A Two Level Hierarchical Control Strategy Applied to an Intelligent House

João M. G. Figueiredo José M. G. Sá da Costa Mechatronics Eng. Dept. Instituto Superior Técnico. University Évora, IDMEC Technical University Lisbon, IDMEC

R. Romão Ramalho, 59, 7000-671 Évora, Portugal Av. Rovisco Pais, 1049-001 Lisboa, Portugal jfig@uevora.pt sadacosta@dem.ist.utl.pt

Abstract - Houses are becoming intelligent in our days. It is expected that in the near future the instrumentation and automation solutions that have been entering in our cars, in the last decade, are going to expand into the house market. The future intelligent house is a system that interacts with its inhabitants showing information about the system state, alarm occurrences and making suggestions to overcome emerging problems. In this paper a two-level hierarchical control strategy is applied to an intelligent house. The developed control architecture is based on an industrial PLC network that is managed by a SCADA supervisor system. The first control loop is managed locally by each PLC, controlling its own process. The second control loop integrates the local information from each slave-PLC and manages the system globally, through the developed SCADA supervisor control algorithm. Tests on a prototype are presented in this paper and the experiments show the capability of this two-level hierarchical control strategy to monitor and control intelligent houses, preventing accidents and improving home comfort. Systems Automation, Supervisory Control.

I. INTRODUCTION

In the last decade, with the development of communications, especially mobile phones and internet-connected systems, the motivation to instrument the private homes has been growing.

In the past, due to the huge investment required in equipment and qualified personal, the well known heavy automatic systems were owned only by big public buildings such as banks, companies, commercial surfaces… These old typical automatic systems usually had the capability to control the electric power system, the air conditioning and the fire and assault alarms. Today, thanks mainly to communications and system automation developments these automatic systems have become much more cheaper, flexible and powerful, what promote their spread into other domain areas, traditionally not automated, [1]- [3], as small private property [4]-[7].

In the last years simpler and cheaper automation solutions have been introduced in the market, promoting its use in generalized private property. The intense automation that has been introduced in our cars, in the last decade, is now becoming to enter in our homes. Alarm systems for temperature, gas or water leakages, property violation… can now be controlled, exclusively by the owner, using remote capabilities such as mobile phones (GSM services) or Internet-connected systems.

Today homes rely on several electric/electronic

equipments and networks that provide quality of living, safety and comfort. A Scada system is developed to integrate these different types of information giving a quick overview about the state of the overall system and having also the capability to dive deep into the system showing each sensor state located in one specified control area. The SCADA supervisor system is developed as an integrator for the several technologies present in modern houses. Today the main domotic systems are grouped into 9 groups [8]:

- flow and discharge systems; - electric energy systems; - ventilation and temperature control systems; - computer network; - lightning control systems; - fire control systems; - property safety systems; - entrance control systems; - mechanical control systems.

In this paper a two-level hierarchical control strategy is

developed and applied to a complex building, named as intelligent building. In this strategy the first control loop is assigned to locally independent PLCs (Programmable Logic Controllers) connected in a network, exchanging control signals with an upper level control loop, managed by a SCADA supervisor system (Supervisor Control And Data Acquisition). The developed strategy is internet available permitting the remotely management of the entire system [9].

II. SYSTEM MODEL

The instrumented building is modeled as tree-structure, composed by two main sub-structures: Floors and Rooms.

This structure provides flexibility to this model as it can deal with both small/simple and big/complex buildings (Fig. 1).

Following this modular model structure, each room is considered as an autonomous unit with its independent monitoring and control activities. The SCADA application is developed according the model-tree structure presented above, allowing the user to go down from the general state view of the entire Building/House to each floor, descending continuously to each room and reaching the final elementary chain devices such as sensors or actuators.

Each floor is modeled as a set of rooms and each room has

1-3 September, 2005 • Faro, Portugal

92 0-7803-9030-X/05/$20.00 ©2005 IEEE.

several sensors and actuators.

Fig. 1 – Intelligent Home as a tree-structure Model

The set of actuators and sensors considered in each room are mainly grouped into 2 groups:

- Group A: leaving rooms and bedrooms; - Group B: kitchen and toilets.

The set of sensors and actuators considered in both groups are shown in fig. 2.

Group A: sensors and actuators

Group B: sensors and actuators

Fig. 2 – Intelligent Home: Sensors and Actuators An additional feature of this model that provides the

system huge accessibility to their users is the SCADA platform where this model was implemented which allows the system to be connected trough the internet. This feature will be presented in chapter III.

