MIR DayMIR Day International WorkshopInternational Workshop

GHRCE, GHRCE, NagpurNagpur

Presentation on Presentation on Identification of Vehicle Class & Speed for Mixed Sensor TechnolIdentification of Vehicle Class & Speed for Mixed Sensor Technology using Hybrid ogy using Hybrid

fuzzy neural genetic algorithmfuzzy neural genetic algorithm: A Design ApproachPrashantPrashant Sharma , Dr. Sharma , Dr. PreetiPreeti BajajBajaj , Dr. , Dr. AjithAjith AbrahimAbrahim

ByByPrashantPrashant SharmaSharmaResearch ScholarResearch ScholarGHRCE, GHRCE, NagpurNagpur

Mentor/GuideMentor/GuideDr. Dr. PreetiPreeti BajajBajaj

Principal GHRCE, Principal GHRCE, NgpNgpCoordinator MIR Coordinator MIR Labs(IndiaLabs(India))

ContentsContents

1.1. INTRODUCTIONINTRODUCTION2.2. LITREATURE SURVEYLITREATURE SURVEY3.3. DESIGN AND IMPLEMENTATION DESIGN AND IMPLEMENTATION 4.4. RESULTSRESULTS5.5. CONCLUSION AND FUTURE SCOPECONCLUSION AND FUTURE SCOPE

1.Introduction1.Introduction

Identification of vehicle class and speed is very important paraIdentification of vehicle class and speed is very important parameter for road meter for road traffic management. It is highly essential for deciding traffic management. It is highly essential for deciding

Type of vehicle to be allowed on the roadType of vehicle to be allowed on the road

Speed limit of the vehicle passing through the roadSpeed limit of the vehicle passing through the road

Presently a very huge infrastructure and large manpower is involPresently a very huge infrastructure and large manpower is involved for ved for extracting the above two informationextracting the above two information

In order to provide lucid solution for the above difficulties, In order to provide lucid solution for the above difficulties, work has been work has been going on very intensively all around the world. going on very intensively all around the world.

This paper tries to present one of the solution for the above taThis paper tries to present one of the solution for the above task using Fuzzy sk using Fuzzy Logic ControllerLogic Controller

it also suggest the design approach to improve the performance it also suggest the design approach to improve the performance of the of the controller by hybridizing fuzzy logic with neural network and controller by hybridizing fuzzy logic with neural network and

finally optimization of hybrid fuzzy finally optimization of hybrid fuzzy neuroneuro controller using genetic algorithm.controller using genetic algorithm.

ContdContd ……

For achieving the above task the signals are obtained from For achieving the above task the signals are obtained from the sensorsthe sensors

Inductive loop sensor provides information forInductive loop sensor provides information foraxle distance axle distance height of the chassisheight of the chassis

Microwave and infrared sensors provides information for Microwave and infrared sensors provides information for length of the vehicle andlength of the vehicle andspeedspeed

These conclusions are based on the report submitted to These conclusions are based on the report submitted to Federal Highway Administrations (FHWA) Intelligent Federal Highway Administrations (FHWA) Intelligent

Transportation Systems Joint Program Office.Transportation Systems Joint Program Office.

Contd..Contd..

The inputs to controller areThe inputs to controller are1.1. Axle DistanceAxle Distance2.2. Height of ChassisHeight of Chassis3.3. Vehicle body length Vehicle body length 4.4. Occupancy timeOccupancy timeThe outputs are The outputs are 1.1. Vehicle Class Vehicle Class 2.2. Speed of vehicleSpeed of vehicle

2.Literature Survey2.Literature Survey

Sensor technologySensor technology

Fuzzy Logic ControllerFuzzy Logic Controller

Fuzzy neural networkFuzzy neural network

Fuzzy neural controller optimized by Fuzzy neural controller optimized by genetic algorithmgenetic algorithm

Block diagram of fuzzy logic Block diagram of fuzzy logic controllercontroller

Fuzzy Logic Controller GUI window Fuzzy Logic Controller GUI window in MATLABin MATLAB

NeuroNeuro Fuzzy SystemFuzzy System

Genetic Algorithm for optimization Genetic Algorithm for optimization of NF Systemof NF System

Design approach uses hybrid genetic algorithm which combine BP wDesign approach uses hybrid genetic algorithm which combine BP with GA ith GA to improve searching speed and convergence speed. to improve searching speed and convergence speed.

