Analysis of Logged OBDII Data - Report Example

Table of Contents 6.Analysis of Logged OBDII Data 4 6 Separating Logged Data into Specific Events 6 6.2.Determining Gear Usage via OBDII Data 7 6.3.Calculated Load to Monitor Fuel Consumption 9 6.4.Analysing Datalogged Events 11 6.5.Journey Gradient Profiling using OBDII Data 13 6.6.Annexure A – Analysis of two real drive cycles 17 List of Figures Figure 6.1 - Drive cycle analysis plotted data for the engine speed, vehicle speed and acceleration obtained from Nexiq 5 Figure 6.2 - Calculated Vehicle to Engine Speed Ratio Distribution 8 Figure 6.3 - Plots of engine speed and vehicle speed for urban drive cycle 11 Figure 6.4 - Map of hilly course 14 Figure 6.5 – Plot of engine load against cruise data for hilly course 14 Figure 6.6 - Inclination factor for hilly course 15 Figure 6.7 - Map of undulating course 15 Figure 6.8 - Plot of engine load against cruise data for undulating course 16 Figure 6.9 - Inclination factor for undulating course 16 Chapter 6 6. Analysis of Logged OBDII Data The post-acquisition analysis of the data logged via OBDII on a Golf Mk4 1.9 TDi (130 PS) was done exclusively in Matlab. Four PIDs were monitored: Engine Speed (rpm), Vehicle Speed (km/h), Calculated Engine Load (%) and Absolute Throttle Position (%). Each PID value logged is time-stamped by the Nexiq on arrival. However the transmission interval is non-deterministic and dependant on the ECU. In addition the each PID arrives independently, so the relationship between PIDs is also non-deterministic. The first stage of analysis was to homogenise the four PID streams acquired so they all have a constant sample rate. Linear interpolation was used to map all the PID data to a fixed sample rate (1 Hz). From this various parameters are easily calculated. The total distance is the trapezoidal piecewise integral of the vehicle speed. From this average speed can be calculated. Acceleration is calculated by piecewise differentiation of the vehicle speed data. Figure 6.1 - Drive cycle analysis plotted data for the engine speed, vehicle speed and acceleration obtained from Nexiq 6.1. Separating Logged Data into Specific Events The rest of the analysis follows the method of Holmén & Niemeier (1998) whereby the logged data is separated in events. Each event was determined to last for one linearised sample interval (1 second). Events are separated firstly into cruise, positive acceleration and negative acceleration. A cruise event is defined as occurring when the vehicle is non-stationary and the acceleration is below a set threshold. This was set to 0.5 ms-2 which provided a good balance of cruise/acceleration events in the datasets considered. Cruise events are further classified as high speed cruise (V > 40 mph), mid speed cruise (25 < V ≤ 40 mph) and low speed cruise (≤ 25 mph). An idle event was defined as occurring when the vehicle was stationary. In order to provide useful parameters for analysis the events where normalised to the total number of events logged. 6.2. Determining Gear Usage via OBDII Data Information on gear selection is not directly available via OBDII PIDs. It is possible however consider the ratio of vehicle speed to engine speed. There are however three potential unknowns required in addition to derive the gear result. These are gear ratios, differential ratio (for rear wheel drive) and tyre radius. Even with a priori knowledge of a given car gearbox/differential, there is no way of deriving tyre radius as these may be changed as aftermarket items. A more generic solution to determining gear ratio is thus preferential. It is possible to use a cluster type analysis in order to derive the current gear. An analysis of the distribution of vehicle to engine speed ratio will yield peaks at each valid gear. There will be some noise associated with actions during clutch depression (e.g. revving, coasting) or wheel spinning caused by traction loss. In order to establish gear positions a normalised distribution of vehicle to engine speed ratio was calculated (200 bin histogram). Calculating the ratio this way avoids the divide by zero that occurs when the vehicle is stationary. A threshold was then used effectively mask the random noise which is spread fairly across the distribution. Any value below this threshold is zeroed. Experimentally, a threshold value of 0.05 worked well for this purpose. The resulting maxima where then detected by for a sign change in the first order differential of the data. In this way it is possible to detect the current gear used during the drive cycle, although not the actual gear ratio. In the following example the four gears were detected and the vehicle-engine speed ratios were calculated as 0.0087, 0.0162, 0.0264 and 0.0359 for first to fourth gear respectively. The relatively small size of the first gear peak indicates little use which corresponds to the presence of a high torque (at low revs) diesel engine. The majority of pull away from stationary is done in second gear. Figure 6.2 - Calculated Vehicle to Engine Speed Ratio Distribution 6.3. Calculated Load to Monitor Fuel Consumption The ability to monitor fuel consumption is something that would be very useful to do via OBDII data. Unfortunately no quantifiable data is generally available for fuel flow in OBDII implementations. It is necessary therefore to try and infer the fuel consumed using other available parameters. Mass air flow can be used to try and determine fuel use in