Decoration supply chain management system and method based on cost prediction

Abstract

The invention relates to the technical field of intelligent management, and particularly discloses a decoration supply chain management system and method based on cost prediction, and the method comprises the steps: carrying out the semantic analysis of to-be-decorated project information through employing an artificial intelligence technology based on deep learning, and extracting the semantic features of the to-be-decorated project information, and meanwhile, a large amount of historical decoration project information is analyzed, semantic association information between the project scale of the historical decoration project and the decoration cost is mined, and the semantic association information is utilized to query the cost characteristics of the to-be-decorated project information, so that the cost prediction of the to-be-decorated project information is realized. Therefore, more accurate and efficient cost prediction and control can be provided for decoration enterprises and users, the decoration cost is reduced, and the profitability of decoration projects and the decoration experience of the users are improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent management technology, and more specifically, to a decoration supply chain management system and method based on cost forecasting. Background Art

[0002] The renovation industry, a complex industry involving multiple processes and supply chains, has always posed a significant challenge to businesses and users regarding cost control and management. From design to construction and final acceptance, every stage of a renovation project involves controlling both input and output costs. This is particularly true in supply chain management, where cost control at every stage, from material procurement to construction team coordination, is directly impactful on project profitability and the user experience. Therefore, effective cost control and forecasting to improve the profitability and user experience of renovation projects is crucial.

[0003] However, traditional renovation supply chain management methods often rely on manual experience and lack data support, which can lead to inaccurate cost control and inefficient supply chains, affecting project progress and quality, and causing economic losses and dissatisfaction for companies and users. Therefore, a renovation supply chain management system and method based on cost forecasting is desired. Summary of the Invention

[0004] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a decoration supply chain management system and method based on cost forecasting, which uses artificial intelligence technology based on deep learning to perform semantic analysis on the information of the decoration project to be decorated, extracts the semantic features of the information of the project to be decorated, and analyzes a large amount of historical decoration project information at the same time, digs out the semantic correlation information between the project scale and decoration cost of the historical decoration project, and uses this semantic correlation information to query the cost features of the information of the project to be decorated, thereby realizing the cost forecast of the information of the project to be decorated. In this way, it can provide decoration companies and users with more accurate and efficient cost forecasting and control, reduce decoration costs, and improve the profitability of decoration projects and user decoration experience.

[0005] Accordingly, according to one aspect of the present application, a decoration supply chain management system based on cost forecasting is provided, which includes:

[0006] A decoration project information acquisition module is used to obtain information on pending decoration projects and information on multiple historical decoration projects, wherein the pending decoration project information includes the type of decoration project, the decoration area, and decoration demand information; the historical decoration project data includes the type of decoration project, the decoration area, decoration material price data, the size and cost of the construction team, and equipment rental costs;

[0007] A decoration project information semantic encoding module, configured to perform semantic understanding on the information of the project to be decorated and the information of the plurality of historical decoration projects to obtain a sequence of semantic feature vectors of the information of the project to be decorated and a sequence of semantic feature vectors of the information of the historical decoration projects;

[0008] A historical renovation project semantic association coding module, configured to perform semantic association coding on the sequence of the historical renovation project information semantic feature vectors to obtain a historical renovation project information semantic association feature matrix;

[0009] The cost prediction module is used to generate a cost prediction value of the project to be renovated based on the association interaction information between the semantic association feature matrix of the historical renovation project information and the semantic feature vector of the project to be renovated information.

[0010] In the above-mentioned decoration supply chain management system based on cost forecasting, the decoration project information semantic encoding module is used to: respectively pass the information of the project to be decorated and the multiple historical decoration project information through the decoration project information semantic encoder based on the Transformer model to obtain a sequence of the semantic feature vector of the information of the project to be decorated and the semantic feature vector of the historical decoration project information.

[0011] In the above-mentioned decoration supply chain management system based on cost forecasting, the historical decoration project semantic association encoding module includes: a semantic enhancement unit, which is used to perform feature adaptive enhancement processing on the sequence of semantic feature vectors of the historical decoration project information to obtain a sequence of essential semantic feature vectors of the historical decoration project information; a semantic association feature extraction unit, which is used to extract the semantic association features of the sequence of essential semantic feature vectors of the historical decoration project information to obtain the semantic association feature matrix of the historical decoration project information.

[0012] In the above-mentioned decoration supply chain management system based on cost forecasting, the semantic enhancement unit is used to: pass the sequence of semantic feature vectors of the historical decoration project information through a feature filter based on the self-attention layer to obtain a sequence of essential semantic feature vectors of the historical decoration project information.

[0013] In the above-mentioned decoration supply chain management system based on cost forecasting, the semantic association feature extraction unit is used to: arrange the sequence of the essential semantic feature vectors of the historical decoration project information into an essential semantic feature matrix of the historical decoration project information and then pass it through a semantic association feature extractor based on a convolutional neural network model to obtain the semantic association feature matrix of the historical decoration project information.

