29 results about "High-throughput computing" patented technology

The invention relates to an iron-based high-temperature alloy design method based on multi-model machine learning and an alloy thereof, and the method comprises the following steps: 1, obtaining iron-based high-temperature alloy components, carrying out high-throughput calculation, and constructing a data set; 2, correlation analysis is carried out on the data set, feature importance sorting is carried out based on a random forest model, and key components are screened out; 3, constructing a machine learning algorithm model, training by using the key components, and adjusting and optimizing hyper-parameters of the machine learning algorithm model through an optimization algorithm to obtain a trained prediction model; 4, constraint conditions are constructed and optimized, an optimal solution set is obtained, the optimal solution set is input into the trained prediction model, and optimal alloy components are obtained; and 5, laser powder bed melting forming is conducted according to the optimal alloy components, and the iron-based high-temperature alloy is obtained. A data-driven machine learning method is adopted to replace a traditional trial and error method, the alloy research and development period is greatly shortened, and the research and development cost is reduced.

The application discloses an alloy component optimization design method for additive manufacturing and belongs to the technical field of additive manufacturing. The alloy component optimization design method for additive manufacturing expands the content numerical range of the alloy component to include all existing alloy component contents on the basis of existing alloy components, combines thermodynamic calculation and high-throughput calculation, optimizes suitable alloy components according to a strain rate hot cracking criterion based on a hot cracking sensitivity index, prepares alloy powder for additive manufacturing according to the optimized alloy components, performs laser additive manufacturing, observes the microstructure of the sample after additive manufacturing and tests the performance of the sample, and selects the component optimization meeting the actual alloy performance. The application takes the alloy component optimization design as a main influencing factor, adopts thermodynamic software and computer language, optimizes the component through the hot cracking sensitivity index to reduce the hot cracking sensitivity of the additive manufacturing alloy, and is beneficial to industrial large-scale production and popularization and use.

The invention discloses a multifunctional catalyst performance prediction method based on high-throughput calculation and machine learning, and relates to the technical field of catalytic material prediction. The method comprises the following steps: acquiring basic structure information of a target catalyst, correspondingly calculating catalytic performance data and related characteristic data of the target catalyst, and pairing the basic structure information and the related characteristic data to establish a data set; importing the data set into a plurality of machine learning models for training, and performing hyper-parameter tuning by using multi-target Bayesian optimization; based on the screened optimal model, sorting and discriminating key features influencing the performance of the catalyst by using an SHAP value, and importing the key features into an interpretable machine learning model SISSO for multi-task training; based on an interpretable machine learning model SISSO and through a symbol regression method, obtaining an explicit mathematical relationship between the feature combination and the catalytic performance as a descriptor formula for application; the multi-aspect catalytic performance of the catalyst can be rapidly predicted, and the research and development efficiency of the catalyst is greatly improved.

According to the time-space-spectrum coding high-throughput calculation imaging architecture based on lithium niobate interference and electro-optical regulation, space inconsistency interference spectrum modulation and time dynamic controllable coding are achieved in a single device, an intelligent reconstruction algorithm is combined, the dimension limitation of an existing scheme is broken through, and the space inconsistency interference spectrum modulation and time dynamic controllable coding are achieved. And a brand new technical path is provided for astronomical snapshot spectral imaging and high-speed industrial detection.

The present application relates to a kind of data-driven deformation nickel-based superalloy multi-objective performance reverse design optimization method, belong to metal material design development technical field, including the following steps: based on first principle calculation stable high-strength deformation nickel-based superalloy system, and based on this, based on nickel-based superalloy field knowledge, construct deformation nickel-based superalloy composition design space;Based on deformation nickel-based superalloy composition design space, using empirical formula is combined with thermodynamics high-throughput calculation to reduce deformation nickel-based superalloy composition design space, obtain the reduced deformation nickel-based superalloy composition design space, and based on this, based on machine learning and genetic algorithm, establish deformation nickel-based superalloy reverse design model, filter target performance deformation nickel-based superalloy.Compared with prior art, the present application proposes a kind of fusion cross-scale calculation, field knowledge constraint and machine learning reverse design, realize the directional development method of high-strength and high-toughness nickel-based superalloy.

