5results about How to "Predictive performance" patented technology

This invention relates to the field of road engineering technology, and more particularly to a multi-objective performance coupling design method for warm-mix synchronously recycled SBS modified asphalt. The method determines the aging degree of aged SBS modified asphalt through testing, and accordingly determines the optimal dosage of the two-component synchronous recycling agent and the warm-mix agent. Subsequently, a diffusion model and a fusion degree calculation model for new and old SBS modified asphalt are established to quantitatively analyze the fusion degree of new and old asphalt under different temperatures and aged asphalt dosages. Based on this, a fusion degree spectrum is constructed, and by defining thresholds for each performance index, a feasible design scheme range for balanced performance is obtained, thereby determining the optimal mixing temperature and the maximum dosage of aged SBS modified asphalt. This invention can quickly determine the maximum allowable dosage of aged SBS modified asphalt and its corresponding optimal mixing temperature while considering multiple performance objectives, providing a reliable theoretical basis for the durability design of high-performance recycled SBS modified asphalt pavements.

The application relates to a UNet-based flow field prediction model training method and device, computer equipment and a storage medium. The method comprises the following steps: constructing grid data of a target cascade, and generating simulation flow field data corresponding to the target cascade according to the grid data and working condition conditions; inputting the grid data and the working condition conditions into a flow field prediction model constructed based on UNet to obtain predicted flow field data; determining a first loss value according to the gradient difference between the predicted flow field data and the simulation flow field data, determining a second loss value according to the difference between the predicted flow field data and the simulation flow field data of the surface area of each blade in the target cascade, and determining a target loss value according to the first loss value and the second loss value; and adjusting the flow field prediction model according to the target loss value to obtain a trained flow field prediction model. The method can improve the prediction accuracy of the model.

A rectifier blade welding repair method, comprising the following steps: step S1: based on the finite element method, a thermal coupling numerical model of the rectifier welding process is established, and a heat source model of argon arc welding and laser welding is established; step S2: setting the argon arc welding numerical simulation parameters and the laser welding numerical simulation parameters, respectively adopting the argon arc welding and the laser welding mode to carry out numerical simulation welding on the rectifier; step S3: comparing the temperature field at the moment of welding completion, the residual stress field after welding and the residual axial deformation after welding by using the two welding modes; step S4: according to the comparison result in step S3, the suitable welding mode is determined; step S5: through the actual welding test, the feasibility of the suitable welding mode determined in step S4 is verified; step S6: if the feasibility verification in step S5 is passed, the rectifier is clamped by using the welding tooling, and the blade to be welded is welded on the outer ring by using the welding mode determined in step S4.

The invention belongs to the crossing field of artificial intelligence and geophysics, and particularly discloses a transient electromagnetic forward modeling method and system considering an induced polarization effect based on deep learning, and the method comprises the steps: inputting Cole-Cole key parameters representing the induced polarization effect into a CNN-LSTM mixed deep learning network model for forward modeling, and obtaining transient electromagnetic response data; wherein Cole-Cole key parameters representing the induced polarization effect serve as multi-dimensional input features of the CNN-LSTM mixed deep learning network model, transient electromagnetic response data of a corresponding time channel serve as output, and the CNN-LSTM mixed deep learning network model is trained by adopting a segmented weighted mean square error loss function. According to the method, the segmented weighted mean square error loss function is adopted in the training stage, higher weights are given to the symbol inversion points and the prediction errors of the adjacent time windows, and the reliability of all-time response prediction is guaranteed.