation data based approach

4. Deriving optimal time series of the manipulations with a numerical optimization and combining an advanced control, for example MPC

5. Eccentric approach by using misfire

An Example of Design Results

time (s)0 1 2 3 4 5

0

200

400

600

800reference

Engi

ne S

peed

(rpm

)

A Key for Success The succeeded control designs have the below architecture.

Engine speed*/

Reference air fuel ratio

Inverse fuel model

Fuel injection

Crank angle

Reference engine speed Spark advance

Throttle angle

*/

Advancedcontrol

Air charge

estimationAir flow rate

Desired cylinder fuel

Contents

1. Motivation and purpose

2. Provided engine model

3. Benchmark problem

4. Overview of intermediate results

5. Summary

Summary

1. Only a single framework didn’t solve the problem.

2. A carefully worked-out plan was essential to succeed.

3. The control architecture would be a key to succeed.

Observations

Intermediate results were reported by six challengers in the SICE symposium on control theory on September 7th.

Almost all could succeed starting the engine model but the robustness of the designed controls weren’t validated yet.

新体制と部門間連携

技術専門委員会(ES)

システムインテグ

レーション部門

部門協議会

技術専門委員会（ﾗｲﾌｻｲｴﾝｽ）

部門連携・活性化専門委員会

計測部門

制御部門

システム情報部門

産業部門

先端融合部門

委員選出

連携の可能性

2008年度予算案

0.本年度活動費 \200,000

1.会議費 会議費(\2,500×4回) \10,000

2.研究会 講師謝礼（1名×\40,000×3回) \120,000

参加費(\3,000×50名） \150,000 講演会小計 \185,000会場費 \40,000アルバイト \25,000講師謝礼（3名×\40,000） \120,000

参加費(\3,000×50名） \150,000 講演会小計 \185,000会場費 \40,000アルバイト \25,000講師謝礼（3名×\40,000） \120,000

収入合計 \500,000 支出合計 \500,000

3.第１回MBD技術講演会 （H20.5）

4.第2回MBD技術講演会（H20.12）

H20年度(活動１年目〕予算案2007年11月24日

制御部門:組み込み制御システムのモデルベース調査

収入 支出

MBD会議計画1.会議開催計画： 1月7日

２月（合宿強羅ﾚﾍﾞﾙ合せ、組織と運営方法承認など）５月（MBD定義と目標）：８月（MBD定義と目標決定）：１１月(研究領域と各種連携定義）：

２．研究会計画：年３回JMAAB嶋田氏, IAV, A&D,未定3月：UEC、９月、12月

３．MBDフォーラム：年２回5月,8月,12月（参加者80名程度を予定）１回目UEC会告（1.5hr×2レクチャー）

４．開催形態：調査研究会委員のよる議論５．開催場所：運営委員の大学、会社会議室など６．SICE Annual Conference MBD-OS, Tutorial

MBD調査研究会ホームページ１．早稲田大学大貝先生がサイト提供とメンテナンスを了承２．ホームページ掲載項目案① 趣意書（目的、背景など）② 組織③ 委員名簿④ MBDの定義⑤ 用語集⑥ 主査挨拶⑦ 活動計画（イベント）⑧ 関連リンク（SICE HP, JMAAB HP, EHS,

先進パワトレ調査研究会, Maplesoft, TMW, dSPACE, ADT,ETS, tool venders, CATS, ニッキ,サンリツ，GAIO, VAST, etc）

ツールベンダー連合会としての位置づけ！