2results about How to "Avoid simplification" patented technology

The present application belongs to the field of microwave technology and artificial intelligence application, and specifically provides a neural network-based microwave filter coupling matrix extraction method to solve the problems of complexity and insufficient accuracy in existing Cauchy method and optimization method, and can be used in the design and tuning of microwave filters. In the process of extracting the microwave filter coupling matrix based on S parameters, a neural network is introduced, which only needs a large amount of data for training, avoids complex formula derivation and simplification, and is convenient and fast. After training, the trained convolutional neural network model can quickly extract the coupling matrix of the microwave filter according to the S parameters, and has strong real-time performance. In addition, based on the data set construction method proposed in the present application, the data acquisition speed is fast, thereby avoiding the consumption of a large amount of time in the data acquisition stage.

This application provides a predictive maintenance method for industrial equipment based on digital twins and spatiotemporal data, relating to the field of data processing. In this method, an equipment identifier is constructed and a uniquely bound digital twin object identifier is generated, establishing a maintenance right pool; multi-source spatiotemporal observation data is collected to form a spatiotemporal evidence film, constructing an irreversible response evidence set and a reversible response evidence set, and establishing an irreversible evidence fidelity quota; a set of maintenance right certificates is generated by combining structural topology, control topology, and constraints; target maintenance right certificates are extracted within a rolling cycle and a secondary revocation check is performed; upon successful revocation, maintenance actions are executed, and irreversible evidence is compared and replayed to generate a fidelity assessment result, which is then output as an executable maintenance recommendation in the form of a maintenance right certificate. Implementing the technical solution provided in this application facilitates the prediction model in maintaining its predictive capabilities while suppressing the mutual adaptation deviation between the digital twin and the real system.