An intelligent engineering cost assessment method based on big data analysis

Abstract

The present invention relates to the technical field of engineering cost assessment, and discloses an intelligent engineering cost assessment method based on big data analysis, the method comprising: performing sentiment analysis on building material price comment text data to obtain building material price sentiment sequence data, and forming building material price evolution sequence data together with building material historical transaction price data; performing multi-scale decomposition on building material price evolution sequence data; calculating the information gain of building material price evolution subsequences and generating weights, and performing subsequence weighting; predicting to obtain building material price prediction values ​​and performing engineering cost assessment. The present invention combines the spectral representation of building material price evolution sequence data, performs multi-scale adaptive decomposition processing, obtains multi-scale price evolution characteristics, uses a building material price prediction model to perform forward information extraction and backward information extraction on the weighted building material price evolution subsequence, captures the complex global dependency of price information, realizes global prediction of building material prices, and then performs engineering cost assessment.

Description

Technical Field

[0001] The present invention relates to the technical field of engineering cost assessment, and in particular to an intelligent engineering cost assessment method based on big data analysis. Background Art

[0002] In modern society, the scale and complexity of engineering projects are increasing. Whether it is infrastructure construction, real estate development, or industrial projects, they all involve multiple stakeholders, complex resource allocation, and a dynamic market environment. Traditional engineering cost assessment methods often rely on historical data and empirical judgments, which are difficult to adapt to rapidly changing market demands and technological advances. This method is not only inefficient, but also easily leads to deviations in cost forecasts, which in turn affects the overall benefits of the project. Summary of the invention

[0003] In view of this, the present invention proposes an intelligent evaluation method for engineering cost based on big data analysis, which can more accurately predict engineering cost, optimize resource allocation and improve project management efficiency through analysis of massive data.

[0004] To achieve the above object, the present invention provides a method for intelligent evaluation of engineering costs based on big data analysis, comprising the following steps:

[0005] S1: Collecting the historical transaction prices of building materials and text data of building material price comments of the project, performing sentiment analysis on the collected text data of building material price comments to obtain building material price sentiment sequence data, which together with the historical transaction price data of building materials constitute building material price evolution sequence data, wherein the text representation processing based on multi-level pooling is the implementation method of the sentiment analysis;

[0006] S2: Perform multi-scale decomposition on the building material price evolution sequence data to obtain building material price evolution sub-sequences under different scale conditions, wherein variational mode decomposition based on swarm optimization is an implementation method of the multi-scale decomposition;

[0007] S3: Calculate the information gain of the building material price evolution subsequence, generate the weight of the building material price evolution subsequence, and weight the building material price evolution subsequence;

[0008] S4: construct a building materials price prediction model, wherein the building materials price prediction model takes weighted building materials price evolution subsequences under different scale conditions as input and takes building materials price prediction values ​​as output, wherein bidirectional LSTM is an implementation method of the price prediction model;

[0009] S5: Use the building materials price prediction model to predict the weighted building materials price evolution subsequences under different scale conditions to obtain the building materials price prediction value, and conduct engineering cost evaluation based on the building materials price prediction value.

[0010] As a further improvement method of the present invention:

[0011] Optionally, the historical transaction prices of building materials and building materials price comment text data of the project collected in step S1 include:

[0012] Collect the historical transaction prices and price review text data of building materials for the project:

[0013] ;

[0014] ;

[0015] in:

[0016] Represents the historical transaction prices of building materials and building materials price comment text data of the collected projects, represents the historical transaction price of building materials and the text data of building materials price comments at the nth historical moment, where N represents the total number of historical moments of the collected data;

[0017] Indicates the historical transaction price of building materials at the nth historical moment, Represents the text data of building material price comments at the nth historical moment, where the text data of building material price comments is the building material price comments of the M groups with the most likes. Indicates the price review of the mth component at the nth historical moment;

[0018] Perform sentiment analysis on the collected building materials price review text data to obtain building materials price sentiment sequence data :

[0019] ;

[0020] in:

[0021] represents the price sentiment of building materials at the nth historical moment;

[0022] Building materials price sentiment series data Together with the historical transaction price data of building materials, it constitutes the building materials price evolution sequence data:

[0023] ;

[0024] ;

[0025] ;

[0026] in:

[0027] Represents the building materials price evolution series data;

[0028] represents the price series data of building materials, Represents the sentiment evolution sequence data corresponding to the building materials price series data.

[0029] Optionally, the performing sentiment analysis on the collected building material price review text data to obtain building material price sentiment sequence data includes:

[0030] The sentiment analysis of the collected building materials price review text data is performed to obtain the building materials price sentiment sequence data, where the sentiment analysis process is as follows:

[0031] S11: Perform word vector encoding processing on any building material price comment in the building material price comment text data to form a word vector sequence corresponding to the building material price comment, where the building material price comment The corresponding word vector sequence is ; In the embodiment of the present invention, the BERT model is the model used for generating word vector sequences;

[0032] S12: Perform nonlinear dimensionality reduction processing on the word vector sequence to obtain the word vector dimensionality reduction sequence corresponding to the building materials price review, where the building materials price review The corresponding word vector dimensionality reduction sequence is , word vector sequence The nonlinear dimensionality reduction process is:

[0033] S121: Select word vector sequences from all word vector sequences The K nearest neighbor vectors of , where the nearest neighbor vector is represented as a distance word vector sequence The most recent sequence of word vectors;

[0034] S122: Initialize word vector sequence The mapping weight sequence is constructed with the minimum mapping difference as the goal, and the objective function is solved to obtain the word vector sequence The mapping weight sequence of , where the objective function is:

[0035] ;

[0036] in:

[0037] Representing word vector sequence The mapping weights corresponding to the kth nearest neighbor vector of the word constitute the word vector sequence. The mapping weight sequence, where the constraints corresponding to the mapping weights are ;

[0038] represents the L1 norm;

[0039] Representing word vector sequence The kth nearest neighbor vector of ;

[0040] S123: Set the dimension of the word vector reduction sequence to dim, according to the word vector sequence The mapping weight sequence of word vectors is constructed with the goal of minimizing the mapping difference of the word vector reduction sequence, and the objective function is solved to obtain the word vector sequence. The dimension of the word vector is dim. , the constructed objective function is:

[0041] ;

[0042] S13: Use the pooling window with a step size of stride to perform average pooling and maximum pooling on the word vector dimensionality reduction sequence, and obtain the average pooling result and maximum pooling result corresponding to the word vector dimensionality reduction sequence, where the word vector dimensionality reduction sequence The corresponding average pooling result is , the maximum pooling result is ;

[0043] S14: Calculate the price sentiment of building materials at different historical moments, where the price sentiment of building materials at the nth historical moment is:

[0044] ;

[0045] in:

[0046] represents the price sentiment of building materials at the nth historical moment, The larger it is, the greater the proportion of positive sentiment in building materials price sentiment;

[0047] represents an exponential function with a natural constant as base;

[0048] Represents the convolution matrix corresponding to emotion i, , Indicates negative, neutral, and positive emotions, respectively;

[0049] represents the L2 norm.

