An LLM-based cross-domain knowledge retrieval augmented generation method for bio-inspired solution design

The application of design knowledge determines the innovativeness of a technical scheme obtained by trimming (a tool for problem analysis and solving in TRIZ). However, limitations in the knowledge, experience and expertise of designers constrain the range of design knowledge that they can apply, thus reducing the effectiveness of trimming. In this paper, biological strategies are introduced to the trimming process to compensate for limitations imposed by the insufficient professional knowledge of designers, thereby improving design innovation. Therefore, this paper proposes a new design method that combines the trimming method and bio-inspired design (BID). First, a trimming analysis of the target system is carried out. Taking the missing functions of the trimmed system as a potential breakthrough point, a keyword search mode based on “V(verb)O(object)P(property) + the effect/features of the associated function” is used to search for biological prototypes in the biological knowledge base. Second, a fuzzy comprehensive evaluation method is used to analyze the biological prototypes from three dimensions, namely, compatibility, completeness and feasibility, and the best-matching biological prototype is selected. Finally, the biological solution is transformed into an engineering design scheme through a resource derivation process based on structure–function–attribute analogies. The proposed method can expand the range of design solutions by adding biological strategies as a new resource to solve trimming problems. The feasibility and effectiveness of the method are verified by redesigning a steel tape armoring machine.

This paper introduces a novel approach, namely Variable-Periodic Voronoi Tessellation (VPVT), for the bio-inspired design of porous structures. The method utilizes distributed points defined by a variable-periodic function to generate Voronoi tessellation patterns, aligning with a wide diversity of artificial or natural cellular structures. In this VPVT design method, the truss-based architecture can be fully characterized by design variables, such as frequency factors, thickness factors. This approach enables the optimal design of porous structures for both mechanical performance and functionality. The varied, anisotropic cell shapes and sizes of VPVT porous structures provide significantly greater design flexibility compared to typical isotropic porous structures. In addition, the VPVT method not only can design micro-macro multiscale materials, but is also applicable for the design of meso-macro scale truss-based porous structures, such as architecture constructions, biomedical implants, and aircraft frameworks. This work employs a Surrogate-assisted Differential Evolution (SaDE) method to perform the optimization process. Numerical examples and experiments validate that the proposed design achieves about 51.1% and 47.8% improvement in compliance performance and damage strength, respectively, than existing studies.

Bio-inspired design for resilient water distribution networks

The obstacle deflector is installed at the front end of trains to sweep obstacles from railways. Conventional obstacle deflectors neglect the obstacles caused by debris flows. To enhance the sweep and resistance reduction performance against debris flows, soil animal claw profiles served as references for the bio-inspired design of a novel obstacle deflector. A parametrization method for claw profiles was proposed based on a Bezier curve of three control points, which was applied to the bio-inspired profile design of obstacle deflectors, thereby proposing a design method for bio-inspired obstacle deflectors (BODs) defined by design parameters α and β. Based on an experimentally validated numerical method which coupled the smoothed particle hydrodynamics (SPH), the discrete element method (DEM) and the finite element method (FEM), the effects of scenario parameters (impact speed and debris depth), design parameters (α and β) and debris flow types on the obstacle deflector-debris flow interaction were investigated. The results indicate that higher impact speed causes greater resistance. Higher debris depth reduces the sweep performance. Through changing the geometric shape of BODs, decreasing α and increasing β enhance the sweep performance, but increasing α and decreasing β reduce platform resistances. Changing the debris flow type does not affect correlations between design parameters and platform resistances. The final BOD configuration (α = 80°, β = 5°) reduces Z-platform resistance by 56.72% while increasing X-platform resistance by only 2.67%, demonstrating a superior performance of resisting the vertical impact of debris flows compared to the prototype obstacle deflector.

Observer-Based Residual Generation for Linear Differential-Algebraic Equation Systems

This paper presents the design and optimization method of near space intelligent target generator to simulate the physical characteristics of the near space vehicle. Combined with High Level Architecture distributed simulation technology, a common, repeatable and verified platform for the near space vehicle has been provided. This method used 3D modeling software Creator and 3D visual rendering software Vega, two-dimensional map and three-dimensional vision were constructed to form a simulation environment, which enhanced the authenticity of the simulation. Based on particle swarm optimization, the intelligent path planning study of near space vehicle was conducted in this environment to make up for the inadequate intelligence of traditional target generators.

