In the power system, the encryption of privacy information needs to ensure high security to prevent data leakage or unauthorized access. Multiple embedding technology requires encryption algorithms to be complex enough to resist various attacks, such as brute force cracking and side channel attacks. However, when privacy is aggregated at one level, its risk is not dispersed and the likelihood of being compromised is greater. To this end, a multi-level partitioning encryption method for multiple embedded privacy information in the encrypted text domain of power grid is studied. In the study, the K-means algorithm was used to partition privacy information at multiple levels, which can disperse risks to different levels. If security measures at one level are compromised, data at other levels can still be protected. This risk diversification mechanism can reduce the impact of a single point of failure on the overall system security. A key sequence is generated through Logistic mapping, and the key is used to embed privacy information multiple times in all power grid encrypted text domains for initial encryption, obtaining the encrypted text. The Paillier semi-homomorphic encryption algorithm is used to achieve two-layer encryption for multi-embedding privacy information encrypted texts in high partition sensitive power grid encrypted text domains. The results indicate that in the face of different attacks, the security level of the studied method is higher, indicating that the encrypted privacy information of the studied method is less likely to be cracked and stolen.