[0014] In the above-mentioned decoration supply chain management system based on cost prediction, the cost prediction module includes: an association coding unit, used to multiply the semantic feature vector of the information of the project to be decorated with the semantic association feature matrix of the historical decoration project information to obtain a cost query feature vector of the project to be decorated; an interference compensation unit, used to perform feature fine-grained internal structure compensation based on intrinsic decomposition on the cost query feature vector of the project to be decorated to obtain a compensated cost query feature vector of the project to be decorated; and a decoding regression unit, used to pass the compensated cost query feature vector of the project to be decorated through a decoration cost predictor based on a decoder to obtain the cost prediction value.

[0015] In the above-mentioned decoration supply chain management system based on cost prediction, the interference compensation unit is used to: calculate the global fine-grained autocorrelation topology matrix of the cost query feature vector of the project to be decorated; perform eigendecomposition on the global fine-grained autocorrelation topology matrix to obtain a set of fine-grained intrinsic component coding vectors for the cost query of the project to be decorated; perform information compression on each fine-grained intrinsic component coding vector of the cost query of the project to be decorated in the set of fine-grained intrinsic component coding vectors for the cost query of the project to be decorated to obtain a set of fine-grained intrinsic component modulation coding vectors for the cost query of the project to be decorated; calculate the internal structure significant modulation factor of each fine-grained intrinsic component modulation coding vector of the cost query of the project to be decorated in the set of fine-grained intrinsic component modulation coding vectors for the cost query of the project to be decorated to obtain a set of internal structure significant modulation factors; regularize the set of internal structure significant modulation factors to obtain a set of internal structure significant modulation weight factors; and based on the set of internal structure significant modulation weight factors, perform fine-grained fusion on the set of fine-grained intrinsic component modulation coding vectors for the cost query of the project to be decorated to obtain a compensated cost query feature vector for the project to be decorated.

[0016] According to another aspect of the present application, a decoration supply chain management method based on cost forecasting is provided, which includes:

[0017] Obtaining information on pending renovation projects and information on multiple historical renovation projects, wherein the pending renovation project information includes renovation project type, renovation area, and renovation demand information, and the historical renovation project data includes renovation project type, renovation area, renovation material price data, construction team size and cost, and equipment rental cost;

[0018] Semantically understanding the information of the project to be renovated and the information of the plurality of historical renovation projects to obtain a sequence of semantic feature vectors of the information of the project to be renovated and a sequence of semantic feature vectors of the information of the historical renovation projects;

[0019] Performing semantic association coding on the sequence of the historical renovation project information semantic feature vectors to obtain a historical renovation project information semantic association feature matrix;

[0020] Based on the association interaction information between the semantic association feature matrix of the historical renovation project information and the semantic feature vector of the information of the project to be renovated, a cost prediction value of the project to be renovated is generated.

[0021] Compared to existing technologies, the cost-forecasting-based renovation supply chain management system and method provided in this application uses deep learning-based artificial intelligence technology to perform semantic analysis on renovation project information, extracting the semantic features of the information. Simultaneously, it analyzes a large amount of historical renovation project information to uncover semantic correlations between the project scale and renovation costs of historical renovation projects. This semantic correlation information is then used to query the cost features of the information on the projects to be renovated, thereby enabling cost prediction for the information on the projects to be renovated. This enables more accurate and efficient cost prediction and control for renovation companies and users, reducing renovation costs and improving the profitability of renovation projects and the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The above and other purposes, features, and advantages of the present application will become more apparent through a more detailed description of the embodiments of the present application in conjunction with the accompanying drawings. The accompanying drawings are intended to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0023] Figure 1 This is a block diagram of a decoration supply chain management system based on cost forecasting according to an embodiment of the present application.

[0024] Figure 2 Schematic diagram of the architecture of a decoration supply chain management system based on cost forecasting according to an embodiment of the present application.

[0025] Figure 3 This is a block diagram of a semantic association coding module for historical decoration projects in a decoration supply chain management system based on cost prediction according to an embodiment of the present application.

[0026] Figure 4 It is a block diagram of a cost prediction module in a decoration supply chain management system based on cost prediction according to an embodiment of the present application.

[0027] Figure 5 This is a flowchart of a decoration supply chain management method based on cost forecasting according to an embodiment of the present application. DETAILED DESCRIPTION

[0028] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described herein.