III. TWO-LEVEL CONTROL STRATEGY

A. Description of the Control Strategy

The strategy developed in this paper is commonly known as a two-level hierarchical control as it integrates a first control loop that is managed by local PLCs with a second loop which is performed by a SCADA supervisor system

that monitors globally the several distributed local systems. Figures 3 and 4 illustrate this control strategy: The inner control (first loop) and the outer control loop.

Fig. 3 – First Loop Control – PLC Local Control

Fig. 4 – Second Loop Control – SCADA Supervisor Control Applying this strategy to a complex building that is

instrumented and monitored through a SCADA supervisor system, we can manage globally the entire net of field PLCs that control locally each own process.

The upper level control law, having a global system overview, generates the set of references for each local process (PLC) avoiding possible conflicts in emergency situations. The input functions for the upper control loop are mentioned as comfort laws, safety laws (F1(t), … ,Fj(t)).

B. PLC Network and SCADA supervisor

The developed strategy to cope with complex buildings with a huge set of geographically distributed actuators and sensors is implemented through a PLC network (fig. 5) consisting of several slaves PLCs connected to a master PLC via Profibus/ DP network [9].

Each slave PLC hosts several control programs which selection is made either locally, via an HMI (Human Machine Interface) or remotely, via the master PLC (PLC

Intelligent House

Floor

Floor

Room 1.1

Room 1.2

Room 2.2

Room 2.1

Group A

Lights sensors

Mov. sensors

Light actuators

Alarm actuators

Temp. sensors Air Cond. actuators

Smoke sensors Exhaust actuators

Group B

Lights sensors

Mov. sensors

Light actuators

Alarm actuators

Temp. sensors Air Cond. actuators

Smoke sensors Exhaust actuators

Water sensors

Gas sensors

Central Water supply

Central Gas supply

PLC

Ref 1.1

1

Ref 1.2

Ref 1.n q 1.n

q 1.2

q 1.1 +

+

+

-

-

-

PLC

Ref 1.i

1

q 1.i +

-

PLC

Ref k.i

k

q k.i +

-

q 1.i q k.i F1(t)

Fj(t)SCADA Supervisor

Ref 1.i

Ref k.i

1-3 September, 2005 • Faro, Portugal

93

0), which is connected to the server PC, via RS232/ MPI Siemens protocol, where the SCADA application is running.

The server PC is simultaneous a SCADA server and an internet server, as the implemented SCADA application is web enabled. All process variables are available at the SCADA PC as these variables are on-line available through a Profibus/ DP connection protocol.

C. SCADA Application Development

The SCADA system used to implement this monitoring and control strategy was a commercial platform: Axeda Supervisor Wizcon for Windows & Internet V 8.2. This application permits the selective access to the application depending on the user’s responsibility degree. In this paper we developed three user levels: Operators, Supervisors and Administrators. Among the main functionalities of a SCADA system there is the so called “Tag”. A “Tag” is a defined variable that permits the change of information between the PLC network and the SCADA system, in a real-time environment. There are usually three types of Tags: PLC, Dummy and Compound. In the PLC Tags the PLC sets the variable values that are directly transferred to the Scada program. In the Dummy Tags the value is set by the user on the Scada interface and transferred to the PLC address. Finally Compound Tags are set by the Scada program, following the programmed operations.

Fig. 5 – PLC Network and SCADA supervisor

When defining the Tags a set of information is required, namely: Tag name, Description, Tag source, Driver, Sample rate, Address and Tag Type. An example of such Tag definition is illustrated in Fig. 6.

The Driver used in this application to establish the communication Scada system - PLC network was the Siemens PC/MPI interface, via the Siemens software Simatic S7 Prodave. According to this communication Protocol, a PLC digital variable address must be defined as NNTXAAAABB, where:

NN = PLC address (from 0 to 32); T = variable type (I for Input, Q for Output and M for memory); X = variable length (B for Byte, W for Word); AAAA = slot address (from 0000 to 9999);

BB = variable address (00 to 17). According this protocol the address 02MB000101

represents the memory M0.1 in the developed Scada application.

Fig. 6 – Tag definition The Scada main interface between the system and the

user are the application images. The image building is a functionality of all SCADA systems and its main function is to permit the user a quick visual identification of all system functional characteristics. An easy identification of the system Inputs and Outputs permits the user an effective monitoring and a rapid actuation on the process when it is necessary.

The application developed used important features that we named animated images. These images change geometry characteristics and colors when their digital variables change the state (on/ off). In figure 7 the configuration menu for the animated images is shown.