[0050] Optionally, in step S2, multi-scale decomposition of the building materials price evolution sequence data is performed, including:

[0051] The building materials price evolution sequence data P is subjected to multi-scale decomposition, wherein the variational mode decomposition based on swarm optimization is the implementation method of the multi-scale decomposition, wherein the multi-scale decomposition process is:

[0052] S21: Determine the number of scales for scale decomposition using group optimization And the penalty parameter ;

[0053] S22: Perform Fourier transform on the building materials price evolution sequence data P to obtain the building materials price sequence data And emotional evolution sequence data The frequency domain representation of is:

[0054] ;

[0055] ;

[0056] in:

[0057] Represents building materials price series data Frequency domain representation of the Fourier transform point number e;

[0058] Representing emotional evolution sequence data Frequency domain representation of the Fourier transform point number e;

[0059] j represents the imaginary unit;

[0060] S23: Calculate the building materials price series data And emotional evolution sequence data Center frequency:

[0061] ;

[0062] ;

[0063] ;

[0064] in:

[0065] Represents building materials price series data The center frequency, Representing emotional evolution sequence data The center frequency of

[0066] Represents the time interval between adjacent historical moments; The frequency axis represents the number of Fourier transform points e;

[0067] S24: Set the current building materials price series data And emotional evolution sequence data The number of subsequences obtained by segmentation is u, the initial value of u is 0, and the maximum value is U-1;

[0068] Initialize and generate building materials price series data The corresponding u-th group of subsequence data , emotion evolution sequence data The corresponding u-th group of subsequence data , and the iterative coefficient sequence ;

[0069] S25: Set the current number of iterations of the sequence to d, the initial value of d is 0, then the subsequence data And the iterative coefficient sequence The d-th iteration results are ;

[0070] S26: Perform Fourier transform on the subsequence data and the iterative coefficient sequence, and iterate the spectrum representation corresponding to the subsequence data, wherein the subsequence data The iterative formula corresponding to the spectrum representation is:

[0071] ;

[0072] in:

[0073] Represents subsequence data The spectrum at Fourier transform point number e represents the iteration result;

[0074] represents the spectrum representation of the i-th group of subsequence data obtained by segmentation at Fourier transform point number e;

[0075] Represents the iterative coefficient sequence Spectral representation at Fourier transform point number e;

[0076] Represents subsequence data The d-th iteration result of the center frequency of For subsequence data The center frequency of

[0077] Iterate the frequency spectrum representation of the center frequency associated with the subsequence data and the iterative coefficient sequence, where the subsequence data The center frequency of the correlation And the spectrum representation of the iterative coefficient sequence The iteration formula is:

[0078] ;

[0079] ;

[0080] in:

[0081] represents the differential of the Fourier transform point number e;

[0082] S27: If the iterative result of the spectrum representation converges, perform inverse Fourier transform on the iterative result of the spectrum representation of the subsequence data to obtain the u-th group of subsequence data of the sequence data, and set u=u+1, return to step S25, until the U-th group of subsequence data is obtained; otherwise, set d=d+1, return to step S26; if the iterative result of the spectrum representation converges If the following equation is satisfied, it means convergence:

[0083] ;

[0084] in:

[0085] Indicates the preset convergence threshold;

[0086] S28: Construct U sets of material price evolution subsequences, where the uth set of material price evolution subsequences for:

[0087] ;

[0088] ;

[0089] in:

[0090] Represents the building materials price evolution subsequence The nth data value in Represents subsequence data The nth data value of Represents subsequence data The nth data value of ;

[0091] Represents subsequence data The standard deviation of Represents subsequence data The mean of .

[0092] Optionally, the information gain of the building material price evolution subsequence is calculated in step S3, including:

[0093] Calculate the information gain of the building material price evolution subsequence, where the building material price evolution subsequence The information gain calculation process is as follows:

[0094] S31: Subsequence of building materials price evolution Perform sequence reconstruction to obtain a set of reconstructed sequences :

[0095] ;

[0096] ;

[0097] ;

[0098] in:

[0099] Represents a set of reconstructed sequences The reconstruction sequence of the space dimension is t, q represents the space dimension parameter, Reconstruction sequence The mean parameter of

[0100] S32: Calculate the similarity of any two sets of reconstructed sequences, where the reconstructed sequence and The similarity between them is:

[0101] ;

[0102] ;

[0103] in:

[0104] Reconstruction sequence and The distance between ;

[0105] Reconstruction sequence and The similarity between

[0106] represents the distance control parameter, Represents the similarity control parameter, set ;

[0107] S33: Calculate the building materials price evolution subsequence Information gain:

[0108] ;

[0109] ;

[0110] in:

[0111] Represents the building materials price evolution subsequence The information gain of Represents an intermediate process function;

[0112] The weight of the building material price evolution subsequence is generated based on the information gain of the building material price evolution subsequence, and the building material price evolution subsequence is weighted.

[0113] Optionally, the step of generating the weight of the building material price evolution subsequence and weighting the building material price evolution subsequence includes:

[0114] Building materials price evolution subsequence The weight calculation formula is:

[0115] ;

[0116] in:

[0117] Represents the building materials price evolution subsequence The weight of

[0118] Building materials price evolution subsequence The corresponding weighted building materials price evolution subsequence is:

[0119] ;

[0120] in:

[0121] Represents the building materials price evolution subsequence The corresponding weighted building materials price evolution subsequence.

