An intelligent interaction method for personalized design based on topology optimization

Abstract

The application discloses a kind of personalized design intelligent interaction methods based on topology optimization, including S1, construct the user behavior monitoring system based on multi-modal data acquisition, generate high-dimensional data set containing time sequence characteristics;S2, based on adaptive deep neural network model, dynamically predict and hierarchical modeling to user personalized demand;S3, the optimal solution of design parameter is generated in real time by dynamic topology reconstruction algorithm;S4, through reinforcement learning mechanism, iteratively update design scheme generation rule;S5, construct neural network supported multi-objective design space exploration and optimization framework, quickly filter and optimize design parameter;S6, through multidimensional data mapping and parameterization control means, realize the dynamic update and local optimization of design scheme;S7, establish data-driven continuous learning and evolution mechanism, constantly optimize user demand prediction model and design optimization algorithm.The application has the advantages of strong dynamic adaptability, high intelligent level and high personalized satisfaction precision.

Description

Technical Field

[0001] This invention relates to the field of intelligent interaction technology, and in particular to a personalized intelligent interaction design method based on topology optimization. Background Technology

[0002] With the rapid development of intelligent technologies, personalized design has become an important requirement in many fields, especially in user experience, product customization, and intelligent interaction systems. However, existing personalized design methods often struggle to effectively address complex and ever-changing user needs and dynamic design optimization scenarios, resulting in inefficient design processes and an inability to accurately meet users' personalized requirements.

[0003] In existing technologies, traditional personalized design methods mainly rely on static rules or predefined templates, combined with limited user behavior data for optimization. These methods have significant shortcomings in terms of dynamism and adaptability, specifically manifested in the following problems:

[0004] 1. Lack of dynamic adaptability: Traditional personalized design methods usually use predefined design templates or fixed optimization rules, which cannot be dynamically adjusted according to real-time changes in user needs, making it difficult to meet complex and diverse personalized needs.

[0005] 2. Lack of intelligent interaction mechanism: Most existing design systems rely on manual user input or a single data source, lacking an intelligent feedback mechanism based on multimodal interaction data. This makes it difficult to accurately capture users' implicit needs and behavioral characteristics, resulting in inaccurate design adjustments.

[0006] 3. Low design optimization efficiency: In the design optimization process, existing methods often rely on single-objective or static optimization algorithms, which cannot fully explore and quickly optimize high-dimensional complex design spaces, resulting in low design generation efficiency and insufficient accuracy.

[0007] 4. Difficulty in achieving continuous learning and optimization: Traditional methods usually do not make full use of users' historical interaction data and design optimization records, lack continuous learning mechanisms, and cannot dynamically optimize user demand prediction models and design optimization algorithms based on new user behavior and historical data, thus limiting the system's adaptability and evolutionary capabilities.

[0008] Therefore, how to provide a personalized intelligent interaction method based on topology optimization is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0009] One objective of this invention is to propose a personalized design intelligent interaction method based on topology optimization. This invention fully integrates topology optimization, deep learning, multimodal interaction and continuous learning technologies, and describes in detail the implementation methods of dynamic personalized design adjustment, user demand prediction, multi-objective optimization and intelligent feedback mechanism. It has the advantages of strong dynamic adaptability, high optimization efficiency, high level of intelligence and high accuracy in personalization.

[0010] A personalized intelligent interaction design method based on topology optimization according to an embodiment of the present invention includes the following steps:

[0011] S1. Construct a user behavior monitoring system based on multimodal data acquisition to obtain the user's operation trajectory, preference features and multi-source feedback information in real time during the interaction process, and generate a high-dimensional dataset containing time-series features;

[0012] S2. Based on an adaptive deep neural network model, by integrating user behavior features and contextual semantic information, dynamic prediction and hierarchical modeling of user personalized needs are performed.

[0013] S3. The personalized design adjustment module driven by topology optimization transforms the user demand prediction results into a multi-objective optimization problem and generates the optimal solution of design parameters in real time through a dynamic topology reconstruction algorithm.

[0014] S4. Through reinforcement learning mechanism, design intelligent feedback algorithm based on users’ real-time operation and interaction behavior, and iteratively update the design scheme generation rules;

[0015] S5. Construct a multi-objective design space exploration and optimization framework supported by neural networks, and quickly screen and optimize design parameters through high-dimensional nonlinear solution and parallel computing strategies;

[0016] S6. Employ dynamic visualization and multimodal interaction technologies to construct a real-time design scheme presentation and interactive interface. Through multidimensional data mapping and parameterized control, achieve dynamic updates and local optimization of the design scheme.

[0017] S7. Establish a data-driven continuous learning and evolution mechanism, and continuously optimize user demand prediction models and design optimization algorithms by integrating new user interaction data and historical design optimization information.

[0018] Optionally, S1 specifically includes:

[0019] S11. Based on multimodal sensing devices, collect user operation behavior data and combine it with interaction context information to generate a dynamic dataset, including operation trajectory T. op ={t i |=1,2,…,n}、Biometric signal B sig ={b j|=1,2,…,m} and environmental parameter E ctx ={e k |=1,2,…,l};

[0020] S12. Construct a modality fusion algorithm, using a dynamic weight allocation mechanism to normalize and align the multimodal data, forming a fused feature matrix C. fusion =[c ijk ], where c ijk This represents the normalized value of the i-th trajectory dimension and the j-th biological signal at the k-th time step;

[0021] S13. Based on the adaptive time series model, extract the dynamic feature vector F of user behavior. time =[f m1 ,f m2 ,…,f mp ], where f mp This represents the weight of user behavior patterns over the time dimension p;

