Vehicles Using Data Acquisition And Fuzzy Logic Computer Science Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Automotive industries are working on improving fuel efficiency, emphasizing to develop hybrid vehicles that are better in performance as well as fuel efficient. Focus on renewable alternate sources of energy, is the key feature. The research work in this paper is aimed at improving the fuel efficiency in Hybrid vehicles by capturing the data from the ECU on RPM, Speed, Fuel Pressure, Load, Timing advance, etc, from the CAN port of the vehicles. Strict emphasis on the ECU data from the engine is the unique feature of this research. The tests were conducted on different models from different manufacturers like Toyota PRIUS Toyota HIGHLANDER, and Ford ESCAPE hybrid vehicles. Dewetron Data analyzer was used to capture the data, and processed for precision using logical rules with mathematical formulae in Dewesoft, and processed in MATLAB where the fuzzy logic conditional rules were implemented to improve the fuel efficiency by online data monitoring

Keywords: Online Data Analysis, Fuzzy logic, Matlab, Dewesoft, Offline data processing


The automotive manufacturers are expanding their hybrid vehicle segment elaborately and extensively. Each manufacturer has their own specific functional hybrid vehicle. For an example, The Toyota Prius and the Highlander have an entirely different mode of operation when compared to the Honda Insight and CRZ. The Toyota segment of vehicles have a strategy that enables vehicles to run on battery till the vehicle achieves a 20mph and above speed for the engine to take the load and charge. This again is dependent upon the passenger load in the vehicle. Unlike this the Honda segment of vehicles have the engine programmed to start first and take up the initial load, and after a certain speed have the motors assist the engine and subsequently cut off until further power is in demand. In either of these ways there is a strategy and computer program behind to govern the working. These complex computer programs can be simplified and made to work using fuzzy logic rules after the ECU data has been captured and analyzed.

The fuzzy logic rules are based upon certain specifications that manufacturers have set in the vehicle. Previous research in this topic does not get in to great details of ECU data acquisition. In this research work, more number of input parameters, which are Fuel pressure, intake air temperature, load, RPM, Speed, Exhaust gas temperature, oxygen sensor reading, absolute throttle position, short and long term fuel trim, manifold absolute pressure, and coolant temperature, and to give a precise decision of the commonly aimed output parameter called 'efficiency monitor' based on IF THEN conditional rules, is where the use of fuzzy logic is legitimated, and emphasized. 'Milton Martinez [1] in his paper 'Fuzzy Logic Controller for Two modes Parallel Hybrid Electric Vehicle' says that fuzzy logic is a way of thinking. As an enhancement to the work done in my previous paper titled 'Use of Fuzzy Logic in Hybrid Vehicles for Improving Fuel Efficiency [2] by Individual Component Control' the engine data is also captured analyzed and then control strategy is implemented.

Previous work was curtailed only along with lesser input parameters for monitoring efficiency. Work in upbringing complex algorithms for providing feedbacks to the driver was also conducted. Analyzing the data from the ECU of the engine with specific functional parameters which can further enhance the control strategy of the hybrid vehicle has been the key area of research work. The experimental setup and execution of work in phases, its need and justification, and the outcome of the experiment, are described in detail in the following sections of this paper.

2. Experimental Setup

2.1 Calibration and Setup of DEWE 43V data analyzer

The DEWETRON DEWE43V data analyzer is an instrument which is capable of capturing data from different sensors. The core working system of this instrument is software called DEWESOFT. The software is capable of data capturing, formatting and analysis by itself. Data in the form of electrical signals from the sensors are acquired in to the DEWE 43V and stored in to the hard disk of the computer. The experiment is conducted in 3 phases.

The first one is to check the self capability of the DEWETRON and DEWESOFT. The second phase is to check the readability of the acquired data in to MATLAB and imply fuzzy logic rules in the offline mode. The third phase is to do the second phase of the experiment on the data which is coming out live from the ECU of the vehicle, which is called online data analysis.

In the first phase the data inputs that were available were stored in the form of channels. Each channel gives an output of a certain voltage range from each sensor on the engine, and this is converted in to graphical data for better visualization. In the first stage, the DEWE 43V was calibrated to the requirements of this experiment. The key requirements which were needed are that it should provide us with values of specific parameters like Manifold Absolute pressure, Absolute Throttle Position, Intake air flow Temperature, Exhaust gas temperature, Coolant temperature, Timing advance, Short and long term Fuel Trim, Engine RPM, and Vehicle speed.

The numerical values of these parameters were obtained in the maximum and minimum ranges. The CAN (control area network port) was connected to the OBD (out bound data) port of the vehicle. After the channels were assigned their titles the hybrid vehicle was turned on. The turn on mode is the point where the data from the ECU Starts flowing in through the CAN port.

