Intelligent construction site digital platform implementation method and system based on full life cycle management and medium

Abstract

The invention provides an intelligent construction site digital platform implementation method based on full-life-cycle management, and aims to realize full-life-cycle comprehensive digital management of a construction project by integrating advanced technologies such as cloud computing, big data, Internet of Things, mobile Internet and artificial intelligence. The platform comprises a data integration module, a real-time monitoring module, an intelligent decision-making module and a collaborative management module, can collect, store, analyze and apply various data in real time in each stage of project planning, design, construction, operation maintenance and the like, and supports intelligent decision-making and optimization. The specific application method comprises the steps of data acquisition and transmission, data processing and analysis, intelligent early warning and optimization decision making, multi-party collaboration and information sharing and the like. Through the platform, the management efficiency and safety of the construction site are remarkably improved, the construction cost and risk are greatly reduced, and the construction industry is promoted to develop in a more intelligent, green and efficient direction.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence, and in particular to a method for realizing a digital platform for a smart construction site based on full life cycle management. Background Art

[0002] With the rapid development of information technology, the construction industry is facing great opportunities and challenges in digital transformation. Traditional construction site management methods rely on manual operations and decentralized information systems, which leads to problems such as information islands, inconsistent data, and inefficient decision-making, which seriously affects the control of construction progress, quality, and safety risks. In order to improve the management efficiency and safety of construction sites, the construction industry urgently needs to introduce advanced information systems to promote digital management of the entire life cycle of projects.

[0003] In recent years, cutting-edge technologies such as cloud computing, the Internet of Things, big data, mobile Internet, and artificial intelligence have been gradually applied to the construction industry, promoting the construction of smart buildings. Smart buildings not only rely on automation technology, but also combine the Internet of Things, artificial intelligence, and data analysis technology to improve the energy efficiency, comfort, and safety of buildings. However, current smart building solutions often focus on the building operation stage and lack digital coverage of the entire life cycle of construction projects, especially in the information system integration during the project design and construction stages. Summary of the invention

[0004] The present invention aims to at least solve the technical problems existing in the prior art, and in particular innovatively proposes a method for implementing a digital platform for smart construction sites based on full life cycle management, including:

[0005] Distributed edge computing architecture: distributes data processing tasks to edge nodes at the construction site;

[0006] Adaptive fusion algorithm: integrates information from multiple sources to enhance data representativeness and decision-making accuracy;

[0007] Real-time processing and decision-making model: Combine state space model with real-time optimization algorithm to achieve dynamic monitoring and decision support for construction progress and safety risks;

[0008] Data transmission technology based on compressed sensing: recovering high-dimensional signals from a small amount of sampled data.

[0009] In a preferred embodiment of the present invention, the distributed edge computing architecture includes:

[0010] N edge nodes are deployed at the construction site. Each node i is responsible for processing data from multiple sensors and recording data including temperature, humidity, and construction progress:

[0011] V ij(t) = s V (t)+∈ V (t);

[0012] V ij (t) is the temperature data collected by sensor j of the i-th edge node at time t, s V (t) is the actual temperature value collected by the sensor at time t, ∈ V (t) is the error of temperature data;

[0013] U ij (t) = s U (t)+∈ U (t);

[0014] U ij (t) is the humidity data collected by sensor j of the i-th edge node at time t, s U (t) is the actual humidity value collected by the sensor at time t, ∈ U (t) is the error of humidity data;

[0015]

[0016] M ij (t) is the construction progress data collected by sensor j of the i-th edge node at time t, V complete (t) is the completed volume of the building obtained by image recognition and scanning, V total The total volume of the building to be completed;

[0017] Integrate the data:

[0018] D ij (t) = [V ij (t),U ij (t),M ij (t)];

[0019] {D i1 (t),D i2 (t),.....,D ij (t),......D im (t)};

[0020] Among them, D ij (t) is the data collected by sensor j at time t processed by edge node i, D im (t) is the data collected by sensor m at time t processed by edge node i;

[0021] The processing capacity determines the distribution of tasks among different edge nodes. When resources are limited, tasks can be distributed to nodes with stronger capabilities based on processing capacity to avoid system overload.

