Smart home quick matching method and system

Abstract

This invention discloses a method and system for rapid smart home selection, relating to the field of rapid selection technology. The method includes: feature encoding of user input information and matching it with historical selection data to generate implicit demand tags; spatial modeling of the apartment layout based on spatial analysis algorithms to obtain spatial constraint results; screening candidate home furnishing product combinations from a preset product library based on the spatial constraint results and implicit demand tags; allocating budgets to the candidate home furnishing product combinations according to budget information, based on the security, comfort, and entertainment function identifiers corresponding to the home furnishing products, and allocating budget information according to preset function budget weights; and generating an initial selection list after obtaining the function budget amount. This invention reduces the possibility of irrelevant home furnishing products being replaced and minimizes the overwriting of user operation results by retaining home furnishing products explicitly added or retained by the user and making local adjustments around home furnishing products with high degree of constraint conflict.

Description

Technical Field

[0001] This invention relates to the field of rapid selection and matching technology, and in particular to a method and system for rapid selection and matching of smart homes. Background Technology

[0002] With the development of IoT, machine learning, home network communication, and smart terminal control technologies, smart homes have gradually shifted from single-product configurations to multi-category collaborative configurations. Home products no longer simply meet user needs with independent functions; instead, a comprehensive selection result needs to be formed based on factors such as apartment layout, power supply conditions, communication protocols, budget, and usage preferences. To improve selection efficiency, related technologies are beginning to employ user profiling analysis, historical selection data matching, product database retrieval, and rule-based filtering to assist in generating selection lists. Furthermore, data-driven methods are used to record the functional, installation, and communication attributes of different home products, thus providing fundamental data support for rapid smart home selection.

[0003] Existing smart home selection methods typically focus on matching explicit user needs, with insufficient consideration given to the interplay between apartment size, rated circuit power, network communication protocols, budget information, and potential needs. Furthermore, after users add, delete, or replace target selection items, existing methods often directly delete or replace home products according to fixed rules, lacking protection for newly added home products and products retained after replacement. They also struggle to dynamically determine the scope of removal and the order of repair based on the degree of constraint conflict, the impact of insertion, and historical adjustment results, easily expanding the scope of list changes and covering the results of user operations. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for rapid selection and configuration of smart home systems to solve the problem of difficulty in dynamically determining the removal range and repair sequence based on the degree of constraint conflict, insertion impact, and historical adjustment results.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for rapid selection and configuration of smart home systems, comprising, Obtain user input information, including apartment dimensions, rated circuit power, network communication protocol, and budget information; The implicit demand labels are generated by encoding features of user input information based on machine learning models and combining them with historical matching data. The spatial modeling of the apartment size is performed based on the spatial analysis algorithm to obtain spatial constraint results. Based on the spatial constraint results and implicit demand tags, candidate home furnishing product combinations are selected from the preset product library. Based on the budget information, the budget is allocated to the candidate home furnishing product combination. The budget allocation is based on the security, comfort and entertainment function labels of the home furnishing products. The budget information is allocated according to the preset function budget weights. After obtaining the function budget amount, an initial selection list is generated. Perform power matching calculations and network communication protocol matching calculations on the initial configuration list; When the power matching calculation result meets the circuit rated power constraint and the network communication protocol matching calculation result meets the consistency constraint, the target selection list is output. If one condition is not met, the corresponding replacement operation is performed until both conditions are met simultaneously, and a target configuration list is generated. The system receives user operations to add, delete, and replace items in the target selection list, generates an updated selection list, and sets user operation retention flags for newly added and replaced furniture products. The updated selection list is used as the current solution, keeping the furniture products with user operation retention flags unchanged. An adaptive large neighborhood search algorithm is used to remove some furniture products according to the degree of constraint conflict, and the current solution is repaired according to the regret insertion value of the replacement candidate products. The weights of the destruction operator and the repair operator are updated based on the repair results, and the updated target selection list is generated iteratively.

[0007] As a preferred embodiment of the smart home rapid selection method of the present invention, the machine learning model includes an input embedding layer, a feature encoding layer, an attention fusion layer, a historical memory matching layer, and an output layer.

[0008] As a preferred embodiment of the smart home rapid selection method described in this invention, the generation of implicit demand tags specifically includes: User input information is fed into the input embedding layer, and the user input information is vectorized to obtain the initial feature vector; A multidimensional feature representation is obtained by performing a nonlinear feature transformation on the initial feature vector based on the feature coding layer. The multidimensional feature representation is input into the attention fusion layer, and weights are assigned and fused for different types of input features to obtain the fused feature representation. In the historical memory matching layer, the fused feature representation is compared with the historical matching data to perform vector similarity retrieval, thereby obtaining historical association features; By performing semantic summarization and labeling of historical correlation features through the output layer, implicit demand labels are formed.

[0009] As a preferred embodiment of the smart home rapid selection and configuration method of the present invention, wherein obtaining the spatial constraint result specifically includes: Convert the apartment dimensions into a uniform-scale apartment coordinate matrix, and write the room boundaries, door and window positions, and fixed structure positions from the apartment dimensions into the apartment coordinate matrix; The coordinate matrix of the apartment type is divided into grids according to the spatial analysis algorithm, and the installable markers, occlusion markers, passage and avoidance markers and boundary distances corresponding to each grid are calculated. Based on the installable markings, obstruction markings, passage and avoidance markings, and boundary distances, calculate the available area, effective installation length, coverage radius, and product capacity within the functional space area; The available area, effective installation length, coverage radius, and product capacity are summarized into regional adaptation parameters. Based on the regional adaptation parameters, the system is matched with the preset product installation rules to determine the types of products that can be installed in each functional space area, the upper limit of the number of products, the range of installation locations, and the coverage threshold. These parameters are then summarized into spatial constraint results.

[0010] As a preferred embodiment of the smart home rapid selection method of the present invention, the step of screening candidate home product combinations in a preset product library based on spatial constraint results and implicit demand tags specifically includes: Read the product type, product quantity limit, installation location range, and coverage threshold from the spatial constraint results, and read the product type, installation size, coverage, and requirement tags of home furnishing products from the preset product library; Match implicit demand tags with demand tags to obtain demand-suitable products; The demand-adapted products are compared with product type, installation location range and coverage threshold. Demand-adapted products that do not meet the spatial constraints are filtered out to obtain spatially adapted products. Combine space-adaptive products according to the maximum number of products, and calculate the completeness of space coverage and the degree of matching with needs after the combination. The combined space-adaptive products are sorted according to the completeness of space coverage and the degree of matching with needs. The space-adaptive product combinations that meet the preset screening conditions are determined as candidate home furnishing product combinations.

[0011] As a preferred embodiment of the smart home rapid selection method described in this invention, the generation of the initial selection list specifically includes: Read the total budget amount, single item price limit, and installation service budget from the budget information, and read the product price, installation cost, and security, comfort, and entertainment function labels of each home product in the candidate home product combination; Establish a functional budget allocation order based on the functional identification of security, comfort and entertainment functions, and set the budget weights for security, comfort and entertainment functions according to the functional budget allocation order. Allocate the budget information according to the order of priority of security function identification, second of comfort function identification, and supplement of entertainment function identification, as well as the budget weights of security, comfort and entertainment functions, to obtain the functional budget amount. The product price and installation cost of each home furnishing product are compared with the corresponding functional budget. Home furnishing products that exceed the corresponding functional budget are eliminated to obtain budget-suitable products. The budget-compatible products are combined and verified according to the total budget amount, the upper limit of the price of each item and the installation service budget to determine the product combination that meets the budget information. The product mix that meets the budget information is processed into a list according to functional space area, home furnishing products, product price, installation location and function label, generating an initial selection list.

[0012] As a preferred embodiment of the smart home rapid selection and matching method of the present invention, the step of performing power matching calculation and network communication protocol matching calculation specifically includes: Based on the home furnishing products in the initial selection list, the power and supported network communication protocols of each home furnishing product are read from the preset product library, and the rated power of the circuit and the network communication protocol are read from the user input information. The power consumption of each home furnishing product is added up according to the functional space area to obtain the area power value, and the power values ​​of each area are added up to obtain the total power value of the list; The total power value in the list is compared with the rated power of the circuit, the power difference and the power occupancy ratio are calculated, and the power matching calculation results are generated based on the power difference and the power occupancy ratio. The network communication protocols supported by each home furnishing product are compared with the network communication protocols in the user input information to obtain a protocol consistency mark for each home furnishing product. Based on the protocol consistency markers, the number of products with consistent protocols and the number of products with inconsistent protocols are counted, and the network communication protocol matching calculation results are generated.

[0013] As a preferred embodiment of the smart home rapid selection and configuration method of the present invention, the step of performing the corresponding replacement operation specifically includes: Using the initial selection list as the current selection list, when the power matching calculation result does not meet the circuit's rated power constraint, replace the power-over-limit home products with low-power replacement products that meet the space constraint result, implicit requirement label, and budget information, and update the current selection list; When the network communication protocol matching calculation result does not meet the consistency constraint, replace the home products with incompatible network communication protocols with protocol replacement products that support the network communication protocol, and update the current selection list.