[0022] S14. Construct a user preference feature clustering model and extract user preference features P through recursive hierarchical clustering analysis. user =[p a1 ,p a2 ,…,p aq ], where p aq This represents the user's preference feature vector under category q;

[0023] S15. Design a multi-source feedback sensing module to collect user interaction feedback on the current design scheme in real time and generate a feedback matrix R. feedback =[r st ], where r st This represents the weight value of feedback category s at time step t;

[0024] S16. Transfer the behavioral feature vector F time Preference features user and feedback matrix R feedback Fusion, generating a dynamic feature set D through a high-dimensional feature interaction model. combined =[d uvw ], where d uvw This represents the combined weight value of the behavioral dimension u, the preference dimension v, and the feedback category w;

[0025] S17. Based on sparse optimization and deep feature extraction algorithms, D combined Dimensionality reduction is performed to generate a user behavior optimization feature vector V. optimized =[v z1 ,v z2 ,…,v zn ], where v zn This represents the nth key feature.

[0026] Optionally, S2 specifically includes:

[0027] S21. Construct a deep neural network model M based on adaptive parameter control. adaptive The model structure includes multiple layers of nonlinear activation units, recursive enhancement modules, and context-weighted feature units. The model weights W are updated in real time through a dynamic optimization strategy. m and bias B m ;

[0028] S22. Optimize the user behavior feature vector Q behavior =[q1,q2,…,q u ] and contextual semantic feature matrix U context =[u ij Enter into M adaptive , where q u The weight of the behavioral feature dimension u is represented by u. ij The weighted value of context category i on semantic dimension j;

[0029] S23. Through the feature cross-fusion module, Q behavior and U context Perform joint embedding to generate a high-dimensional joint feature representation matrix:

[0030] Z joint =[z kl ];

[0031] z kl =g(q) k ,u il );

[0032] Where g is the feature fusion function, representing the interaction between behavioral and semantic features;

[0033] S24. Based on the demand-layered forecasting unit, and using Z... joint Generate a demand vector through multi-channel feature extraction:

[0034] D predict =[d1,d2,…,d p ];

[0035] Where, d p =h(W d ·z kl +B d ), W d and B d represent the prediction layer weights and biases, respectively, and h is a non-linear activation function;

[0036] S25. In the time-step recursive analysis module, based on the user's historical behavior sequence matrix H...history =[h mn ] and the current context feature matrix U context A time-step prediction model is constructed using a recurrent neural network to generate the dynamic demand distribution T at time step t. dynamic =[t1,t2,…,t v ], where t v =k(h mn ,u ij W t ), where k is the recursive relation mapping function;

[0037] S26. Construct the optimization objective function:

[0038]

[0039] Where Φ represents the set of model parameters, λ i and μ i These are the demand weights and time correlation coefficients, respectively, which are solved using an adaptive gradient optimization algorithm.

[0040] S27. Generate the demand forecast result matrix P output =[p xy ], where p xy This represents the predicted value of demand category x at time step y. The matrix is ​​then optimized and iterated using a dynamic update mechanism, and the result is output to the topology optimization module as a basis for dynamic adjustment of design parameters.

[0041] Optionally, S3 specifically includes:

[0042] S31, Based on the user demand prediction result matrix P output =[p ab Construct a multi-objective optimization model and define the set of objective functions F. opt ={f1(Y),f2(Y),…,f u (Y)}, where f i (Y) represents the design parameter set Y = {y1, y2, ..., y} v} Optimize performance under objective dimension i, while setting the set of constraints G. constraints ={g1(Y),g2(Y),…,g w (Y)} limits the search scope of the design space;

[0043] S32. Using a non-uniform topological mesh generation method, the design parameter set Y is discretized into a multi-scale mesh structure M. topo =[m ij ], where m ij The state variables of the i,j-th grid cells are represented by dynamic topology variables φ. ij Represents the grid cell mij Based on the weight distribution, construct the mesh optimization function:

[0044]

[0045] Where, β k The objective function f k The priority weight, δC(φ) ij φ is the grid complexity regularization term, which is dynamically optimized using a nonlinear constraint function. ij ;

[0046] S33. Design a sensitivity analysis module, calculate the sensitivity of the topological state to the optimization objective based on higher-order partial derivatives, and define the sensitivity matrix S. topo =[s pq ],in The sensitivity matrix S is optimized through a dynamic adjustment strategy. topo And enhance the global search capability of the optimization model;

[0047] S34. Using a distributed computing framework combined with a dynamic topology reconfiguration algorithm for M topo Perform iterative optimization to generate the optimization result matrix R. opt =[r ij ], where r ij This represents the final state of the mesh cells after multi-objective optimization;

[0048] S35. By using the inverse mapping relationship from mesh to design parameters, the topology optimization result matrix R... opt Mapping to the design parameter set generates the final optimized design parameter set. As an output of personalized design solutions.

[0049] Optionally, S4 specifically includes:

[0050] S41. Construct a deep reinforcement learning model F based on multi-source interaction data. RL , containing state space Z state ={z1,z2,…,z m Action space K action ={k1,k2,…,k n} and reward function V reward (z,k), where state z i Represents real-time user interaction characteristics, action k j This indicates the optimization strategy for system adjustments;

[0051] S42. Construct a state transition probability model based on user interaction behavior, with the state transition relationship being P. change (z t+1 |,k t ), and through the policy network σω (k|z) output in state z t The optimal action To dynamically adjust design parameters;

[0052] S43. Design a multi-layered composite reward function:

[0053] V reward (z t ,k t )=κ1·L efficiency +κ2·L precision -κ3·R overhead -κ4·D latency ;

[0054] Among them, L efficiency L represents the optimized efficiency of interactive operations. precision This indicates improved accuracy in user feedback, R overhead D represents the cost of resource consumption. latency κ1, κ2, κ3, and κ4 represent response delays, and are reward weighting coefficients.

