Fundamental Evaluation of Data Clustering Approaches for Driving Cycle-Based Machine Design Optimization

Salameh, Mohamad; Brown, Ian; Krishnamurthy, Mahesh

doi:10.1109/tte.2019.2950869

Cited by 32 publications

(7 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Determining the representative points from the load energy distribution can be done in multiple ways. In [1,[9][10][11][12]27], the energy center of gravity (ECG) method is used, which is based on the energy distribution of points within a region of the driving cycle. This method considers the amount of energy used within each region and the weight of the energy for the representative points.…”

Section: Determining Representative Points From the Driving Cyclementioning

confidence: 99%

“…Another method is the use of clustering methods. Different types of clustering techniques can be used in order to determine the representative points [1,4,12,13,28]. An evaluation of different methods of choosing the representative points and their respective impact on the design optimisations is reported in [1,12].…”

Section: Determining Representative Points From the Driving Cyclementioning

confidence: 99%

“…Electric motors are one of the critical components in electric vehicle (EV) power trains. The design of these electric motors can be a challenging task as it involves multiple design considerations such as the efficiency, cost, size, weight, and torque quality or a combination of them [1][2][3][4]. Traditionally, the EV motor design mainly focuses on one or two key working points of the torque-speed envelope [5][6][7].…”

Section: Introductionmentioning

confidence: 99%

“…In [1,[9][10][11][12], the energy representative points are defined through an energy center of gravity approach from the driving cycle, which is used to maximise the energy efficiency of an EV's traction motor over the driving cycle. The energy centre of gravity is similar to the center of gravity method; however, it uses the equivalent energy distribution over the driving cycle instead of the speed-torque points.…”

Section: Introductionmentioning

confidence: 99%

“…This method helps to define the most important working points that the motor operates at compared to a segmented approach. Further, an evaluation is conducted in [1] for different clustering and grouping methods of representative points from the driving cycle. It was shown that by considering these representative operating points, an improved efficiency can be found over a single operating point method.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Evaluation of Drive Cycle-Based Traction Motor Design Strategies Using Gradient Optimisation

et al. 2022

View full text Add to dashboard Cite

In this paper, two design optimisation methods are evaluated using gradient-based optimisation for electric vehicle traction applications. A driving cycle-based approach is used to evaluate specific operational points for the design optimisation procedure. To determine the operational points, an energy centre of gravity (ECG) approach is used. Both optimisation methods are described, namely the point based method and the flux mapping method, with a focus on the flux mapping procedure. Within the flux mapping approach, an inner optimisation loop is defined in order to maintain the stability of gradient calculation for the gradient-based optimisation. An emphasis is placed on the importance of how the optimisation problem is defined, in terms of the objective function and constraints, and how it affects a gradient based optimisation. Based on a design case study conducted in the paper, it is found that the point-based strategy realised motor designs with a slightly lower overall cost (5.66% lower than that of the flux mapping strategy with 8 ECG points), whereas the flux mapping strategy found motor designs with a lower input energy (1.48% lower than that of the point-based strategy with 8 ECG points). This may be attributed to the difference in the definition and interpretation of constraints between these two methods. It is also shown that including more operational points from the driving cycle in the design optimisation leads to designs with reduced total input energy and thus better drive-cycle energy efficiency. This paper further illustrates the significant computational advantages of a gradient-based optimisation over a global optimisation method as it can be completed within a fraction of the time while still finding a global optimum, as long as the problem definition is correctly determined.

show abstract