[0122] Optionally, constructing a building materials price prediction model in step S4 includes:

[0123] Constructing a building materials price prediction model, wherein the building materials price prediction model takes weighted building materials price evolution subsequences under different scale conditions as input and takes building materials price prediction values ​​as output, wherein bidirectional LSTM is an implementation method of the price prediction model;

[0124] The building materials price prediction model includes an input layer, a forward information extraction layer, a backward information extraction layer, an information fusion layer, and a prediction calculation layer;

[0125] The input layer is used to receive weighted building material price evolution subsequences under different scale conditions;

[0126] The forward information extraction layer is a long short-term memory unit in a bidirectional long short-term memory network, which is used to extract forward information according to the order of the forward data values ​​of the weighted building material price evolution subsequence;

[0127] The backward information extraction layer is a long short-term memory unit in a bidirectional long short-term memory network, which is used to extract backward information according to the reverse data value order of the weighted building material price evolution subsequence;

[0128] The information fusion layer is used to fuse the forward information and backward information of the multi-scale weighted building material price evolution subsequence to obtain the building material price fusion information;

[0129] The prediction calculation layer is a fully connected layer structure in the neural network, which is used to map the building materials price fusion information into the building materials price prediction value.

[0130] Optionally, in step S5, the weighted building material price evolution subsequences under different scale conditions are predicted using the building material price prediction model to obtain the building material price prediction value, including:

[0131] The building materials price prediction model is used to predict the weighted building materials price evolution subsequences under different scale conditions, where the prediction process is as follows:

[0132] S51: The input layer receives the weighted building material price evolution subsequences under different scale conditions;

[0133] S52: The forward information extraction layer extracts forward information according to the order of the forward data values ​​of the weighted building material price evolution subsequence, where the weighted building material price evolution subsequence The forward information extraction formula is:

[0134] ;

[0135] ;

[0136] in:

[0137] Represents the weighted building materials price evolution subsequence The nth data value in ;

[0138] Represents the weighted building materials price evolution subsequence forward information;

[0139] represents the activation function;

[0140] Represents the forward convolution matrix parameters;

[0141] S53: The backward information extraction layer extracts the backward information according to the reverse data value order of the weighted building material price evolution subsequence, where the weighted building material price evolution subsequence The backward information extraction formula is:

[0142] ;

[0143] ;

[0144] in:

[0145] Represents the weighted building materials price evolution subsequence backward information;

[0146] Represents the backward convolution matrix parameters;

[0147] S54: The information fusion layer fuses the forward information and backward information of the multi-scale weighted building material price evolution subsequence to obtain the building material price fusion information :

[0148] ;

[0149] in:

[0150] Represents the fused convolution matrix parameters;

[0151] S55: The prediction calculation layer integrates building material prices with information Mapped to building materials price forecast :

[0152] ;

[0153] in:

[0154] Represents the mapping matrix parameters;

[0155] represents the price control parameter;

[0156] The project cost is evaluated based on the predicted price of building materials, where the project cost is the sum of the number of building materials and the predicted price of building materials. The product of .

[0157] Optionally, in step S21, the number of scales for scale decomposition is determined by using a group optimization method. And the penalty parameter ,include:

[0158] S211: Initialize and generate H groups of position coordinates, where each group of position coordinates includes the number of scales and a penalty parameter. The hth group of position coordinates generated by initialization is ,in The corresponding scale number, Corresponding penalty parameter;

[0159] And initialize the speed parameters for generating each set of position coordinates, where the position coordinates The corresponding speed parameter is ;

[0160] S212: Set the current iteration number of the position coordinates and speed parameters to , the maximum number of iterations is Max, then the hth group of position coordinates The result of the iteration is ;

[0161] S213: Construct the fitness function of the position coordinates:

[0162] ;

[0163] in:

[0164] Represents location-based coordinates After decomposing the scale, the sequence mean of the b-th group of material price evolution subsequences is obtained. Represents location-based coordinates Perform scale decomposition and obtain the mean of the sequence after Hilbert demodulation of the obtained b-th group of material price evolution subsequences;

[0165] represents the fitness function, Indicates location coordinates The corresponding fitness function value;

[0166] S214: The H sets of position coordinates obtained in the iteration are used as the input of the fitness function, and the position coordinate with the minimum fitness function value is used as the first The optimal position coordinates after iterations ;

[0167] S215: Iterate the speed parameters and position coordinates in sequence, wherein the iteration formula is:

[0168] ;

[0169] ;

[0170] in:

[0171] represents the iteration weight parameter;

[0172] S216: Order , until the maximum number of iterations is reached, and the H groups of position coordinates obtained by the iteration at this time are used as the input of the fitness function, and the position coordinates with the smallest fitness function value are decomposed into the number of scales And the penalty parameter .

[0173] In order to solve the above problem, the present invention provides an electronic device, the electronic device comprising:

[0174] A memory storing at least one instruction;

[0175] Communication interface, enabling electronic equipment to communicate; and

[0176] The processor executes the instructions stored in the memory to implement the above-mentioned intelligent engineering cost assessment method based on big data analysis.

[0177] In order to solve the above problems, the present invention also provides a computer-readable storage medium, in which at least one instruction is stored. The at least one instruction is executed by a processor in an electronic device to implement the above-mentioned intelligent engineering cost assessment method based on big data analysis.

[0178] Compared with the prior art, the present invention proposes an intelligent engineering cost assessment method based on big data analysis, which has the following advantages:

[0179] Firstly, this scheme collects building materials transaction prices and building materials price comments at different historical moments, performs multi-level pooling representation on building materials price comments, captures the multi-scale features of the comment text at the word level, sentence level and overall text scale, obtains richer and more diverse text feature representation, and improves the accuracy of sentiment analysis. The sequence data describing the sentiment evolution of price comments and the price sequence data are combined to form the building materials price evolution sequence data, and combined with the spectral representation of the sequence data, the building materials price evolution sequence data is subjected to multi-scale adaptive decomposition processing, and building materials price evolution subsequences of different frequencies are obtained as the sequence decomposition results under different scale conditions.

[0180] At the same time, this scheme calculates the complexity of the building materials price evolution subsequence as information gain, where the higher the complexity, the more effective price information contained in the subsequence. By converting the information gain into sequence weights, weighted processing of the subsequences is achieved, and the building materials price prediction model is used to perform forward information extraction and backward information extraction on the weighted building materials price evolution subsequences, thereby capturing the complex global dependencies of price information, and more comprehensively understanding the contextual relationship of price information. The engineering cost is evaluated based on the predicted value of building materials prices, thereby realizing engineering cost evaluation prediction combining price review data and price sequences. BRIEF DESCRIPTION OF THE DRAWINGS

[0181] Figure 1 A schematic diagram of a process flow of an intelligent engineering cost assessment method based on big data analysis provided by an embodiment of the present invention;

[0182] Figure 2 A schematic diagram of the structure of an electronic device for implementing an intelligent engineering cost assessment method based on big data analysis provided by an embodiment of the present invention.