[0055] S44. Using a dual-policy network structure based on deep reinforcement learning, define the evaluation network Q. eval (z,k) and the target network Q target (z,k), using a recursive time difference update strategy for parameter ω:

[0056]

[0057] Where ν is the learning rate, ρ is the discount factor, and k ′ The optimal action in the next state;

[0058] S45, Optimize the policy network Dynamic generation rules applied to design schemes, combined with a real-time updated state set Z state and action set K action Adjust the design parameter space to output personalized design solutions that meet the user's real-time needs.

[0059] Optionally, S5 specifically includes:

[0060] S51. Construct a design space exploration model T based on multi-objective nonlinear expression. explore Define the design parameter set Ω params ={ω1,…,ω m}, Objective function set Λ objectives ={λ1(Ω),…,λ p (Ω)} and constraint set Γ constraints ={γ1(Ω),…,γ q(Ω)}, where λ i (Ω) represents the optimization objective of the parameter set in dimension i;

[0061] S52. Constructing a high-dimensional nonlinear target prediction model N based on deep neural networks. optimize The input is the design parameter set Ω params The output is a set of predicted target values. By optimizing the network weight set Θ optimize Minimize the prediction loss function:

[0062]

[0063] Where, λ i (Ω) represents the true target value;

[0064] S53. Employ a multi-objective optimization algorithm based on the dynamic Pareto front, and define the Pareto boundary solution set Δ. Pareto ={δ1,δ2,…,δ r}, where δ k Indicates a non-inferior solution:

[0065]

[0066] Among them, O multi_goal For a multi-objective optimization function, ζ i This represents the target weight, and the solution set is optimized according to a dynamic update strategy.

[0067] S54. Employ a distributed parallel optimization strategy for the design parameter set Ω. params Partition the dataset into computational subsets P. k ={ω k1 ,ω k2 ,…,ω kn In a multi-threaded environment, the objective function value and Pareto boundary conditions are computed in parallel, and the results are aggregated to generate an optimized solution set.

[0068] S55, Based on Pareto solution set Δ Pareto A global update generates an optimized set of design parameters. This is used as the output of a personalized design scheme.

[0069] Optionally, S6 specifically includes:

[0070] S61. Construct a multimodal interactive data acquisition and fusion module Q interaction Collect user operation behavior set U action ={u1,u2,…,u s Real-time feedback set V feedback ={v1,v2,…,vt} and environmental parameter set W context ={w1,w2,…,w r}, fusion generates dynamic data stream Q dynamic ={U action V feedback W context};

[0071] S62. Design a real-time visualization engine R based on dynamic multidimensional data mapping. visualize Define the set of mapping functions Ψ map ={ψ1(Q),ψ2(Q),…,ψ h (Q)}, where ψ k (Q) represents the interaction data Q. dynamic To the visualization dimension R view The dynamic mapping relationship generates a multi-dimensional visualization scheme set R in real time. design ={ρ1,ρ2,…,ρ h};

[0072] S63, via parameterized control module P adaptive Dynamically adjust the set of design parameters Θ parameters ={θ1,θ2,…,θ z Construct a local optimization objective function:

[0073]

[0074] Where, τ j To optimize the weighting coefficients, H(θ) j Q dynamic ) represents the parameter θ j With dynamic data Q dynamic Nonlinear relationship;

[0075] S64. Utilize the user's set of local optimization requirements K local ={k1,k2,…,k l Establish a mapping matrix Φ between requirements and parameters. adjus t={φ ij |=G(k i ,θ j )}, where υ ij Indicates demand k i For parameter θ j The influence value, G is the dynamic mapping function, through matrix Φ adjust The design parameters are partially updated to generate an optimized parameter set.

[0076] S65, through the real-time visualization engine R visualize The optimized design scheme Θoptimal Presented to the user interface I interactive It provides a dynamic adjustment interface, allowing users to perform secondary personalized optimization based on real-time input of local parameters.

[0077] Optionally, S7 specifically includes:

[0078] S71. Construct a continuous learning module L based on data fusion. fusion Define the user real-time interaction data set Ξ realtime ={ξ1,ξ2,…,ξ a} and historical optimization record set Θ history ={θ1,θ2,…,θ b}, fusion to generate learning dataset Π training ={Ξ realtime ,Θ history} is used for dynamic model updates;

[0079] S72. Design a user demand prediction model U based on deep recursive networks. predict The input is the learning dataset Π training The output is a set of user demand prediction results Λ output ={λ1,λ2,…,λ c}, by optimizing the model parameter set Γ weights Minimize the prediction error function:

[0080]

[0081] Where, λ i This is the actual demand value. These are the model's predicted values;

[0082] S73. Employ a dynamic incremental learning mechanism for dataset Π training Continuous expansion is carried out, and the parameter set Γ of the demand forecasting model is adjusted based on the multi-step gradient descent method. weights :

[0083]

[0084] Where η is the learning rate and t is the number of update steps;

[0085] S74. Constructing a design optimization and evolution module G based on genetic algorithms. optimize Define the set of optimization parameters Δ parameters ={δ1,δ2,…,δ d} and the set of objective functions Φ targets ={υ1,υ2,…,υ e The objective function is:

[0086]

[0087] Where, ζ j For the target weight, T(δ) j ,Π training The ) indicates the fitness of the optimization parameters for the comprehensive dataset;

[0088] S75. Based on the optimized user demand prediction model and design optimization parameter set The rules for generating personalized designs are dynamically adjusted, and the optimization results are applied to the design output module in real time to achieve adaptive evolution and continuous optimization.

