Detecting software requirements defects is crucial in reducing the risk of software project failures. Existing methods for automatic detection, especially during requirements changes, are limited in coverage and often lack robust tool support. Addressing this gap, we define the four most common types of requirements defects (incompleteness, inconsistency, redundancy, and ambiguity) and present algorithms for their detection. We propose a novel behaviour engineeringbased approach, translating software requirements into behaviour trees and then into the Web Ontology Language (OWL). We developed 'requirements defects identifier', a tool that queries the OWL knowledge base to identify potential defects during requirements analysis and change. Validated on three final-year student projects, our approach demonstrated success in detecting all four defect types, offering broader coverage compared to existing tools. A real-world case study has been used to validate the applicability of the proposed approach. Our experiments demonstrate that the tool can successfully detect all four different types of requirement defects during both requirements analysis and requirements change.