This paper proposes an channel equalization method based on artificial neural network(CD-net), which for deals with the channel equalization problem of coherent optical filter bank multicarrier with offset quadrature amplitude modulation (CO-FBMC-OQAM)communication system. CD-net combines the advantages of convolutional neural network and deep neural network to perform the channel equalization task for the CO-FBMC-OQAM communication system. In CD-net, the convolutional neural network modular first performs feature extraction of the distorted signal at the receiving end. Then the deep neural network modular performs signal demodulation and channel equalization, and recovers the original information. The numerical simulation results show that compared with the traditional channel equalization method, the CD-net method has a greater advantage in bit error rate performance, which can better solve the channel distortion problem of the CO-FBMC-OQAM communication system.