The application discloses a method for high-throughput calculation of electronic structures of different temperatures of semi-Hasler materials. The calculation is mainly based on a MatHub-3d database. There are 274 semi-Hasler materials in the database. First, the structures of the 274 materials are optimized, and then the electronic structures and phonon spectra are calculated, and 109 stable semiconductor structures with no virtual frequency and a band gap greater than 0.1 eV are screened. Then, the electronic structures at 11 temperatures from 0 to 1000K are calculated by applying Allen-Heine-Cardona (AHC) theory. The method can improve the efficiency by high-throughput calculation, achieve the purpose of quickly calculating the target material system, and provide a theoretical reference for the design of photoelectric and thermoelectric materials.

The invention relates to the technical field of steel and iron material research and development, in particular to an intelligent steel and iron material research and development system and method, and the system comprises a high-throughput calculation data module, a laboratory data module, a machine learning model training module, a material performance prediction module and a multi-objective optimization module. By integrating multi-source data of the whole life cycle of steel and iron material research and development, production and application, information islands are eliminated, comprehensive display and comprehensive analysis of steel and iron material research and development data are achieved, deep mining and application are conducted on the data through a machine learning algorithm, the efficiency and accuracy of new steel grade research and development are improved, a multi-target optimization function is provided, and the research and development efficiency is improved. Various performance indexes of the material can be optimized at the same time, the requirements of complex application scenes are met, a large model auxiliary research and development function is provided, help and guidance can be provided at any time in the steel material data analysis and component process design process by research and development personnel, and the material development process is accelerated.

The application relates to the technical field of simulation calculation, in particular to a high-throughput calculation method and system for catalytic materials based on machine learning. The method comprises the following steps: establishing a two-dimensional sparse matrix by taking a catalyst identifier and a condition parameter as indexes and taking a known performance index value as a filling value; generating a mixed feature vector by mapping through a multilayer perception network based on non-empty coordinates and the catalyst identifier; obtaining a sample score by performing feature fusion on the mixed feature vector through a neural collaborative filtering network and a full-connection neural network, intercepting samples based on the condition parameter, and outputting a to-be-verified list; calculating performance data by using a simulation calculation model, backfilling the performance data to the two-dimensional sparse matrix, and updating the non-empty coordinates; judging based on a termination condition, and when the termination condition is not met, iteratively predicting the non-empty coordinates; and when the termination condition is met, outputting a catalytic material list. The application deeply couples structured space indexing, double-channel deep learning and adaptive closed-loop iteration, and improves data utilization.

The invention discloses a transition state initial guess structure generation method, and belongs to the technical field of computational chemistry and computer-aided molecular design. The method comprises the following steps: inputting a substrate three-dimensional template file containing reaction priori knowledge and a two-dimensional topological sequence of a catalyst ligand, and automatically identifying ligand donor atoms and completing coordination assembly of a metal center by utilizing an algorithm; and then connecting the catalyst with the substrate template by adopting a conformation generation algorithm based on chemical constraint, executing restrictive conformation optimization under the condition of freezing the core coordinates of the substrate template, and screening and outputting a transition state initial guess three-dimensional structure with the lowest energy. The method not only can greatly improve the efficiency of establishing the transition state, but also can effectively avoid atomic space conflict and molecular fragmentation, and ensures the stability and rationality of the generated initial guess structure; meanwhile, high-throughput batch processing and automatic generation of quantum chemical calculation software with constraint interfaces are supported, and the method is particularly suitable for molecular design in the fields of reaction mechanism research, catalyst performance prediction and high-throughput calculation catalysis.

The application provides a material design method and system based on agent cooperation, and relates to the technical field of the cross of artificial intelligence and material science. The method comprises: collaborative work of an automatic task planning decomposition module, an automatic thermoelectric knowledge retrieval module, a material system prediction module, a high-throughput calculation module, a result analysis module and a feedback optimization module. The core is that the material system prediction module couples a microscopic physical and chemical rule constraint mechanism when generating a structure sequence of a candidate material by using an autoregressive generation model. The mechanism generates a dynamic word table mask, hard blocks candidate words that do not conform to the physical and chemical rules at each word prediction step, and guarantees the rationality of the generated structure from the source. At the same time, the feedback optimization module compares and analyzes the results and the research and development demand, generates an adjustment instruction feedback to the upstream module, and forms a closed loop optimization. The application can improve the design efficiency from the source, and realizes the automation and intelligent iteration of the whole process of material research and development.