Fig. 7 – Animated Images

SCADA-PC Workstation

SCADA-PC Workstation

SCADA PC-Server

PLC Master

PLC 1Slave

PLC 2Slave

PLC 3Slave

Profibus/ DP

(RS 232/ MPI)

WISP 2005

94

D. PLC Application Development The PLCs used to implement this monitoring and control strategy were the Siemens S7-3**. Each modular PLC used was composed by:

Slot1 = Power Supply Module Slot2 = CPU Module Slot4 = Communications Module (Profibus net) Slot5 = Digital Input/Output Module Slot6 = Digital Input/Output Module Slot7 = Analog Input/Output Module

The Siemens software used to implement the PLC application was the Simatic Manager. The program language used was the Ladder Diagram (LAD).

Several algorithms have been developed for temperature control, gas and water leakages and property violation.

D1. Property Violation Control The control algorithm developed for property violation,

was implemented following the Grafcet methodology, which diagram is presented in Fig. 8. Variables si represent electric switches assembled to doors and windows. Variables zi represent proximity sensors detecting movements or unexpected volumes in prescribed areas. Variables Ai and Bi represent acoustic alarm actuators, variables Ci and Di represent visual alarm actuators. A Reset switch turns off the alarm system.

Fig. 8 – Grafcet for Property Violation According with the Tag definition mentioned in III.B, the

input variables si and zi were defined as PLC-Tags. The output variables Ai and Ci were defined as PLC-Tags, which activation depends on the PLC program variables. The output variables Bi and Di were defined as Dummy-Tags as the user can switch on or switch off the alarm actuators via the Scada system, locally on the Scada PC-server or remotely if the PC-server is connected via internet.

In Fig. 9 the LAD program developed for property violation is presented.

D2. Water Leakage Control

The PLC control algorithm developed for water leakage, was also implemented following the Grafcet methodology, which diagram is presented in Fig. 10. Variables li represent electric logic sensors to detect water presence and variables Wi represent the central electro-valve that feeds all the water network in the house. Output variables Mi represent

alarm messages to the system.

Fig. 9 – LAD program for Property Violation

Fig. 10 – Grafcet for Water Leakage

The main difference between this algorithm and the previous algorithm for property violation is that the system cannot be reset remotely, by the operator, because the water network must be prior repaired, before the system can be operated again. In this way, the control system when a leakage problem is detected, closes immediately the primary water distribution electro-valve and sends a failure message to the operator and remains on failure mode until the system is restarted. To restart the system the command is

Ci on

0 Ai off/ Ci off

Bi off/ Di off

1 2 Bi on Di on

Ai on

zi si

Reset

+

R

m0.0 A i / B i

R

R

S

-

m0.0

C i / D i

z i m0.0

m0.1

S

m0.1 A i

S

C i

R

S

m0.0 s i m0.0

m0.2

S

m0.2 B i

S

D i

S

Reset m0.0

R

m0.1

R

m0.2

0 Wi on

1 Wi off

l i

Local Reset

Mi on

Mi off

1-3 September, 2005 • Faro, Portugal

95

only accepted if it is given locally, through the Scada server-PC, in order that the system can be locally repaired.

In Fig. 11 the LAD program developed for water leakage control is presented.

Similar Grafcet projects and LAD programs had been developed for the other house control attributes: temperature control, gas leakage and energy control.

Fig. 11 – LAD program for Water Leakage Control

IV. EXPERIMENTAL RESULTS

The developed application to monitor and control intelligent homes have been implemented in an experimental setup with the following software and hardware requirements:

A. Software Requirements

The Scada system was developed over the platform Axeda Supervisor Wizcon for Windows & Internet V8.2.

The PLCs programming used the Siemens Simatic Manager V5.2, with the Simatic S7 Prodave V5.5 needed for the communication between the Scada system and the PLC network.

B. Hardware requirements The PLC network implemented had two PLCs Siemens

S7-3** (one PLC controlling each home floor). Each Siemens PLC was composed by the following modules:

Slot1 = Power supply PS 307 2A Slot2 = CPU 315 Slot4 = CP 342 - 5 Slot5 = DI8/ DO8xDC24V/0,5A Slot6 = DI8/ DO8xDC24V/0,5A Slot7 = AI4/ AO2x8/ 8bit

The PLC which controls the Ground-Floor was set as the

master PLC and it was connected to the Scada System via the Siemens PC/MPI protocol.

The sensors used were standard industrial or domestic sensors for proximity, temperature, gas, water and movement. Their main technical characteristics are shown below:

- Temperature: Pt100, Labfacility, (-50oC …+250oC); - Safety: micro-switch RS V123 1C5 (250V/ 16A) and movement sensor Gardiner Techn. 110o/ 12m and proximity sensor Telemecanique XULM06031; - Gases: RS 286-636 for CO and Butane; - Water level: Crydom SSF212XP.

