11results about How to "Shorten the optimization cycle" patented technology

The invention provides a neural network multi-objective optimization and FPGA hardware acceleration collaborative design method, and belongs to the field of deep learning model compression and hardware collaborative design. The method comprises the following steps: constructing a joint optimization space containing a neural network compression parameter and an FPGA hardware design parameter; a multi-target Bayesian optimization search strategy is adopted, iterative search is carried out in the joint optimization space, model precision, FPGA resource occupation and reasoning delay are synchronously optimized, and optimal candidate configuration is obtained; matching the compressed network structure with the FPGA parallel architecture by using a hardware-perceived pruning and quantification strategy; a multi-task performance prediction model is adopted to quickly predict the precision, resource occupation and delay of the optimal candidate configuration so as to accelerate the search process; according to the optimal configuration, a hardware accelerator code facing the target FPGA is automatically generated, and integration and implementation are completed. According to the method, collaborative optimization of neural network compression and hardware design is achieved, FPGA resource occupation can be remarkably reduced, the reasoning speed can be increased, and meanwhile the model precision is kept.

The invention belongs to the technical field of simulation optimization, and particularly relates to a centrifugal pump impeller multi-physics field coupling simulation optimization method which comprises the following steps: constructing a geometric model of a centrifugal pump by adopting three-dimensional drawing software according to an actual structure of the centrifugal pump; finite element modeling is conducted, and a finite element analysis model of the centrifugal pump is obtained; hexahedral grids are divided for the impeller, a shape variable space of the impeller is built, and the impeller deformation serves as a design variable; solving the finite element analysis model after grid division through a solver; an experimental design sample is generated by adopting a Latin hypercube sampling method, a global response surface model is constructed through quadratic polynomial regression fitting, an optimal impeller deformation amount is iteratively solved by adopting a global response surface optimization algorithm, a finite element analysis model is adjusted, and an optimization model is formed. According to the method, fluid-solid coupling simulation, a global response surface method and an automatic iteration process are fused, and the efficiency of the centrifugal pump impeller is maximized and optimized on the basis of accurately matching actual working conditions.

This application relates to the field of wireless communication technology, providing an antenna robustness design method and apparatus based on hybrid deep learning. This invention simulates manufacturing process errors by combining sampling methods within the range of process errors, obtaining design variables and... S Sensitivity analysis was performed on the Gaussian distribution curves of the statistical mapping relationship between parameter responses to identify variables more sensitive to manufacturing errors, thereby reducing the dimensionality of variables and decreasing the complexity of subsequent antenna robustness optimization. A hybrid deep learning model was used to construct an antenna response substitution model, replacing traditional electromagnetic simulation, significantly shortening the optimization cycle while ensuring the accuracy of response prediction and avoiding distortion of optimization results due to model errors. By constructing a robustness objective function and employing a genetic algorithm to optimize antenna parameter robustness, a highly robust optimal solution was found, avoiding premature convergence of the gradient method in multi-peaked environments caused by random errors.

The invention belongs to the field of cell culture automation and intelligent regulation and control, and discloses an optogenetics-based adherent cell culture optimization method and system. The method comprises the following steps: inputting an adherent cell type and a culture target, and calling an adherent cell-optical genetic parameter joint database; performing targeted pretreatment on the target adherent cells; processing through a multi-parameter optimization algorithm hardware platform to generate an executable customized photostimulation scheme; performing accurate light stimulation on the positive adherent cells; comparing the obtained detection data with normal adherent cell parameters in a joint database, and evaluating the influence of a photostimulation scheme on the cells; updating the parameters of the multi-parameter optimization algorithm and the prediction model; and inputting the adherent cell type and the culture target to the updated prediction model, and outputting an optimal light stimulation scheme. The core problems of low photostimulation accuracy and poor targeting in the prior art are solved, and high-activity and high-consistency light-sensitive adherent cells are stably produced.

Embodiments of the present application provide a client running method, a computing device, a computer readable storage medium and a computer program product. The client running method comprises: receiving stutter information sent by a first client, the stutter information being generated in the case that the first client detects a stutter event; the stutter information comprising function information of a target function corresponding to the stutter event; analyzing the stutter information to obtain the function information of the target function; generating stutter prompt information based on the function information; and sending the stutter prompt information to a target client in a silent push manner, so that the target client adjusts an execution operation of the target function based on the stutter prompt information, wherein the target client comprises the first client and / or a second client. The technical solution provided by the embodiments of the present application improves the stutter optimization efficiency of the client.

