The widespread integration of nonlinear equipment in distribution networks leads to the coexistence of multiple harmonic sources with the network. A prerequisite for harmonic mitigation is to obtain the harmonic information through the installation of power quality monitoring devices. However, when making decisions on the placement of monitoring devices, it is common to encounter scenarios where harmonic source information is unknown. This paper proposes an optimal placement method for monitoring devices that balances the accuracy of harmonic source identification and the cost-effectiveness of monitoring devices placement, thereby achieving optimal deployment under such conditions. First, a multi-scenario harmonic current sample generation method is proposed, which builds a low-complexity node harmonic current correlation model to enable rapid and large-scale generation of harmonic current samples, thereby addressing the problem of missing samples when harmonic source information is unavailable. Second, a node harmonic source sensitivity calculation method is presented to characterize node correlations under the interaction of multiple harmonic sources. Then, an optimal placement method for monitoring devices based on the proposed sensitivity index is introduced, incorporating centralized and decentralized sensitivity constraints to avoid redundant placements. Finally, case studies based on the IEEE33-bus system show that the number of monitoring devices configured by this method accounts for only 21.8% of the total nodes, verifying its correctness and applicability.