[0022] Among them, the processing capacity of each edge node can be expressed as:

[0023]

[0024] Among them, P i (t) is the processing capacity of edge node i at time t, which reflects the comprehensive processing capability of the node for sensor data;

[0025] m is the number of sensors processed by each edge node, α ij is the processing weight of edge node i on sensor j data, indicating the importance of the sensor data to the comprehensive processing capability, F ij (t) is the load of sensor j of edge node i at time t.

[0026] In a preferred embodiment of the present invention, the adaptive fusion algorithm comprises:

[0027] Assume that the data collected from each edge node is:

[0028] I(t)={I 1 (t),I 2 (t),.....,I k (t)};

[0029] Among them, k is the dimension of the integrated data, and the integrated data It can be expressed as:

[0030]

[0031] Among them, w l (t) is the weight of the lth dimension data at time t, reflecting the relative importance of different data dimensions in the comprehensive data and satisfying the normalization condition I l (t) is the l-th dimension data at time t;

[0032] Weight w l (t) Adaptive adjustment based on historical and real-time data through machine learning models:

[0033]

[0034] w(t) is the weight vector at time t, W is the model parameter matrix, b is the model parameter bias vector, It is a comprehensive data The feature extraction function of , Softmax(*) is used to ensure that the weights meet the normalization conditions.

[0035] In a preferred embodiment of the present invention, the real-time processing and decision-making model includes:

[0036] Construct a state space model to describe the evolution of the system state x(t):

[0037]

[0038] x(t+1) is the state vector of the system at time t+1, x(t) is the state vector of the system at time t, u(t) is the control input, and ζ(t) is the process noise, which indicates the instability within the system;

[0039] A is the system matrix, which describes the state transition relationship, B is the control input matrix, which describes the impact of the control input on the state, and C is the output matrix, which describes how the state is converted into the observed value;

[0040] y(t) is the observation vector, v(t) is the observation noise, which represents the uncertainty in the measurement process;

[0041] Based on this model, Kalman filtering is used for real-time state estimation:

[0042] Kalman filter prediction:

[0043]

[0044] is the state at time t predicted based on time t-1, is the state estimate at time t-1, and u(t-1) is the control input at time t-1;

[0045] Kalman filter prediction error covariance, reflecting the uncertainty of state prediction:

[0046] P(t|t-1)=AP(t-1|t-1)A T +Q;

[0047] P(t|t-1) is the error covariance at time t predicted based on time t-1, P(t-1|t-1) is the updated error covariance, Q is the process noise covariance matrix, and A T is the transpose of the system matrix;

[0048] By calculating the Kalman gain, the update method of the state estimate can be dynamically adjusted to minimize the estimation error:

[0049]

[0050] K(t) is the Kalman gain, which is used to update the state estimate, R is the observation noise covariance matrix, C T is the transpose of the output matrix;

[0051] Kalman filter update:

[0052]

[0053] is the updated state estimate, y(t) is the observation vector, is the predicted observation vector;

[0054] P(t|t)=(EK(t)C)P(t|t-1);

[0055] P(t|t) is the updated error covariance, and E is the identity matrix.

[0056] In a preferred embodiment of the present invention, the data transmission technology based on compressed sensing includes:

[0057] Assume the original data vector is Compression via Compressed Sensing:

[0058] g = ΦD;

[0059] in, is a compressed matrix, satisfying M<N, It is the compressed data;

[0060] At the receiving end, the original data is restored through the sparse reconstruction algorithm:

[0061]

[0062] is the reconstructed original data vector, Ensure that the reconstructed signal is consistent with the compressed observation, ‖*‖ 2 is the L2 norm, which is used to measure the error between the reconstructed signal and the observed signal, λ‖ΨD‖ 1 is a sparse regularization term that controls the sparsity of non-sparse data in the transform domain Ψ, λ is a regularization parameter, ‖*‖ 1 Indicates l 1 norm, used for sparse representation, g = ΦD is the constraint condition for compressed data;

[0063] In order to improve the recovery accuracy, deep learning methods are used to optimize the reconstruction process, and deep neural networks are used to reconstruct the signal, where the input of the network is compressed sensing sampling data;

[0064]

[0065] f θ (*) is the mapping function of the deep neural network with parameter θ, which can recover the signal from the compressed sampling data after training. The network is trained with the following loss function:

[0066] Training is performed by combining reconstruction error with sparsity constraints:

[0067]

[0068] in, is the loss function, which represents the training error of the model.