Generating designs via machine learning has been an on-going challenge in computer-aided design. Recently, deep learning methods have been applied to randomly generate images in fashion, furniture and product design. However, such deep generative methods usually require a large number of training images and human aspects are not taken into account in the design process. In this work, we seek a way to involve human cognitive factors through brain activity indicated by electroencephalographic measurements (EEG) in the generative process. We propose a neuroscience-inspired design with a machine learning method where EEG is used to capture preferred design features. Such signals are used as a condition in generative adversarial networks (GAN). First, we employ a recurrent neural network Long Short-Term Memory as an encoder to extract EEG features from raw EEG signals; this data are recorded from subjects viewing several categories of images from ImageNet. Second, we train a GAN model conditioned on the encoded EEG features to generate design images. Third, we use the model to generate design images from a subject’s EEG measured brain activity. To verify our proposed generative design method, we present a case study, in which the subjects imagine the products they prefer, and the corresponding EEG signals are recorded and reconstructed by our model for evaluation. The results indicate that a generated product image with preference EEG signals gains more preference than those generated without EEG signals. Overall, we propose a neuroscience-inspired artificial intelligence design method for generating a design taking into account human preference. The method could help improve communication between designers and clients where clients might not be able to express design requests clearly.

Nowadays, with open information and the development of digital technology, designers combine computer and graphic designs. The powerful computing power of computers can transform previously inexpressible graphic ideas into reality and replace manual labor to improve design productivity. Due to the randomness and uncontrollability of graphics, the application of fractal graphics in design is mostly limited. Therefore, practice is conducted through generative and exploratory design methods to discuss the theory of fractal generation technology and basic composition methods and simulates them. The research proves that the fractal graph automatically generated by the computer breaks the thinking of traditional graphic design, broadens the field of graphic design, and promotes the development of the digital graphic design. More importantly, the computer helps designers visualize the interpretation of many rational theorems, and the design method can be profoundly changed.

This paper explores the development and optimization of artificial reefs by introducing a novel generative design method incorporating Computational Fluid Dynamics (CFD) and Optimization (BESO). Since the 1950s, efforts to create artificial reefs have been pursued to improve marine ecosystems. Our study first surveyed the existing design of artificial reefs. Addressing the limitations in existing design methods, this research employs a topological optimization strategy, focusing on optimizing space allocation for polyphony expansion within these reefs. By analyzing fluid dynamics and connecting it with an iterative optimization process, we assess the effectiveness of material exchange in these artificial structures. This is critical for the design process, considering constraints from advanced manufacturing to allow for quick production with natural-like geometries. To advance the design of artificial reefs and explore new possibilities, we introduce a novel generative design approach, where the design of the reef emerges from the interaction between CFD and BESO, through an iterative process of material removal and addition in response to the external loading condition. The artificial reef generated is compared against benchmarks of current artificial design to assess its material efficiency, structure performance, and geometrical characteristics in complex underwater conditions.

Along with the urban renewal and development, the urban living environment has given rise to various problems that need to be solved. With an eye on the future development model of residential communities, an experimental preliminary design for the construction of architectural space, public space, and landscape space based on people’s actual needs is carried out in an attempt to alleviate the more urgent symbiotic relationship between people and urban environment. To this end, this paper proposes a planning and design generation framework for the constructed external spatial environment of building groups based on a recursive double-adversarial network model. Firstly, we extract the features of the constructed external spatial environment of the building group in depth and generate the expression feature map, which is used as a supervisory signal to generate an expression seed image of the constructed external spatial environment of the building group; then we use the generated seed image together with the constructed external spatial environment of the original target building group as the input to generate a feature-holding image as the output of the current frame, and the feature-holding image is also used as the input for the next. Finally, the seed image generation network and the feature-holding image generation network are recursively used to generate the next frame, and the video sequence of the expressions of the constructed external spatial environment of the building group with the same feature-holding expressions as the original input is recursively obtained several times. The experimental results on the building group database show that the proposed method can generate clear and natural video frames of the constructed external spatial environment of the building group, which can be gradually derived from the design of building units to the construction of the building group and penetrate into the planning and design of the external spatial environment in order to comprehensively improve the living environment of urban population and provide a design method and theoretical support for the design of future urban residential communities.

