Decision management method and system based on machine learning

Abstract

The invention belongs to the technical field of data management, and discloses a decision management method and system based on machine learning. The method comprises the following steps: constructing an engineering construction decision analysis model and an engineering construction decision optimization model based on a cloud data center; acquiring real-time engineering construction decision data, and performing engineering construction decision analysis by using the engineering construction decision analysis model; performing engineering construction decision optimization on the real-time engineering construction decision data by using an engineering construction decision optimization model according to a real-time engineering construction decision analysis result; according to the optimized real-time engineering construction decision data, an engineering construction decision analysis model is used to carry out analysis after an engineering construction decision; and performing uplink storage on the optimized real-time engineering construction decision data and the optimized real-time engineering construction decision analysis result by using a preset decision storage block chain system. According to the method, the problems of low data security, low analysis accuracy and low practicability in the prior art are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of data management, and in particular relates to a decision management method and system based on machine learning. Background Art

[0002] In the field of engineering construction, a large amount of engineering construction decision-making data will be generated during the normal progress of engineering projects, including the early investment stage of the project, the construction period, the operation period after completion, and the later maintenance period of the project, etc. These engineering construction decision-making data are very critical to engineering construction, operation, maintenance, etc. and need to be properly managed.

[0003] The existing engineering construction decision-making management methods have the following problems: First, they often use local storage, which is prone to hardware crashes and data loss, and have low security; second, they rely on expert experience and qualitative analysis, lack in-depth mining and utilization of decision-making data, and have low analysis accuracy; third, there is a lack of optimization methods for engineering construction decision-making data, and their practicality is low. Summary of the invention

[0004] In order to solve the problems of low data security, low analysis accuracy and low practicality in the prior art, the present invention aims to provide a decision management method and system based on machine learning.

[0005] The technical solution adopted by the present invention is:

[0006] A decision management method based on machine learning comprises the following steps:

[0007] Based on the cloud data center, according to a number of historical engineering construction decision data, using machine learning algorithms, we build engineering construction decision analysis models and engineering construction decision optimization models;

[0008] Collect real-time engineering construction decision data, and use the engineering construction decision analysis model to perform engineering construction decision analysis based on the real-time engineering construction decision data to obtain real-time engineering construction decision analysis results;

[0009] According to the real-time engineering construction decision analysis results, the engineering construction decision optimization model is used to optimize the real-time engineering construction decision data to obtain the optimized real-time engineering construction decision data;

[0010] According to the optimized real-time engineering construction decision data, the engineering construction decision analysis model is used to perform post-engineering construction decision analysis to obtain the optimized real-time engineering construction decision analysis results;

[0011] Use the preset decision storage blockchain system to store the optimized real-time engineering construction decision data and the optimized real-time engineering construction decision analysis results on the chain.

[0012] Furthermore, based on the cloud data center and according to a number of historical engineering construction decision data, a machine learning algorithm is used to construct an engineering construction decision analysis model and an engineering construction decision optimization model, including the following steps:

[0013] Based on the cloud data center, historical engineering construction decision data of several historical engineering construction projects are collected and preprocessed to obtain several preprocessed historical engineering construction decision data;

[0014] Based on a number of pre-processed historical engineering construction decision data, a deep learning algorithm is used to build an engineering construction decision analysis model and generate a number of historical engineering construction decision analysis results;

[0015] Based on several pre-processed historical engineering construction decision data and historical engineering construction decision analysis results, a reinforcement learning algorithm is used to construct an engineering construction decision optimization model.

[0016] Furthermore, the engineering construction decision analysis model is constructed based on the N-GAN-CNN-Attention-RF-MLP algorithm, and the engineering construction decision analysis model includes N attention dimension feature extraction modules constructed based on the GAN algorithm, a data feature extraction module constructed based on the CNN algorithm, a feature splicing module constructed based on the Attention mechanism, a key feature screening module constructed based on the RF algorithm, and a feature fusion analysis module constructed based on the MLP algorithm.

[0017] Furthermore, based on a number of pre-processed historical engineering construction decision data, a deep learning algorithm is used to construct an engineering construction decision analysis model, and a number of historical engineering construction decision analysis results are generated, including the following steps:

[0018] Clustering is performed on some pre-processed historical engineering construction decision data to obtain some cluster centers and cluster clusters;

[0019] A true analysis label is set for each cluster center, and the true analysis label of the cluster center is diffused to the corresponding cluster clusters, so as to obtain a number of historical engineering construction decision analysis samples with true analysis labels;

[0020] Several historical engineering construction decision analysis samples were divided into model training set and model test set in a ratio of 7:3, and the initial engineering construction decision analysis model was constructed using a deep learning algorithm.