[0029] Figure 1 This is a block diagram of a decoration supply chain management system based on cost forecasting according to an embodiment of the present application. Figure 2 FIG is a schematic diagram of the architecture of a decoration supply chain management system based on cost forecasting according to an embodiment of the present application. Figure 1 and Figure 2 As shown, according to an embodiment of the present application, a decoration supply chain management system 100 based on cost forecasting includes: a decoration project information acquisition module 110, which is used to acquire information on a project to be decorated, and information on a plurality of historical decoration projects, wherein the information on the project to be decorated includes information on the type of decoration project, the area of decoration, and decoration demand information, and the historical decoration project data includes information on the type of decoration project, the area of decoration, price data of decoration materials, the size of the construction team and its cost, and equipment rental cost; a decoration project information semantic encoding module 120, which is used to perform semantic understanding on the information on the project to be decorated and the plurality of historical decoration project information respectively to obtain a sequence of semantic feature vectors of the information on the project to be decorated and a sequence of semantic feature vectors of the semantic feature vectors of the historical decoration project information; a historical decoration project semantic association encoding module 130, which is used to perform semantic association encoding on the sequence of semantic feature vectors of the historical decoration project information to obtain a semantic association feature matrix of the historical decoration project information; and a cost prediction module 140, which is used to generate a cost prediction value for the project to be decorated based on the association interaction information between the semantic association feature matrix of the historical decoration project information and the semantic feature vectors of the information on the project to be decorated.

[0030] As mentioned in the above background technology, traditional decoration supply chain management methods often rely on manual experience and lack data support, which may lead to inaccurate cost control, low supply chain efficiency, affect the progress and quality of the project, and bring economic losses and dissatisfaction to enterprises and users. In response to the above technical problems, the technical concept of this application is to use artificial intelligence technology based on deep learning to perform semantic analysis on the information of decoration projects to be renovated, extract the semantic features of the information of the projects to be renovated, and at the same time analyze a large amount of historical decoration project information to dig out the semantic correlation information between the project scale and decoration cost of historical decoration projects, and use this semantic correlation information to query the cost features of the information of the projects to be renovated, thereby realizing the cost prediction of the information of the projects to be renovated. In this way, it can provide decoration companies and users with more accurate and efficient cost prediction and control, reduce decoration costs, and improve the profitability of decoration projects and user decoration experience.

[0031] In the aforementioned cost-forecasting-based renovation supply chain management system 100, the renovation project information acquisition module 110 is used to acquire information on pending renovation projects and multiple historical renovation project information. The pending renovation project information includes project type, renovation area, and renovation requirements. The historical renovation project data includes project type, renovation area, renovation material pricing data, construction team size and cost, and equipment rental costs. It should be understood that the pending renovation project information includes project type, renovation area, and renovation requirements information. Different types of renovation projects (e.g., residential, office, and commercial spaces) have different renovation characteristics and cost structures. The renovation area refers to the size of the space to be renovated and is a significant factor influencing renovation costs. Furthermore, the client's renovation requirements (e.g., specific requirements for renovation materials, construction techniques, and design) also influence renovation costs. Therefore, acquiring the project type, renovation area, and renovation requirements information for pending renovation projects provides the necessary data foundation for subsequent cost forecasting. Furthermore, the historical renovation project data includes project type, renovation area, renovation material pricing data, construction team size and cost, and equipment rental costs. This information reflects the relationship between the type, scale and cost of historical renovation projects, and can provide valuable reference for cost prediction of upcoming renovation projects.

[0032] In the above-mentioned decoration supply chain management system 100 based on cost forecasting, the decoration project information semantic encoding module 120 is used to perform semantic understanding on the information of the project to be decorated and the information of the multiple historical decoration projects respectively to obtain a sequence of semantic feature vectors of the information of the project to be decorated and a sequence of semantic feature vectors of the information of the historical decoration project. In a specific example of the present application, the encoding method for performing semantic understanding on the information of the project to be decorated and the information of the multiple historical decoration projects respectively is to pass the information of the project to be decorated and the information of the multiple historical decoration projects respectively through a decoration project information semantic encoder based on a Transformer model to obtain a sequence of semantic feature vectors of the information of the project to be decorated and a sequence of semantic feature vectors of the information of the historical decoration project. It should be understood that the Transformer model is a deep learning model based on the self-attention mechanism, and its powerful feature extraction and sequence encoding capabilities enable it to effectively capture semantic features and contextual relationships in text data. In the technical solution of the present application, the information of the project to be renovated and the information of the multiple historical renovation projects are input into a renovation project information semantic encoder based on a Transformer model. The renovation project information semantic encoder utilizes the multi-layer self-attention mechanism of the Transformer model to perform word-by-word semantic encoding on the information of the project to be renovated and the multiple historical renovation project information, and performs parallel processing on each word in the text information to capture the long-distance dependency relationship between each word, and respectively mines out the deep semantic features of the information of the project to be renovated and the multiple historical renovation project information, and encodes them into a vector representation to obtain a sequence of semantic feature vectors of the information of the project to be renovated and the semantic feature vectors of the historical renovation project information, thereby providing an effective data representation for subsequent cost prediction.