[0183] In the figure: 1 electronic device, 10 processor, 11 memory, 12 program, 13 communication interface.

[0184] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0185] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0186] The embodiment of the present application provides an intelligent evaluation method for engineering costs based on big data analysis. The execution subject of the intelligent evaluation method for engineering costs based on big data analysis includes but is not limited to at least one of the electronic devices such as a server and a terminal that can be configured to execute the method provided by the embodiment of the present application. In other words, the intelligent evaluation method for engineering costs based on big data analysis can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc.

[0187] Embodiment 1:

[0188] S1: Collect the historical transaction prices of building materials and text data of building materials price comments of the project, perform sentiment analysis on the collected text data of building materials price comments to obtain building materials price sentiment sequence data, which together with the historical transaction price data of building materials constitute the building materials price evolution sequence data.

[0189] The historical transaction prices of building materials and building materials price comment text data of the project collected in step S1 include:

[0190] Collect the historical transaction prices and price review text data of building materials for the project:

[0191] ;

[0192] ;

[0193] in:

[0194] Represents the historical transaction prices of building materials and building materials price comment text data of the collected projects, represents the historical transaction price of building materials and the text data of building materials price comments at the nth historical moment, where N represents the total number of historical moments of the collected data;

[0195] Indicates the historical transaction price of building materials at the nth historical moment, Represents the text data of building material price comments at the nth historical moment, where the text data of building material price comments is the building material price comments of the M groups with the most likes. Indicates the price review of the mth component at the nth historical moment;

[0196] Perform sentiment analysis on the collected building materials price review text data to obtain building materials price sentiment sequence data :

[0197] ;

[0198] in:

[0199] represents the price sentiment of building materials at the nth historical moment;

[0200] Building materials price sentiment series data Together with the historical transaction price data of building materials, it constitutes the building materials price evolution sequence data:

[0201] ;

[0202] ;

[0203] ;

[0204] in:

[0205] Represents the building materials price evolution series data;

[0206] represents the price series data of building materials, Represents the sentiment evolution sequence data corresponding to the building materials price series data.

[0207] The sentiment analysis of the collected building material price comment text data to obtain building material price sentiment sequence data includes:

[0208] The sentiment analysis of the collected building materials price review text data is performed to obtain the building materials price sentiment sequence data, where the sentiment analysis process is as follows:

[0209] S11: Perform word vector encoding processing on any building material price comment in the building material price comment text data to form a word vector sequence corresponding to the building material price comment, where the building material price comment The corresponding word vector sequence is ; In the embodiment of the present invention, the BERT model is the model used for generating word vector sequences;

[0210] S12: Perform nonlinear dimensionality reduction processing on the word vector sequence to obtain the word vector dimensionality reduction sequence corresponding to the building materials price review, where the building materials price review The corresponding word vector dimensionality reduction sequence is , word vector sequence The nonlinear dimensionality reduction process is:

[0211] S121: Select word vector sequences from all word vector sequences The K nearest neighbor vectors of , where the nearest neighbor vector is represented as a distance word vector sequence The most recent sequence of word vectors;

[0212] S122: Initialize word vector sequence The mapping weight sequence is constructed with the minimum mapping difference as the goal, and the objective function is solved to obtain the word vector sequence The mapping weight sequence of , where the objective function is:

[0213] ;

[0214] in:

[0215] Representing word vector sequence The mapping weights corresponding to the kth nearest neighbor vector of the word constitute the word vector sequence. The mapping weight sequence, where the constraints corresponding to the mapping weights are ;

[0216] represents the L1 norm;

[0217] Representing word vector sequence The kth nearest neighbor vector of ;

[0218] S123: Set the dimension of the word vector reduction sequence to dim, according to the word vector sequence The mapping weight sequence of word vectors is constructed with the goal of minimizing the mapping difference of the word vector reduction sequence, and the objective function is solved to obtain the word vector sequence. The dimension of the word vector is dim. , the constructed objective function is:

[0219] ;

[0220] S13: Use the pooling window with a step size of stride to perform average pooling and maximum pooling on the word vector dimensionality reduction sequence, and obtain the average pooling result and maximum pooling result corresponding to the word vector dimensionality reduction sequence, where the word vector dimensionality reduction sequence The corresponding average pooling result is , the maximum pooling result is ;

[0221] S14: Calculate the price sentiment of building materials at different historical moments, where the price sentiment of building materials at the nth historical moment is:

[0222] ;

[0223] in:

[0224] represents the price sentiment of building materials at the nth historical moment, The larger it is, the greater the proportion of positive sentiment in building materials price sentiment;

[0225] represents an exponential function with a natural constant as base;

[0226] Represents the convolution matrix corresponding to emotion i, , Indicates negative, neutral, and positive emotions, respectively;

[0227] represents the L2 norm.

[0228] S2: Perform multi-scale decomposition on the building materials price evolution sequence data to obtain building materials price evolution sub-sequences under different scale conditions.

[0229] In the step S2, the building materials price evolution sequence data is decomposed at multiple scales, including:

[0230] The building materials price evolution sequence data P is subjected to multi-scale decomposition, wherein the variational mode decomposition based on swarm optimization is the implementation method of the multi-scale decomposition, wherein the multi-scale decomposition process is:

[0231] S21: Determine the number of scales for scale decomposition using group optimization And the penalty parameter ;

[0232] S22: Perform Fourier transform on the building materials price evolution sequence data P to obtain the building materials price sequence data And emotional evolution sequence data The frequency domain representation of is:

[0233] ;

[0234] ;

[0235] in:

[0236] Represents building materials price series data Frequency domain representation of the Fourier transform point number e;

[0237] Representing emotional evolution sequence data Frequency domain representation of the Fourier transform point number e;

[0238] j represents the imaginary unit;

[0239] S23: Calculate the building materials price series data And emotional evolution sequence data Center frequency:

[0240] ;

[0241] ;

[0242] ;

[0243] in:

[0244] Represents building materials price series data The center frequency, Representing emotional evolution sequence data The center frequency of

[0245] Represents the time interval between adjacent historical moments; The frequency axis represents the number of Fourier transform points e;

[0246] S24: Set the current building materials price series data And emotional evolution sequence data The number of subsequences obtained by segmentation is u, the initial value of u is 0, and the maximum value is U-1;

[0247] Initialize and generate building materials price series data The corresponding u-th group of subsequence data , emotion evolution sequence data The corresponding u-th group of subsequence data , and the iterative coefficient sequence ;

[0248] S25: Set the current number of iterations of the sequence to d, the initial value of d is 0, then the subsequence data And the iterative coefficient sequence The d-th iteration results are ;

[0249] S26: Perform Fourier transform on the subsequence data and the iterative coefficient sequence, and iterate the spectrum representation corresponding to the subsequence data, wherein the subsequence data The iterative formula corresponding to the spectrum representation is:

[0250] ;

[0251] in:

[0252] Represents subsequence data The spectrum at Fourier transform point number e represents the iteration result;

[0253] represents the spectrum representation of the i-th group of subsequence data obtained by segmentation at Fourier transform point number e;

[0254] Represents the iterative coefficient sequence Spectral representation at Fourier transform point number e;

[0255] Represents subsequence data The d-th iteration result of the center frequency of For subsequence data The center frequency of

[0256] Iterate the frequency spectrum representation of the center frequency associated with the subsequence data and the iterative coefficient sequence, where the subsequence data The center frequency of the correlation And the spectrum representation of the iterative coefficient sequence The iteration formula is:

[0257] ;

[0258] ;

[0259] in:

[0260] represents the differential of the Fourier transform point number e;

[0261] S27: If the iterative result of the spectrum representation converges, perform inverse Fourier transform on the iterative result of the spectrum representation of the subsequence data to obtain the u-th group of subsequence data of the sequence data, and set u=u+1, return to step S25, until the U-th group of subsequence data is obtained; otherwise, set d=d+1, return to step S26; if the iterative result of the spectrum representation converges If the following equation is satisfied, it means convergence:

[0262] ;

[0263] in:

[0264] Indicates the preset convergence threshold;

[0265] S28: Construct U sets of material price evolution subsequences, where the uth set of material price evolution subsequence for:

[0266] ;

[0267] ;

[0268] in:

[0269] Represents the building materials price evolution subsequence The nth data value in Represents subsequence data The nth data value of Represents subsequence data The nth data value of ;

[0270] Represents subsequence data The standard deviation of Represents subsequence data The mean of .

[0271] S3: Calculate the information gain of the building material price evolution subsequence, generate the weight of the building material price evolution subsequence, and weight the building material price evolution subsequence.

[0272] The information gain of the building material price evolution subsequence is calculated in step S3, including:

[0273] Calculate the information gain of the building material price evolution subsequence, where the building material price evolution subsequence The information gain calculation process is as follows:

[0274] S31: Subsequence of building materials price evolution Perform sequence reconstruction to obtain a set of reconstructed sequences :

[0275] ;

[0276] ;

[0277] ;

[0278] in:

[0279] Represents a set of reconstructed sequences The reconstruction sequence of the space dimension is t, q represents the space dimension parameter, Reconstruction sequence The mean parameter of

[0280] S32: Calculate the similarity of any two sets of reconstructed sequences, where the reconstructed sequence and The similarity between them is:

[0281] ;

[0282] in:

[0283] Reconstruction sequence and The distance between ;

[0284] Reconstruction sequence and The similarity between

[0285] represents the distance control parameter, Represents the similarity control parameter, set ;

[0286] S33: Calculate the building materials price evolution subsequence Information gain:

[0287] ;

[0288] ;

[0289] in:

[0290] Represents the building materials price evolution subsequence The information gain of Represents an intermediate process function;

[0291] The weight of the building material price evolution subsequence is generated based on the information gain of the building material price evolution subsequence, and the building material price evolution subsequence is weighted.

[0292] The step of generating the weight of the building material price evolution subsequence and weighting the building material price evolution subsequence includes:

[0293] Building materials price evolution subsequence The weight calculation formula is:

[0294] ;

[0295] in:

[0296] Represents the building materials price evolution subsequence The weight of

[0297] Building materials price evolution subsequence The corresponding weighted building materials price evolution subsequence is:

[0298] ;

[0299] in:

[0300] Represents the building materials price evolution subsequence The corresponding weighted building materials price evolution subsequence.

[0301] S4: Construct a building material price prediction model, wherein the building material price prediction model takes weighted building material price evolution subsequences under different scale conditions as input and takes building material price prediction values ​​as output.

[0302] The construction material price prediction model is constructed in step S4, including:

[0303] Constructing a building materials price prediction model, wherein the building materials price prediction model takes weighted building materials price evolution subsequences under different scale conditions as input and takes building materials price prediction values ​​as output, wherein bidirectional LSTM is an implementation method of the price prediction model;

[0304] The building materials price prediction model includes an input layer, a forward information extraction layer, a backward information extraction layer, an information fusion layer, and a prediction calculation layer;

[0305] The input layer is used to receive weighted building material price evolution subsequences under different scale conditions;

[0306] The forward information extraction layer is a long short-term memory unit in a bidirectional long short-term memory network, which is used to extract forward information according to the order of the forward data values ​​of the weighted building material price evolution subsequence;

[0307] The backward information extraction layer is a long short-term memory unit in a bidirectional long short-term memory network, which is used to extract backward information according to the reverse data value order of the weighted building material price evolution subsequence;

[0308] The information fusion layer is used to fuse the forward information and backward information of the multi-scale weighted building material price evolution subsequence to obtain the building material price fusion information;

[0309] The prediction calculation layer is a fully connected layer structure in the neural network, which is used to map the building materials price fusion information into the building materials price prediction value.

[0310] S5: Use the building materials price prediction model to predict the weighted building materials price evolution subsequences under different scale conditions to obtain the building materials price prediction value, and conduct engineering cost evaluation based on the building materials price prediction value.