[0021] According to the model training set, the initial engineering construction decision analysis model is optimized and trained to obtain an optimized engineering construction decision analysis model;

[0022] According to the model test set, the optimized engineering construction decision analysis model is tested to obtain several historical prediction analysis labels;

[0023] Analyze and compare several historical prediction analysis labels with several real analysis labels to obtain the test accuracy of the optimized engineering construction decision analysis model;

[0024] If the test accuracy is greater than the accuracy threshold, the final engineering construction decision analysis model is output and several historical engineering construction decision analysis results are generated. Otherwise, training optimization continues.

[0025] Furthermore, the engineering construction decision optimization model is constructed based on the DQN algorithm.

[0026] Furthermore, based on a number of pre-processed historical engineering construction decision data and historical engineering construction decision analysis results, a reinforcement learning algorithm is used to construct an engineering construction decision optimization model, including the following steps:

[0027] Based on the engineering construction decision optimization problem, define the simulation environment of the reinforcement learning algorithm, and build the intelligent agent and experience replay pool of the reinforcement learning algorithm;

[0028] Based on the pre-processed historical engineering construction decision data and historical engineering construction decision analysis results, define the state space and action space of the reinforcement learning algorithm;

[0029] Based on the simulation environment, the impact of each execution action in the action space on the pre-processed historical engineering construction decision data is analyzed, and the reward function of the reinforcement learning algorithm is defined;

[0030] Construct the input layer, several hidden layers, and output layer of the deep Q network, connect the state space to the input layer, and connect the action space and reward function to the output layer;

[0031] Integrate deep Q networks, intelligent agents, and experience replay pools to build an initial engineering construction decision optimization model;

[0032] According to a number of pre-processed historical engineering construction decision data and historical engineering construction decision analysis results, the initial engineering construction decision optimization model is optimized and trained to obtain an optimized engineering construction decision optimization model, and a number of historical engineering construction decision optimization experiences are generated;

[0033] Several historical engineering construction decision optimization experiences are stored in the experience playback pool, and the optimized engineering construction decision optimization model is evaluated. If several evaluations meet the requirements, the final engineering construction decision optimization model is output.

[0034] Furthermore, real-time engineering construction decision data is collected, and based on the real-time engineering construction decision data, an engineering construction decision analysis model is used to perform engineering construction decision analysis to obtain real-time engineering construction decision analysis results, including the following steps:

[0035] Collecting real-time engineering construction decision data, preprocessing it, obtaining preprocessed real-time engineering construction decision data, and inputting the preprocessed real-time engineering construction decision data into the engineering construction decision analysis model;

[0036] Use the attention dimension feature extraction module to extract N real-time attention dimension features of the pre-processed real-time engineering construction decision data;

[0037] Use the data feature extraction module to extract the real-time data features of the pre-processed real-time engineering construction decision data;

[0038] Use the feature splicing module to perform feature splicing on N real-time attention dimension features and real-time data features according to the preset attention weights to obtain real-time splicing features;

[0039] Use the key feature screening module to screen the real-time splicing features for key features and obtain M real-time key features;

[0040] Use the feature fusion analysis module to fuse M real-time key features to obtain real-time fusion features;

[0041] Based on the feature fusion analysis module, engineering construction decision analysis is performed according to real-time fusion features to obtain real-time engineering construction decision analysis results.

[0042] Further, according to the real-time engineering construction decision analysis results, the engineering construction decision optimization model is used to optimize the real-time engineering construction decision data to obtain the optimized real-time engineering construction decision data, including the following steps:

[0043] According to the real-time engineering construction decision analysis results, a number of corresponding target historical engineering construction decision optimization experiences are matched in the experience playback pool of the engineering construction decision optimization model;

[0044] According to several target historical engineering construction decision optimization experiences, the action space of the engineering construction decision optimization model is updated to obtain an updated action space;

[0045] According to the real-time engineering construction decision data, the state space of the engineering construction decision optimization model is updated to obtain an updated state space;

[0046] Input the updated state space into the input layer of the deep Q network of the engineering construction decision optimization model, and input the updated action space into the output layer of the deep Q network;

[0047] Using the intelligent agent of the engineering construction decision optimization model, the deep Q network is controlled to generate updated Q values ​​for each possible action in the action space;

[0048] Use the reward function to obtain the reward value of each possible action in the updated action space, and update the Q value of the possible action according to the reward value to obtain the updated Q value of the possible action;

[0049] Repeat the above steps until the number of iterations reaches the iteration threshold, and take the possible action corresponding to the highest updated Q value as the execution action for engineering construction decision optimization;

[0050] According to the execution action, the real-time engineering construction decision data is optimized to obtain the optimized real-time engineering construction decision data.