The invention discloses an amorphous nanocrystalline magnetically soft alloy powder grading method, which comprises the following steps: based on a powder close packing model, carrying out high-throughput calculation through a computer program, establishing the correlation between a grading scheme and a packing density parameter, and quickly and accurately obtaining a grading scheme with a high close packing degree. The porosity of the amorphous nanocrystalline soft magnetic composite material can be remarkably reduced, the high-frequency soft magnetic performance is comprehensively improved, and the contradictory relation between high magnetic conductivity and high direct current bias performance is overcome. The amorphous nanocrystalline soft magnetic alloy powder grading method provided by the invention has the advantages of accuracy, high efficiency and the like, and is of great significance to research and development of soft magnetic composite materials with high magnetic conductivity and low loss.

The invention provides a DLP equipment and material collaborative process, and relates to the technical field of high-end electronic material design and intelligent manufacturing, and the DLP equipment and material collaborative process comprises the following steps: based on high-throughput calculation, machine learning and material genome engineering technologies, constructing a multi-target generative AI model fusing a polymerization activity discriminator and printability scoring; the method has the advantages that the problem of compatibility of material chemical rationality and DLP process feasibility is solved by constructing a generative AI and printability dual-threshold screening collaborative system fusing physical constraints, and the sample piece precision, surface quality and performance consistency are remarkably improved by means of dynamic optimization manufacturing and online real-time evaluation; and finally, multi-target collaborative optimization is realized within 30 rounds of iteration through data closed-loop feedback, the research and development period is greatly shortened, the cost is reduced, a portable platform and a database are formed, and a full-chain technical support is provided for autonomous controllability and industrialization of high-end electronic materials.

The invention discloses a CF4 / N2 separation-oriented molecular sieve performance cross-temperature zone rapid prediction and optimization method. The method mainly comprises the following steps: calculating single-component and mixed adsorption amounts of CF4 and N2 in a certain temperature range of a zeolite database based on giant regular Monte Carlo simulation high throughput; fitting a plurality of adsorption isotherm models, and screening the optimal adsorption isotherm model to construct an adsorption capacity-temperature-pressure data matrix; data enhancement is carried out in combination with CCR, and rapid and continuous generation of the adsorption isotherm at any temperature within a wide temperature range is realized; predicting the CF4 / N2 adsorption selectivity based on an IAST interpolation method, and comparing the CF4 / N2 adsorption selectivity with a GCMC calculation result for verification; finally, within the applicable temperature range, the adsorption capacity and the selectivity performance target of the mixed components are integrated and optimized, and intelligent screening of the high-performance molecular sieve is achieved. The method can significantly improve the acquisition efficiency and accuracy of adsorption data at different temperatures, and is suitable for rational design and intelligent optimization of the adsorbent under a wide temperature condition.

The embodiment of the invention relates to an atomic spatial distribution analysis method and device based on an ADF-STEM graph. The method comprises the following steps: constructing a first GAN model, a second GAN model and a first prediction model; constructing a cross-system and cross-experiment style first data set; performing joint training on the two GAN models according to the first data set according to a training mechanism of a cyclic generative adversarial network; after the training is finished, constructing a second data set by using a first generator of the first GAN model to train a first prediction model; after training is finished, atomic-scale three-dimensional space distribution analysis is conducted on the sample ADF-STEM graph of any system through the first prediction model, and monatomic density analysis, dimer distribution analysis and trimer distribution analysis are conducted on the basis of an analyzed atom set. According to the method, atomic-scale three-dimensional space distribution analysis can be carried out, cross-system cross-experiment style analysis can be carried out, and the method can adapt to a high-throughput calculation scene.

The invention discloses a flat region research and design method for a medium-wave high-temperature infrared detector, and relates to the technical field of semiconductors, and the method comprises the steps: constructing an infrared detector simulation model; setting parameter ranges of the absorption layer and the barrier layer; under the fixed working temperature and the fixed working bias voltage, high-throughput calculation is carried out on a plurality of infrared detector simulation models formed by combining different parameter values, and the dark current magnitude and the flat region range corresponding to each infrared detector simulation model are obtained; and analyzing an influence mechanism of parameter changes of the absorption layer and the barrier layer on the performance of the detector based on the size of the dark current and the range of the flat region. By changing the thickness and doping concentration of the absorption layer and the barrier layer, the detector structure under various different parameters is formed through combination, and by calculating the dark current and the flat region range of detectors of different structures, the quantitative mapping relation among dark current, flat regions and material parameters is established, and theoretical support is provided for dynamic range optimization.