[0033] In the above-mentioned renovation supply chain management system 100 based on cost forecasting, the historical renovation project semantic association coding module 130 is used to perform semantic association coding on the sequence of the historical renovation project information semantic feature vectors to obtain a historical renovation project information semantic association feature matrix. Specifically, Figure 3 FIG is a block diagram of a semantic association coding module for historical decoration projects in a decoration supply chain management system based on cost prediction according to an embodiment of the present application. Figure 3 As shown, the historical decoration project semantic association encoding module 130 includes: a semantic enhancement unit 131, which is used to perform feature adaptive enhancement processing on the sequence of semantic feature vectors of the historical decoration project information to obtain a sequence of essential semantic feature vectors of the historical decoration project information; a semantic association feature extraction unit 132, which is used to extract the semantic association features of the sequence of essential semantic feature vectors of the historical decoration project information to obtain the semantic association feature matrix of the historical decoration project information.

[0034] Specifically, the semantic reinforcement unit 131 is used to perform feature adaptive reinforcement processing on the sequence of the historical renovation project information semantic feature vectors to obtain a sequence of the historical renovation project information essential semantic feature vectors. In a specific example of the present application, the processing method for performing feature adaptive reinforcement processing on the sequence of the historical renovation project information semantic feature vectors is to pass the sequence of the historical renovation project information semantic feature vectors through a feature filter based on a self-attention layer to obtain a sequence of the historical renovation project information essential semantic feature vectors. It should be understood that, considering that the sequence of the historical renovation project information semantic feature vectors may contain some redundant and minor information, it is not important for subsequent cost prediction and may even introduce interference. Therefore, in order to strengthen the information in the sequence of the historical renovation project information semantic feature vectors that is more critical for cost prediction, a feature filter based on a self-attention layer is further introduced to perform feature adaptive reinforcement on the sequence of the historical renovation project information semantic feature vectors. Among them, the self-attention layer is a deep learning layer that strengthens key information in the sequence data by calculating the correlation between different positions in the input sequence data. In the technical solution of the present application, the feature filter performs self-attention weight calculation by learning the correlation and dependency between each semantic feature vector of the historical decoration project information in the sequence of the semantic feature vectors of the historical decoration project information, and performs feature weighted screening on the sequence of the semantic feature vectors of the historical decoration project information based on the self-attention weight calculation results to strengthen the key information that is more relevant to cost prediction, suppress redundant and minor information, and provide a more effective data representation for subsequent cost prediction, so as to improve the accuracy and stability of subsequent cost prediction.

[0035] Specifically, the semantic association feature extraction unit 132 is configured to extract semantic association features from the sequence of intrinsic semantic feature vectors of the historical renovation project information to obtain the semantic association feature matrix of the historical renovation project information. In a specific example of the present application, the encoding method for extracting semantic association features from the sequence of intrinsic semantic feature vectors of the historical renovation project information to obtain the semantic association feature matrix of the historical renovation project information is to arrange the sequence of intrinsic semantic feature vectors of the historical renovation project information into a historical renovation project information essential semantic feature matrix and then pass the matrix through a semantic association feature extractor based on a convolutional neural network model to obtain the semantic association feature matrix of the historical renovation project information. It should be understood that different historical renovation projects may have certain associations or patterns. For example, even for the same renovation project type, different decoration styles, material selections, and engineering complexity may affect the project cost. Therefore, in order to mine potential associations and patterns between the scale and cost of renovation projects from a large number of historical renovation projects, thereby providing more accurate cost predictions for pending renovation projects, the technical solution of the present application introduces a convolutional neural network model to extract semantic association features from the sequence of intrinsic semantic feature vectors of the historical renovation project information. Specifically, the sequence of intrinsic semantic feature vectors of historical renovation project information is first arranged into a 'historical renovation project information essential semantic feature matrix' to integrate the semantic feature information of each historical renovation project, facilitating the capture of potential relationships between them in a matrix format. A semantic association feature extractor based on a convolutional neural network model is then used to perform sliding convolution and pooling operations on the 'historical renovation project information essential semantic feature matrix' to extract local association patterns within the 'historical renovation project information essential semantic feature matrix', capturing the semantic association information between each historical renovation project and providing an important reference for subsequent cost prediction.

[0036] In the above-mentioned renovation supply chain management system 100 based on cost forecasting, the cost forecasting module 140 is used to generate a cost forecast value for the project to be renovated based on the association interaction information between the semantic association feature matrix of the historical renovation project information and the semantic feature vector of the project to be renovated information. Specifically, Figure 4 FIG is a block diagram of a cost prediction module in a decoration supply chain management system based on cost prediction according to an embodiment of the present application. Figure 4As shown, the cost prediction module 140 includes: an association coding unit 141, which is used to multiply the semantic feature vector of the information of the project to be renovated with the semantic association feature matrix of the historical renovation project information to obtain a cost query feature vector of the project to be renovated; an interference compensation unit 142, which is used to perform feature fine-grained internal structure compensation based on intrinsic decomposition on the cost query feature vector of the project to be renovated to obtain a compensated cost query feature vector of the project to be renovated; and a decoding regression unit 143, which is used to pass the compensated cost query feature vector of the project to be renovated through a decoration cost predictor based on a decoder to obtain the cost prediction value.