[0089] The beneficial effects of this invention are:

[0090] (1) By combining topology optimization, deep learning, adaptive neural networks and multimodal interaction technology, this invention can capture user behavior and demand changes in real time, dynamically transform demand into a multi-objective optimization problem and generate optimal design parameters through topology reconstruction algorithm, which solves the shortcomings of traditional methods in dynamic adaptability and personalized satisfaction, and effectively improves the accuracy and efficiency of design optimization.

[0091] (2) This invention uses intelligent feedback mechanism and reinforcement learning algorithm to dynamically adjust the design scheme generation rules based on real-time user interaction data, making the design process more intelligent and humanized. At the same time, it supports multimodal interaction and dynamic local optimization, overcomes the problem of lack of intelligent interaction mechanism in traditional systems, and significantly improves user satisfaction and participation.

[0092] (3) This invention utilizes a genetic algorithm-based design evolution module and a continuous learning mechanism to integrate new user data and historical design optimization records, continuously optimize the user demand prediction model and design optimization algorithm, and has a high degree of adaptability and evolutionary capability. This not only reduces the reliance on manual intervention, but also enables the system to continuously learn and adapt to complex and ever-changing personalized needs, thus realizing the long-term optimization and intelligent upgrading of the design system. Attached Figure Description

[0093] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0094] Figure 1 This is a diagram illustrating the overall framework of a personalized intelligent interaction method based on topology optimization proposed in this invention. Detailed Implementation

[0095] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0096] refer to Figure 1 A personalized intelligent interaction method based on topology optimization, characterized by the following steps:

[0097] S1. Construct a user behavior monitoring system based on multimodal data acquisition to obtain the user's operation trajectory, preference features and multi-source feedback information in real time during the interaction process, and generate a high-dimensional dataset containing time-series features;

[0098] In this embodiment, S1 specifically includes:

[0099] S11. Based on multimodal sensing devices, collect user operation behavior data and combine it with interaction context information to generate a dynamic dataset, including operation trajectory T. op ={t i |=1,2,…,n}、Biometric signal B sig ={b j |=1,2,…,m} and environmental parameter E ctx ={e k |=1,2,…,l};

[0100] S12. Construct a modality fusion algorithm, using a dynamic weight allocation mechanism to normalize and align the multimodal data, forming a fused feature matrix C. fusion =[c ijk ], where c ijk This represents the normalized value of the i-th trajectory dimension and the j-th biological signal at the k-th time step;

[0101] S13. Based on the adaptive time series model, extract the dynamic feature vector F of user behavior. time =[f m1 ,f m2 ,…,f mp ], where f mp This represents the weight of user behavior patterns over the time dimension p;

[0102] S14. Construct a user preference feature clustering model and extract user preference features P through recursive hierarchical clustering analysis. user =[p a1 ,p a2 ,…,p aq ], where p aq This represents the user's preference feature vector under category q;

[0103] S15. Design a multi-source feedback sensing module to collect user interaction feedback on the current design scheme in real time and generate a feedback matrix R. feedback =[r st ], where r st This represents the weight value of feedback category s at time step t;

[0104] S16. Transfer the behavioral feature vector F time Preference features user and feedback matrix R feedback Fusion, generating a dynamic feature set D through a high-dimensional feature interaction model. combined =[d uvw ], where d uvw This represents the combined weight value of the behavioral dimension u, the preference dimension v, and the feedback category w;

[0105] S17. Based on sparse optimization and deep feature extraction algorithms, D combined Dimensionality reduction is performed to generate a user behavior optimization feature vector V. optimized =[v z1 ,v z2 ,…,v zn ], where v zn This represents the nth key feature.

[0106] S2. Based on an adaptive deep neural network model, by integrating user behavior features and contextual semantic information, dynamic prediction and hierarchical modeling of user personalized needs are performed.

[0107] In this embodiment, S2 specifically includes:

[0108] S21. Construct a deep neural network model M based on adaptive parameter control. adaptive The model structure includes multiple layers of nonlinear activation units, recursive enhancement modules, and context-weighted feature units. The model weights W are updated in real time through a dynamic optimization strategy. m and bias B m ;

[0109] S22. Optimize the user behavior feature vector Q behavior =[q1,q2,…,q u ] and contextual semantic feature matrix U context =[u ij Enter into M adaptive , where q u The weight of the behavioral feature dimension u is represented by u. ij The weighted value of context category i on semantic dimension j;

[0110] S23. Through the feature cross-fusion module, Q behavior and U contextPerform joint embedding to generate a high-dimensional joint feature representation matrix:

[0111] Z joint =[z kl ];

[0112] z kl =g(q) k ,u il );

[0113] Where g is the feature fusion function, representing the interaction between behavioral and semantic features;

[0114] S24. Based on the demand-layered forecasting unit, and using Z... joint Generate a demand vector through multi-channel feature extraction:

[0115] D predict =[d1,d2,…,d p ];

[0116] Where, d p =h(W d ·z kl +B d ), W d and B d represent the prediction layer weights and biases, respectively, and h is a non-linear activation function;

[0117] S25. In the time-step recursive analysis module, based on the user's historical behavior sequence matrix H... history =[h mn ] and the current context feature matrix U context A time-step prediction model is constructed using a recurrent neural network to generate the dynamic demand distribution T at time step t. dynamic =[t1,t2,…,t v ], where t v =k(h mn ,u ij W t ), where k is the recursive relation mapping function;

[0118] S26. Construct the optimization objective function:

[0119]

[0120] Where Φ represents the set of model parameters, λ i and μ i These are the demand weights and time correlation coefficients, respectively, which are solved using an adaptive gradient optimization algorithm.

[0121] S27. Generate the demand forecast result matrix P output =[p xy ], where pxy This represents the predicted value of demand category x at time step y. The matrix is ​​then optimized and iterated using a dynamic update mechanism, and the result is output to the topology optimization module as a basis for dynamic adjustment of design parameters.

