The sparse communication network exposes the AC microgrid (MG) with higher risk under malicious cyber attacks. Intrusion detection of malicious cyber attacks becomes a critical issue in operations of isolated AC MGs. In this work, a general framework for detecting cyber intrusion deception attacks is proposed by using localized dual variables of the partial primal-dual (PPD) consensus-based distributed droop control. In additional to individual link attacks, node attacks, or stealthy deception attacks considered in conventional studies, concurrent deception attacks, which refer to both link attacks and node attacks will take place in various remote locations of the MG simultaneously, are also investigated. Impacts of concurrent cyber deception attacks on the closed-loop stability of the entire MG will be analyzed comprehensively. Redundancy relationships between the local information and localized dual variables of the PPD algorithm are utilized to derive analytical indices for detecting these concurrent deception attacks. By enhancing our previous cyber defense mechanism with only considering link attacks or node attacks, a new cyber defense mechanism for concurrent deception attacks is presented. Tasks in cyber defense mechanisms including (i) attack localization, (ii) attack isolation, and (iii) resilience enhancement are also presented in details. Numerical experiments on the real-time simulator OPAL-RT are conducted to validate the effectiveness of the proposed detection mechanism.