[0014] As a preferred embodiment of the smart home rapid selection method of the present invention, the step of generating and updating the target selection list specifically includes: Read the updated optional list, and re-execute the budget allocation, power matching calculation, network communication protocol matching calculation, and spatial constraint comparison on the updated optional list to obtain the updated budget allocation results, updated power matching calculation results, updated network communication protocol matching calculation results, and updated spatial constraint comparison results; The budget overrun items in the updated budget allocation results are associated with the home furnishing products within the corresponding functional budget limits. The regional power overrun items and the total power overrun items in the updated power matching calculation results are associated with the home furnishing products within the corresponding functional space area. The protocol inconsistency items in the updated network communication protocol matching calculation results are associated with the corresponding home furnishing products. The spatial mismatch items in the updated spatial constraint comparison results are associated with the home furnishing products within the corresponding functional space area. Home furnishing products that do not have user operation retention marks and are associated with at least one non-satisfied item are collected to form a set of conflicting home furnishing products. Remove the home furnishing products from the conflicting home furnishing product set from the updated selection list to form a constraint restoration list, and re-perform budget allocation, power matching calculation, network communication protocol matching calculation, and spatial constraint comparison on the constraint restoration list; When there are still unmet items in the constraint recovery list, the home furnishing products with user operation retention tags and associated with the unmet items are identified as retention conflict home furnishing products, and constraint conflict results are generated based on the retention conflict home furnishing products and the unmet items; When the constraint recovery list meets the budget information, circuit rated power constraints, network communication protocol consistency constraints, and space constraints, the related home furnishing products are selected one by one from the conflicting home furnishing product set and written back to the constraint recovery list in order of preset functional budget weight from high to low and demand matching degree from high to low. After each write-back of related home products, the budget allocation, power matching calculation, network communication protocol matching calculation, and space constraint comparison are re-executed. Related home products that still meet the budget information, circuit rated power constraints, network communication protocol consistency constraints, and space constraint results after the write-back are retained, and related home products that generate unsatisfactory items after the write-back are canceled. Related home products that have not been written back after all write-backs have been completed are collected into a set of products that cannot be further reduced or replaced. Read the product identifier, product type, functional space area, function identifier, product price, product power, coverage area, and supported network communication protocols of home products in the set of products that cannot be further reduced or replaced. Retrieve products from the preset product library that have the same product type, functional space area, and function identifier, and whose product identifier is different from the product identifier of the home products deleted by the user. This satisfies the spatial constraint results and implicit requirement tags, and supports replacement candidate products for network communication protocols in the user input information. Based on the correspondence between home furnishing products and replacement candidate products in the set of irreducible replacement products, the replacement candidate products are combined to form replacement candidate combinations. The replacement candidate combinations are written into the constraint recovery list, and the budget allocation, power matching calculation, network communication protocol matching calculation and spatial constraint comparison are re-executed. Replacement candidate combinations that simultaneously meet the budget information, circuit rated power constraints, network communication protocol consistency constraints and spatial constraint results are retained. Calculate the relative differences in product price, power, and coverage between the candidate replacement products and their corresponding home furnishing products in the candidate replacement combinations. Sum these differences to obtain the replacement disturbance value. Select candidate replacement combinations from smallest to largest according to the replacement disturbance value. Write the selected candidate replacement combinations as automatic replacement products into the constraint recovery list and generate the updated target selection list.

[0015] Secondly, this invention provides a smart home rapid selection system, including, The data acquisition module obtains user input information, including apartment dimensions, rated circuit power, network communication protocol, and budget information. The encoding module encodes user input information based on a machine learning model and performs matching processing in conjunction with historical selection data to generate implicit demand labels. The filtering module performs spatial modeling of the apartment size based on spatial analysis algorithms to obtain spatial constraint results, and filters candidate home furnishing product combinations from a preset product library based on the spatial constraint results and implicit demand tags. The optional module allocates budget to candidate home furnishing product combinations based on budget information. The budget allocation is based on the security, comfort and entertainment function labels of the home furnishing products and allocates the budget information according to the preset function budget weights. After obtaining the function budget amount, an initial optional list is generated. The calculation module performs power matching calculations and network communication protocol matching calculations on the initial selection list; The matching module outputs a target selection list when the power matching calculation result meets the circuit rated power constraint and the network communication protocol matching calculation result meets the consistency constraint. The replacement module performs the corresponding replacement operation when one condition is not met, until both conditions are met simultaneously, and then generates the target configuration list. The update module receives user operations to add, delete, and replace items in the target selection list, generates an updated selection list, and sets user operation retention flags for newly added and replaced furniture products. The updated selection list is used as the current solution, keeping the furniture products with user operation retention flags unchanged. An adaptive large neighborhood search algorithm is used to remove some furniture products according to the degree of constraint conflict, and the current solution is repaired according to the regret insertion value of the replacement candidate products. The weights of the destruction operator and the repair operator are updated based on the repair results, and the updated target selection list is generated iteratively.

[0016] The beneficial effects of this invention are as follows: By comprehensively considering factors such as apartment size, rated circuit power, network communication protocols, budget information, and historical selection data, the selection of home furnishing products is no longer limited to explicit preference matching. It helps to take into account potential functional needs, actual installation conditions, and budget allocation, reducing situations such as insufficient budget for core functions, excessive product power, inconsistent network communication protocols, and spatial incompatibility, thereby improving the rationality and feasibility of the initial selection list. After the user adds, deletes, or replaces items on the target selection list, by retaining the home furnishing products that the user explicitly adds or retains, and making local adjustments around home furnishing products with a high degree of constraint conflict, the possibility of irrelevant home furnishing products being replaced can be reduced, the scope of changes to the selection list can be narrowed, and the user's operation results can be reduced from being overwritten. At the same time, the updated target selection list maintains good coordination between budget, power, network communication protocols, and space limitations. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart for a quick selection method for smart home systems.

[0019] Figure 2 Generate a flowchart for implicit requirement tags.

[0020] Figure 3 A flowchart for screening spatial constraints and candidate furniture product combinations.

[0021] Figure 4 Generate a flowchart for budget allocation and initial selection list.

[0022] Figure 5 Flowchart for updating the target optional list. Detailed Implementation

[0023] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0024] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0025] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0026] Reference Figures 1-5 This is one embodiment of the present invention, which provides a method for rapid selection and configuration of smart home systems, including the following steps: S1: Obtain user input information, including apartment dimensions, rated circuit power, network communication protocol, and budget information; The user interface includes fields for entering floor plan dimensions, rated circuit power, network communication protocol selection, and budget information. It receives user-entered information such as room length, width, height, boundaries, door and window locations, fixed structure locations, and room connectivity. Based on room identifiers, it assigns room length, width, height, and boundaries to corresponding rooms. It determines the area use of each room based on its functional space area name. It records doorway connections, corridor connections, and open boundary connections between different rooms to form the floor plan dimensions. It also receives user-entered rated power for distribution circuits, socket circuits, and lighting circuits to form the rated circuit power. Finally, it receives user selections. The system receives the network communication protocol name, network communication protocol version, and network connection method to form the network communication protocol; it receives the total budget amount, single item price limit, and installation service budget filled in by the user to form budget information; it performs integrity and format checks on the apartment size, rated circuit power, network communication protocol, and budget information according to the field name. The integrity check confirms that the apartment size, rated circuit power, network communication protocol, and budget information are all filled in and that there are no missing fields. The format check confirms that the apartment size uses the size value format, the rated circuit power uses the power value format, the network communication protocol uses the protocol name and protocol version format, and the budget information uses the amount value format. After the checks pass, they are merged to form the user input information.

[0027] S2: Based on the machine learning model, the user input information is feature-encoded and matched with historical selection data to generate implicit demand labels.

[0028] S2.1: The training set for the machine learning model is extracted from completed historical selection data. This historical selection data includes historical user input information, historical candidate home furnishing product combinations, historical initial selection lists, historical target selection lists, and historical user add / delete / replace operations. Functional space areas, home furnishing products, and function identifiers are read from the historical target selection lists. Budget information is read from historical user input information. Home furnishing products retained, deleted, and replaced by users are read from historical user add / delete / replace operations. An implicit demand labeling system is pre-established, including functional space area labels, function category labels, product type labels, and budget matching labels. The functional space area label is determined by the functional space area itself; the function category label is determined by security, comfort, and entertainment function identifiers; and the product type label is determined by the product type of the home furnishing product. Budget-matching labels are determined by comparing the product price of the home furnishing product with the maximum price limit of a single item in the budget information. Positive labels are assigned to home furnishing products in the historical target selection list, home furnishing products retained by the user, and home furnishing products replaced in the user's replacement operation. Negative labels are assigned to home furnishing products deleted by the user and home furnishing products replaced in the user's replacement operation. Positive labels with the same functional space area label, functional category label, product type label, and budget-matching label under the same historical user input information are merged, and label combinations with more negative labels than positive labels are removed. The remaining label combinations form the implicit demand label labeling results corresponding to the historical user input information. The historical user input information and the implicit demand label labeling results are combined to form a training set.

[0029] The machine learning model comprises an input embedding layer, a feature encoding layer, an attention fusion layer, a history memory matching layer, and an output layer. The input embedding layer converts historical user input information into initial training feature vectors. Key parameters of the input embedding layer include the field encoding dimension and the embedding vector dimension; the field encoding dimension is 32-dimensional, and the embedding vector dimension is 128-dimensional. The feature encoding layer performs non-linear feature transformations on the initial training feature vectors. Key parameters of the feature encoding layer include the number of hidden layers, the hidden layer dimension, and the activation method. There are three hidden layers with dimensions of 256, 128, and 64 dimensions respectively, and the activation method is rectified linear unit activation. The attention fusion layer performs weight allocation and fusion processing on different types of input features. Key parameters of the attention fusion layer include the attention weight matrix and the number of attention heads. The attention weight matrix is ​​a 128-dimensional matrix, with 4 attention heads. The weight normalization method is normalized exponential function normalization. The historical memory matching layer is used to perform vector similarity retrieval between the training fused feature representation and historical selected data. The key parameters of the historical memory matching layer include the historical associated feature vector library, the similarity calculation method, and the number of matches. A single historical associated feature vector in the historical associated feature vector library is 128-dimensional. The similarity calculation method is cosine similarity calculation, and the number of matches is 20. The output layer is used to output the training implicit demand label prediction results. The key parameters of the output layer include the number of implicit demand label categories, the label probability threshold, and the accuracy threshold. The number of implicit demand label categories is 12, the label probability threshold is 0.6, and the preset accuracy threshold is 0.6.

[0030] To further explain, the label probability threshold is set based on the range of label probability output by the output layer for implicit demand labels, and is used to filter out implicit demand labels with low confidence, and is set to 0.6; the accuracy threshold is set based on the consistency between the training implicit demand label prediction results and the implicit demand label annotation results, and is used to determine whether the machine learning model training is complete, and is set to 0.6.

[0031] The historical association feature vector library is pre-built from historical configuration data. For each piece of historical configuration data, it reads the functional space area, home furnishing products, and function identifiers from the historical target configuration list, and reads the budget information from the historical user input information. The functional space area is coded as a region, the corresponding product type of the home furnishing products is coded as a product type, the function identifier is coded as a function category, and the matching relationship between the budget information and the product price of the home furnishing products is coded as a budget matching. The region code, product type code, function category code, and budget matching code are each set to 32 dimensions, and are concatenated in the order of region code, product type code, function category code, and budget matching code to form a 128-dimensional historical association feature vector. The historical association feature vector library stores the correspondence between the historical association feature vector number, the historical user input information number, the historical target configuration list number, the 128-dimensional historical association feature vector, and the implicit demand label annotation results.