[0122] S3. The personalized design adjustment module driven by topology optimization transforms the user demand prediction results into a multi-objective optimization problem and generates the optimal solution of design parameters in real time through a dynamic topology reconstruction algorithm.

[0123] In this embodiment, S3 specifically includes:

[0124] S31, Based on the user demand prediction result matrix P output =[p ab Construct a multi-objective optimization model and define the set of objective functions F. opt ={f1(Y),f2(Y),…,f u (Y)}, where f i (Y) represents the design parameter set Y = {y1, y2, ..., y} v} Optimize performance under objective dimension i, while setting the set of constraints G. constraints ={g1(Y),g2(Y),…,g w (Y)} limits the search scope of the design space;

[0125] S32. Using a non-uniform topological mesh generation method, the design parameter set Y is discretized into a multi-scale mesh structure M. topo =[m ij ], where m ij The state variables of the i,j-th grid cells are represented by dynamic topology variables φ. ij Represents the grid cell m ij Based on the weight distribution, construct the mesh optimization function:

[0126]

[0127] Where, β k For the objective function f k The priority weight, δC(υ) ij ) represents the grid complexity regularization term, which is dynamically optimized using a nonlinear constraint function. ij ;

[0128] S33. Design a sensitivity analysis module, calculate the sensitivity of the topological state to the optimization objective based on higher-order partial derivatives, and define the sensitivity matrix S. topo =[s pq ],in The sensitivity matrix S is optimized through a dynamic adjustment strategy. topo And enhance the global search capability of the optimization model;

[0129] S34. Using a distributed computing framework combined with a dynamic topology reconfiguration algorithm for M topo Perform iterative optimization to generate the optimization result matrix R. opt =[r ij ], where r ij This represents the final state of the mesh cells after multi-objective optimization;

[0130] S35. By using the inverse mapping relationship from mesh to design parameters, the topology optimization result matrix R... opt Mapping to the design parameter set generates the final optimized design parameter set. As an output of personalized design solutions.

[0131] S4. Through reinforcement learning mechanism, design intelligent feedback algorithm based on users’ real-time operation and interaction behavior, and iteratively update the design scheme generation rules;

[0132] In this embodiment, S4 specifically includes:

[0133] S41. Construct a deep reinforcement learning model F based on multi-source interaction data. RL , containing state space Z state ={z1,z2,…,z m Action space K action ={k1,k2,…,k n} and reward function V reward (z,k), where state z i Represents real-time user interaction characteristics, action k j This indicates the optimization strategy for system adjustments;

[0134] S42. Construct a state transition probability model based on user interaction behavior, with the state transition relationship being P. change (z t+1 |,k t ), and through the policy network σ ω (k|z) output in state z t The optimal action To dynamically adjust design parameters;

[0135] S43. Design a multi-layered composite reward function:

[0136] V reward (z t ,k t )=κ1·L efficiency +κ2·L precision -κ3·R overhead -κ4·D latency ;

[0137] Among them, L efficiencyL represents the optimized efficiency of interactive operations. precision This indicates improved accuracy in user feedback, R overhead D represents the cost of resource consumption. latency κ1, κ2, κ3, and κ4 represent response delays, and are reward weighting coefficients.

[0138] S44. Using a dual-policy network structure based on deep reinforcement learning, define the evaluation network Q. eval (z,k) and the target network Q target (z,k), using a recursive time difference update strategy for parameter ω:

[0139]

[0140] Where ν is the learning rate, ρ is the discount factor, and k ′ The optimal action in the next state;

[0141] S45, Optimize the policy network Dynamic generation rules applied to design schemes, combined with a real-time updated state set Z state and action set K action Adjust the design parameter space to output personalized design solutions that meet the user's real-time needs.

[0142] S5. Construct a multi-objective design space exploration and optimization framework supported by neural networks, and quickly screen and optimize design parameters through high-dimensional nonlinear solution and parallel computing strategies;

[0143] In this embodiment, S5 specifically includes:

[0144] S51. Construct a design space exploration model T based on multi-objective nonlinear expression. explore Define the design parameter set Ω params ={ω1,…,ω m}, Objective function set Λ objectives ={λ1(Ω),…,λ p (Ω)} and constraint set Γ constraints ={γ1(Ω),…,γ q (Ω)}, where λ i (Ω) represents the optimization objective of the parameter set in dimension i;

[0145] S52. Constructing a high-dimensional nonlinear target prediction model N based on deep neural networks. optimize The input is the design parameter set Ω params The output is a set of predicted target values. By optimizing the network weight set Θ optimize Minimize the prediction loss function:

[0146]

[0147] Where, λ i (Ω) represents the true target value;

[0148] S53. Employ a multi-objective optimization algorithm based on the dynamic Pareto front, and define the Pareto boundary solution set Δ. Pareto ={δ1,δ2,…,δ r}, where δ k Indicates a non-inferior solution:

[0149]

[0150] Among them, O multi_goal For a multi-objective optimization function, ζ i This represents the target weight, and the solution set is optimized according to a dynamic update strategy.

[0151] S54. Employ a distributed parallel optimization strategy for the design parameter set Ω. params Partition the dataset into computational subsets P. k ={ω k1 ,ω k2 ,…,ω kn In a multi-threaded environment, the objective function value and Pareto boundary conditions are computed in parallel, and the results are aggregated to generate an optimized solution set.

[0152] S55, Based on Pareto solution set Δ Pareto A global update generates an optimized set of design parameters. This is used as the output of a personalized design scheme.