[0037] Specifically, the association coding unit 141 is configured to multiply the semantic feature vector of the project to be renovated information with the semantic association feature matrix of the historical renovation project information to obtain a cost query feature vector for the project to be renovated. It should be understood that the semantic association feature matrix of the historical renovation project information captures the semantic association information between historical renovation projects and reflects the association pattern between the scale and cost of renovation projects. The semantic feature vector of the project to be renovated information represents the project scale and demand of the project to be renovated. Therefore, the semantic association feature matrix of the historical renovation project information can be used as a reference feature, and the semantic feature vector of the project to be renovated information can be used to query the cost information of historical renovation projects similar to the project to be renovated from the semantic association feature matrix of the historical renovation project information, thereby providing an accurate cost forecast for the project to be renovated. In the technical solution of the present application, the semantic association feature matrix of the historical renovation project information and the semantic feature vector of the project to be renovated information are effectively associated and combined by multiplication to generate the cost query feature vector for the project to be renovated, which reflects the similarity between the historical renovation projects and the project to be renovated in terms of decoration type, area, and demand, providing a more accurate and comprehensive feature representation for subsequent cost forecasting.

[0038] Specifically, the interference compensation unit 142 is used to perform feature fine-grained internal structure compensation based on eigendecomposition on the feature vector of the cost query of the project to be renovated to obtain the compensated feature vector of the cost query of the project to be renovated. Since the feature vector of the cost query of the project to be renovated often contains a large amount of redundant information or is contaminated by noise during the generation process, especially in high-dimensional space, if the feature vector of the cost query of the project to be renovated is directly used for cost prediction, it may lead to increased model complexity, increased risk of overfitting, and decreased generalization ability. Based on this, in the technical solution of the present application, the feature vector of the cost query of the project to be renovated is compensated based on eigendecomposition to obtain the compensated feature vector of the cost query of the project to be renovated.

[0039] Specifically, the interference compensation unit 142 is configured to:

[0040] First, the global fine-grained autocorrelation topology matrix of the query feature vector of the cost of the project to be renovated is calculated, which can be expressed as follows:

[0041]

[0042] M=D1⊙D2

[0043] v i ,v j ∈V

[0044] Among them, V represents the set of query feature vectors of the cost of the renovation project to be completed, v i and v j They represent the i-th and j-th eigenvalues of the feature vector of the cost query of the project to be renovated, respectively. w1, w2, w3 and w4 represent different weight hyperparameters, ⊙ represents matrix dot product, D1 represents the forward weighted matrix of the cost query feature of the project to be renovated, and D2 represents the reverse weighted matrix of the cost query feature of the project to be renovated. Represents the value of the (i, j)th position of the forward weighted matrix of the query feature of the cost of the project to be renovated, It represents the value of the (i, j)th position of the inverse weighted matrix of the query feature of the cost of the renovation project, and M represents the global fine-grained autocorrelation topology matrix.

[0045] That is, by conducting a global analysis of the feature vector of the cost query of the renovation project, covering all dimensional elements and going deep into the basic constituent unit level, the originally implicit structural information within the cost query feature of the renovation project is transformed into an explicit global fine-grained autocorrelation topological matrix, and the degree of dependence or correlation pattern between the dimensions is clearly quantified, providing a directly usable structured data foundation for the subsequent modulation and cost prediction based on the structure within the cost query feature of the renovation project.

[0046] Secondly, the global fine-grained autocorrelation topological matrix is subjected to eigendecomposition to obtain a set of fine-grained eigencomponent encoding vectors for querying the cost of the project to be renovated, which is expressed as follows:

[0047]

[0048] Where Λ represents a diagonal matrix, λ1 and λ m denote the first and mth eigenvalues of the diagonal matrix, respectively, (·) T represents the transpose of the vector, U represents the set of fine-grained intrinsic component encoding vectors for querying the cost of the renovation project, x1, x2 and x m They respectively represent the first, second and mth fine-grained intrinsic component coding vectors for querying the cost of the project to be renovated in the set of fine-grained intrinsic component coding vectors for querying the cost of the project to be renovated.

[0049] Specifically, the complex, interconnected structure within the features is decoupled and converted into a series of independent, prioritized infrastructure patterns. This results in a fine-grained intrinsic component encoding vector for querying the cost of a renovation project, guided by the data's internal structure. This vector represents the distinct, independent patterns of variation within the features, facilitating subsequent targeted adjustments to each independent structural aspect. This approach allows the model to focus more closely on important feature structural patterns, improving the accuracy and efficiency of cost predictions.

