As the number of communication terminals in power IoT continues to grow, the significance of Ultra-Dense Networks (UDN) interference management for maintaining quality of service is becoming increasingly critical. Nevertheless, the deployment of base stations for a multitude of communication needs also engenders considerable interference in both the frequency and power domains. This paper addresses the interference issues in power UDNs by proposing a multi-base station cooperative interference management method. The method is centered around terminals service, establishing an interference-transmission rate model and service satisfaction optimization problem. It divides interference management into an access carrier allocation phase and a power allocation phase. To address the high-dimensional, discrete and continuous space coupled, large-scale search problem, a novel Dual-Color Bat Algorithm (DCBA) has been designed. DCBA is capable of achieving an efficient joint search of discrete and continuous spaces, significantly enhancing the convergence speed. Experimental results of different algorithms indicate promising advancements of DCBA in addressing the complexities of interference management in UDNs, which significantly improves service satisfaction rates and convergence efficiency in comparison to other algorithms.