petrol engines, but a stochiometric air/fuel ratio must be assumed. This is invalid for high load (open loop) conditions. Furthermore this method is only viable for petrol engines, diesel engines are not throttled thus the airflow is proportional to only engine speed (simple volumetric displacement). OBDII provides a parameter for ‘Calculate Engine Load’ (Mode 01h, PID 04h) which yields a percentage figure which relates to the current engine load. For petrol engines it is calculated as ratio of airflow to maximum airflow (at any given engine speed). In diesel engines it is calculated as the ratio of fuel flow to peak fuel flow (at any given engine speed). The diesel method of determining calculated engine load means that is should give a good indication of fuel consumption. The petrol method is still limited by the potential for open loop excursions to result in under-reading. It may be possible to resolve open loop conditions through the monitoring of absolute throttle position (Mode 01, PID 11). A threshold may be set (e.g. > 80%) above which the engine is considered to be in open loop operation. Of course this would still not allow for the fuel flow to be corrected. The integrating the calculated engine load can provide an indication of overall fuel consumption for any single journey. This is not a relative value and is useful for comparing multiple drive cycles on any single car, but not able to derive an absolute ‘mpg’ value. 6.4. Analysing Datalogged Events The project aim is to try and reconstruct parts of the NEDC drive cycle from real logged data. To provide a starting point for analysis the NEDC data captured from the dyno rig was first analysed. To simplify matters at this early stage, only the EDE15 portion was looked at (this corresponds to the urban fuel consumption calculation). The logged engine speed and vehicle speed data is shown below. Figure 6.3 - Plots of engine speed and vehicle speed for urban drive cycle Two real drive cycles were also analysed having a similar duration to the EDE15. Both drive cycles were done from a cold start and are the routes were urban in nature. The first was a hilly course, the second more mildly undulating. The analysed results are shown in Annexure ‘A’. A cruise to acceleration ratio was also calculated to try and quantify what is likely to be an important driving style which impacts fuel consumption. The mpg values for the real drive cycles are taken from the onboard trip computer. It is very likely that this calculation will under read, some form of calibration will be necessary to confirm this. The mpg values appears to be higher (especially if under-read is assumed) for both the real drive cycles. This is explained by the fact average speeds are higher and the cruise to acceleration ratio is higher. Clearly the driving was faster and less ‘smooth’ than in an EDE15 cycle. 6.5. Journey Gradient Profiling using OBDII Data On major impact on fuel consumption is the gradients involved on a journey. Clearly a hilly course is likely to result in a worse mpg value for a given drive time. OBDII does not have access to any accelerometer or inclination data (assuming these sensors even exist on a particular model). A point of weakness in any comparison to the NEDC drive cycle is this lack of information. One potential method for determining hilliness is by using the engine load and derived cruise data. For any cruise condition (no acceleration) the engine load will be higher for a positive (uphill) gradient. Conversely it might be anticipated that any negative gradients will result in no engine load as the vehicle coasts. The situation is complicated slightly by the fact that engine load goes up slightly (~20% observed) if the engine is idling (clutch in) as the wheels are no longer turning the engine block and as such fuel is being injected to maintain engine operation. In order to test whether it is possible to gauge any gradient information from engine load and cruise data, two journeys were monitored. One was on a hilly course, the other mildly undulating. Ordnance survey mapping software was used to determine the actual gradient profiles of these two journeys. Both journeys started from the same point and the route/gradient profiles are hilly and mildly undulating respectively. An inclination factor, was then calculated for all categorised cruise data such that: where L is the calculated engine load and V is the vehicle speed. The denominator attempts to account for the approximately square law relationship between power and vehicle speed to do aerodynamic drag. Figure 6.4 - Map of hilly course Figure 6.5 – Plot of engine load against cruise data for hilly course Figure 6.6 - Inclination factor for hilly course Figure 6.7 - Map of undulating course Figure 6.8 - Plot of engine load against cruise data for undulating course Figure 6.9 - Inclination factor for undulating course There is without doubt some correlation between the results for and the calculated inclination factor for both examples. The peak values do correlate with gradients and the peak excursions are lower for the mildly undulating course. Clearly there is a lot of noise in the data, and a lot of missing data due to acceleration events. There may be scope using more advanced signal processing to derive better data. A more advanced correction factor for the effects of drag with vehicle speed would be a good starting point. 6.6. Annexure A – Analysis of two real drive cycles Read More