The invention provides a positive electrode material reverse design method, device and equipment, a medium and a vehicle, and the method comprises the steps: responding to a reverse design request for a lithium-rich manganese-based positive electrode material, and carrying out the high-throughput calculation of a plurality of candidate element doping combinations based on a target performance parameter through a machine learning potential function, determining an optimal element formula and atomic scale key parameters meeting target performance parameter constraints; based on the optimal element formula and the atomic scale key parameters, performing phase field simulation by adopting an anisotropic lattice strain model containing Jahn-Teller distortion correction, and determining bulk phase gradient structure features and surface coating layer configuration parameters; and based on the bulk phase gradient structure characteristics and the surface coating layer configuration parameters, calling a mapping model of a process and a structure stored in a preset knowledge graph to perform process inversion, and determining a candidate coating material system and a corresponding synthesis process path thereof, so that the success efficiency and research and development capability of the positive electrode material design are improved.

The application belongs to the technical field of data processing, and discloses a cable insulation material selection method and device, electronic equipment and medium. The method comprises the following steps: obtaining candidate material data, performing high-throughput calculation on the candidate material data to obtain corresponding target micro features, inputting the material composition and the corresponding target micro features into a target prediction model to obtain a corresponding performance prediction vector, taking the macro performance represented by the performance prediction vector as an optimization target, performing multi-objective iterative optimization solving on the candidate material data based on a non-dominated sorting genetic algorithm, introducing a dynamic coupling constraint rule in the optimization process to obtain a Pareto optimal solution set, and screening target material data from the Pareto optimal solution set. In this way, the micro features of a large amount of material data can be generated through high-throughput calculation, and the multi-objective optimization algorithm is used to realize the collaborative adaptation of the cable insulation material and the working condition, thereby solving the problems of low efficiency and multi-objective coupling optimization difficulty in traditional selection.

The invention discloses a visual intelligent analysis platform for collaborative characterization of a magnetic structure and a lattice structure of a magnetic material, and relates to the technical field of visual platforms, and the scheme comprises experimental verification and core performance of the visual intelligent analysis platform for collaborative characterization of the magnetic structure and the lattice structure of the magnetic material. And platform function and performance verification is completed based on adaptive software and hardware environments and a 150000 + material structure database. According to the scheme, experiments comprise material loading, unit cell switching, period completion, magnetic moment visualization, view angle and parameter adjustment, CIF export and Agent calling simulation, each link is quick in response, accurate in result and high in consistency with an authoritative tool, and Agent calling feasibility is also verified. Results show that the platform is complete in function, convenient to operate and standard in performance, magnetocrystalline structure characteristics can be visually presented, the problems of inaccurate characterization visualization of magnetic materials and tedious operation are solved, and the platform is suitable for scientific research, teaching and high-throughput calculation scenes.

A control / data processor architecture is disclosed which merges data flow with control-flow into a unified control / data flow, enabling thread-level program execution—traditionally bound to a single-core hardware context—to transcend thread, task, and processor boundaries. The architecture supports software-pipelining of entire high-performance computing (HPC) kernels of arbitrary size and complexity, achieving high throughput and instruction-level parallelism. Additionally, the architecture accelerates sparse artificial intelligence workloads with very high sustained throughput and performance per watt.

The application provides a method for designing high-stability multi-element lithium alloy through artificial intelligence, aiming to optimize the calendar life and interface stability of lithium metal batteries. By constructing a multi-dimensional database containing alloy components, physical and chemical characteristics, and experimental characteristics, machine learning algorithms are used for feature selection and prediction model optimization. In particular, Pearson correlation coefficient and random forest regression are used for preliminary feature selection, combined with forward feature selection optimization to identify key influencing factors. High-throughput calculation is used to predict the calendar life, and the composition space is further optimized through the expected improvement maximization strategy. The model forms a closed loop through continuous iteration and experimental data feedback, improving prediction accuracy and design precision. This design method not only effectively prolongs the calendar life of lithium alloy anodes, but also ensures their high performance and industrial feasibility.