2.2 Evaluation and Basic Formulation

Dewesoft has an option called offline math, which actually helps us in setting up rules. The logic rules are assigned based on the recorded input values, and numerical limitations are provided. The maximum limit to which the incurred value can elevate is shown here. Based on this value we can assign a control rule which will give is a triggered signal when the sensors obtain a signal that is outside the limit of our boundary. The logic rule assigned to govern the vehicle speed v/s the Engine RPM are


The rule here triggers a linear function in the Graphic screen of dewesoft when the vehicle is in motion and is being propelled in BATTERY alone mode. The 0 and 1 are set trigger values which can be manually adjusted. Similarly the fuzzy rule implemented to indicate the point where the engine takes over is represented with

IF "ENG RPM">0AND "VEHICLE SPEED" > 20, 1 ,0 (2)

This rule triggers a linear function on the graphic screen of dewesoft that will show the ENGINE taking over mode. Here again the trigger values are set between 0 and 1.

The Figure below shows the graphic display scene monitoring the vehicle in the static and moving conditions. Rule1 is shoed when triggered by the solid Green trigger line and Rule 2 is shown by the solid orange line respectively as indicated in Fig.1.

Fig.1 Dewesoft Graphic Display with Rule 1 and Rule 2 triggered

2.3 Advanced formulation & Experimentation in Offline mode

Once the work foundation is set up to trigger the basic modes of operation of the hybrid vehicle, the advanced data is programmed in the Dewesoft for evaluation. Here the parameters are set to obtain triggered functions which exceed the set values at different point of time during vehicle runtime.

To achieve this, the vehicle was operated in Driver alone mode, driver and single passenger, driver and two passengers, driver and three passengers, and driver and four passengers respectively. This was done so as to obtain the various load governed conditions for the engine take over time. The rules assigned for other conditions are

If TA <1deg AND ENGINE RPM < 1500,0,1 (3)

Engine Power (IHP) = (2pNT)/60 Nm (4) [6]

Rule (3) and Formula (4) are assigned to calculate the spark timing, as all vehicles Gasoline (or) hybrid, are equipped with a mechanism which will automatically advance the current excitation given by the distributor to the spark plug of the engine, so as to reduce the phenomenon of knock. All manufacturers have their own specific timing advance and retard mechanism according to the road and load conditions of the vehicle. Also the driver behavior is an important aspect.

Torque output from the engine is available from the engine CAN data, but then in general the formula for calculating the torsional force is explained by Formula (4), Where p = 3.14 which is a constant, N= Engine RPM, T = torque in NM and IHP = indicated horse power. The indicated horse power is a form of Brake horse power and the deductible frictional horse power. Brake Horse power is the Power available at the flywheel of the engine. Frictional horse power is always a negative value. The peak values of various sensor readings which are preset by the manufacturers are shown in Table 1.

The minimum value and maximum values are indicated. This is a further justification for the rules assigned hereafter. Rules assigned are within the boundaries of maximum and minimum values. Any numerical value which exceeds these rule boundaries at any point of operation of the vehicle, due to any circumstances shall trigger a linear graph on the display which is an error indication or a warning.

Parameter With Units Minimum Value Maximum Value

Engine RPM 0 7500

Vehicle Speed MPH 0 150

Intake Air Temperature Deg C -40 115

Coolant Temperature Deg c -40 115

Manifold Air Flow gm/sec

Load value %

Fuel trim %

Timing Advance Deg

Absolute Throttle Position %

Oxygen Sensor voltage V -50





-5 655.35






Table1. Parameters with minimum and maximum readings

Further the rules including other parameters are

IF RPM>3000AND Speed<40 and MAF<-50, 0,1 (5)

IF Load>100AND fuel trim< >-1 and ATP>100 , 0,1 (6)

IF OSV>= -1 AND ATP> <25 0,1 (7)

Rule (5) justifies for defects occurring at the time of lower air intake or a faulty air intake system. Rule (6) is used to detect the point where the engine delays to start up at maximum load condition. This may result in over discharge of battery current and thereby this rule helps in avoiding the same. Rule (7) maps the oxygen sensor voltage. This is needed for detecting the exhaust levels from the Engine and also helps in detecting if the oxygen sensor has failed. At this time the ECU shall not allow the engine to start and thus the battery will have to take the entire load and it may discharge at a very short running period.

The limitation of Dewesoft is that on the application of logic rules, it can only trigger a linear function which can make the operator understand that at this point the distortion or deviation from the usual running cycle of the Internal combustion Engine is taking place, and further it cannot directly communicate with the ECU of the engine to readjust parameters at the point of defect. Taking this in to consideration, we now incorporate the same data obtained from Dewesoft and import it in to MATLAB. The fuzzy logic toolbox is used to further enhance the efficiency of the research work, which is covered in the following section.