[0153] S6. Employ dynamic visualization and multimodal interaction technologies to construct a real-time design scheme presentation and interactive interface. Through multidimensional data mapping and parameterized control, achieve dynamic updates and local optimization of the design scheme.

[0154] In this embodiment, S6 specifically includes:

[0155] S61. Construct a multimodal interactive data acquisition and fusion module Q interaction Collect user operation behavior set U action ={u1,u2,…,u s Real-time feedback set V feedback ={v1,v2,…,v t} and environmental parameter set W context ={w1,w2,…,w r}, fusion generates dynamic data stream Q dynamic ={U action Vfeedback W context};

[0156] S62. Design a real-time visualization engine R based on dynamic multidimensional data mapping. visualize Define the set of mapping functions Ψ map ={ψ1(Q),ψ2(Q),…,ψ h (Q)}, where ψ k (Q) represents the interaction data Q. dynamic To the visualization dimension R view The dynamic mapping relationship generates a multi-dimensional visualization scheme set R in real time. design ={ρ1,ρ2,…,ρ h};

[0157] S63, via parameterized control module P adaptive Dynamically adjust the set of design parameters Θ parameters ={θ1,θ2,…,θ z Construct a local optimization objective function:

[0158]

[0159] Where, τ j To optimize the weighting coefficients, H(θ) j Q dynamic ) represents the parameter θ j With dynamic data Q dynamic Nonlinear relationship;

[0160] S64. Utilize the user's set of local optimization requirements K local ={k1,k2,…,k l Establish a mapping matrix Φ between requirements and parameters. adjust ={υ ij |=G(k i ,θ j )}, where φ ij Indicates demand k i For parameter θ j The influence value, G is the dynamic mapping function, through matrix Φ adjust The design parameters are partially updated to generate an optimized parameter set.

[0161] S65, through the real-time visualization engine R visualize The optimized design scheme Θ optimal Presented to the user interface I interactive It provides a dynamic adjustment interface, allowing users to perform secondary personalized optimization based on real-time input of local parameters.

[0162] S7. Establish a data-driven continuous learning and evolution mechanism, and continuously optimize user demand prediction models and design optimization algorithms by integrating new user interaction data and historical design optimization information.

[0163] In this embodiment, S7 specifically includes:

[0164] S71. Construct a continuous learning module L based on data fusion. fusion Define the user real-time interaction data set Ξ realtime ={ξ1,ξ2,…,ξ a} and historical optimization record set Θ history ={θ1,θ2,…,θ b}, fusion to generate learning dataset Π training ={Ξ realtime ,Θ history} is used for dynamic model updates;

[0165] S72. Design a user demand prediction model U based on deep recursive networks. predict The input is the learning dataset Π training The output is a set of user demand prediction results Λ output ={λ1,λ2,…,λ c}, by optimizing the model parameter set Γ weights Minimize the prediction error function:

[0166]

[0167] Where, λ i This is the actual demand value. These are the model's predicted values;

[0168] S73. Employ a dynamic incremental learning mechanism for dataset Π training Continuous expansion is carried out, and the parameter set Γ of the demand forecasting model is adjusted based on the multi-step gradient descent method. weights :

[0169]

[0170] Where η is the learning rate and t is the number of update steps;

[0171] S74. Constructing a design optimization and evolution module G based on genetic algorithms. optimize Define the set of optimization parameters Δ parameters ={δ1,δ2,…,δ d} and the set of objective functions Φ targets ={υ1,υ2,…,υ e The objective function is:

[0172]

[0173] Where, ζ j For the target weight, T(δ) j ,Π training The ) indicates the fitness of the optimization parameters for the comprehensive dataset;

[0174] S75. Based on the optimized user demand prediction model and design optimization parameter set The rules for generating personalized designs are dynamically adjusted, and the optimization results are applied to the design output module in real time to achieve adaptive evolution and continuous optimization.

[0175] Example 1:

[0176] To verify the feasibility of this invention, it was applied to a well-known domestic smart home customization company, which mainly engages in the design and production of personalized customized products. Due to the complexity and diversity of customer needs, traditional design methods have significant limitations in terms of dynamic response capabilities, design efficiency, and user satisfaction, making it difficult to meet market demands. Therefore, the company introduced the topology optimization-based personalized design intelligent interaction method proposed in this invention to optimize its design process, improve customer satisfaction, and reduce design costs.

[0177] In a real-world application scenario, the company tested a customized design task for a smart wardrobe. The client wanted a wardrobe with highly personalized features, including dynamically adjusting its size according to the room layout, providing a modular storage design, and integrating a smart lighting system. Traditionally, designers would need to communicate with the client multiple times, manually sketching and refining design models to complete the task. This process typically takes more than two weeks, and the accuracy of the design depends heavily on the designer's subjective understanding of the client's needs, making it prone to errors.

[0178] Through the implementation of this invention, the company has established a user interaction system based on multimodal data acquisition, capturing customer actions on the interface in real time, including clicks, voice input, and text descriptions. This data is integrated into a deep neural network model for requirement modeling, generating hierarchical, personalized requirement prediction results. Subsequently, based on a topology optimization algorithm, the system transforms the requirements into a multi-objective optimization problem, generating design parameters that meet the requirements through topology reconstruction. Simultaneously, an intelligent feedback mechanism dynamically captures user adjustment behaviors, such as adjusting module size or adding new features, updating design rules in real time through a reinforcement learning model.

[0179] In practice, the customer first selects the initial style and color of the wardrobe through the system interface and enters the dimensions and layout of their room. The system generates a preliminary design within 10 seconds, including a recommended modular storage layout and smart lighting configuration. The customer then adjusts the wardrobe's internal structure through the interface, such as increasing the number of drawers and changing the door opening mechanism. The system captures these adjustments in real time, updates the design using a topology optimization algorithm, and finally generates a final design that meets the customer's needs after two rounds of adjustments. The data collected and optimized by the system in this case is shown in Table 1.