[0311] In the step S5, the weighted building material price evolution subsequences under different scale conditions are predicted using the building material price prediction model to obtain the building material price prediction value, including:

[0312] The building materials price prediction model is used to predict the weighted building materials price evolution subsequences under different scale conditions, where the prediction process is as follows:

[0313] S51: The input layer receives the weighted building material price evolution subsequences under different scale conditions;

[0314] S52: The forward information extraction layer extracts forward information according to the order of the forward data values ​​of the weighted building material price evolution subsequence, where the weighted building material price evolution subsequence The forward information extraction formula is:

[0315] ;

[0316] ;

[0317] in:

[0318] Represents the weighted building materials price evolution subsequence The nth data value in ;

[0319] Represents the weighted building materials price evolution subsequence forward information;

[0320] represents the activation function;

[0321] Represents the forward convolution matrix parameters;

[0322] S53: The backward information extraction layer extracts the backward information according to the reverse data value order of the weighted building material price evolution subsequence, where the weighted building material price evolution subsequence The backward information extraction formula is:

[0323] ;

[0324] ;

[0325] in:

[0326] Represents the weighted building materials price evolution subsequence backward information;

[0327] Represents the backward convolution matrix parameters;

[0328] S54: The information fusion layer fuses the forward information and backward information of the multi-scale weighted building material price evolution subsequence to obtain the building material price fusion information :

[0329] ;

[0330] in:

[0331] Represents the fused convolution matrix parameters;

[0332] S55: The prediction calculation layer integrates building material prices with information Mapped to building materials price forecast :

[0333] ;

[0334] in:

[0335] Represents the mapping matrix parameters;

[0336] represents the price control parameter;

[0337] The project cost is evaluated based on the predicted price of building materials, where the project cost is the sum of the number of building materials and the predicted price of building materials. The product of .

[0338] Embodiment 2:

[0339] like Figure 2 , is a schematic diagram of the structure of an electronic device for implementing an intelligent engineering cost assessment method based on big data analysis provided by an embodiment of the present invention.

[0340] The electronic device 1 may include a processor 10 , a memory 11 , a communication interface 13 and a bus, and may also include a computer program stored in the memory 11 and executable on the processor 10 , such as a program 12 .

[0341] The memory 11 includes at least one type of readable storage medium, and the readable storage medium includes a flash memory, a mobile hard disk, a multimedia card, a card-type memory (e.g., SD or DX memory, etc.), a magnetic memory, a disk, an optical disk, etc. In some embodiments, the memory 11 may be an internal storage unit of the electronic device 1, such as a mobile hard disk of the electronic device 1. In other embodiments, the memory 11 may also be an external storage device of the electronic device 1, such as a plug-in mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device 1. Further, the memory 11 may also include both an internal storage unit of the electronic device 1 and an external storage device. The memory 11 may not only be used to store application software and various types of data installed in the electronic device 1, such as the code of the program 12, etc., but also be used to temporarily store data that has been output or is to be output.

[0342] The processor 10 may be composed of an integrated circuit in some embodiments, for example, a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and combinations of various control chips, etc. The processor 10 is the control core (ControlUnit) of the electronic device, and uses various interfaces and lines to connect various components of the entire electronic device, and executes or executes programs or modules stored in the memory 11 (such as a program 12 for realizing intelligent evaluation of engineering costs based on big data analysis), and calls data stored in the memory 11, so as to execute various functions of the electronic device 1 and process data.

[0343] The communication interface 13 may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device 1 and other electronic devices, and to achieve connection and communication between internal components of the electronic device.

[0344] The bus may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. The bus is configured to realize connection and communication between the memory 11 and at least one processor 10, etc.

[0345] Figure 2 Only an electronic device with components is shown, and those skilled in the art will understand that Figure 2 The structure shown does not constitute a limitation on the electronic device 1, and may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0346] For example, although not shown, the electronic device 1 may also include a power source (such as a battery) for supplying power to various components. Preferably, the power source may be logically connected to the at least one processor 10 through a power management device, so that the power management device can realize functions such as charging management, discharging management, and power consumption management. The power source may also include one or more DC or AC power sources, recharging devices, power failure detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device 1 may also include a variety of sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be repeated here.

[0347] Optionally, the electronic device 1 may further include a user interface, which may be a display, an input unit (such as a keyboard), or a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, and an OLED (Organic Light-Emitting Diode) touch device. The display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the electronic device 1 and to display a visual user interface.

[0348] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited to this structure.

[0349] The program 12 stored in the memory 11 in the electronic device 1 is a combination of multiple instructions. When running in the processor 10, it can achieve:

[0350] Collect the historical transaction prices of building materials and text data of building materials price comments of the project, perform sentiment analysis on the collected text data of building materials price comments to obtain building materials price sentiment sequence data, and together with the historical transaction price data of building materials, form the building materials price evolution sequence data;

[0351] Perform multi-scale decomposition on the building materials price evolution sequence data to obtain the building materials price evolution subsequences under different scale conditions;

[0352] Calculate the information gain of the building material price evolution subsequence, generate the weight of the building material price evolution subsequence, and weight the building material price evolution subsequence;

[0353] The building materials price prediction model is used to predict the weighted building materials price evolution subsequences under different scale conditions to obtain the building materials price prediction values, and the project cost is evaluated based on the building materials price prediction values.

[0354] Specifically, the specific implementation method of the processor 10 for the above instructions can refer to Figure 1 to Figure 2 The description of the relevant steps in the corresponding embodiments will not be repeated here.

