Abstract: The optimal filter design criteria for PMU measurement and its theoretical basis are derived and thoroughly discussed in Part I of the three-part paper set. Based on the optimal design criteria, Part II deals with the practical design of the optimal algorithms for P class measurement. To meet the P class requirements of high precision and very short response time, the form of the discrete Fourier transform（DFT） based dynamic algorithm was extended and select as the prototype filter, which is intrinsically immune to harmonics. The free parameter set to be optimized consists of Nw（the length of rectangular self-convolutional window）, d（the number of DFT values） and K（the Taylor series order）. The optimization results of both constant efficient filter bank and frequency-tracking based time variant filter banks（with sub-band number Q=5 and 9） are provided, where each filter bank constructs a complete PMU measurement algorithm. The excellent performance of the optimal algorithms are verified with simulation results, along with its comparison with traditional algorithms, including windowed DFT, dynamic phasor estimate algorithm（DPEA） and Taylor-model weighted least squares（TWLS） algorithms.