In the second phase of the research, as mentioned above, the data acquired from Dewesoft was imported to Matlab. Using the fuzzy logic toolbox, and Mamdani method, logic rules were assigned to the input parameters retaining their values exactly the same as that acquired. A program was written to call the function from Dewesoft. From dewesoft the files are exported to matlab by the 'export' function which makes the file readable by matlab. The program would then execute the fuzzy logic rules on the input parameters. Two output parameters are assigned to the function which are, Control, and Execute Respectively. Gaussian membership function was assigned to all the input as well as output parameters. The ranges of each function in the input variables are as provided by the manufacturers, but the output parameters are ranged between 0 and 1. The control [3] and execute mode has three respective membership functions which are NIL LOWER AND ACCELERATE, and NIL, CONTROL, AND STOP. These MF's are assigned for the ease of program to take a logic decision at the time of error and is set to give a feedback as well as execute the function in the simulation mode.

Fig.2 Fuzzy Logic interface, and surface view

Fig.2 shows the results of fuzzification of the input parameters in MATLAB and the surface diagram for the same. The defuzzified results are in the output parameters which are control and Execute. The program is capable of reading the specific file which has the stored data in the offline mode. The tests were conducted within the range starting at 20mph speed to 80mph speed and the readings of all parameters that changed or showed a warning signal as mentioned earlier were noted graphically. To test the efficiency of the warning function, the vehicle was accelerated to speeds of 80mph for a small interval of time. The function, as desired showed warning signals triggered on the graphic screen of dewesoft, as well as a written prompt shown on the Fuzzy logic screen. The factor which prompts the use of logic rules integrated with the acquired data from dewesoft is the view interface which is not an added feature in dewesoft. This again justifies the use of Matlab.

The Fuzzy logic rules that were assigned on MATLAB are shown in Fig.3 Here all the input parameters assigned are shown in the form of yellow columns and the defuzzified result is shown with a blue marking intend at the right corner of the figure. The Red colored indent is the display parameter which shows output respectively. And thus the offline data processing using dewesoft (independently), and matlab integrated data processing was achieved.

Fig.3 Fuzzy Logic Rules With Limits and Output.

Therefore the experiment has so far proven success in capturing the offline data and processing it. Further as mentioned is the Third and final phase of the experiment which is designed to import the exported data live into matlab and perform the tests on the hybrid vehicle at various speed and load conditions. Apart from this , the test has been carried out on different models like the sports utility hybrid vehicle FORD ESCAPE Hybrid, and also the Toyota HIGHLANDER hybrid respectively. For the same set of rules mentioned earlier, and on the same test conditions.

4. Processing Online Data

In the third phase of the experiment, as data keeps flowing in to the dewetron data analyzer, it is simultaneously exported to matlab. This data is then linked to the matlab program as a live streaming file. This is then delivered to the set of fuzzy logic rules assigned to each input parameter. All this is very similar to the work done in previous stages, but for data being readily captured from the OBD2 port of the hybrid vehicle to the CAN port of the data analyzer and exported to Matlab. During this test phase the same set of experimental runs are conducted on the vehicle starting from a running speed of 20Mph to 70 Mph. This covers the vehicle starting from the parking space to driving in the street conditions, and then entering the highway. This is done to successively increase the speed of the vehicle. The vehicle was intensely tested on the highway where the merging speed was 45mph and the minimum speed was 55Mph. the maximum limit allowed on highway conditions is 70Mph.Here when the vehicle is subjected to sudden acceleration to take the required speed from 55Mph to 70Mph hard acceleration is required and in this speed the hybrid vehicle is designed to operate in dual mode, where in the electric motor powered by the battery provides extra torsion force to the wheels along with the torsion force of the engine. When the live data is captured, the rules assigned here would assist the system to adapt the throttle level to lower itself according to the assigned rules. Fig.4 shows the online data that is flowing along with the rules assigned to reduce the engine effort for propelling the vehicle with respect to various driver behaviors. These include normal low and hard acceleration patterns.

Graph 1. Eng RPM, and Speed for 15sec runtime

Graph.1 shows the first 400 sec run time of the vehicle where in after implication of fuzzy rules, the rpm v/s speed was non fluctuation. This is in the simulated mode where the fuzzy rule would automatically adjust the RPM if the driver over accelerates the vehicle. In the case of over acceleration, more fuel flow to the engine is obvious. Reducing the same is the prime motive.