The application discloses a high-throughput design method for low-dimensional materials, which utilizes existing crystal structures and performs element recombination to construct a structure library for high-throughput calculation, and optimizes a target material according to a calculation result, wherein, the structure is subjected to dimension reduction processing and / or dimension reduction calculation. The application solves the problem that the existing method cannot quickly and efficiently design low-dimensional materials by performing dimension reduction processing and / or dimension reduction calculation on the structure, improves the screening and prediction efficiency, and improves the design accuracy, and can be used to further design high-performance optoelectronic materials and devices.

The invention relates to the technical field of pure copper additive manufacturing, in particular to a laser melting forming pure copper microstructure parameter optimization method. According to the method, specially-made sensing powder is used for carrying out a high-throughput experiment, molten pool sensing signals and microstructure data are synchronously collected, and a basic agent model is trained; in the process, the consistency of microstructure quantification is ensured by using a double-sensitization corrosive agent; migration fine tuning can be carried out on the basic model only through verification data of a small amount of commercially available pure copper powder, and a calibration model is obtained; performing high-throughput calculation by using the calibration model, and quickly searching and determining a robust process window meeting a target; the invention provides a data-driven optimization method, which solves the two technical problems of low signal-to-noise ratio of a molten pool signal and difficulty in micro-structure quantification in pure copper laser melting forming through the self-made sensing powder and corrosive agent.

The invention belongs to the technical field of aluminum alloy component design and computers, and provides a high-strength low-quenching-sensitivity aluminum alloy design method based on high-throughput calculation, which comprises the following steps: acquiring a plurality of aluminum alloy component combinations; the TTT curve of the multiple aluminum alloy component combinations is calculated in a high-throughput mode through a parallel computing pool, and inoculation time is extracted; the loss strength of the strengthening phase of each aluminum alloy component combination is calculated by combining the TTT curve and the quenching cooling curve of each aluminum alloy component combination in a high-flux mode through a parallel computing pool; based on a Seidman model, the yield strength of the multiple aluminum alloy component combinations is calculated in a high-throughput mode through a parallel computing pool and corrected, and the corrected yield strength is obtained; aluminum alloy component combinations meeting the screening criterion are screened out and output; screening criteria: correcting that the yield strength is greater than a preset strength threshold value; the invention further provides a computer device and a computer readable storage medium, and the high-strength low-quenching-sensitivity aluminum alloy component combination can be reliably screened out.

The invention relates to the technical field of simulation calculation, in particular to a catalytic material high-throughput calculation method and system based on machine learning. Comprising the following steps: establishing a two-dimensional sparse matrix by using a catalyst identifier and a condition parameter as indexes and using a known performance index value as a filling value; based on the non-empty coordinates and the catalyst identifier, mapping through a multi-layer perceptron network to generate a mixed representation vector; performing feature fusion on the mixed representation vector through a neural collaborative filtering network and a full-connection neural network to obtain a sample score, intercepting a sample based on a condition parameter, and outputting a to-be-verified list; performing calculation through a simulation calculation model to obtain performance data, backfilling the performance data to the two-dimensional sparse matrix, and updating non-empty coordinates; judging based on a termination condition, and performing iterative prediction on a non-empty coordinate when the termination condition is not satisfied; and if yes, outputting a catalytic material list. According to the method, structured spatial indexing, dual-channel deep learning and adaptive closed-loop iteration are deeply coupled, and the data utilization rate is increased.

The invention discloses a high-throughput calculation method and system adaptive to multiple scientific calculation software, and relates to the technical field of computational chemistry, electrochemistry and material science. According to the method, parameter recognition and summarization are carried out on multiple scientific calculation software input files, an execution template and a calculation work station matched with target software are generated, parameters are classified in a differentiated mode and combined to generate a complete parameter configuration set so as to automatically render the target input files, task scheduling and result structured conversion and storage are achieved by means of a unified scheduling platform, and the target input files are automatically rendered. The method effectively solves the problems of high learning operation cost and difficulty in reuse caused by the fact that input files are manually compiled and modified in an existing calculation process and lack of unified standards, avoids repeated labor during parameter or model change, unifies output data formats to simplify analysis management, and improves the efficiency. According to the method, operations such as input file generation and task submission in high-throughput calculation do not need to be manually processed one by one, the error risk caused by tedious operation is greatly reduced, unification of input file automatic generation, task automatic submission, result analysis and high-throughput batch execution is achieved, and scientific research efficiency and data reuse capacity are remarkably improved.