[0032] When pre-training the machine learning model using the training set, historical user input information is input into the input embedding layer and converted into an initial training feature vector. The initial training feature vector is then fed into the feature encoding layer and converted into a multi-dimensional training feature representation. This multi-dimensional training feature representation is fed into the attention fusion layer to form a fused training feature representation. The fused training feature representation is then fed into the history memory matching layer to extract training historical association features from historical matching data. Finally, the output layer generates a training implicit demand label prediction result. The training implicit demand label prediction result is compared with the implicit demand label annotation result, and the key parameters of the input embedding layer, feature encoding layer, attention fusion layer, history memory matching layer, and output layer are adjusted according to the comparison result. This process of inputting historical user input information, generating training implicit demand label prediction results, comparing differences, and adjusting key parameters is repeated until the training implicit demand label prediction result reaches a preset accuracy threshold, resulting in a fully trained machine learning model.

[0033] S2.2: Input user input information into the input embedding layer of the trained machine learning model. The input embedding layer reads the apartment dimensions, rated circuit power, network communication protocol, and budget information according to the field names. It encodes the room length, width, height, boundaries, door and window locations, and fixed structure locations within the apartment dimensions; it normalizes the rated power of the distribution circuit, socket circuit, and lighting circuit within the rated circuit power; it discretizes the protocol name, protocol version, and network connection method within the network communication protocol; and it performs a numeric encoding on the total budget amount, single-item price ceiling, and installation service budget within the budget information. The data is normalized, and the field encoding results, discrete encoding results, numerical normalization results, and amount normalization results are arranged into an initial feature vector according to the embedding vector dimension. The feature encoding layer receives the initial feature vector and performs layer-by-layer linear transformation and nonlinear activation on the apartment size features, circuit rated power features, network communication protocol features, and budget information features in the initial feature vector according to the number of hidden layers, hidden layer dimensions, and activation methods determined in the trained machine learning model. This transforms the combination relationship between apartment size, circuit rated power, network communication protocol, and budget information into a computable high-dimensional feature expression, resulting in a multi-dimensional feature representation for subsequent weight allocation and fusion processing.

[0034] S2.3: The multi-dimensional feature representation is fed into the attention fusion layer of the trained machine learning model. The attention fusion layer performs linear mapping on the apartment size feature, rated power circuit feature, network communication protocol feature, and budget information feature in the multi-dimensional feature representation according to the determined attention weight matrix, obtaining the weight scores for apartment size, rated power circuit, network communication protocol, and budget information, respectively. These weight scores are then used as corresponding importance values. The attention fusion layer normalizes the importance values ​​to obtain the attention weights corresponding to the apartment size feature, rated power circuit, network communication protocol feature, and budget information feature. The attention fusion layer applies these attention weights to the apartment size feature, rated power circuit, network communication protocol feature, and budget information feature, respectively, to obtain weighted features for apartment size, rated power circuit, network communication protocol, and budget information. These weighted features are then weighted according to the apartment size, rated power circuit, network communication protocol, and budget information features. The weighted features of apartment size, rated power circuit, network communication protocol, and budget information are concatenated according to the field order of the budget information to form a weighted feature sequence. Following the field order of apartment size, rated power circuit, network communication protocol, and budget information, the weighted feature sequence is split into weighted feature vectors for apartment size, rated power circuit, network communication protocol, and budget information. These vectors are then weighted and summed dimension-wise, and the summation results are normalized to obtain a fused feature representation. The historical memory matching layer receives the fused feature representation and calculates its cosine similarity with historical associated feature vectors in the historical matching data. Historical associated feature vectors within the matching range are selected according to their similarity from high to low. The corresponding historical target matching list and implicit demand labeling results are read and merged into historical associated features.

[0035] S2.4: The output layer receives historical association features, reads the historical association feature vector, historical target selection list, and implicit demand label annotation results from the historical association features, and performs a fully connected mapping on the historical association feature vector according to the number of implicit demand label categories already determined in the trained machine learning model to obtain the label score value corresponding to each implicit demand label category; the output layer normalizes the label score value to obtain the label probability value corresponding to each implicit demand label category, and compares the label probability value with the label probability threshold to filter implicit demand label categories with label probability values ​​not lower than the label probability threshold; the filtered implicit demand label categories are matched with the functional space areas, home furnishing products, and functional identifiers in the historical target selection list to verify their correspondence, and the implicit demand label categories with corresponding relationships are retained to form implicit demand labels.

[0036] S3: Based on spatial analysis algorithms, spatial modeling of apartment dimensions is performed to obtain spatial constraint results. Based on the spatial constraint results and implicit demand tags, candidate home furnishing product combinations are selected from a preset product library.

[0037] The obtained spatial constraint result is specifically as follows: S3.1: Read the room identifiers, functional space area names, room lengths, room widths, room heights, room boundaries, door and window positions, fixed structure positions, and room connectivity relationships from the apartment layout dimensions. Using the starting corner point of the room boundary as the coordinate origin, convert the room length, width, and height of each room into coordinate ranges along the coordinate axes according to a unified scale, forming the row, column, and height ranges of the apartment layout coordinate matrix. The apartment layout coordinate matrix adopts a three-dimensional index structure composed of row, column, and height indices. The row index corresponds to the length direction grid under a unified scale, the column index corresponds to the width direction grid under a unified scale, and the height index corresponds to the height direction hierarchy under a unified scale. Each matrix element in the apartment layout coordinate matrix records the corresponding grid number, coordinate range, room identifier, functional space area name, and edge... Boundary markers, door and window markers, fixed structure markers, and connectivity markers are used. Coordinate intervals are represented by the grid start coordinates and grid end coordinates. The grid number is determined by the position of the coordinate interval in the row index, column index, and height index, establishing a correspondence between the coordinate positions in the apartment size and the grid positions in the apartment coordinate matrix. Then, the room boundaries are converted into boundary coordinate sequences and written into the apartment coordinate matrix, the door and window positions are converted into door and window coordinate intervals and written into the apartment coordinate matrix, the fixed structure positions are converted into fixed structure coordinate intervals and written into the apartment coordinate matrix, and the room identifiers, functional space area names, and room connectivity relationships are written into the corresponding coordinate intervals. This allows the apartment coordinate matrix to record the boundary range, area use, and connectivity relationships of each functional space area, and simultaneously record the spatial range, opening positions, and immovable structure positions in the apartment size.

[0038] S3.2: The spatial analysis algorithm divides the room into continuous grids within the row and column range of the floor plan coordinate matrix using a preset grid side length, and assigns grid coordinates to each grid. The spatial analysis algorithm reads the coordinate relationship between the grid coordinates of each grid and the room boundary, door and window positions, and fixed structure positions. When the grid coordinates are within the room boundary and do not fall into the door / window or fixed structure positions, the corresponding grid's installable mark is marked as "installable". When the grid coordinates fall into the fixed structure positions, the corresponding grid's occlusion mark is marked as "occlusion exists". When the grid coordinates fall into the opening range and passage range corresponding to the door / window positions, the corresponding grid's passage avoidance mark is marked as "avoidance required". The spatial analysis algorithm uses the grid center coordinates of each grid as the calculation reference to calculate the distance between the grid center coordinates and the nearest room boundary coordinates, and records the distance as the boundary distance of the corresponding grid, thereby obtaining the installable mark, occlusion mark, passage avoidance mark, and boundary distance corresponding to each grid.

[0039] To further explain, the grid side length is set based on the measurement accuracy of the apartment size, the minimum resolution requirement for the installation size of home furnishing products, and the computational load control requirements of the spatial analysis algorithm. It can be 0.1 meters. This is because when the apartment size is measured at the centimeter level, a grid side length of 0.1 meters can distinguish the positions of doors and windows, fixed structures, and the boundaries of the installable area, and control the number of grids in the apartment coordinate matrix.

[0040] S3.3: The functional space area is determined by the name of the functional space area entered by the user in the apartment size input field, and a corresponding relationship is established with the room identifier. When one room identifier corresponds to only one functional space area name, the coordinate interval enclosed by the room boundary is determined as the functional space area. When one room identifier corresponds to multiple functional space area names, the coordinate interval enclosed by the room boundary is divided into multiple sub-coordinate intervals according to the functional space area boundary lines entered by the user, and each sub-coordinate interval is determined as the corresponding functional space area. The spatial analysis algorithm reads the installable markers, obstruction markers, passage avoidance markers, and boundary distances corresponding to each grid in the functional space area based on the room identifiers, functional space area names, and corresponding coordinate intervals in the apartment coordinate matrix. Grids with the installable marker marked as "allowed for installation," obstruction markers not marked as "obstructed," and passage avoidance markers not marked as "needs to be avoided" are determined as valid grids. The number of valid grids is counted, and the usable area in the functional space area is calculated in combination with the preset grid side length. Valid grids are continuously read along the room boundary direction, and the boundary distances corresponding to consecutive valid grids are determined to meet the installation distance. The required length is accumulated to form the effective installation length. The installation distance is the minimum interval distance that needs to be maintained between the installation location of the home furnishing product and the room boundary, door and window location, and fixed structure location. The installation distance requirement is determined according to the installation location restrictions and avoidance distance requirements of the corresponding home furnishing products in the preset product installation rules. Using the center coordinates of the effective grid as candidate installation locations, the farthest distance from the candidate installation location to the grid that can be covered within the functional space area is calculated, and grids marked as having obstruction or requiring avoidance are excluded to obtain the coverage radius. The number of products that can be accommodated is determined according to the distribution of the effective installation length, coverage radius, and candidate installation locations. The available area, effective installation length, coverage radius, and number of products that can be accommodated are summarized into regional adaptation parameters. The regional adaptation parameters are matched with the installation area requirements, installation length requirements, coverage range requirements, and installation quantity requirements in the preset product installation rules to determine the product types, product quantity limits, installation location ranges, and coverage range thresholds that can be installed in each functional space area. The product types, product quantity limits, installation location ranges, and coverage range thresholds are summarized into spatial constraint results.