[0050] Then, information compression is performed on each fine-grained intrinsic component code vector of the query on the cost of the project to be renovated in the set of fine-grained intrinsic component code vectors to obtain a set of fine-grained intrinsic component modulation code vectors for query on the cost of the project to be renovated, which is expressed as follows:

[0051]

[0052] Among them, x i represents the i-th fine-grained intrinsic component encoding vector of the cost query of the renovation project in the set of fine-grained intrinsic component encoding vectors, ||·|| represents the Euclidean norm, y i Represents the fine-grained intrinsic component modulation coding vector for querying the cost of the i-th project to be renovated.

[0053] That is, each decoupled structural component is finely modulated through nonlinear transformations. In the process of compressing the dynamic range, normalizing the norm, or scaling according to rules, more complex semantic dependencies are captured and feature discrimination capabilities are enhanced. At the same time, irrelevant noise is suppressed, so that the feature representation shifts from pure linear decomposition to a fine-grained intrinsic component modulation coding vector for querying the cost of the project to be renovated that is more in line with the factors affecting the actual renovation cost. Specifically, by introducing nonlinear processing capabilities, the limitations of linear decomposition are broken, so that the generated fine-grained intrinsic component modulation coding vector for querying the cost of the project to be renovated can more accurately reflect the complex mapping relationship between the attributes of the renovation project and the cost, reduce redundant information interference, and increase the effective information content of the features for cost prediction, providing more discriminative and robust input for the subsequent cost prediction model.

[0054] Next, the internal structure significant modulation factor of each fine-grained intrinsic component modulation coding vector of the cost query of the project to be renovated in the set of fine-grained intrinsic component modulation coding vectors is calculated to obtain a set of internal structure significant modulation factors, which is expressed as follows:

[0055]

[0056] Among them, α and β represent different weight parameters, ||·||1 represents the first norm, ||·||2 represents the second norm, L represents the length of the fine-grained intrinsic component modulation coding vector for querying the cost of the renovation project to be completed, and a i represents y i The corresponding internal structure significant modulation factor.

[0057] That is, the fine-grained intrinsic component modulation coding vector of the cost query of each project to be renovated is calculated to generate a scalar value, and the significance of the structural information it carries for cost prediction is dynamically evaluated, thereby providing a differentiated importance basis for subsequent weighted fusion, realizing adaptive reconstruction of features, and enabling the model to more accurately capture the key mapping relationship between the attributes of the renovation project and the cost, thereby improving the pertinence and effectiveness of feature representation.

[0058] Then, the set of the internal structure significant modulation factors is regularized to obtain a set of internal structure significant modulation weight factors, which is expressed as follows:

[0059] w i =Softmax(a i )

[0060] Among them, Softmax(·) represents the regularization function, w i Indicates a i The corresponding internal structure significantly modulates the weight factor.

[0061] That is, by imposing constraints, the intrinsic structure significant modulation factors are converted into a set of appropriate intrinsic structure significant modulation weight factors, so that the weight distribution is more reasonable, numerical instability such as gradient explosion or disappearance is prevented, excessive favoritism for a few components is avoided, and the subsequent feature fusion process is ensured to be stable and controllable, so that the generated intrinsic structure significant modulation weight factors have a reasonable scale and distribution, thereby improving the model stability and generalization ability.

[0062] Finally, based on the set of the significant modulation weight factors of the internal structure, the set of fine-grained intrinsic component modulation code vectors of the query cost of the project to be renovated is fine-grainedly fused to obtain the compensated query cost feature vector of the project to be renovated, which is expressed as follows:

[0063]

[0064] Among them, V' represents the query feature vector of the cost of the renovation project after compensation.

[0065] Specifically, the method utilizes the significant internal structure modulation weight factor to perform a weighted combination of the modulation coding vectors of the fine-grained intrinsic components of each project cost query. This concentrates scattered information, highlights key information, and suppresses secondary or noisy information, achieving effective information integration and reconstruction. The resulting compensated project cost query feature vector condenses the key structural information within the project cost query feature. By enhancing its representational capabilities through nonlinear modulation, it is more robust and discriminative than the project cost query feature vector, enabling it to better accomplish downstream machine learning tasks such as cost prediction.

[0066] Specifically, the decoding regression unit 143 is used to pass the compensated cost query feature vector of the project to be renovated through a decoder-based decoration cost predictor to obtain the cost prediction value. It should be understood that the decoder is a neural network model used to decode the encoded feature representation into a target value. In the technical solution of the present application, the decoder is used to decode the compensated cost query feature vector of the project to be renovated into a cost prediction value. Specifically, the compensated cost query feature vector of the project to be renovated contains similar features between the project to be renovated and historical projects and cost reference information of historical projects. The compensated cost query feature vector of the project to be renovated is input into the decoder-based decoration cost predictor, and the decoder can perform decoding regression based on the feature information in the compensated cost query feature vector of the project to be renovated, map the compensated cost query feature vector of the project to be renovated to the numerical space of the cost prediction value, and output the cost prediction result of the project to be renovated, thereby providing effective decision support for the cost management of the decoration supply chain.