Fig.4 Online Data processed with rules triggered

Along with implying formulae and rules for the Engine, for calculating values on both offline as well as online modes, formulae to calculate the torque of the electric motor was used in order to balance the control strategy on motor and battery current operated propulsion side, as well as the engine propelled side respectively. The torque of the electric motor is calculated by using the relation

t=(5252*HP)/N (8) [7]

Where t = torque, 5252 is a numerical constant, HP = power of the motor in Horsepower, N= revolutions per minute of the motor. Also the torque is calculated using another relation

t=(120*F)/P (9)[8]

Where t = torque, F = supply frequency in cycles/sec, P = number of motor winding poles. Different formulas are used because some manufacturers will have different data output, and using different formulae eliminates the factor of error in calculations.

5. Result and Discussion

The experimental setup was also simulated using LABVIEW software where manually adjustable parameters were set to simulate the real time situation of passengers occupying the vehicle. This was done to view the change in load conditions and study the effect of increased load over performance of the vehicle under various road conditions and speeds respectively. Variables were also assigned on the LABVIEW deck to validate the driver behavior inputs like over acceleration, under acceleration, hard breaking, and reverse motion. Boolean function was used to indicate the error if shown in the simulation mode. Fig.5 shows the deck where in all controls and variables were assigned to simulate the situation. The Dials shown in the deck are similar to those in the real automobile.[3] Additional dials are assigned to monitor the motor torque and load conditions which are not available as standard equipments in the instrument panel of vehicles.

Fig.5 LABVIEW deck

Fig. 6 shows the DEWESOFT deck where in at time intervals the rules implemented on the online data takes place and the graphical output display is obtained. Here the experimental test run on FORD ESCAPE HYBRID is recorded for various runs starting from 20mph and successively increasing the speed in 10 MPH increments. The maximum speed travelled was 75mph on the FORD vehicle. It is not a safe practice even for the purpose of testing to accelerate the hybrid vehicle above the speed limits because it may cause unwanted damage to the motor control and propulsion system. Hence the 75mph was maintained as the peak limit of acceleration during the entire range of experiments conducted on each class of vehicle respectively.

Fig.6 Fuzzy rules implemented on live data obtained from FORD ESCAPE

Fig. 7 shows the same experiment with same set of procedures and same set of logic rules applied for the Toyota Prius [5] which belongs to the Economy segment. Here the blue lines depict the error trigger and the red is the automatic correction in the simulation mode. This is further for communication with the driver to have a check on his mode of operation as well as to auto correct the system to stabilize the values obtained.

Fig.7 Toyota Prius with trigger and correction graphs

In the final step of the research conducted the same set of procedures were repeated on the Toyota HIGHLANDER. The results obtained after the tests on SUV's and Economy size cars of different manufacturers are discussed in detail in the following section.

Fig.8 shows the Trigger mode and correction graphs which are for the rests conducted on the Toyota Highlander. Being a mid size SUV [4] in it s segment, the Toyota Highlander hybrid has potentially lesser use of battery propulsion and the motor has lesser torque output compared to the small segment and the ford segment of vehicles.[10] This is done by the manufacturers, for keeping the vehicle performance curves to the maximum. Yet use of the electric motor propulsion system in the HIGHLANDER does help in reducing the fuel consumption especially during the cruise mode where the operation is carried out on a large scale in the dual operating mode which lowers the propulsion load to the engine.

Fig.8 Toyota Highlander with trigger and correction graphs

Here the white lines represent use of engine propulsion. It is visible that at lower instances in comparison to the Prius, and Ford escape, the engine has to take command of propelling the vehicle. The blue lines are trigger for the system to identify over acceleration, and the green trigger lines depict the control. Control is triggered during the beginning and at the end of the test run of the vehicle respectively. The yellow and pink bars are excitation current triggers that are shown when the vehicle is overloaded and the LED display at the bottom of the screen shows a warning of damage to the system if operated above these rule ranges. This is done to provide a feedback to the operator and caution the control system simultaneously. The response time of the vehicle in simulated mode with rules applied, is much larger when compared to the normal operation. Thus it justifies the use of fuzzy logic rules in the governing of fuel economy, and energy conservation.

6. Conclusion

By using the data acquisition system and implementing the Fuzzy rules the control of fuel consumption can be significantly reduced. The battery propulsion time can be maximized. Small segment vehicles have lesser errors triggered when compared to the SUV range of vehicles. The load value plays a very vital role in determining the take over time for engine in propelling the vehicle. Simulated results show increase in fuel economy when compared to the standard mode of operation of hybrid vehicles under various driver and load conditions. The novel achievements are that the data can be processed as and when it is acquired and corrective rules can be applied for simulation instantaneously which minimizes the errors, and maximizes the performance and fuel efficiency. The work carried out here is in a simulated mode which limits direct communication with the ECU of the vehicle. Further work can be done in this aspect where the rules assigned may directly be communicated to the running vehicle.