[0041] To further explain, the preset product installation rules are stored in correspondence with the home furnishing products in the preset product library. This is used to convert the regional adaptation parameters into the installation access conditions for home furnishing products within the functional space area, thereby determining the allowed product types, product quantity limits, installation location range, and coverage threshold.

[0042] S3.4: From the spatial constraint results, read the product type, product quantity limit, installation location range, and coverage threshold according to the functional space area, and search for home furnishing products of the same product type in the preset product library using the product type as the search condition; for the found home furnishing products, read the product type, installation size, coverage area, and demand tags, and compare the demand tags with the implicit demand tags in terms of tag name and functional space area; when the demand tag name of the home furnishing product is consistent with the implicit demand tag name, and the functional space area applicable to the home furnishing product is consistent with the functional space area associated with the implicit demand tag, the home furnishing product is determined as a demand-matched product.

[0043] S3.5: Compare the product type of the demand-adapted product with the product type in the space constraint results for consistency. Map the installation dimensions of the demand-adapted product to the installation location range, determine whether the occupied area corresponding to the installation dimensions falls within the installation location range, and compare the coverage range of the demand-adapted product with the coverage range threshold. When the product type is consistent, the occupied area corresponding to the installation dimensions falls within the installation location range, and the coverage range is not lower than the coverage range threshold, retain the demand-adapted product as a space-adapted product. Combine the space-adapted products in the same functional space area according to the maximum number of products. Specifically, using the functional space area as the combination unit, group the space-adapted products in the same functional space area according to product type, and sort them from high to low according to the matching degree between the demand label and the implicit demand label. Select products sequentially from the sorted space-adapted products, and... During the selection process, it is verified whether the number of selected space-adaptive products exceeds the product quantity limit, whether the occupancy range corresponding to the installation size overlaps, and whether the coverage range covers the functional space area. When the number of selected space-adaptive products reaches the product quantity limit, or when the remaining space-adaptive products cannot improve the space coverage completeness and demand matching degree, the current combination search is stopped, forming a space-adaptive product combination. The coverage ratio of the functional space area of ​​each space-adaptive product in the space-adaptive product combination is calculated to obtain the space coverage completeness, and the matching ratio between the demand tags and implicit demand tags of each space-adaptive product in the space-adaptive product combination is calculated to obtain the demand matching degree. The space-adaptive product combinations are sorted from high to low according to the space coverage completeness and demand matching degree, and the space-adaptive product combinations that meet the preset screening conditions are determined as candidate home furnishing product combinations.

[0044] To further explain, the screening criteria are pre-set based on the coverage requirements of functional space areas and the matching requirements of implicit demand tags. These criteria include a spatial coverage integrity threshold, a demand matching degree threshold, a product quantity limit, and a ranking range. The spatial coverage integrity threshold is 0.85, the demand matching degree threshold is 0.80, and the ranking range is the top 5.

[0045] S4: Allocate budget to the candidate home furnishing product combination based on the budget information. The budget allocation is based on the security, comfort and entertainment function labels of the home furnishing products. The budget information is allocated according to the preset function budget weight. After obtaining the function budget amount, an initial selection list is generated.

[0046] S4.1: Budget information enters the budget allocation process. The total budget amount, single item price limit, and installation service budget are read according to the field names, and the amount format of the total budget amount, single item price limit, and installation service budget are validated. Then, according to the order of the home furnishing products in the candidate home furnishing product combination, the product price, installation fee, and security, comfort, and entertainment function identifiers of each home furnishing product are read one by one. The product price is used to determine whether each home furnishing product exceeds the single item price limit, the installation fee is used to calculate the installation service budget usage, and the security, comfort, and entertainment function identifiers are used to determine the corresponding function category of each home furnishing product when participating in the budget allocation.

[0047] S4.2: Based on the security, comfort, and entertainment function labels, classify the home products in the candidate home product portfolio according to their functions. Home products with security function labels are ranked first in the function budget allocation order, home products with comfort function labels are ranked second, and home products with entertainment function labels are ranked as a supplementary position. Set the budget weights for security, comfort, and entertainment functions according to the function budget allocation order, with security function budget weight being greater than comfort function budget weight, and comfort function budget weight being greater than entertainment function budget weight. Deduct the installation service budget from the total budget amount in the budget information to obtain the product's allocable budget. Then, allocate the product's allocable budget in segments according to the security, comfort, and entertainment function budget weights. For example, security function labels correspond to 50% of the product's allocable budget, comfort function labels correspond to 30%, and entertainment function labels correspond to 20%, thus obtaining the security function budget amount, comfort function budget amount, and entertainment function budget amount.

[0048] S4.3: Based on the security, comfort, and entertainment function labels corresponding to each home furnishing product, determine the corresponding function budget for each home furnishing product. Add the product price and installation cost of each home furnishing product to obtain the budget allocation amount for each home furnishing product. According to the order of the candidate home furnishing product combinations, accumulate the budget allocation amounts of the home furnishing products already retained under the same function budget allocation. If the budget allocation amount of a home furnishing product does not exceed the single item price limit and the accumulated budget allocation amount does not exceed the corresponding function budget allocation, retain the home furnishing product. If the budget allocation amount of a home furnishing product exceeds the single item price limit, filter out the home furnishing product. If the accumulated budget allocation amount exceeds the corresponding function budget allocation, filter out the home furnishing products that cause the excess of the corresponding function budget allocation amount, and use the retained home furnishing products as budget-suitable products.

[0049] S4.4: From the budget-matching products, form product combinations to be verified according to functional space areas and functional identifiers. Accumulate the product prices of each home furnishing product in the product combination to obtain the total product combination price. Accumulate the installation costs of each home furnishing product in the product combination to obtain the total installation cost. Compare the total product combination price with the total budget amount, compare the product price of each home furnishing product with the upper limit of the individual price, and compare the total installation cost with the installation service budget. When the total product combination price does not exceed the total budget amount, the product price of each home furnishing product does not exceed the upper limit of the individual price, and the total installation cost does not exceed the installation service budget, the product combination is determined to meet the budget information. Organize the product combinations that meet the budget information into a list according to the order of functional space area, home furnishing product, product price, installation location, and functional identifier, forming an initial selection list including functional space area, home furnishing product, product price, installation location, and functional identifier.

[0050] S5: Perform power matching calculations and network communication protocol matching calculations on the initial selection list.

[0051] S5.1: Following the order of functional space areas and home furnishing products in the initial configuration list, search for the same home furnishing products in the preset product library item by item, read the product power and supported network communication protocols corresponding to each home furnishing product, and establish a correspondence between the read product power and the functional space areas and home furnishing products in the initial configuration list, and establish a correspondence between the read supported network communication protocols and the home furnishing products in the initial configuration list; at the same time, read the rated power of the circuit and the network communication protocol from the user input information, and use the rated power of the circuit as the comparison benchmark for subsequent power matching calculations, and use the network communication protocol as the comparison benchmark for subsequent network communication protocol matching calculations.

[0052] S5.2: According to the functional space areas in the initial selection list, the product power of each home furnishing product in the same functional space area is accumulated to obtain the regional power value of each functional space area. Then, the regional power values ​​of each functional space area are accumulated to obtain the total power value of the initial selection list. The total power value of the list is compared with the rated power of the circuit in the user input information. The rated power of the circuit is subtracted from the total power value of the list to obtain the power difference. The total power value of the list is divided by the rated power of the circuit to obtain the power occupancy ratio. When the power difference is not less than zero and the power occupancy ratio is not greater than one, the power matching mark is marked as meeting the rated power constraint of the circuit. When the power difference is less than zero and the power occupancy ratio is greater than one, the power matching mark is marked as not meeting the rated power constraint of the circuit. The power difference, power occupancy ratio and power matching mark are combined into the power matching calculation result.

[0053] S5.3: Encode the network communication protocols supported by each home furnishing product according to the protocol name, protocol version, and network connection method to obtain the product protocol code corresponding to each home furnishing product; convert the network communication protocol in the user input information into the user protocol code according to the same encoding method; compare the product protocol code with the user protocol code item by item. When the protocol name, protocol version, and network connection method are all the same, mark the corresponding home furnishing product as protocol consistent; when at least one of the protocol name, protocol version, and network connection method is different, mark the corresponding home furnishing product as protocol inconsistent; count the number of home furnishing products marked as protocol consistent to obtain the number of protocol consistent products; count the number of home furnishing products marked as protocol inconsistent to obtain the number of protocol inconsistent products; summarize the number of protocol consistent products, the number of protocol inconsistent products, and the protocol consistent marks of each home furnishing product to generate the network communication protocol matching calculation result.

[0054] S6: Output the target selection list when the power matching calculation result meets the circuit rated power constraint and the network communication protocol matching calculation result meets the consistency constraint.

[0055] Read the power difference, power occupancy ratio, and power matching flag from the power matching calculation results, and read the number of products with consistent protocols, the number of products with inconsistent protocols, and the protocol consistency flag of each home furnishing product from the network communication protocol matching calculation results; when the power difference is not less than zero, the power occupancy ratio is not greater than one, and the power matching flag meets the circuit rated power constraint, and the number of products with inconsistent protocols is zero and the protocol consistency flag of each home furnishing product is consistent, confirm that the initial selection list meets the circuit rated power constraint and consistency constraint, and output the initial selection list as the target selection list.

[0056] S7: If one condition is not met, perform the corresponding replacement operation until both conditions are met simultaneously, and generate the target configuration list.

[0057] Using the initial configuration list as the current configuration list, the system reads the information for each home furnishing product in the current configuration list, including its power, supported network communication protocols, installation location, price, and functional identifiers. It also reads the power matching calculation results and the network communication protocol matching calculation results. When the power matching calculation results do not meet the circuit's rated power constraint, the system locates the over-power home furnishing products in the current configuration list according to their power from highest to lowest. It then searches the preset product library for home furnishing products with the same product type and requirement tags, meeting space constraints, meeting budget information, and having lower power than the over-power home furnishing products. These retrieved home furnishing products are identified as low-power replacement products, and the over-power home furnishing products are replaced with low-power replacement products. The current configuration list is then updated. When the network communication protocol matching calculation result does not meet the consistency constraint, the home furnishing product with the incompatible network communication protocol is located according to the protocol consistency mark. Then, the home furnishing products with the same product type and the same requirement tag, which meet the spatial constraint result, meet the budget information, and support the network communication protocol are searched in the preset product library. The searched home furnishing products are identified as protocol replacement products, and the home furnishing products with the incompatible network communication protocol are replaced with protocol replacement products. The current selection list is updated. The power matching calculation and network communication protocol matching calculation are re-executed on the updated current selection list until the power matching calculation result meets the circuit rated power constraint and the network communication protocol matching calculation result meets the consistency constraint. The updated current selection list is then used to generate the target selection list.