[0069] The present invention also discloses a computer system, comprising:

[0070] a memory for storing processor-executable instructions;

[0071] Among them, the processor is configured to implement the method for implementing the smart construction site digital platform based on full life cycle management when executing the executable instructions.

[0072] The present invention also discloses a computer-readable storage medium, comprising:

[0073] a memory having a computer program stored thereon;

[0074] A processor is used to execute the program in the memory to implement the method for implementing the smart construction site digital platform based on full life cycle management.

[0075] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0076] The digital platform of the present invention covers the entire life cycle of a construction project from planning, design, construction to operation and maintenance, ensuring seamless connection of data and management processes at each stage, avoiding information islands, and improving overall management efficiency.

[0077] Through advanced cloud computing and edge computing technologies, efficient integration and real-time processing of massive multi-source data at construction sites can be achieved, ensuring the accuracy and timeliness of the data and providing reliable data support for management decisions.

[0078] By combining intelligent algorithms with historical construction data, we have successfully achieved accurate prediction of construction progress. By combining data from different construction stages, we can predict the completion time of specific tasks within a 5% error. This improves the accuracy of project progress prediction, reduces additional costs caused by construction delays, and ensures that the project is delivered on time.

[0079] By using the Internet of Things technology and deploying 100 sensors to monitor the operating status of equipment, the system issued an early warning three hours before equipment failure occurred, preventing 10 potential equipment failure accidents. By implementing this early warning system, the project's equipment failure rate dropped by 15%, and the impact of failures on construction progress was reduced by 30%.

[0080] By analyzing the data flow in real time at the construction site, the spatial layout of the construction site and the safety arrangements of personnel were optimized. Using sensors and cloud computing technology to collect temperature and humidity data in the construction area, the system proposed targeted construction adjustment plans by analyzing these data, reducing the time construction workers spent in high-risk areas by 20%, effectively improving overall safety.

[0081] Through comprehensive multi-source data analysis, the system provides efficient decision support for project managers. Based on real-time data from the construction site, the system can automatically generate risk assessment reports and construction schedule adjustment suggestions. Through these intelligent decision-making supports, the budget overrun of a project was reduced by 12%, and the construction period was 10% ahead of schedule.

[0082] The digital platform of the present invention has good scalability and compatibility, can adapt to the needs of construction projects of different scales and complexities, supports the integration and upgrading of future technologies, and has broad application prospects.

[0083] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0085] Figure 1 It is a flow chart of the method of the present invention.

[0086] Figure 2 It is a schematic diagram of a digital platform of the present invention. DETAILED DESCRIPTION

[0087] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0088] like Figure 1 and Figure 2 As shown, a method for implementing a digital platform for a smart construction site based on full life cycle management includes:

[0089] 1. Distributed edge computing architecture

[0090] Distribute data processing tasks to edge nodes (such as sensor nodes and mobile terminals) at the construction site to reduce data transmission delays and improve real-time processing capabilities.

[0091] N edge nodes are deployed at the construction site. Each node i is responsible for processing data from multiple sensors and recording data including temperature, humidity, and construction progress:

[0092] V ij (t) = s V (t)+∈ V (t);

[0093] V ij (t) is the temperature data collected by sensor j of the i-th edge node at time t, s V (t) is the actual temperature value collected by the sensor at time t, ∈ V (t) is the error of temperature data;

[0094] U ij (t) = s U (t)+∈ U (t);

[0095] U ij (t) is the humidity data collected by sensor j of the i-th edge node at time t, s U (t) is the actual humidity value collected by the sensor at time t, ∈ U (t) is the error of humidity data;

[0096]

[0097] M ij (t) is the construction progress data collected by sensor j of the i-th edge node at time t, V complete (t) is the completed volume of the building obtained by scanning through image recognition, V total The total volume of the building to be completed;

[0098] Integrate the data:

[0099] D ij (t) = [V ij (t),U ij (t),M ij (t)];

[0100] {D i1 (t),D i2 (t),.....,D ij (t),......D im (t)};

[0101] Among them, D ij (t) is the data collected by sensor j at time t processed by edge node i, D im (t) is the data collected by sensor m at time t processed by edge node i;

[0102] In smart construction sites or other IoT scenarios, data collected by sensors (such as temperature, humidity, and construction progress) need to be analyzed in real time, and edge nodes must have sufficient processing power to cope with sudden increases in data volume.