The application relates to a composition processing method of a nickel-based powder high-temperature alloy, which comprises the following steps: determining a mass percentage range and a step length of a target alloy element in the high-temperature alloy based on preset target performance parameters; performing high-throughput calculation based on the mass percentage range and the step length of the target alloy element to obtain a high-throughput calculation result, wherein the high-throughput calculation result comprises all candidate alloy composition combinations, and a volume fraction of a gamma prime phase and a balanced precipitation proportion of a harmful TCP phase corresponding to each candidate alloy composition combination; and screening a target alloy composition combination from all candidate alloy composition combinations. The application can accurately calculate the volume fraction of the gamma prime phase, the balanced precipitation proportion of the harmful TCP phase and all preset mechanical performance parameters corresponding to each candidate alloy composition combination, and then can quickly screen the target alloy composition combination which can simultaneously meet excellent high-temperature performance and effectively inhibit harmful phase precipitation from all candidate alloy composition combinations.

The application discloses a high-throughput calculation method of electric transport performance based on deformation potential approximation. In the method, the electron relaxation time involved in the calculation of the electric transport performance is only considered as the electric-acoustic interaction, and the electric-acoustic interaction is described by using the deformation potential approximation. A high-throughput calculation method of the deformation potential constant is developed, the average value of the first band is used as a reference state, and the obtained deformation potential constant dataset is used for high-throughput electric transport performance calculation. Compared with the constant relaxation time approximation and the constant mean free path approximation method, the method has higher precision, and compared with the precise electric-acoustic coupling method, the method is more suitable for high-throughput calculation in terms of cost and time. The method is based on the MatHub-3d database, 11993 semiconductor materials are screened from the database, the deformation potential constants of the 11993 semiconductor materials are high-throughput calculated, and the electric transport performance of 10195 semiconductor materials is high-throughput predicted by using the deformation potential constant dataset.

The invention provides a phase parameter constraint method and system for machine learning design of a high-entropy high-temperature alloy. The method comprises the following steps: constructing a high-entropy high-temperature alloy initial data set and a high-entropy high-temperature alloy property prediction agent model; generating a virtual alloy data set, calculating a precipitated phase regulation factor, and determining an optimal value range of components and related parameter constraint conditions; calculating data based on the high-entropy high-temperature alloy property prediction agent model according to the constraint; and an initial population is constructed in combination with relevant parameter constraint conditions and phase constraint parameters, and high-entropy high-temperature alloy components are obtained through optimization. The problems that in alloy design based on machine learning, mesoscopic phase parameter constraints are few, and components and phases are difficult to design cooperatively are solved, rapid screening and coupling design of the alloy components are achieved, the method is suitable for high-throughput calculation and high-performance alloy design in the material field, design efficiency is high, cost is low, expandability is high, and the method is suitable for large-scale popularization and application. And a calculation means and engineering application support are provided for the rational design of the high-performance high-entropy alloy.

The present application belongs to the technical field of neutron diffraction analysis, and proposes a neutron diffraction vectorization generation and identification method. The generation method comprises the following steps: specifying a chemical system, generating all single elements and compounds by means of high-throughput calculation; removing repeated and unstable structures to generate neutron diffraction spectra of stable structures; mixing different concentrations of components to generate neutron diffraction spectra of mixed phases; generating one-dimensional vectors according to rules; performing half-peak broadening and noise enhancement processing on the simulation spectrum; extracting the features of the strongest diffraction peak to form a feature vector; and embedding the vector database to generate a simulation diffraction vector database. The method has the following beneficial effects: improving analysis efficiency through high-throughput calculation and vector database; enhancing generalization by not relying on specific chemical system training data; improving accuracy through data processing and correction; reducing manual intervention due to high automation; improving data quality through data preprocessing and feature extraction; and realizing rapid prediction through similarity query in the vector database.