[0355] It should be noted that the serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. And the terms "including", "comprising" or any other variants thereof in this article are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "including a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0356] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0357] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. An intelligent engineering cost assessment method based on big data analysis, characterized in that: The method comprises: S1: Collect the historical transaction prices of building materials and text data of building materials price comments of the project, perform sentiment analysis on the collected text data of building materials price comments to obtain building materials price sentiment sequence data, and together with the historical transaction price data of building materials, form the building materials price evolution sequence data; Collect the historical transaction prices of building materials and text data of building materials price comments for the project, including: data=(data1,data2,...,data n ,...,data N ) in: data represents the historical transaction price of building materials and building materials price comment text data of the collected project. n represents the historical transaction price of building materials and the text data of building materials price comments at the nth historical moment, where N represents the total number of historical moments of the collected data; v n Indicates the historical transaction price of building materials at the nth historical moment, Represents the text data of building material price comments at the nth historical moment, where the text data of building material price comments is the building material price comments of the M groups with the most likes. Indicates the price review of the mth component at the nth historical moment; The sentiment analysis of the collected building materials price review text data is performed to obtain the building materials price sentiment sequence data R: R＝(r1,r2,...,r n ,...,r N ) in: r n represents the price sentiment of building materials at the nth historical moment; The building materials price sentiment series data R and the building materials historical transaction price data together constitute the building materials price evolution series data: P = [P1; P2] P1=(v1,v2,...,v n ,...,v N ) P2=R in: P represents the building materials price evolution series data; P1 represents the building materials price series data, and P2 represents the sentiment evolution series data corresponding to the building materials price series data; The sentiment analysis process is: S11: Perform word vector encoding processing on any building material price comment in the building material price comment text data to form a word vector sequence corresponding to the building material price comment, where the building material price comment The corresponding word vector sequence is S12: Perform nonlinear dimensionality reduction processing on the word vector sequence to obtain the word vector dimensionality reduction sequence corresponding to the building materials price review, where the building materials price review The corresponding word vector dimensionality reduction sequence is Word vector sequence The nonlinear dimensionality reduction process is: S121: Select word vector sequences from all word vector sequences The K nearest neighbor vectors of , where the nearest neighbor vector is represented as a distance word vector sequence The most recent sequence of word vectors; S122: Initialize word vector sequence The mapping weight sequence is constructed with the minimum mapping difference as the goal, and the objective function is solved to obtain the word vector sequence The mapping weight sequence of , where the objective function is: in: Representing word vector sequence The mapping weights corresponding to the kth nearest neighbor vector of the word constitute the word vector sequence. The mapping weight sequence, where the constraints corresponding to the mapping weights are ||·|| represents the L1 norm; Representing word vector sequence The kth nearest neighbor vector of ; S123: Set the dimension of the word vector reduction sequence to dim, according to the word vector sequence The mapping weight sequence of word vectors is constructed with the goal of minimizing the mapping difference of the word vector reduction sequence, and the objective function is solved to obtain the word vector sequence. The dimension of the word vector is dim. The constructed objective function is: S13: Use the pooling window with a step size of stride to perform average pooling and maximum pooling on the word vector dimensionality reduction sequence, and obtain the average pooling result and maximum pooling result corresponding to the word vector dimensionality reduction sequence, where the word vector dimensionality reduction sequence The corresponding average pooling result is The maximum pooling result is S14: Calculate the price sentiment of building materials at different historical moments, where the price sentiment of building materials at the nth historical moment is: in: r n represents the building materials price sentiment at the nth historical moment, r n The larger it is, the greater the proportion of positive sentiment in building materials price sentiment; exp(·) represents an exponential function with a natural constant as the base; W i Represents the convolution matrix corresponding to emotion i, i∈{-1,0,1}, where -1,0,1 represent negative, neutral, and positive emotions respectively; ||·||2 represents the L2 norm; S2: Perform multi-scale decomposition on the building materials price evolution sequence data to obtain the building materials price evolution subsequences under different scale conditions; S3: Calculate the information gain of the building material price evolution subsequence, generate the weight of the building material price evolution subsequence, and weight the building material price evolution subsequence; S4: constructing a building materials price prediction model, wherein the building materials price prediction model takes weighted building materials price evolution subsequences under different scale conditions as input and takes building materials price prediction values ​​as output; S5: Use the building materials price prediction model to predict the weighted building materials price evolution subsequences under different scale conditions to obtain the building materials price prediction value, and conduct engineering cost evaluation based on the building materials price prediction value.

2. The method for intelligent engineering cost assessment based on big data analysis according to claim 1, characterized in that: In the step S2, the building materials price evolution sequence data is decomposed at multiple scales, including: The building materials price evolution sequence data P is decomposed at multiple scales, and the multi-scale decomposition process is as follows: S21: Determine the number of scales U and penalty parameter τ of scale decomposition by using group optimization; S22: Perform Fourier transform processing on the building material price evolution sequence data P to obtain the frequency domain representation of the building material price sequence data P1 and the emotion evolution sequence data P2: in: Y1(e) represents the frequency domain representation of the building materials price series data P1 at Fourier transform point e; Y2(e) represents the frequency domain representation of the emotion evolution sequence data P2 at the Fourier transform point number e; j represents the imaginary unit; S23: Calculate the central frequencies of the building materials price series data P1 and the sentiment evolution series data P2: in: f1 represents the central frequency of the building materials price series data P1, and f2 represents the central frequency of the sentiment evolution series data P2; Δ represents the time interval between adjacent historical moments; f1(e) represents the frequency axis of the Fourier transform point number e; S24: Set the number of subsequences obtained by dividing the current building material price sequence data P1 and the emotion evolution sequence data P2 to u, where the initial value of u is 0 and the maximum value is U-1; Initialize and generate the u-th group of subsequence data P1(u) corresponding to the building materials price sequence data P1, the u-th group of subsequence data P2(u) corresponding to the emotion evolution sequence data P2, and the iteration coefficient sequences s1(u), s2(u); S25: Set the current iteration number of the sequence to d, the initial value of d is 0, then the d-th iteration results of the subsequence data P1(u), P2(u) and the iteration coefficient sequence s1(u), s2(u) are S26: Performing Fourier transform on the subsequence data and the iteration coefficient sequence, and iterating the frequency spectrum representation corresponding to the subsequence data; S27: If the iterative result of the spectrum representation converges, then perform inverse Fourier transform on the iterative result of the spectrum representation of the subsequence data to obtain the u-th group of subsequence data of the sequence data, and set u = u + 1, return to step S25, until the U-th group of subsequence data is obtained; otherwise, set d = d + 1, return to step S26; if the iterative result of the spectrum representation converges If the following equation is satisfied, it means convergence: in: ε represents the preset convergence threshold; S28: Construct U sets of material price evolution subsequences, where the u-th set of material price evolution subsequence P(u) is: P(u)=(P(u,1),P(u,2),...,P(u,n),...,P(u,N)) in: P(u,n) represents the nth data value in the building materials price evolution subsequence P(u), P1(u,n) represents the nth data value of the subsequence data P1(u), and P2(u,n) represents the nth data value of the subsequence data P2(u); represents the standard deviation of the subsequence data P2(u), Represents the mean of the subsequence data P2(u).

