Aspect-based sentiment analysis (ABSA) focuses on accurately classifying the sentiment polarity of various aspects within a sentence. In recent years, graph convolutional networks leveraging syntactic dependency trees have gained popularity in ABSA tasks due to their exceptional ability to capture syntactic structures. However, the challenge lies in effectively integrating both syntactic and semantic information without introducing excessive noise interference. This paper addresses this issue by proposing a novel syntax-enhanced multi-channel graph convolutional network model for ABSA.To enhance the model's understanding of grammatical structures, we have devised a multi-channel graph structure. This structure employs syntactic dependency types, positional information, and tree-based distances as adjacency matrices in various channel graphs to represent different types of relationships between words. Additionally, to accurately capture aspect-related information, we've incorporated an aspect attention module, complemented by a mask matrix to filter out non-aspect word features. Our experimental results, based on three benchmark datasets, demonstrate that our proposed model outperforms existing approaches, achieving the highest level of performance.