[0103] The processing power determines the distribution of tasks among different edge nodes. When resources are limited, tasks can be distributed to nodes with stronger capabilities based on processing power to avoid system overload.

[0104] The processing capacity of each edge node can be expressed as:

[0105]

[0106] Among them, P i (t) is the processing capacity of edge node i at time t, which reflects the comprehensive processing capability of the node for sensor data;

[0107] m is the number of sensors processed by each edge node, α ij is the processing weight of edge node i on sensor j data, indicating the importance of the sensor data to the comprehensive processing capability, F ij (t) is the load of sensor j of edge node i at time t.

[0108] 2. Adaptive Fusion Algorithm

[0109] Assume that the data collected from each edge node is:

[0110] I(t)={I 1 (t),I 2 (t),.....,I k (t)};

[0111] Among them, k is the dimension of the integrated data, and the integrated data It can be expressed as:

[0112]

[0113] Among them, w l (t) is the weight of the lth dimension data at time t, reflecting the relative importance of different data dimensions in the comprehensive data and satisfying the normalization condition I l (t) is the l-th dimension data at time t;

[0114] Weight w l (t) Adaptive adjustment based on historical and real-time data through machine learning models:

[0115]

[0116] w(t) is the weight vector at time t, W is the model parameter matrix, b is the model parameter bias vector, It is a comprehensive data The feature extraction function of , Softmax(*) is used to ensure that the weights meet the normalization conditions.

[0117] 3. Real-time processing and decision-making model

[0118] Combining the state space model with the real-time optimization algorithm, dynamic monitoring and decision support for construction progress and safety risks can be achieved.

[0119] Construct a state space model to describe the evolution of the system state x(t):

[0120]

[0121] x(t+1) is the state vector of the system at time t+1, x(t) is the state vector of the system at time t, u(t) is the control input, and ζ(t) is the process noise, which indicates the instability within the system;

[0122] A is the system matrix, which describes the state transition relationship, B is the control input matrix, which describes the impact of the control input on the state, and C is the output matrix, which describes how the state is converted into the observed value;

[0123] y(t) is the observation vector and v(t) is the observation noise, which represents the uncertainty in the measurement process.

[0124] Based on this model, Kalman filtering is used for real-time state estimation:

[0125] Kalman filter prediction:

[0126]

[0127] is the state at time t predicted based on time t-1, is the state estimate at time t-1, and u(t-1) is the control input at time t-1;

[0128] Kalman filter prediction error covariance, reflecting the uncertainty of state prediction:

[0129] P(t|t-1)=AP(t-1|t-1)A T +Q;

[0130] P(t|t-1) is the error covariance at time t predicted based on time t-1, P(t-1|t-1) is the updated error covariance, Q is the process noise covariance matrix, and A T is the transpose of the system matrix;

[0131] By calculating the Kalman gain, the update method of the state estimate can be dynamically adjusted to minimize the estimation error:

[0132]

[0133] K(t) is the Kalman gain, which is used to update the state estimate, R is the observation noise covariance matrix, C T is the transpose of the output matrix;

[0134] Kalman filter update:

[0135]

[0136] is the updated state estimate, y(t) is the observation vector, is the predicted observation vector;

[0137] P(t|t)=(EK(t)C)P(t|t-1);

[0138] P(t|t) is the updated error covariance, and E is the identity matrix.

[0139] 4. Data transmission and communication optimization

[0140] Compressed Sensing is a signal processing technique used to recover high-dimensional signals from a small amount of sampled data. In this formulation, the original high-dimensional data D is converted into low-dimensional data g through a compression matrix Φ, thereby reducing the amount of data transmission and storage requirements. The compression matrix Φ is designed to meet certain sparsity conditions to ensure that the original data can be accurately restored in the subsequent reconstruction step.