[0058] When no low-power replacement or protocol replacement product that meets the spatial constraints, implicit requirement tags, budget information, circuit rated power constraints, and network communication protocol consistency constraints is found in the preset product library, the current replacement process is stopped. The home products that do not meet the constraints, the types of constraints that do not meet the constraints, and the reasons for the constraints that do not meet the constraints are read, and an exception prompt message is generated. The types of constraints that do not meet the constraints include spatial constraints not met, budget information not met, circuit rated power constraints not met, and network communication protocol consistency constraints not met. The reasons for the constraints that do not meet the constraints include inconsistent product types, installation location range not met, coverage threshold not met, product price exceeding budget information, product power still exceeding circuit rated power constraints, and inconsistent supported network communication protocols. The current selection list is marked as a selection list to be adjusted, and adjustment suggestions are output according to the types of constraints that do not meet the constraints. The adjustment suggestions include reducing the number of home products in the same functional space area, adjusting the installation location range, reducing power consumption, changing the network communication protocol, and adjusting the budget information.

[0059] S8: Receive user operations to add, delete, and replace items in the target selection list, form an updated selection list, and set user operation retention marks for newly added home furnishing products and home furnishing products retained after replacement. Use the updated selection list as the current solution, keep the home furnishing products with user operation retention marks unchanged, use an adaptive large neighborhood search algorithm to remove some home furnishing products according to the degree of constraint conflict, and repair the current solution according to the regret insertion value of the replacement candidate products. Update the weights of the destruction operator and the repair operator according to the repair results, and iteratively generate an updated target selection list.

[0060] S8.1: Following the order of the home furnishing products in the updated optional list, read the product type, price, installation cost, power, installation location, dimensions, coverage, supported network communication protocols, and function identifiers for each home furnishing product, and establish the correspondence between these items using the home furnishing product identifiers; record the updated optional list with the established correspondence as the current solution, which retains the arrangement and attribute content of the home furnishing products; check whether each home furnishing product has a user operation reserved mark, and write the home furnishing products with the user operation reserved mark into the prohibited removal state, so that when the subsequent destruction operator selects home furnishing products, it excludes home furnishing products in the prohibited removal state and only determines the removal object from home furnishing products that have never been in the prohibited removal state.

[0061] S8.2: Recalculate the total budget allocation and the budget allocation for each function based on the product prices, installation costs, and function identifiers in the current solution, and calculate the budget overrun amount exceeding the budget information; temporarily deduct the product prices and installation costs corresponding to the home furnishing products item by item. If the original budget overrun amount is greater than zero, calculate the ratio of the reduction in the budget overrun amount before and after deduction to the original budget overrun amount, and determine the ratio as the budget overrun contribution. If the original budget overrun amount is equal to zero, record the budget overrun contribution as zero; re-accumulate the product power according to the functional space area, calculate the regional power overrun value and the total power overrun value of the list, temporarily deduct the product power corresponding to the home furnishing products item by item, calculate the regional power overrun reduction ratio if the original regional power overrun value is greater than zero, and calculate the total power overrun reduction ratio of the list if the original total power overrun value of the list is greater than zero. When the original power over-limit value is equal to zero, the corresponding reduction ratio is recorded as zero. The sum of the regional power over-limit reduction ratio and the total power over-limit reduction ratio in the list is determined as the power over-limit contribution. The number of home furnishing products is counted according to functional space area and product type to form the product quantity. The supported network communication protocol is compared with the network communication protocol in the user input information. If there is an inconsistency in the protocol name, protocol version, or network connection method, the protocol inconsistency mark is recorded as one, otherwise it is recorded as zero. The product type, product quantity, installation location, installation size, and coverage area are compared with the spatial constraint results. If any content is not met, the spatial mismatch mark is recorded as one, otherwise it is recorded as zero. The budget over-limit contribution, power over-limit contribution, protocol inconsistency mark, and spatial mismatch mark are added together to obtain the constraint conflict degree corresponding to each home furnishing product.

[0062] S8.3: Remove furniture products that are prohibited from being removed from the current solution, obtaining movable furniture products. If any movable furniture product has a constraint conflict level greater than zero, identify it as a candidate for destruction. If no movable furniture product has a constraint conflict level greater than zero, identify all movable furniture products as candidate for destruction, and record the number of movable furniture products as [missing information]. When N equals zero, the removal of home furnishing products is not performed, and the current solution is directly used as a candidate solution for evaluation; when N is greater than zero, the number of products removed in a single operation is determined according to the following formula: ; in, Indicates the quantity to be removed in a single operation. , The pre-set damage ratio, for example, 20%.

[0063] The selection rule of "randomly arranging the candidate furniture products to be damaged and selecting the top q" is defined as the random damage operator. The selection rule of "arranging the candidate furniture products to be damaged from high to low according to the degree of constraint conflict and selecting the top q" is defined as the conflict damage operator. The selection rule of "selecting the furniture product with the highest degree of constraint conflict as the benchmark furniture product, arranging the remaining candidate furniture products to be damaged from high to low according to the number of consistent items in product type, functional space area, and functional label, and selecting the benchmark furniture product and the top q" is defined as the conflict damage operator. The selection rule for "home furnishing products" is determined as the similarity destruction operator; the initial weights of the random destruction operator, conflict destruction operator, and similar destruction operator are all set to one. The selection probability of the destruction operator is determined according to the proportion of the current weight of each destruction operator to the sum of the current weights of all destruction operators, and a destruction operator is selected according to the selection probability of the destruction operator; the selected destruction operator is executed, and the selected home furnishing product and its corresponding product price, installation cost, product power, installation location, coverage, supported network communication protocols, and function identifiers are removed from the current solution. At the same time, the original functional space area and installation location of the removed home furnishing product are recorded to form the solution to be repaired.

[0064] S8.4: Following the order in which the home furnishing products are removed, read the product type, functional space area, and functional identifier item by item, and combine the product type, functional space area, and functional identifier into search conditions. Search the preset product library for home furnishing products with the same product type, applicable functional space area, and functional identifier. Read the demand tags corresponding to the found home furnishing products, compare the demand tags with the implicit demand tags in terms of tag name and functional space area, and retain home furnishing products with the same tag name and associated functional space area as replacement candidate products for the corresponding removed home furnishing products. According to the product identifier of the replacement candidate products, read the product price, installation cost, product power, installation size, coverage area, and supported network communication protocols from the preset product library, and establish a correspondence with the original functional space area and installation location of the removed home furnishing products.

[0065] If no replacement candidate product is found, cancel the current home furnishing product removal operation, restore the current solution, and reselect the destruction operator; if one replacement candidate product is found, mark the corresponding removed home furnishing product as a single candidate to be repaired; if two replacement candidate products are found, mark the corresponding removed home furnishing product as a second-order regret to be repaired; if three or more replacement candidate products are found, mark the corresponding removed home furnishing product as a second-order and third-order regret to be repaired.

[0066] S8.5: Based on the correspondence between the removed furniture products and the replacement candidate products, select one replacement candidate product in sequence, keeping the other furniture products in the solution unchanged. Temporarily write the replacement candidate product into the original functional space area and installation location of the removed furniture product, and associate it with the product price, installation cost, product power, installation size, coverage area, supported network communication protocols, function identifiers, and requirement tags. Before and after the temporary writing, calculate the budget occupancy rate, power occupancy rate, spatial coverage integrity, requirement matching degree, and network communication protocol inconsistency ratio. The budget occupancy rate is the larger of the total budget occupancy ratio in the list and the budget occupancy ratio of the corresponding function identifier; the power occupancy rate is the larger of the power occupancy ratio of the corresponding functional space area and the total power occupancy ratio in the list; spatial coverage integrity is determined by the ratio of the effective coverage area to the required coverage area in the spatial constraint results; if the calculated result exceeds one, record it as one; the requirement matching degree is determined by the ratio of the number of consistent tags in the requirement tags of the replacement candidate product to the total number of implicit requirement tags; the network communication protocol inconsistency ratio is determined by the proportion of inconsistent items in the protocol name, protocol version, and network connection method; let the spatial coverage integrity be... Spatial coverage missing degree is Let the degree of matching between requirements be . The degree of mismatch between demand and supply is The spatial coverage completeness is converted into spatial coverage missingness, and the demand matching degree is converted into demand mismatch degree. The expression is: ; ; The conversion was completed so that the proportions of budget occupancy, power occupancy, spatial coverage gap, demand mismatch, and network communication protocol inconsistency were all evaluated using the smaller values, indicating less impact from insertion.

[0067] After unifying the evaluation direction, the five values ​​before temporary writing are arranged in the order of budget occupancy rate, power occupancy rate, spatial coverage gap, demand mismatch, and network communication protocol inconsistency ratio to form the pre-insertion evaluation sequence, and the five values ​​after temporary writing are arranged to form the post-insertion evaluation sequence. The difference between the post-insertion value and the pre-insertion value for the same evaluation item is calculated, and the five differences are weighted and summarized using preset evaluation weights to obtain the insertion increment value corresponding to the replacement candidate product. The expression is: ; in, Indicates replacement candidate product Insert the removed home furnishing products The insertion increment value generated at the corresponding position, It indicates that no items were removed from the home furnishing products. The corresponding position is temporarily written when replacing the candidate product. Item evaluation value, Indicates the first The replacement candidate product is temporarily written to the first... The first [item] obtained after the home furnishing product was removed from its corresponding functional space area and installation location. Item evaluation value, Indicates the first The preset evaluation weights corresponding to the evaluation values ​​of each item.

