Evolutionary Optimization for Inverse Problem in Engineering: The Case Study of Defects Shape Reconstruction

Complex mechanical engineering problems usually have the following stages: problem formulation, conceptual design, embodiment development, detailed design, testing, manufacturing, and quality control. The many tasks in the quality control phase often require significant time commitments of highly skilled personnel to complete successfully. Therefore, the application of evolutionary optimization at this stage of production can reduce the engineer’s involvement in the development process, thereby reducing the time and cost of the task. This paper explores the possibility of evolutionary algorithms to solve the inverse problem of flaw detection, specifically the reconstruction of defect shapes in a product. The proposed solution addresses one of the most challenging technical problems in flaw detection. Experimental studies on a synthetic example demonstrate that the proposed approach can provide a qualitative reconstruction of various defect shapes, even with a limited number of sound pressure receivers. The results show that sound pressure measurements at 64 points are sufficient to correctly detect a defect located anywhere in space and fully recover its shape. Thus, using the defect shape reconstruction problem, the study confirms that evolutionary algorithms can be applied to various engineering problems.