Abstarct: Cybersecurity in smart power grids has become one of the fundamental challenges in the electric power industry due to the extensive reliance on measurement data. One of the most critical threats in this domain is false data injection (FDI) attacks, which, by manipulating data, can disrupt the state estimation and system control processes, leading to consequences such as reduced reliability, widespread blackouts, and equipment damage. However, conventional bad data detection methods, including the chi-square (χ²) statistical test, have limited effectiveness against sophisticated attacks and are highly dependent on the quality of input data. In this study, an innovative approach baxsed on the optimal selection of a subset of buses equipped with trusted monitoring devices and its integration with the chi-square test is proposed. The main novelty of the work lies in selecting a sensitive and effective subset from already protected buses, rather than increasing the protective infrastructure, in order to achieve maximum attack detection capability with minimum cost. To evaluate the performance of the proposed method, standard IEEE test systems were modeled in MATLAB, and various false data injection attack scenarios were implemented. Simulation results indicate that the proposed method achieves a detection rate of over 98% in large-scale attack scenarios and about 85% in less extensive attack scenarios. Furthermore, by optimally selecting trusted buses, the required protected infrastructure is reduced by approximately 40%, implying lower cost and improved protection efficiency. These results demonstrate that the targeted integration of statistical tests with trusted data can provide an effective and cost-efficient approach for enhancing the cyber resilience of smart power grids.