The learning process as follow:The learning process as follow:

(1) Produce populations that have S individuals.(1) Produce populations that have S individuals.(2) Calculate every individuals(2) Calculate every individuals’’ fitness value.fitness value.(3) Select S(3) Select S--s individuals by gambling model, and placed in selection pool, s individuals by gambling model, and placed in selection pool,

then the optimum individual is learned for s times by BP algoritthen the optimum individual is learned for s times by BP algorithm with s hm with s different learning speed . S new individuals are produced.different learning speed . S new individuals are produced.

(4) S(4) S--s individuals are operated by crossover and mutation, while s ins individuals are operated by crossover and mutation, while s individuals dividuals is added to produce new populations.is added to produce new populations.

(5) if new populations is desired, the optimum individuals is ch(5) if new populations is desired, the optimum individuals is chosen, else go osen, else go (2).(2).

3.Design and Implementation 3.Design and Implementation

Design and simulation of fuzzy logic controllerDesign and simulation of fuzzy logic controller

Implementing fuzzyImplementing fuzzy--neural networkneural network

Applying genetic algorithm to optimize fuzzyApplying genetic algorithm to optimize fuzzy-- neural networkneural network

Step 1: FLC designStep 1: FLC design

Cont Cont ……..

Output 1 (Vehicle Class)Output 1 (Vehicle Class)

Small car Small car

VanVan

LorryLorry

Mid size carMid size car

Big size carBig size car

Truck/busTruck/bus

TrolleyTrolley

Output 2 (Speed)Output 2 (Speed)

Very lowVery low

LowLow

MediumMedium

FastFast

Very fastVery fast

Step 2 : Designing Fuzzy Neural Step 2 : Designing Fuzzy Neural NetworkNetwork

Step 3:Optimization of Fuzzy Step 3:Optimization of Fuzzy NeuroNeuro systemsystem

FuzzyFuzzy--neural system, combines neural system, combines qualitative reasoning ability of fuzzy logic qualitative reasoning ability of fuzzy logic quantitative numeric processing of ANN. quantitative numeric processing of ANN.

A problem of the fuzzyA problem of the fuzzy--neural system is the dimensionality.neural system is the dimensionality.As the input dimension increases the fuzzy rule base As the input dimension increases the fuzzy rule base increase exponentially, which increases the increase exponentially, which increases the

computational costcomputational costmemory &memory &training data requirementstraining data requirements

ContdContd ……

This property limits the practical application of This property limits the practical application of fuzzyfuzzy--neural system to low input dimension neural system to low input dimension problem.problem.

Genetic algorithm provides approach to adjustGenetic algorithm provides approach to adjust

the control points i.e. placement (the control points i.e. placement (base)andbase)and apex of membership functions and apex of membership functions and

the weightings of fuzzythe weightings of fuzzy--neural networks neural networks Genetic Algorithm would be used to optimized the Genetic Algorithm would be used to optimized the

hybrid controllerhybrid controller

4.RESULT4.RESULT

5.CONCLUSION AND FUTURE 5.CONCLUSION AND FUTURE SCOPESCOPE

Fuzzy logic controller is giving two outputs i.e. vehicle class Fuzzy logic controller is giving two outputs i.e. vehicle class and speed . and speed .

The designing and simulation is done with the help of The designing and simulation is done with the help of MATLAB fuzzy logic toolbox.MATLAB fuzzy logic toolbox.

In the future scope part the above controller would be In the future scope part the above controller would be implemented by hybridizing fuzzy with neural then network is implemented by hybridizing fuzzy with neural then network is optimized using genetic algorithm. optimized using genetic algorithm.

These approach will certainly improve the efficiency of the These approach will certainly improve the efficiency of the controller.controller.