[0180] Table 1: Comparison of Design Efficiency and Effectiveness of Smart Home Customization Companies

[0181]

[0182] As shown in Table 1, traditional design methods require 15 days to complete a design, while this invention shortens the design cycle to 3 days. Simultaneously, user satisfaction scores increased from 6.8 points with the traditional method to 9.4 points; the number of design adjustments decreased from an average of 7 to 2; and the accuracy of the design improved from 70% to 95%. Furthermore, the system's response time was reduced from the traditional 2 minutes to 8 seconds, significantly improving the interactive experience.

[0183] During three months of application testing, the company completed over 200 personalized design tasks, covering multiple scenarios such as smart wardrobes, bookshelves, and kitchen storage systems. User surveys showed that over 93% of clients were highly satisfied with the design solutions generated by the system. Furthermore, the company saved nearly 30% on design costs due to shorter design cycles and improved accuracy.

[0184] This invention provides an intelligent, efficient, and dynamically adaptive solution for the field of personalized design by combining topology optimization, multimodal interaction, deep learning, and continuous learning technologies. This significantly improves design efficiency and user satisfaction, creating substantial economic benefits and market competitive advantages for enterprises.

[0185] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A personalized intelligent interaction design method based on topology optimization, characterized in that, Includes the following steps: S1. Construct a user behavior monitoring system based on multimodal data acquisition to obtain the user's operation trajectory, preference features and multi-source feedback information in real time during the interaction process, and generate a high-dimensional dataset containing time-series features; S2. Based on an adaptive deep neural network model, by integrating user behavior features and contextual semantic information, dynamic prediction and hierarchical modeling of user personalized needs are performed. S3. A personalized design adjustment module driven by topology optimization transforms user demand predictions into a multi-objective optimization problem, generating optimal solutions for design parameters in real time through a dynamic topology reconstruction algorithm; specifically including: S31. User Demand Prediction Result Matrix Construct a multi-objective optimization model and define a set of objective functions. ,in Represents the set of design parameters In the target dimension Optimize performance while setting a set of constraints. This limits the search scope of the design space; S32. Using a non-uniform topological mesh generation method, the design parameter set is... Discretized into a multi-scale grid structure ,in Indicates the first , The state variables of the grid cells adopt dynamic topology variables. Represents grid cells Based on the weight distribution, construct the mesh optimization function: ； in, For the objective function Priority weight, The grid complexity regularization term is dynamically optimized using a nonlinear constraint function. ; S33. Design a sensitivity analysis module, calculate the sensitivity of the topological state to the optimization objective based on higher-order partial derivatives, and define the sensitivity matrix. ,in The sensitivity matrix is ​​optimized by dynamically adjusting the strategy. And enhance the global search capability of the optimization model; S34. Employing a distributed computing framework combined with a dynamic topology reconfiguration algorithm... Perform iterative optimization to generate an optimization result matrix. ,in This represents the final state of the mesh cells after multi-objective optimization; S35. By using the inverse mapping relationship from mesh to design parameters, the topology optimization result matrix is... Mapping to the design parameter set generates the final optimized design parameter set. As an output of personalized design solutions; S4. Through reinforcement learning mechanism, design intelligent feedback algorithm based on users’ real-time operation and interaction behavior, and iteratively update the design scheme generation rules; S5. Construct a multi-objective design space exploration and optimization framework supported by neural networks, and quickly screen and optimize design parameters through high-dimensional nonlinear solution and parallel computing strategies; S6. Employ dynamic visualization and multimodal interaction technologies to construct a real-time design scheme presentation and interactive interface. Through multidimensional data mapping and parameterized control, achieve dynamic updates and local optimization of the design scheme. S7. Establish a data-driven continuous learning and evolution mechanism, and continuously optimize user demand prediction models and design optimization algorithms by integrating new user interaction data and historical design optimization information.

2. The personalized design intelligent interaction method based on topology optimization according to claim 1, characterized in that, S1 specifically includes: S11. Based on multimodal sensing devices, collect user operation behavior data and combine it with interaction context information to generate a dynamic dataset, including operation trajectories. Biometric signals and environmental parameters ; S12. Construct a modality fusion algorithm, using a dynamic weight allocation mechanism to normalize and align the features of multimodal data, forming a fused feature matrix. ,in Indicates the first The trajectory dimension and the first The biological signal in the first Normalized values ​​at each time step; S13. Extract dynamic feature vectors of user behavior based on an adaptive time series model. ,in Representing the time dimension User behavior pattern weights; S14. Construct a user preference feature clustering model and extract user preference features through recursive hierarchical clustering analysis. ,in Indicates the user in the category The preferred feature vectors under; S15. Design a multi-source feedback sensing module to collect user interaction feedback on the current design scheme in real time and generate a feedback matrix. ,in Indicates feedback category At time step The weight values ​​below; S16. Transfer the behavioral feature vector Preference characteristics and feedback matrix Fusion generates dynamic feature sets through a high-dimensional feature interaction model. ,in Representing behavioral feature dimensions Preference Dimension and feedback categories The overall weight value; S17. Based on sparse optimization and deep feature extraction algorithms Dimensionality reduction is performed to generate optimized feature vectors for user behavior. ,in Indicates the first Key features.