[0068] The smaller the insertion increment value, the lower the degree of increase in budget usage, power consumption, spatial coverage loss, demand mismatch, and network communication protocol inconsistency after inserting the replacement candidate product. After completing the insertion increment value calculation, the temporary write to the replacement candidate product is undone and the solution to be repaired is restored. Then, the next replacement candidate product is selected to perform temporary write and insertion increment value calculation, until the insertion increment value corresponding to each replacement candidate product is obtained.

[0069] To further explain, the preset evaluation weights are set based on the impact of budget information, circuit rated power constraints, network communication protocol consistency constraints, spatial constraints, and implicit requirement tags on the executability of the selection results. For example, the preset evaluation weight for budget occupancy rate is set to 0.25, the preset evaluation weight for power occupancy rate is set to 0.25, the preset evaluation weight for network communication protocol inconsistency ratio is set to 0.2, the preset evaluation weight for spatial coverage gap is set to 0.2, and the preset evaluation weight for requirement mismatch is set to 0.1, with the sum of the five preset evaluation weights being 1. Budget occupancy rate and power occupancy rate directly determine whether the updated selection list can meet the budget information and circuit rated power constraints, therefore they are set with higher weights. The network communication protocol inconsistency ratio affects the communication connection and coordinated operation between home products, and the spatial coverage gap affects the installation adaptation and functional coverage of home products within functional space areas, therefore they are set with the second highest weights. Requirement mismatch is used to distinguish the degree of matching between replacement candidate products that meet basic constraints and implicit requirement tags, and does not directly determine whether replacement candidate products can be written into the solution to be repaired, therefore it is set with a lower weight.

[0070] S8.6: For the same removed home furnishing product, read the insertion increment value of the corresponding replacement candidate product and arrange them in ascending order of insertion increment value; when the insertion increment values ​​are the same, determine the arrangement order according to the replacement candidate product identifier. For a single candidate to be repaired, only the first insertion increment value is recorded, and the regret insertion value is not calculated; for a second-order regret to be repaired item, the insertion increment values ​​corresponding to the first and second ranked replacement candidate products are determined as the first and second insertion increment values, and the second insertion increment value is subtracted from the first insertion increment value to obtain the second-order regret insertion value; for second-order and third-order regret to be repaired items, the insertion increment values ​​corresponding to the first, second, and third ranked replacement candidate products are determined as the first, second, and third insertion increment values, and the second-order regret insertion value and the third-order regret insertion value are calculated respectively. The second-order regret insertion value represents the increased evaluation loss when the second ranked replacement candidate product was selected instead of the first ranked replacement candidate product, and the third-order regret insertion value represents the increased evaluation loss when the third ranked replacement candidate product was selected instead of the first ranked replacement candidate product.

[0071] S8.7: Read the second-order regret to be repaired items and the corresponding second-order regret insertion values ​​for the second-order and third-order regret to be repaired items. Sort the removed home furnishing products according to the second-order regret insertion values ​​in descending order. If the second-order regret insertion values ​​are the same, determine the sorting order according to the first insertion increment value in ascending order. Select the first removed home furnishing product according to the sorting order, and select the replacement candidate product with the smallest insertion increment value from the corresponding replacement candidate products. Combine the sorting rules for the removed home furnishing products and the replacement candidate product selection rules into a second-order regret insertion repair operator. Read the third-order regret insertion values ​​corresponding to the second-order and third-order regret to be repaired items. Sort the removed home furnishing products according to the third-order regret insertion values ​​in descending order. If the third-order regret insertion values ​​are the same, determine the sorting order according to the first insertion increment value in ascending order. Select the first removed home furnishing product according to the sorting order, and select the replacement candidate product with the smallest insertion increment value from the corresponding replacement candidate products. Combine the sorting rules for the removed home furnishing products and the replacement candidate product selection rules into a third-order regret insertion repair operator. When the second-order regret insertion repair operator and the third-order regret insertion repair operator are first formed, the corresponding initial weights are set to one respectively.

[0072] S8.8: When a single candidate item to be repaired exists, first write the unique replacement candidate product into the corresponding empty position, and recalculate the insertion increment value and regret insertion value corresponding to the remaining removed furniture products based on the repair solution after writing. Read the currently executable second-order regret insertion repair operator and third-order regret insertion repair operator; when both repair operators can be executed, determine the ratio of the current weight of each repair operator to the sum of the current weights of the two repair operators as the repair operator selection probability, and select the repair operator according to the repair operator selection probability; when only one repair operator can be executed, directly select the executable repair operator. When selecting the second-order regret insertion repair operator, the unrepaired removed furniture products are arranged in descending order of the second-order regret insertion value. When selecting the third-order regret insertion repair operator, the unrepaired removed furniture products are arranged in descending order of the third-order regret insertion value. If the regret insertion values ​​are the same, the arrangement order is determined by the first insertion increment value from small to large. The removed furniture product at the top of the list is selected, and the replacement candidate product corresponding to the first insertion increment value is selected from the corresponding replacement candidate products and written into the solution to be repaired. This process continues until all the empty positions corresponding to the removed furniture products are repaired. The solution to be repaired after all the products have been written is determined as the candidate solution.

[0073] S8.9: Before the first iteration, the current solution is compared with the budget information, circuit rated power constraint, network communication protocol consistency constraint and space constraint results according to the constraint comparison method of the candidate solution; when the current solution satisfies all four constraints at the same time, the comprehensive evaluation value of the current solution is calculated according to the calculation method of the comprehensive evaluation value of the candidate solution, and the current solution is recorded as the initial optimal solution; when the current solution has any unsatisfied content, only the constraint state of the current solution is recorded, and the current solution is not written into the optimal solution record.

[0074] The process involves: retrieving product prices and installation costs from candidate solutions for home furnishing products; recalculating the total budget allocation and the budget allocation for each function according to the budget allocation rules; comparing this with the total budget and function budget in the budget information; accumulating product power across functional space areas and comparing the power accumulation results for each functional space area with the total power in the list against the circuit rated power constraint; comparing the protocol name, protocol version, and network connection method supported by the home furnishing products with the network communication protocol consistency constraint; and comparing the installation location, installation size, and coverage area of ​​the home furnishing products with the installable area, available installation size, and required coverage area in the space constraint results. If any candidate solution fails to meet these constraints, it is marked as unacceptable and the comprehensive evaluation of that candidate solution is stopped. If a candidate solution simultaneously meets the budget information, circuit rated power constraint, network communication protocol consistency constraint, and space constraint results, it is marked as acceptable and the comprehensive evaluation value continues to be calculated. Candidate solutions that simultaneously meet the budget information, circuit rated power constraint, network communication protocol consistency constraint, and space constraint results are retained. The number of candidate solutions whose demand labels match the implicit demand labels is counted, and the ratio of the number of matching labels to the total number of implicit demand labels is determined as the demand matching degree. The effective coverage area of ​​home furnishing products is counted according to functional space areas, and the ratio of the effective coverage area to the required coverage area is determined as the spatial coverage integrity. Each instance of spatial coverage integrity exceeding one level is recorded. A correspondence is established between the candidate solutions and the updated selection list according to functional space areas and functional identifiers. The number of home furnishing products whose product identifiers have changed is counted, and the ratio of the number of replaced home furnishing products to the total number of home furnishing products in the updated selection list is determined as the replacement home furnishing product ratio. The relative differences in product price, installation cost, product power, installation size, and coverage area before and after replacement are calculated. The consistency of supported network communication protocols and functional identifiers is recorded, and the average of these differences is determined as the product attribute difference before and after replacement. After calculating the four evaluation criteria, the demand matching degree is converted into the demand mismatch degree, and the spatial coverage integrity degree is converted into the spatial coverage gap degree. This ensures that all four evaluation criteria adopt the evaluation direction where smaller values ​​indicate smaller overall deviation of the candidate solution. Following the order of demand mismatch degree, spatial coverage gap degree, proportion of replaced furniture products, and differences in product attributes before and after replacement, the four values ​​are arranged into a candidate solution evaluation sequence. Then, the four evaluation values ​​are weighted and summed using preset comprehensive evaluation weights to obtain the comprehensive evaluation value of the candidate solution, expressed as: ; in, This represents the comprehensive evaluation value of the candidate solution. Indicates the evaluation item number, Indicates the degree of mismatch between requirements. Indicates the degree of spatial coverage loss. Indicates the percentage of home furnishing products replaced. This indicates the difference in product attributes before and after the replacement. Indicates the first The preset comprehensive evaluation weights corresponding to the evaluation values ​​are 1, and the sum of the four preset comprehensive evaluation weights is 1.

[0075] To further explain, the value range of demand matching degree and spatial coverage integrity is uniformly limited to zero to one, with one representing complete matching or complete coverage; one is subtracted from demand matching degree to obtain demand mismatch degree, and one is subtracted from spatial coverage integrity degree to obtain spatial coverage gap degree. Both demand mismatch degree and spatial coverage gap degree adopt the evaluation direction that the smaller the value, the better the candidate solution evaluation result.

[0076] The preset comprehensive evaluation weights are set based on the degree to which candidate solutions retain implicit demand labels, spatial constraints, and the updated selection list. Specifically, the preset comprehensive evaluation weight corresponding to the degree of demand mismatch is set to 0.35, the preset comprehensive evaluation weight corresponding to the degree of spatial coverage deficiency is set to 0.3, the preset comprehensive evaluation weight corresponding to the proportion of replaced furniture products is set to 0.2, and the preset comprehensive evaluation weight corresponding to the difference in product attributes before and after replacement is set to 0.15. The degree of demand mismatch and the degree of spatial coverage deficiency directly affect the adaptability of the selection results to user needs and functional space areas, therefore they are given higher weights; the proportion of replaced furniture products is used to limit the scope of list modifications, therefore it is given a medium weight; the difference in product attributes before and after replacement is used to compare the degree of attribute change before and after replacement, therefore it is given a lower weight.

[0077] S8.10: When a candidate solution is marked as unacceptable, the candidate solution acceptance probability calculation is not performed, and the current solution is retained; when a candidate solution is marked as acceptable and the current solution has any unsatisfied condition, the candidate solution directly replaces the current solution; when both the candidate solution and the current solution satisfy the four restrictions, they are compared according to their comprehensive evaluation values. **When the comprehensive evaluation value of the candidate solution is less than that of the current solution, the candidate solution directly replaces the current solution; when the comprehensive evaluation value of the candidate solution is greater than or equal to that of the current solution, the comprehensive evaluation value of the candidate solution is subtracted from that of the current solution to obtain the difference in comprehensive evaluation values.