3. The method for intelligent engineering cost assessment based on big data analysis according to claim 2, characterized in that: The information gain of the building material price evolution subsequence is calculated in step S3, including: Calculate the information gain of the building material price evolution subsequence, where the information gain calculation process of the building material price evolution subsequence P(u) is: S31: Reconstruct the building materials price evolution subsequence P(u) to obtain the reconstructed sequence set Q: Q={Q(t)|t∈[1,N-q+1]} Q(t)=(P(u,t),P(u,t+1),...,P(u,t+q)μ(t)) in: Q(t) represents the reconstructed sequence with a spatial dimension of t in the reconstructed sequence set Q, q represents the spatial dimension parameter, and μ(t) represents the mean parameter of the reconstructed sequence Q(t); S32: Calculate the similarity between any two groups of reconstructed sequences, where the similarity between the reconstructed sequences Q(t) and Q(z) is: dis(Q(t),Q(z))=max c∈[0,q-1] |P(u,t+c)-μ(t)-(P(u,z+c)-μ(j))|,z∈[1,N-q+1] in: dis(Q(t),Q(z)) represents the distance between the reconstructed sequences Q(t) and Q(z), z≠t; Sim(Q(t),Q(z)) represents the similarity between the reconstructed sequences Q(t) and Q(z); α1 represents the distance control parameter, α2 represents the similarity control parameter, and α2=2 is set; S33: Calculate the information gain of the building materials price evolution subsequence P(u): in: I u represents the information gain of the building materials price evolution subsequence P(u), Represents an intermediate process function; The weight of the building material price evolution subsequence is generated based on the information gain of the building material price evolution subsequence, and the building material price evolution subsequence is weighted.

4. The method for intelligent engineering cost assessment based on big data analysis as claimed in claim 3, characterized in that: The step of generating the weight of the building material price evolution subsequence and weighting the building material price evolution subsequence includes: The weight calculation formula of the building materials price evolution subsequence P(u) is: in: w(u) represents the weight of the building materials price evolution subsequence P(u); The weighted building material price evolution subsequence corresponding to the building material price evolution subsequence P(u) is: L(u)=w(u)P(u) in: L(u) represents the weighted building material price evolution subsequence corresponding to the building material price evolution subsequence P(u).

5. The method for intelligent engineering cost assessment based on big data analysis according to claim 1, characterized in that: The construction material price prediction model is constructed in step S4, including: Constructing a building material price prediction model, wherein the building material price prediction model takes weighted building material price evolution subsequences under different scale conditions as input and takes building material price prediction values ​​as output; The building materials price prediction model includes an input layer, a forward information extraction layer, a backward information extraction layer, an information fusion layer, and a prediction calculation layer; The input layer is used to receive weighted building material price evolution subsequences under different scale conditions; The forward information extraction layer is a long short-term memory unit in a bidirectional long short-term memory network, which is used to extract forward information according to the order of the forward data values ​​of the weighted building material price evolution subsequence; The backward information extraction layer is a long short-term memory unit in a bidirectional long short-term memory network, which is used to extract backward information according to the reverse data value order of the weighted building material price evolution subsequence; The information fusion layer is used to fuse the forward information and backward information of the multi-scale weighted building material price evolution subsequence to obtain the building material price fusion information; The prediction calculation layer is a fully connected layer structure in the neural network, which is used to map the building materials price fusion information into the building materials price prediction value.

6. The method for intelligent engineering cost assessment based on big data analysis according to claim 5, characterized in that: In the step S5, the weighted building material price evolution subsequences under different scale conditions are predicted using the building material price prediction model to obtain the building material price prediction value, including: The building materials price prediction model is used to predict the weighted building materials price evolution subsequences under different scale conditions, where the prediction process is as follows: S51: The input layer receives the weighted building material price evolution subsequences under different scale conditions; S52: The forward information extraction layer extracts forward information according to the order of the forward data values ​​of the weighted building material price evolution subsequence, wherein the forward information extraction formula of the weighted building material price evolution subsequence L(u) is: in: L(u,n) represents the nth data value in the weighted building materials price evolution subsequence L(u); Represents the forward information of the weighted building materials price evolution subsequence L(u); δ(·) represents the activation function; W 1 Represents the forward convolution matrix parameters; S53: The backward information extraction layer extracts backward information according to the reverse data value order of the weighted building material price evolution subsequence, wherein the backward information extraction formula of the weighted building material price evolution subsequence L(u) is: in: Represents the backward information of the weighted building materials price evolution subsequence L(u); W 2 Represents the backward convolution matrix parameters; S54: The information fusion layer fuses the forward information and backward information of the multi-scale weighted building material price evolution subsequence to obtain the building material price fusion information g: in: W 3 Represents the fused convolution matrix parameters; S55: The prediction calculation layer maps the building material price fusion information g into the building material price prediction value in: W 4 Represents the mapping matrix parameters; represents the price control parameter; The project cost is evaluated based on the predicted price of building materials, where the project cost is the sum of the number of building materials and the predicted price of building materials. The product of .

7. The method for intelligent engineering cost assessment based on big data analysis according to claim 2, characterized in that: In the step S21, the number of scales U and the penalty parameter τ of the scale decomposition are determined by using a group optimization method, including: S211: Initialize and generate H groups of position coordinates, where each group of position coordinates includes the number of scales and a penalty parameter. The hth group of position coordinates generated by initialization is in The corresponding scale number, Corresponding penalty parameter; And initialize the speed parameters for generating each set of position coordinates, where the position coordinates The corresponding speed parameter is S212: Set the current iteration number of the position coordinates and speed parameters to γ, and the maximum iteration number to Max. Then the γth iteration result of the hth group of position coordinates is S213: Construct the fitness function of the position coordinates: in: Represents location-based coordinates After decomposing the scale, the sequence mean of the b-th group of material price evolution subsequences is obtained. Represents location-based coordinates Perform scale decomposition and obtain the mean of the sequence after Hilbert demodulation of the obtained b-th group of material price evolution subsequences; F(·) represents the fitness function, Indicates location coordinates The corresponding fitness function value; S214: Take the H group of position coordinates obtained in the γth iteration as the input of the fitness function, and take the position coordinate with the minimum fitness function value as the optimal position coordinate after the γth iteration S215: Iterate the speed parameters and position coordinates in sequence, wherein the iteration formula is: in: θ1, θ2, θ3, θ4 represent iterative weight parameters; S216: Let γ=γ+1 until the maximum number of iterations is reached, and use the H groups of position coordinates obtained by the iteration at this time as the input of the fitness function, and decompose the position coordinate with the minimum fitness function value into the scale number U and the penalty parameter τ.

Images