Abstract： A multi information fusion classification and identification system was proposed in view of the fact that traditional tool identification methods suffered from inefficiency and being susceptible to the bar code failure due to inadequate protection, pollution and other factors in the process of complex production and circulation. First, this system quantized tool features, such as shape, texture, weight and other characteristics, from image sensors and weight sensors. Then, high dimension features vector from both training and testing samples of tool and bar code was extracted. Finally the failure barcode was obtained with the algorithms of support vector machine and DempsterShafer. The experimental results show that the system could classify and identify the tool of destructive bar code accurately and effectively which can satisfy the actual requirement in production.