[0141] The original data vector is Compression via Compressed Sensing:

[0142] g = ΦD;

[0143] in, is a compressed matrix, satisfying M<N, It is the compressed data;

[0144] At the receiving end, the original data is restored through the sparse reconstruction algorithm:

[0145]

[0146] is the reconstructed original data vector, Ensure that the reconstructed signal is consistent with the compressed observation, ‖*‖ 2is the L2 norm, which is used to measure the error between the reconstructed signal and the observed signal, λ‖ΨD‖ 1 is a sparse regularization term that controls the sparsity of non-sparse data in the transform domain Ψ, λ is a regularization parameter, ‖*‖ 1 Indicates l 1 norm, used for sparse representation, g = ΦD is the constraint condition for compressed data;

[0147] In order to improve the recovery accuracy, deep learning methods are used to optimize the reconstruction process, and deep neural networks are used to reconstruct the signal, where the input of the network is compressed sensing sampling data;

[0148]

[0149] f θ (*) is the mapping function of the deep neural network with parameter θ, which can recover the signal from the compressed sampling data after training. The network is trained with the following loss function:

[0150] Training is performed by combining reconstruction error with sparsity constraints:

[0151]

[0152] in, is the loss function, which represents the training error of the model.

[0153] By combining intelligent algorithms with historical construction data, we have successfully achieved accurate prediction of construction progress. By combining data from different construction stages, we can predict the completion time of specific tasks within a 5% error. This improves the accuracy of project progress prediction, reduces additional costs caused by construction delays, and ensures that the project is delivered on time.

[0154] By using the Internet of Things technology and deploying 100 sensors to monitor the operating status of equipment, the system issued an early warning three hours before equipment failure occurred, preventing 10 potential equipment failure accidents. By implementing this early warning system, the project's equipment failure rate dropped by 15%, and the impact of failures on construction progress was reduced by 30%.

[0155] By analyzing the data flow in real time at the construction site, the spatial layout of the construction site and the safety arrangements of personnel were optimized. Using sensors and cloud computing technology to collect temperature and humidity data in the construction area, the system proposed targeted construction adjustment plans by analyzing these data, reducing the time construction workers spent in high-risk areas by 20%, effectively improving overall safety.

[0156] Through comprehensive multi-source data analysis, the system provides efficient decision support for project managers. Based on real-time data from the construction site, the system can automatically generate risk assessment reports and construction schedule adjustment suggestions. Through these intelligent decision-making supports, the budget overrun of a project was reduced by 12%, and the construction period was 10% ahead of schedule.

[0157] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A method for implementing a digital platform for smart construction sites based on full life cycle management, characterized in that: include: Distributed edge computing architecture: distributes data processing tasks to edge nodes at the construction site; Adaptive fusion algorithm: integrates information from multiple sources to enhance data representativeness and decision-making accuracy; Real-time processing and decision-making model: Combine state space model with real-time optimization algorithm to achieve dynamic monitoring and decision support for construction progress and safety risks; Data transmission technology based on compressed sensing: recovering high-dimensional signals from a small amount of sampled data.

2. The method for implementing a smart construction site digital platform based on full life cycle management according to claim 1 is characterized in that: The distributed edge computing architecture includes: N edge nodes are deployed at the construction site. Each node i is responsible for processing data from multiple sensors and recording data including temperature, humidity, and construction progress: V ij (t)=s V (t)+∈ V (t); V ij (t) is the temperature data collected by sensor j of the i-th edge node at time t, s V (t) is the actual temperature value collected by the sensor at time t, ∈ V (t) is the error of temperature data; U ij (t)=s U (t)+∈ U (t); U ij (t) is the humidity data collected by sensor j of the i-th edge node at time t, s U (t) is the actual humidity value collected by the sensor at time t, ∈ U (t) is the error of humidity data; M ij (t) is the construction progress data collected by sensor j of the i-th edge node at time t, V complete (t) is the completed volume of the building obtained by image recognition and scanning, V total The total volume of the building to be completed; Integrate the data: D ij (t)=[V ij (t),U ij (t),M ij (t)]; {D i1 (t),D i2 (t),.....,D ij (t),......D im (t)}; Among them, D ij (t) is the data collected by sensor j at time t processed by edge node i, D im (t) is the data collected by sensor m at time t processed by edge node i; The processing capacity determines the distribution of tasks among different edge nodes. When resources are limited, tasks can be distributed to nodes with stronger capabilities based on processing capacity to avoid system overload. The processing capacity of each edge node can be expressed as: Among them, P i (t) is the processing capacity of edge node i at time t, which reflects the comprehensive processing capability of the node for sensor data; m is the number of sensors processed by each edge node, α ij is the processing weight of edge node i on sensor j data, indicating the importance of the sensor data to the comprehensive processing capability, F ij (t) is the load of sensor j of edge node i at time t.