The actuators used were standard industrial or domestic actuators for water flow, gas flow and current flow. Their main technical characteristics are shown below,

- Water flow: electro-valve Burquert, ½”, 240 VAC; - Gas flow: electro-valve Burquert, ¼”, 240 VAC - Current flow: Telemec contactor LC1D09BL (240VAC/ 4KW);

In Fig. 12 a part of the experimental setup, that was built

to test the developed control strategy is shown (setup for property violation).

Fig. 12 – Experimental Setup for Property Violation

C. Developed Application Menus

Several Scada menus had been built. The main characteristic of a Scada Menu is to be simple, explicit and quick on transmitting the information to the operator or to the house owner. The first Menu developed is a general overview Menu, named “Monitoring” which give us access to the several floors of the intelligent house and permits the operator to exit the Scada control and monitoring application. This Menu is illustrated in Fig. 13.

Climbing the tree structure from the main menu we can reach each anyone of the floors that compose the house. In On Fig. 14, the Ground-Floor is shown. This Menu shows all the rooms located on the Ground-Floor and permits to go

+

S

m0.0

R

R

S

-

m0.0 l i m0.0

m0.1

R

m0.1 W i

S

M i

S

Local Reset m0.0

R

m0.1

W i

M i

WISP 2005

96

to each one of the rooms in this floor or return to the main menu.

Fig. 13 – Scada Main Menu: Overview Intelligent House

Fig. 14 – Scada Menu: Ground Floor

Entering in one specific room we get the information about all the sensors located in this room: lights, water and gas leakages and temperature. The programmed animated images characterize the sensor state by a change of color: green indicates the sensor is activated and red indicates the sensor is inactive. All the sensors used to test this application were binary sensors. In Figure 15 it is shown the Scada Menu developed for the house kitchen.

V. CONCLUSIONS

This paper is a valuable contribution to see the advantages of industrial products when applied to control and monitor several systems that are traditionally not automated. The automation products market is nowadays price fair and accessible to a large number of people, not only the expert companies. The automation of private houses and buildings is today a market reality that is growing fast. This paper is a contribution to show a robust strategy to follow in the automation of private property as it develops the so called

two-level hierarchical control strategy that has already proven its qualities in the industrial environment.

Fig. 15 – Scada Menu: Kitchen (Ground Floor)

REFERENCES

[1] M. Almeida, J. Figueiredo, M. Rijo, “SCADA Configuration and

Control Modes Implementation on an Experimental Water Supply Canal” in Conf. Proc. 2002 10th IEEE Mediterranean Conf. On Control and Automation – MED2002, Lisbon, Portugal.

[2] A. Neto, J. Martins, S. Zolnier, P. Monteiro, F. Baeta “1-Wire ™ System Performance – An Application for Temperature Measurement in Agricultural Buildings” in Conf. Proc. 2004 6th Conf. On Automatic Control – CONTROLO 2004, Faro, Portugal, pp. 445-450.

[3] T. Ratinho, J. Figueiredo, M. Rijo “Modelling, Control and Field Tests on an Experimental Irrigation Canal” in Conf. Proc. 2002 10th IEEE Mediterranean Conf. On Control and Automation – MED2002, Lisbon, Portugal.

[4] H. Christensen “Domestic Robots” in Conf. Proc. 2002 10th IEEE Mediterranean Conf. On Control and Automation – MED2002, Lisbon, Portugal.

[5] B. Brumitt, B. Meyers, J. Krumm, A. Kern, S. Shafer “Multi-Camera Multi-Person Tracking for Easy Living” in Conf. Proc. 2000 3rd IEEE Intl. Workshop on Visual Surveillance, 2000.

[6] M. Coen, B. Phillips, N. Warshawsky, L. Weisman, S. Peters, P. Finin “Meeting the Computation Needs of Intelligent Environments: The Metaglue System” in Conf. Proc. 1999 MANSE’99, 1999.

[7] R. Orr, G. Abowd “The smart Floor: A Mechanism for Natural User Identification and Tracking” in Conf. Proc. 2001 Intl. Conf. On Human Factors in Computing Systems, Netherlands, 2001.

[8] A. Dey, G. Abowd, D. Salber, A Context-based Infrastructure for Smart Environments, Managing Interactions in Smart Environments. Springer-Verlag, 1999, pp. 114-130.

[9] J. Figueiredo, M. Botto “Automatic Control Strategies Implemented on a Water Canal Prototype” in Conf. Proc. 2005 16th IFAC World Congress – Intl. Federation of Automatic Control, Praha, Czech Republic.

1-3 September, 2005 • Faro, Portugal

97