Fitting Heterogeneous Lanchester Models on the Kursk Campaign
share
The battle of Kursk between Soviet and German is known to be the biggest tank battle in the history. The present paper uses the tank and artillery data from the Kursk database for fitting both forms of homogeneous and heterogeneous Lanchester model. Under homogeneous form the Soviet (or German) tank casualty is attributed to only the German(or Soviet) tank engagement. For heterogeneous form the tank casualty is attributed to both tank and artillery engagements. A set of differential equations using both forms have been developed, and the commonly used least square estimation is compared with maximum likelihood estimation for attrition rates and exponent coefficients. For validating the models, different goodness-of-fit measures like R2, sum-of-square-residuals (SSR), root-mean-square error (RMSE), Kolmogorov-Smirnov (KS) and chi-square statistics are used for comparison. Numerical results suggest the model is statistically more accurate when each day of the battle is considered as a mini-battle. The distribution patterns of the SSR and likelihood values with varying parameters are represented using contour plots and 3D surfaces.
READ FULL TEXT