[0078] Read the preset initial iteration temperature, cooling coefficient, and current iteration number to determine the current iteration temperature: ; in, This indicates the preset initial iteration temperature. Indicates the cooling coefficient. This indicates the current iteration number, starting from zero. Indicates the first The temperature corresponding to the current iteration in the next iteration.

[0079] The acceptance probability of a candidate solution is calculated using the difference in comprehensive evaluation values ​​and the current iteration temperature, expressed as: ; in, This represents the probability of accepting a candidate solution. This represents the difference in comprehensive evaluation value between the candidate solution and the current solution.

[0080] Generate a random number with a value between 0 and 1. When the random number is less than or equal to the acceptance probability of the candidate solution, replace the current solution with the candidate solution. When the random number is greater than the acceptance probability of the candidate solution, retain the current solution.

[0081] To further explain, the preset initial iteration temperature is set according to the range of the comprehensive evaluation value of the candidate solution and the need to retain poor candidate solutions in the initial stage. For example, it is set to 0.1 to give the candidate solution a certain probability of acceptance in the initial stage. The cooling coefficient is set according to the need to gradually reduce the probability of acceptance of candidate solutions and maintain stable convergence of the search during the iteration process. For example, it is set to 0.95 to make the current iteration temperature decrease slowly after each iteration, so as to avoid the candidate solution being fixed too early due to the temperature dropping too quickly.

[0082] S8.11: Compare the comprehensive evaluation value of the candidate solution with the minimum comprehensive evaluation value recorded since the start of the iteration and the comprehensive evaluation value of the current solution before replacement, and limit the evaluation score to a range of zero to one; when the comprehensive evaluation value of the candidate solution is less than the recorded minimum comprehensive evaluation value, the candidate solution is determined as the new optimal solution, and 1.0 evaluation score is assigned to the selected destruction operator and repair operator respectively; when the candidate solution does not form a new optimal solution but the comprehensive evaluation value is less than the comprehensive evaluation value of the current solution before replacement, 0.6 evaluation score is assigned to each; when the candidate solution is not better than the current solution before replacement but is accepted according to the candidate solution acceptance probability, 0.2 evaluation score is assigned to each; when the candidate solution is marked as an unacceptable candidate solution or the candidate solution is not accepted, 0 evaluation score is assigned to each.

[0083] After completing the evaluation score allocation, the pre-update weights of the selected damage operator and repair operator are read. The pre-update weights of the damage operator are weighted with their corresponding evaluation scores to obtain the updated weights of the damage operator. Similarly, the pre-update weights of the repair operator are weighted with their corresponding evaluation scores to obtain the updated weights of the repair operator. The expressions are as follows: ; in, This indicates the sequence number of the destruction or repair operator selected in this instance. Indicates the first The pre-update weights of a destructive or repairing operator, Indicates the first The weights updated by a destruction operator or a repair operator, Indicates the first The evaluation score obtained by each destruction operator or repair operator in this instance. This represents the weight update coefficient. Unselected destruction and repair operators retain their original weights.

[0084] To further explain, the weight update coefficient is set to 0.2 based on the balance between the search performance of the destruction and repair operators in previous iterations and the current evaluation score. This ensures that the weight before the update accounts for 80% of the updated result, while the current evaluation score accounts for 20%. This avoids large fluctuations in weight caused by a single search result, and also allows destruction and repair operators that have consistently obtained high evaluation scores to gradually increase their probability of being selected in subsequent iterations.

[0085] S8.12: At the start of each iteration, read the product price, installation cost, product power, product type, product quantity, installation location, installation size, coverage area, supported network communication protocols, and function identifiers from the current solution. Re-execute budget allocation, power matching calculation, network communication protocol matching calculation, and spatial constraint comparison, and redetermine the constraint conflict degree corresponding to each home furnishing product according to the calculation method in S8.2. Summarize the updated weights of each destruction operator and each repair operator. Determine the probability of selecting a destruction operator as the proportion of the updated weight of a single destruction operator to the sum of the updated weights of all destruction operators, and determine the probability of selecting a repair operator as the proportion of the updated weight of a single repair operator to the sum of the updated weights of all repair operators. Divide the probability intervals from 0 to 1 according to the probability intervals of destruction and repair operators, generate random numbers from 0 to 1, and select the destruction and repair operators corresponding to the probability intervals where the random numbers fall.

[0086] Using the selected destruction operator, furniture products not in the prohibited removal state are removed from the current solution. Using the selected repair operator, replacement candidate products are written into the solution to be repaired. Then, candidate solution evaluation, acceptance judgment, and weight update are performed on the formed candidate solutions, and the iteration count is increased by 1. When a candidate solution forms a new optimal solution, the optimal solution record is updated, and the number of consecutive failures to form a new optimal solution is reset to zero. When a candidate solution does not form a new optimal solution, the number of consecutive failures to form a new optimal solution is increased by 1. The process of recalculating constraint conflict, selecting destruction operators, removing furniture products, selecting repair operators, inserting replacement candidate products, evaluating candidate solutions, accepting judgments, and updating weights is repeated until the number of iterations reaches a preset number (e.g., one hundred), or the number of consecutive failures to form a new optimal solution reaches a preset stopping number (e.g., twenty). The optimal solution record is read, and the budget information, circuit rated power constraints, network communication protocol consistency constraints, and space constraint results are checked. The candidate solution that simultaneously satisfies all four constraints and has the smallest comprehensive evaluation value is determined as the updated target selection list.

[0087] This embodiment also provides a smart home quick selection system, including: The data acquisition module obtains user input information, including apartment dimensions, rated circuit power, network communication protocol, and budget information. The encoding module encodes user input information based on a machine learning model and performs matching processing in conjunction with historical selection data to generate implicit demand labels. The filtering module performs spatial modeling of the apartment size based on spatial analysis algorithms to obtain spatial constraint results, and filters candidate home furnishing product combinations from a preset product library based on the spatial constraint results and implicit demand tags. The optional module allocates budget to candidate home furnishing product combinations based on budget information. The budget allocation is based on the security, comfort and entertainment function labels of the home furnishing products and allocates the budget information according to the preset function budget weights. After obtaining the function budget amount, an initial optional list is generated. The calculation module performs power matching calculations and network communication protocol matching calculations on the initial selection list; The matching module outputs a target selection list when the power matching calculation result meets the circuit rated power constraint and the network communication protocol matching calculation result meets the consistency constraint. The replacement module performs the corresponding replacement operation when one condition is not met, until both conditions are met simultaneously, and then generates the target configuration list. The update module receives user operations to add, delete, and replace items in the target selection list, generates an updated selection list, and sets user operation retention flags for newly added and replaced furniture products. The updated selection list is used as the current solution, keeping the furniture products with user operation retention flags unchanged. An adaptive large neighborhood search algorithm is used to remove some furniture products according to the degree of constraint conflict, and the current solution is repaired according to the regret insertion value of the replacement candidate products. The weights of the destruction operator and the repair operator are updated based on the repair results, and the updated target selection list is generated iteratively.

[0088] In summary, this invention, by comprehensively considering factors such as apartment size, rated circuit power, network communication protocols, budget information, and historical selection data, makes home furnishing product selection no longer limited to explicit preference matching. It helps to balance potential functional needs, actual installation conditions, and budget allocation, reducing issues such as insufficient budget for core functions, exceeding product power limits, inconsistent network communication protocols, and spatial incompatibility, thus improving the rationality and feasibility of the initial selection list. After users add, delete, or replace items on the target selection list, by retaining the home furnishing products explicitly added or retained by the user and making local adjustments around home furnishing products with high degree of constraint conflict, it can reduce the possibility of irrelevant home furnishing products being replaced, narrow the scope of changes to the selection list, reduce the overwriting of user operation results, and simultaneously maintain good coordination between budget, power, network communication protocols, and spatial limitations when updating the target selection list.

[0089] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for rapid selection and configuration of smart home systems, characterized by: include, Obtain user input information, including apartment dimensions, rated circuit power, network communication protocol, and budget information; The implicit demand labels are generated by encoding features of user input information based on machine learning models and combining them with historical matching data. The spatial modeling of the apartment size is performed based on the spatial analysis algorithm to obtain spatial constraint results. Based on the spatial constraint results and implicit demand tags, candidate home furnishing product combinations are selected from the preset product library. Based on the budget information, the budget is allocated to the candidate home furnishing product combination. The budget allocation is based on the security, comfort and entertainment function labels of the home furnishing products. The budget information is allocated according to the preset function budget weights. After obtaining the function budget amount, an initial selection list is generated. Perform power matching calculations and network communication protocol matching calculations on the initial configuration list; When the power matching calculation result meets the circuit rated power constraint and the network communication protocol matching calculation result meets the consistency constraint, the target selection list is output. If one condition is not met, the corresponding replacement operation is performed until both conditions are met simultaneously, and a target configuration list is generated. The system receives user operations to add, delete, and replace items in the target selection list, generates an updated selection list, and sets user operation retention flags for newly added and replaced furniture products. The updated selection list is used as the current solution, keeping the furniture products with user operation retention flags unchanged. An adaptive large neighborhood search algorithm is used to remove some furniture products according to the degree of constraint conflict, and the current solution is repaired according to the regret insertion value of the replacement candidate products. The weights of the destruction operator and the repair operator are updated based on the repair results, and the updated target selection list is generated iteratively.

2. The method for rapid selection and configuration of smart home systems as described in claim 1, characterized in that: The machine learning model includes an input embedding layer, a feature encoding layer, an attention fusion layer, a history memory matching layer, and an output layer.

3. The method for rapid selection and configuration of smart home systems as described in claim 2, characterized in that: The generation of implicit requirement tags specifically involves: User input information is fed into the input embedding layer, and the user input information is vectorized to obtain the initial feature vector; A multidimensional feature representation is obtained by performing a nonlinear feature transformation on the initial feature vector based on the feature coding layer. The multidimensional feature representation is input into the attention fusion layer, and weights are assigned and fused for different types of input features to obtain the fused feature representation. In the historical memory matching layer, the fused feature representation is compared with the historical matching data to perform vector similarity retrieval, thereby obtaining historical association features; By performing semantic summarization and labeling of historical correlation features through the output layer, implicit demand labels are formed.