[0067] Specifically, the decoding regression unit 143 is used to: use the decoder-based renovation cost predictor to perform decoding regression on the post-compensation renovation project cost query feature vector using the following decoding formula to obtain the cost prediction value, wherein the decoding formula is Wherein, X is the cost query feature vector of the renovation project after compensation, Y is the cost prediction value, and W is the weight matrix. Represents matrix multiplication.

[0068] In summary, the cost-forecasting-based renovation supply chain management system according to the embodiment of the present application is illustrated. It uses deep learning-based artificial intelligence technology to perform semantic analysis on renovation project information, extracting the semantic features of the renovation project information. At the same time, it analyzes a large amount of historical renovation project information to mine the semantic correlation information between the project scale and renovation cost of historical renovation projects. This semantic correlation information is then used to query the cost features of the renovation project information, thereby realizing cost prediction of the renovation project information. In this way, it can provide renovation companies and users with more accurate and efficient cost prediction and control, reduce renovation costs, and improve the profitability of renovation projects and the user's renovation experience.

[0069] Figure 5 FIG is a flow chart of a decoration supply chain management method based on cost forecasting according to an embodiment of the present application. Figure 5 As shown, the decoration supply chain management method based on cost forecasting according to the embodiment of the present application includes the steps of: S110, obtaining information of a project to be decorated, and information of multiple historical decoration projects, wherein the information of the project to be decorated includes the type of decoration project, the area of decoration, and decoration demand information, and the historical decoration project data includes the type of decoration project, the area of decoration, the price data of decoration materials, the size of the construction team and its cost, and the equipment rental cost; S120, performing semantic understanding on the information of the project to be decorated and the multiple historical decoration project information respectively to obtain a sequence of semantic feature vectors of the information of the project to be decorated and a sequence of semantic feature vectors of the semantic feature vectors of the historical decoration project information; S130, performing semantic association encoding on the sequence of semantic feature vectors of the historical decoration project information to obtain a semantic association feature matrix of the historical decoration project information; S140, generating a cost prediction value of the project to be decorated based on the association interaction information between the semantic association feature matrix of the historical decoration project information and the semantic feature vectors of the information of the project to be decorated.

[0070] Here, those skilled in the art will appreciate that the specific operations of each step in the above-mentioned decoration supply chain management method based on cost forecasting have been described in the above reference. Figures 1 to 4 The above description has been introduced in detail in the description of the decoration supply chain management system based on cost forecasting, and therefore, its repeated description will be omitted.

[0071] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed devices, systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the module division is only a logical function division, and there may be other division methods in actual implementation. The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0072] In addition, the functional modules in various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or hardware plus software functional modules.

[0073] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be encompassed therein. Any reference to a figure in a claim should not be construed as limiting the claim to which it relates.

[0074] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units stated in the system claims can also be implemented by one unit through software or hardware.

[0075] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A decoration supply chain management system based on cost forecasting, characterized in that: include: A decoration project information acquisition module is used to obtain information on pending decoration projects and information on multiple historical decoration projects, wherein the pending decoration project information includes the type of decoration project, the decoration area, and decoration demand information; the historical decoration project data includes the type of decoration project, the decoration area, decoration material price data, the size and cost of the construction team, and equipment rental costs; A decoration project information semantic encoding module, configured to perform semantic understanding on the information of the project to be decorated and the information of the plurality of historical decoration projects to obtain a sequence of semantic feature vectors of the information of the project to be decorated and a sequence of semantic feature vectors of the information of the historical decoration projects; A historical renovation project semantic association coding module, configured to perform semantic association coding on the sequence of the historical renovation project information semantic feature vectors to obtain a historical renovation project information semantic association feature matrix; The cost prediction module is used to generate a cost prediction value of the project to be renovated based on the association interaction information between the semantic association feature matrix of the historical renovation project information and the semantic feature vector of the project to be renovated information.

2. The decoration supply chain management system based on cost forecasting according to claim 1 is characterized in that: The decoration project information semantic encoding module is used to: The information of the project to be renovated and the information of the multiple historical renovation projects are respectively passed through a renovation project information semantic encoder based on a Transformer model to obtain a sequence of the semantic feature vector of the information of the project to be renovated and the semantic feature vector of the historical renovation project information.

3. The decoration supply chain management system based on cost forecasting according to claim 2 is characterized in that: The historical renovation project semantic association coding module includes: a semantic enhancement unit, configured to perform feature adaptive enhancement processing on the sequence of semantic feature vectors of the historical renovation project information to obtain a sequence of essential semantic feature vectors of the historical renovation project information; The semantic association feature extraction unit is used to extract the semantic association features of the sequence of the essential semantic feature vectors of the historical renovation project information to obtain the semantic association feature matrix of the historical renovation project information.