Application of Rhombic Thinking Mode in Highway Slope Greening Design

The design and development process of complex products is a multi-disciplinary collaborative process, which requires the cooperation of distributed teams. However, traditional simulation methods are usually discipline-oriented and there is a lack of efficiency in terms of share and reuse of the existing simulation models in a distributed and collaborative environment. In order to address this problem, we propose a multi-level knowledge framework for simulation model knowledge representation, retrieval and reasoning. Specifically, a simulation knowledge model by the name of “Engineering-APP” is developed to enable simulation knowledge sharing by using Web Service technology. The Engineering-APP model represents an integrated simulation knowledge wrapper, which includes information about the geometric model, design algorithm or analysis codes. Based on the Engineering-APP model, an intelligent collaborative prototype system is developed to support the design process with different stakeholders. Through an engineering case study, it is demonstrated that the proposed framework and Engineering-APP are effective and efficient for the representation and reuse of existing simulation models knowledge.

Research on design day generation method for air-conditioning system design considering the coincidence of hourly variation coefficient

We present a method for the automatic generation of oligonucleotide probe sets for DNA microarrays. This approach is well suited particularly for specificity evaluation of designed probes in large data sets. Algorithms for probe preselection, hybridization prediction and probe selection are presented. Combinatorial techniques are introduced for the selection of probe sets of high differentiation capability even from sequence databases of homologous conserved genes. These techniques include the automatic generation of group specific probes and the design of excluding probes. A basic prototype was implemented including a shared memory parallelization and distribution. The principal applicability of our method to a database of very conserved sequence data was shown and the run-time performance estimated.

본 연구는 인간중심디자인(Human-Centered Design, HCD) 프로세스에서 발생하는 디자인 고착화 문제를 해결하기 위해, AI 기반 생성형 브레인스토밍 시스템인 BrainChat의 개념설계의 제안에 목적이 있다. 인간중심디자인(HCD)은 사용자의 실제 문제를 해결하고 사용자 경험을 개선하기 위한 중요한 접근 방식이며, 브레인스토밍은 창의적인 아이디어를 도출하기 위한 대표적인 방법이다. 그러나 브레인스토밍 과정에서 인간의 정보 한계, 주관적 편견, 정보 처리 속도의 제한, 피로 등으로 인해 창의적인 아이디어 도출이 저해될 수 있다. 이를 해결하고자 AI의 대량 데이터 처리 능력, 반복 작업 효율성, 무작위성과 다양한 지식배경을 활용한다. 이를 통해 인간의 한계를 보완하며, 창의적이고 유연한 아이디어 도출을 지원할 수 있을 것으로 기대된다. 따라서, 본 연구는 인간중심디자인 프로세스 초기 단계에서 창의적이고 혁신적인 아이디어를 도출하여 사용자의 문제를 해결하는데 기여하고자 한다.<br/>선행 문헌 연구를 통해 디자인 프로세스 방법론, 브레인스토밍 기법, 인간과 AI의 장단점을 검토하고, 이를 바탕으로 BrainChat 시스템을 제안하였다. 이 시스템은 인간중심디자인 프로세스의 효과성을 높이고, 사용자 요구를 더욱 잘 반영하는 디자인 결과물을 도출할 수 있을 것으로 기대된다. BrainChat의 주요 기능은 자연어 처리(NLP) 기술을 활용한 OpenAI의 ChatGPT API를 통해 AI 페르소나를 생성하여 사용자와 브레인스토밍을 진행하는 것이다. 본 연구의 초점과 범위는 생성형 AI를 활용한 브레인챗 서비스의 시스템 구축 및 UI/UX 디자인의 개념 설계까지 진행하는 것에 있다. 이에 따른 사용성 테스트와 실제 시스템 구축은 후속 연구를 통해 진행할 예정이다. 이를 통해 AI와 인간의 협업을 통해 더 많은 아이디어를 생성하고, 다양한 관점을 반영하여 사용자 요구를 더욱 잘 반영하는 디자인 결과물을 도출할 수 있을 것으로 기대된다.