3. The method for implementing a digital platform for smart construction sites based on full life cycle management according to claim 1 is characterized in that: The adaptive fusion algorithm comprises: Assume that the data collected from each edge node is: I(t)={I1(t),I2(t),.....,I k (t)}; Among them, I(t) is the original input data, k is the dimension of the integrated data, and I k (t) is the k-th dimension data at time t, the integrated data after fusion It can be expressed as: Among them, w l (t) is the weight of the lth dimension data at time t, reflecting the relative importance of different data dimensions in the comprehensive data and satisfying the normalization condition I l (t) is the l-th dimension data at time t; Weight w l (t) Adaptive adjustment based on historical and real-time data through machine learning models: w(t) is the weight vector at time t, W is the model parameter matrix, b is the model parameter bias vector, It is a comprehensive data The feature extraction function of , Softmax(*) is used to ensure that the weights meet the normalization conditions.

4. The method for implementing a smart construction site digital platform based on full life cycle management according to claim 1 is characterized in that: The real-time processing and decision-making model includes: Construct a state space model to describe the evolution of the system state x(t): x(t+1) is the state vector of the system at time t+1, x(t) is the state vector of the system at time t, u(t) is the control input, and ζ(t) is the process noise, which indicates the instability within the system; A is the system matrix, which describes the state transition relationship, B is the control input matrix, which describes the impact of the control input on the state, and C is the output matrix, which describes how the state is converted into the observed value; y(t) is the observation vector, v(t) is the observation noise, which represents the uncertainty in the measurement process; Based on this model, Kalman filtering is used for real-time state estimation: Kalman filter prediction: is the state at time t predicted based on time t-1, is the state estimate at time t-1 obtained by combining the observation data at time t-1, and u(t-1) is the control input at time t-1; Kalman filter prediction error covariance, reflecting the uncertainty of state prediction: P(t|t-1)=AP(t-1|t-1)A T +Q; P(t|t-1) is the error covariance at time t predicted based on time t-1, P(t-1|t-1) is the updated error covariance after combining the observed data at time t-1, Q is the process noise covariance matrix, and A T is the transpose of the system matrix; By calculating the Kalman gain, the update method of the state estimate can be dynamically adjusted to minimize the estimation error: K(t) is the Kalman gain, which is used to update the state estimate, R is the observation noise covariance matrix, C T is the transpose of the output matrix; Kalman filter update: is the updated state estimate, is the predicted state estimate, y(t) is the observation vector, is the predicted observation vector; P(t|t)=(EK(t)C)P(t|t-1); P(t|t) is the updated error covariance, and E is the identity matrix.

5. The method for implementing a digital platform for smart construction sites based on full life cycle management according to claim 1 is characterized in that: The data transmission technology based on compressed sensing includes: The original data vector is Compression via Compressed Sensing: g = ΦD; in, is a compressed matrix, satisfying M<N, It is the compressed data; At the receiving end, the original data is restored through the sparse reconstruction algorithm: is the reconstructed original data vector, Ensure that the reconstructed signal is consistent with the compressed observation value, ‖*‖2 is the L2 norm, which is used to measure the error between the reconstructed signal and the observed signal, λ‖ΨD‖1 is the sparse regularization term, which controls the sparsity of non-sparse data in the transform domain Ψ, λ is the regularization parameter, ‖*‖1 represents the l1 norm, which is used for sparse representation, and g=ΦD is the data constraint condition after compression; In order to improve the recovery accuracy, deep learning methods are used to optimize the reconstruction process, and deep neural networks are used to reconstruct the signal, where the input of the network is compressed sensing sampling data; f θ (*) is the mapping function of the deep neural network with parameter θ, which can recover the signal from the compressed sampling data after training. The network is trained with the following loss function: Training is performed by combining reconstruction error with sparsity constraints: in, is the loss function, which represents the training error of the model.

6. A computer system, characterized in that: include: processor; a memory for storing processor-executable instructions; Wherein, the processor is configured to implement the method for implementing a smart construction site digital platform based on full life cycle management as described in any one of claims 1 to 5 when executing the executable instructions.

7. A computer-readable storage medium, characterized in that: include: a memory having a computer program stored thereon; A processor is used to execute the program in the memory to implement the method for implementing a smart construction site digital platform based on full life cycle management as described in any one of claims 1 to 5.

Images