3. The personalized design intelligent interaction method based on topology optimization according to claim 1, characterized in that, S2 specifically includes: S21. Construct a deep neural network model based on adaptive parameter control. The model structure includes multiple layers of nonlinear activation units, recursive enhancement modules, and context-weighted feature units, and the model weights are updated in real time through a dynamic optimization strategy. and bias vector ; S22. Optimize user behavior feature vectors and contextual semantic feature matrix Enter to ,in Representing behavioral feature dimensions The weight, For context category In the semantic dimension The weighted value on; S23. Through the feature cross-fusion module, and Perform joint embedding to generate a high-dimensional joint feature representation matrix: ； ； in, This is a feature fusion function that represents the interaction between behavioral and semantic features; S24. Based on the demand-layered forecasting unit, based on Generate a demand vector through multi-channel feature extraction: ； in, , and These represent the prediction layer weights and bias vectors, respectively. It is a non-linear activation function; S25. In the time-step recursive analysis module, based on the user's historical behavior sequence matrix... and the current context feature matrix A time-step prediction model is constructed using a recurrent neural network to generate time steps. Dynamic demand distribution ,in , '' represents the recursive relation mapping function; S26. Construct the optimization objective function: ； in, Represents the set of model parameters. and These are the demand weights and time correlation coefficients, respectively, which are solved using an adaptive gradient optimization algorithm. S27. Generate the demand forecast result matrix ,in Indicates the category of demand The predicted value at time step t is used to optimize and iterate the matrix using a dynamic update mechanism, and the result is output to the topology optimization module as a basis for dynamic adjustment of design parameters.

4. The personalized design intelligent interaction method based on topology optimization according to claim 1, characterized in that, S4 specifically includes: S41. Construct a deep reinforcement learning model based on multi-source interaction data. , including state space Action space and reward function , where the state This indicates the characteristics of real-time user interaction; S42. Construct a state transition probability model based on user interaction behavior, with the state transition relationship as follows: and through policy network Output in state The optimal action To dynamically adjust design parameters; S43. Design a multi-layered composite reward function: ； in, This indicates the optimized efficiency of interactive operations. This indicates an improvement in the accuracy of user feedback. Indicates the cost of resource consumption. This indicates a response delay. , , , This is the reward weighting coefficient; S44. Define the evaluation network using a dual-policy network structure based on deep reinforcement learning. and target network The parameters are updated using a recursive time difference strategy. : ； in, For learning rate, As a discount factor, The optimal action in the next state; S45, Optimize the policy network Dynamic generation rules applied to design schemes, combined with real-time updated state sets. and action set Adjust the design parameter space to output personalized design solutions that meet the user's real-time needs.

5. The personalized design intelligent interaction method based on topology optimization according to claim 1, characterized in that, S5 specifically includes: S51. Constructing a design space exploration model based on multi-objective nonlinear expression. Define the set of design parameters , set of objective functions and constraint set ,in Indicates the parameter set in dimension The following optimization objectives; S52. Constructing a high-dimensional nonlinear target prediction model based on deep neural networks. The input is a set of design parameters. The output is a set of predicted target values. By optimizing the network weight set Minimize the prediction loss function: ； in, Indicates the true target value; S53, adopts a dynamic-based approach Cutting-edge multi-objective optimization algorithms, defined Boundary solution set ,in Indicates a non-inferior solution: ； in, For a multi-objective optimization function, This represents the target weight, and the solution set is optimized according to a dynamic update strategy. S54. Employ a distributed parallel optimization strategy for the design parameter set. Partitioning the data into computational subsets In a multi-threaded environment, the objective function value is computed in parallel. Boundary conditions are aggregated to generate an optimized solution set; S55, based on Solution set A global update generates an optimized set of design parameters. This is used as the output of a personalized design solution.

6. The personalized design intelligent interaction method based on topology optimization according to claim 1, characterized in that, S6 specifically includes: S61. Construct a multimodal interactive data acquisition and fusion module. Collect user operation behavior sets Real-time feedback collection and environmental parameter set , fusion to generate dynamic data stream ; S62. Design a real-time visualization engine based on dynamic multidimensional data mapping. Define the set of mapping functions ,in Indicates data from interaction To the visualization dimension The dynamic mapping relationship generates a set of multi-dimensional visualization solutions in real time. ; S63, via parameterized control module Dynamically adjust the set of design parameters Construct a local optimization objective function: ； in, To optimize the weighting coefficients, Indicates parameters With dynamic data Nonlinear relationship; S64. Utilize the user's set of local optimization requirements. Establish a mapping matrix between requirements and parameters. ,in Indicate demand For parameters The impact value, For dynamic mapping functions, through matrices The design parameters are partially updated to generate an optimized parameter set. ; S65, through a real-time visualization engine The optimized design scheme Presented to the user interface It provides a dynamic adjustment interface, allowing users to perform secondary personalized optimization based on real-time input of local parameters.

7. The personalized design intelligent interaction method based on topology optimization according to claim 1, characterized in that, Specifically, S7 includes: S71. Construct a continuous learning module based on data fusion. Define the set of real-time user interaction data and historical optimization record set , fusion to generate learning datasets , used for dynamic model updates; S72. Design a user demand prediction model based on deep recursive networks. The input is the learning dataset. The output is a set of user demand prediction results. By optimizing the model parameter set Minimize the prediction error function: ； in, This is the actual demand value. These are the model's predicted values; S73. Employ a dynamic incremental learning mechanism for the dataset. Continuously expand the model and adjust the parameter set of the demand forecasting model based on the multi-step gradient descent method. : ； in, For learning rate, To update the step count; S74. Construct a design optimization and evolution module based on genetic algorithms. Define the set of optimization parameters and the set of objective functions The objective function is optimized as follows: ； in, For the target weight, This indicates the fitness of the optimization parameters for the comprehensive dataset; S75. Based on the optimized user demand prediction model and design optimization parameter set The system dynamically adjusts the personalized design generation rules and applies the optimization results to the design output module in real time, achieving adaptive evolution and continuous optimization.

Images