The invention discloses an LDMOS device electrical performance optimization method based on an SAC algorithm, and belongs to the technical field of semiconductor power device design. The method comprises the following steps: sampling in a process parameter range, and extracting electrical characteristics by utilizing TCAD simulation to construct a data set; a deep neural network agent model is constructed and trained; designing a state space, an action space and a reward function, building an SAC deep reinforcement learning framework, and accessing the proxy model to the environment; and training the SAC model to obtain optimal process parameters. According to the method, the DNN proxy model is used for replacing high-time-consumption simulation, and the SAC algorithm is combined for self-adaptive adjustment, so that the problem of difficulty in multi-target collaborative optimization in a high-dimensional parameter space is effectively solved, the optimization period is remarkably shortened, collaborative improvement of static and dynamic performance of a device is realized, and an optimization result has high reliability through circuit verification.

This invention provides a multi-physics coupling test method and system for intelligent vehicle hybrid network architecture, belonging to the field of automotive electronics testing technology. The method includes: applying single-physics coupling, dual-physics coupling, and system-level multi-physics coupling loads through a tiered testing process, simultaneously acquiring electrical signal, optical signal performance parameters, and environmental parameters; calculating a unified signal health index (HI) and degradation rate based on a preset algorithm; analyzing the contribution of each physical field factor to performance degradation when the HI is below a health threshold or the degradation rate exceeds a rate threshold, and identifying the dominant failure factor; and intelligently matching the corresponding optimization strategy from a pre-built optimization strategy library. This invention solves the problems of media isolation, data fragmentation, and inefficient optimization in traditional testing methods, achieving unified cross-media testing, real-time health status assessment, and precise automated optimization, thus improving the reliability verification efficiency of intelligent vehicle hybrid network architecture.

The application provides a material preparation method based on purification and delamination synergistic optimization of montmorillonite, and comprises the following steps: S1, pretreating natural montmorillonite raw materials; S2, centrifugal purification by using a purification parameter linear optimization model; S3, preparing a montmorillonite delamination precursor; S4, ultrasonic delamination treatment by using a delamination effect dynamic regulation model; S5, purification and delamination synergistic optimization adjustment; S6, delamination montmorillonite stabilization treatment; S7, material forming and drying; S8, material sintering and performance characterization. Through the whole-process intelligent synergy of "raw material characteristics-purification parameters-delamination effect", the application breaks through the limitations of traditional processes and realizes efficient and stable preparation of montmorillonite materials.

The invention discloses an online firepower distribution method based on a battlefield situation and an improved genetic algorithm, and relates to the field of firepower distribution, and the method comprises the following steps: collecting battlefield situation core data, and obtaining input information based on the battlefield situation core data; constructing an online firepower distribution model based on the input information, and forming an initial online firepower distribution model through dominant function definition, influence factor standardization, matrix construction, total income determination and constraint condition setting; and optimizing the initial online firepower distribution model by adopting an improved genetic algorithm to obtain an optimized firepower distribution scheme. According to the invention, efficient online distribution of multi-missile cooperative combat can be realized, and the real-time performance of battlefield decision making and cooperative combat effectiveness are improved.

This invention relates to the fields of materials science and intelligent optimization technology, and discloses a multi-objective intelligent optimization method for the composition and heat treatment process of 2xxx series aluminum alloys. This method integrates the XGBoost multi-output prediction model with four intelligent optimization algorithms: Firefly Optimization (FA), Gray Wolf Optimization (GWO), Particle Swarm Optimization (PSO), and Simulated Annealing (SA). Through an automated process of data preprocessing, multi-objective prediction, constraint optimization, and result verification, it achieves the synergistic optimization of the tensile strength (UTS), yield strength (TYS), and elongation (EL) of aluminum alloys. The system uses unique heat encoding to process categorical process parameters, introduces constraint penalty mechanisms and early stopping strategies to ensure that the optimization results meet the requirements of composition ratio and process parameter range. Experimental verification shows that the comprehensive performance of the aluminum alloy optimized by this method is suitable for the performance optimization of 2xxx series aluminum alloys in aerospace, automotive manufacturing, and other fields, significantly reducing R&D costs and time.

This invention relates to the field of samarium iron nitride (SFeNi) permanent magnet material manufacturing technology, and discloses a method and system for optimizing SFeNi injection molding process parameters using digital twins. This method achieves dynamic optimization by integrating a digital twin with the physical production line. Real-time sensor data from the production line is collected and preprocessed. Based on this, a digital twin containing a particle layer, a melt layer, and a magnetic pole layer is constructed and updated. Using this twin, the agglomeration and orientation of SFeNi particles are predicted in the particle layer, temperature, shear, and magnetic flux distribution are simulated in the melt layer, and magnetic performance indicators are evaluated in the magnetic pole layer. Based on these predictions and evaluations, the material temperature, injection speed, pulsating magnetic field waveform, and holding pressure curve are optimized using an artificial intelligence proxy model. Finally, the optimized parameters are sent to the production line for execution via a programmable logic controller (PLC). This method achieves online adaptive adjustment of process parameters, improving the quality and efficiency of magnet molding.