4. The method for rapid selection and configuration of smart home systems as described in claim 1, characterized in that: The obtained spatial constraint result is specifically as follows: Convert the apartment dimensions into a uniform-scale apartment coordinate matrix, and write the room boundaries, door and window positions, and fixed structure positions from the apartment dimensions into the apartment coordinate matrix; The coordinate matrix of the apartment type is divided into grids according to the spatial analysis algorithm, and the installable markers, occlusion markers, passage and avoidance markers and boundary distances corresponding to each grid are calculated. Based on the installable markings, obstruction markings, passage and avoidance markings, and boundary distances, calculate the available area, effective installation length, coverage radius, and product capacity within the functional space area; The available area, effective installation length, coverage radius, and product capacity are summarized into regional adaptation parameters. Based on the regional adaptation parameters, the system is matched with the preset product installation rules to determine the types of products that can be installed in each functional space area, the upper limit of the number of products, the range of installation locations, and the coverage threshold. These parameters are then summarized into spatial constraint results.

5. The method for rapid selection and configuration of smart home systems as described in claim 4, characterized in that: The process of selecting candidate home furnishing product combinations from a pre-set product library based on spatial constraints and implicit demand tags specifically involves: Read the product type, product quantity limit, installation location range, and coverage threshold from the spatial constraint results, and read the product type, installation size, coverage, and requirement tags of home furnishing products from the preset product library; Match implicit demand tags with demand tags to obtain demand-suitable products; The demand-adapted products are compared with product type, installation location range and coverage threshold. Demand-adapted products that do not meet the spatial constraints are filtered out to obtain spatially adapted products. Combine space-adaptive products according to the maximum number of products, and calculate the completeness of space coverage and the degree of matching with needs after the combination. The combined space-adaptive products are sorted according to the completeness of space coverage and the degree of matching with needs. The space-adaptive product combinations that meet the preset screening conditions are determined as candidate home furnishing product combinations.

6. The method for rapid selection and configuration of smart home systems as described in claim 5, characterized in that: The generation of the initial selection list is specifically as follows: Read the total budget amount, single item price limit, and installation service budget from the budget information, and read the product price, installation cost, and security, comfort, and entertainment function labels of each home product in the candidate home product combination; Establish a functional budget allocation order based on the functional identification of security, comfort and entertainment functions, and set the budget weights for security, comfort and entertainment functions according to the functional budget allocation order. Allocate the budget information according to the order of priority of security function identification, second of comfort function identification, and supplement of entertainment function identification, as well as the budget weights of security, comfort and entertainment functions, to obtain the functional budget amount. The product price and installation cost of each home furnishing product are compared with the corresponding functional budget. Home furnishing products that exceed the corresponding functional budget are eliminated to obtain budget-suitable products. The budget-compatible products are combined and verified according to the total budget amount, the upper limit of the price of each item and the installation service budget to determine the product combination that meets the budget information. The product mix that meets the budget information is processed into a list according to functional space area, home furnishing products, product price, installation location and function label, generating an initial selection list.

7. The method for rapid selection and configuration of smart home systems as described in claim 6, characterized in that: The power matching calculation and network communication protocol matching calculation are performed as follows: Based on the home furnishing products in the initial selection list, the power and supported network communication protocols of each home furnishing product are read from the preset product library, and the rated power of the circuit and the network communication protocol are read from the user input information. The power consumption of each home furnishing product is added up according to the functional space area to obtain the area power value, and the power values ​​of each area are added up to obtain the total power value of the list; The total power value in the list is compared with the rated power of the circuit, the power difference and the power occupancy ratio are calculated, and the power matching calculation results are generated based on the power difference and the power occupancy ratio. The network communication protocols supported by each home furnishing product are compared with the network communication protocols in the user input information to obtain a protocol consistency mark for each home furnishing product. Based on the protocol consistency markers, the number of products with consistent protocols and the number of products with inconsistent protocols are counted, and the network communication protocol matching calculation results are generated.

8. The method for rapid selection and configuration of smart home systems as described in claim 7, characterized in that: The specific steps for performing the corresponding replacement operation are as follows: Using the initial selection list as the current selection list, when the power matching calculation result does not meet the circuit's rated power constraint, replace the power-over-limit home products with low-power replacement products that meet the space constraint result, implicit requirement label, and budget information, and update the current selection list; When the network communication protocol matching calculation result does not meet the consistency constraint, replace the home products with incompatible network communication protocols with protocol replacement products that support the network communication protocol, and update the current selection list.

9. The method for rapid selection and configuration of smart home systems as described in claim 1, characterized in that: The process of removing some home furnishing products based on the degree of constraint conflict using an adaptive large neighborhood search algorithm is as follows: Read the updated optional list of home products, product prices, installation costs, product power, installation location, coverage, supported network communication protocols and function identifiers, use the updated optional list as the current solution, and set the home products with user operation retention marks to a prohibited removal state; For the current solution, re-execute budget allocation, power matching calculation, network communication protocol matching calculation and spatial constraint comparison, and calculate the degree of constraint conflict based on the budget over-limit contribution, power over-limit contribution, protocol inconsistency mark and spatial mismatch mark corresponding to each home furnishing product; Establish a set of destruction operators. The destruction operators select home furnishing products that are not in the prohibited removal state according to random order, order of constraint conflict degree from high to low, and similarity of product type, functional space area and functional identifier. Remove the selected home furnishing products from the current solution to form a solution to be repaired. Based on the product type, functional space area, and functional identifier of the removed home products, the system retrieves replacement candidate products that meet the implicit demand tags from the preset product library, and reads the product price, installation cost, product power, installation size, coverage area, and supported network communication protocols of the replacement candidate products. Each replacement candidate product is inserted into the functional space area corresponding to the removed home furnishing product in the solution to be repaired. The budget occupancy, power occupancy, space coverage integrity, demand matching degree and network communication protocol matching results are recalculated. The insertion increment value is determined based on the difference between the calculation results before and after insertion. Arrange the replacement candidate products corresponding to the same removed home furnishing product in ascending order of the insertion increment value. The difference between the second insertion increment value and the first insertion increment value is used as the second-order regret insertion value, and the difference between the third insertion increment value and the first insertion increment value is used as the third-order regret insertion value. The removed home furnishing products are sorted according to the second-order regret insertion value and the third-order regret insertion value, respectively. The corresponding sorting results are combined with the replacement candidate product selection rule with the smallest insertion increment value to form the second-order regret insertion repair operator and the third-order regret insertion repair operator. Based on the current weight of the repair operator, the corresponding regret insertion value is selected for repair. Priority is given to the removed furniture products with larger regret insertion values. The replacement candidate products corresponding to the first insertion increment value are written into the solution to be repaired. The replacement candidate product insertion is repeated until the repair of the vacant position corresponding to the removed furniture products is completed, and a candidate solution is formed. The budget allocation, power matching calculation, network communication protocol matching calculation and spatial constraint comparison are re-executed for the candidate solutions. The candidate solutions that meet the budget information, circuit rated power constraints, network communication protocol consistency constraints and spatial constraint results are retained. The comprehensive evaluation value of the candidate solutions is calculated based on the demand matching degree, spatial coverage integrity, number of replaced home furnishing products and differences in product attributes before and after replacement. When the comprehensive evaluation value of a candidate solution is better than that of the current solution, the candidate solution replaces the current solution. When the comprehensive evaluation value of a candidate solution is not better than that of the current solution, the acceptance probability of the candidate solution is determined based on the difference in comprehensive evaluation values ​​between the candidate solution and the current solution and the current iteration temperature. The acceptance probability of the candidate solution is then used to determine whether to replace the current solution with the candidate solution. Based on whether the candidate solution forms a new optimal solution, whether it is better than the current solution, and whether it is accepted, evaluation scores are assigned to the destruction and repair operators selected this time, and the corresponding weights are updated according to the evaluation scores and the historical weights of the destruction and repair operators. Based on the updated destruction operator weights and repair operator weights, the destruction operator and repair operator are reselected, and the processes of removing home furnishing products, inserting replacement candidate products, evaluating candidate solutions, and updating weights are repeated until the preset number of iterations is reached or no new optimal solution is formed in consecutive iterations. The candidate solution that satisfies the budget information, circuit rated power constraints, network communication protocol consistency constraints, and space constraints and has the best comprehensive evaluation value is determined as the updated target selection list.

10. A method system for rapid selection and configuration of smart homes, based on the rapid selection and configuration method for smart homes according to any one of claims 1 to 9, characterized in that: include, The data acquisition module obtains user input information, including apartment dimensions, rated circuit power, network communication protocol, and budget information. The encoding module encodes user input information based on a machine learning model and performs matching processing in conjunction with historical selection data to generate implicit demand labels. The filtering module performs spatial modeling of the apartment size based on spatial analysis algorithms to obtain spatial constraint results, and filters candidate home furnishing product combinations from a preset product library based on the spatial constraint results and implicit demand tags. The optional module allocates budget to candidate home furnishing product combinations based on budget information. The budget allocation is based on the security, comfort and entertainment function labels of the home furnishing products and allocates the budget information according to the preset function budget weights. After obtaining the function budget amount, an initial optional list is generated. The calculation module performs power matching calculations and network communication protocol matching calculations on the initial selection list; The matching module outputs a target selection list when the power matching calculation result meets the circuit rated power constraint and the network communication protocol matching calculation result meets the consistency constraint. The replacement module performs the corresponding replacement operation when one condition is not met, until both conditions are met simultaneously, and then generates the target configuration list. The update module receives user operations to add, delete, and replace items in the target selection list, generates an updated selection list, and sets user operation retention flags for newly added and replaced furniture products. The updated selection list is used as the current solution, keeping the furniture products with user operation retention flags unchanged. An adaptive large neighborhood search algorithm is used to remove some furniture products according to the degree of constraint conflict, and the current solution is repaired according to the regret insertion value of the replacement candidate products. The weights of the destruction operator and the repair operator are updated based on the repair results, and the updated target selection list is generated iteratively.

Images