4. The decoration supply chain management system based on cost forecasting according to claim 3 is characterized in that: The semantic enhancement unit is used to: The sequence of the semantic feature vectors of the historical renovation project information is passed through a feature filter based on a self-attention layer to obtain a sequence of the essential semantic feature vectors of the historical renovation project information.

5. The decoration supply chain management system based on cost forecasting according to claim 4 is characterized in that: The semantic association feature extraction unit is used to: The sequence of the essential semantic feature vectors of the historical decoration project information is arranged into an essential semantic feature matrix of the historical decoration project information, and then the matrix is obtained by a semantic association feature extractor based on a convolutional neural network model.

6. The decoration supply chain management system based on cost forecasting according to claim 5 is characterized in that: The cost prediction module includes: An association coding unit, configured to multiply the semantic feature vector of the information of the project to be renovated by the semantic association feature matrix of the historical renovation project information to obtain a query feature vector of the cost of the project to be renovated; An interference compensation unit, configured to perform characteristic fine-grained internal structure compensation based on eigendecomposition on the query feature vector of the cost of the project to be renovated to obtain a compensated query feature vector of the cost of the project to be renovated; The decoding regression unit is used to pass the query feature vector of the cost of the project to be renovated after compensation through a renovation cost predictor based on a decoder to obtain the cost prediction value.

7. The decoration supply chain management system based on cost forecasting according to claim 6 is characterized in that: The interference compensation unit includes: An activation subunit, configured to pass the query feature vector of the cost of the project to be renovated through an activation function to obtain a category probability vector; a maximum class probability value identification subunit, configured to input the class probability vector into a maximum class probability value identifier to obtain a maximum class probability value; a multiplicative interference factor vector calculation subunit, configured to calculate, based on the maximum class probability value, a multiplicative interference factor of a class probability value at each position in the class probability vector to obtain a multiplicative interference factor vector, wherein, based on the maximum class probability value, calculating the multiplicative interference factor of the class probability value at each position in the class probability vector comprises: performing maximum-based normalization processing on the class probability vector based on the maximum class probability value to obtain the multiplicative interference factor vector; a compensation subunit, configured to perform characteristic fine-grained internal structure compensation on the query feature vector of the cost of the project to be renovated based on eigendecomposition to obtain a compensated query feature vector of the cost of the project to be renovated; The splicing and fusion subunit is used to splice and fuse the query feature vector of the cost of the project to be renovated and the compensation feature vector to obtain the compensated query feature vector of the cost of the project to be renovated.

8. The decoration supply chain management system based on cost forecasting according to claim 7 is characterized in that: The interference compensation unit is configured to: Calculating a global fine-grained autocorrelation topological matrix of the query feature vector of the cost of the project to be renovated; Performing eigendecomposition on the global fine-grained autocorrelation topological matrix to obtain a set of fine-grained eigencomponent coding vectors for querying the cost of the project to be renovated; Performing information compression on each fine-grained intrinsic component coding vector for querying the cost of the project to be renovated in the set of fine-grained intrinsic component coding vectors for querying the cost of the project to be renovated to obtain a set of fine-grained intrinsic component modulation coding vectors for querying the cost of the project to be renovated; Calculating the inner structure significant modulation factor of each fine-grained intrinsic component modulation coding vector of the cost query of the project to be renovated in the set of fine-grained intrinsic component modulation coding vectors to obtain a set of inner structure significant modulation factors; Regularizing the set of internal structure significant modulation factors to obtain a set of internal structure significant modulation weight factors; Based on the set of the internal structure significant modulation weight factors, the set of fine-grained intrinsic component modulation coding vectors for querying the cost of the project to be renovated is fine-grainedly fused to obtain a compensated feature vector for querying the cost of the project to be renovated.

9. A decoration supply chain management method based on cost forecasting, characterized in that: include: Obtaining information on pending renovation projects and information on multiple historical renovation projects, wherein the pending renovation project information includes renovation project type, renovation area, and renovation demand information, and the historical renovation project data includes renovation project type, renovation area, renovation material price data, construction team size and cost, and equipment rental cost; Semantically understanding the information of the project to be renovated and the information of the plurality of historical renovation projects to obtain a sequence of semantic feature vectors of the information of the project to be renovated and a sequence of semantic feature vectors of the information of the historical renovation projects; Performing semantic association coding on the sequence of the historical renovation project information semantic feature vectors to obtain a historical renovation project information semantic association feature matrix; Based on the association interaction information between the semantic association feature matrix of the historical renovation project information and the semantic feature vector of the information of the project to be renovated, a cost prediction value of the project to be renovated is generated.

Images