[0051] Furthermore, the decision storage blockchain system includes a blockchain network consisting of several data servers as nodes, a calling interface, a smart contract and an IFPS system.

[0052] A decision management system based on machine learning is used to implement a decision management method. The system is deployed in a cloud data center and includes a model building unit, a decision analysis unit, a decision optimization unit, a post-decision analysis unit, and an on-chain storage unit.

[0053] The beneficial effects of the present invention are:

[0054] The present invention discloses a decision management method and system based on machine learning, which adopts cloud storage to avoid data loss caused by local hardware crashes. At the same time, blockchain technology is used to ensure the decentralization and traceability of engineering construction decision data storage, thereby improving data security; deep learning algorithms are used to analyze data, and decision data is accurately and efficiently mined and utilized; reinforcement learning algorithms are used to optimize decisions, thereby improving the quality and effect of decisions and improving practicality.

[0055] Other beneficial effects of the present invention will be further described in the specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a flowchart of the decision management method based on machine learning in the present invention.

[0057] Figure 2 It is a structural block diagram of the decision management system based on machine learning in the present invention. DETAILED DESCRIPTION

[0058] The present invention will be further explained below in conjunction with the accompanying drawings and specific embodiments.

[0059] Embodiment 1:

[0060] like Figure 1 As shown, this embodiment provides a decision management method based on machine learning, comprising the following steps:

[0061] S1: Based on the cloud data center and some historical engineering construction decision data, the machine learning algorithm is used to build an engineering construction decision analysis model and an engineering construction decision optimization model, including the following steps:

[0062] S1-1: Based on the cloud data center, historical engineering construction decision data of several historical engineering construction projects are collected and preprocessed to obtain several preprocessed historical engineering construction decision data;

[0063] Historical engineering construction decision data includes the initial investment, engineering scheme, construction plan, acceptance criteria, acceptance process and other decision instructions or data that are strongly related to engineering construction;

[0064] S1-2: Based on a number of pre-processed historical engineering construction decision data, a deep learning algorithm is used to construct an engineering construction decision analysis model and generate a number of historical engineering construction decision analysis results, including the following steps:

[0065] S1-2-1: Taking the actual situation of engineering construction decision-making as the clustering target, a fuzzy clustering algorithm, for example, the Fuzzy C-Means (FCM) clustering algorithm, is used to cluster a number of pre-processed historical engineering construction decision data to obtain a number of cluster centers and cluster clusters;

[0066] Since model training requires a large number of training samples, and adding real analysis labels requires a lot of manpower and material resources, which is time-consuming, clustering algorithms are used to pre-analyze data, which reduces the cost of obtaining training samples and improves the efficiency and accuracy of subsequent model construction.

[0067] S1-2-2: Set a true analysis label for each cluster center, and diffuse the true analysis label of the cluster center to the corresponding cluster cluster, to obtain a number of historical engineering construction decision analysis samples with true analysis labels;

[0068] S1-2-3: Divide several historical engineering construction decision analysis samples into a model training set and a model test set in a ratio of 7:3, and use a deep learning algorithm to build an initial engineering construction decision analysis model;

[0069] The engineering construction decision analysis model is constructed based on N-Generative Adversarial Network (GAN)-Convolutional Neural Network (CNN)-Attention-Random Forest (RF)-Multilayer Perceptron (MLP) algorithm, and the engineering construction decision analysis model includes N attention dimension feature extraction modules constructed based on the GAN algorithm, a data feature extraction module constructed based on the CNN algorithm, a feature splicing module constructed based on the Attention mechanism, a key feature screening module constructed based on the RF algorithm, and a feature fusion analysis module constructed based on the MLP algorithm;

[0070] The GAN network is a deep learning model consisting of two parts: a generator and a discriminator. The input of the generator is a random vector (for example, sampled from a standard normal distribution), and the output is a sample of the original data. The purpose is to deceive the discriminator into thinking that the generated sample is real. The input of the discriminator is the original data sample or the sample generated by the generator. The output is a probability value indicating whether the sample comes from real data or the generator. The purpose is to distinguish between real samples and generated samples. This embodiment designs a special discriminator structure so that it only focuses on the features of key dimensions instead of the entire data set. Once the generator and the discriminator are trained, the features of the key dimensions can be extracted because the generator learns how to generate and identify key dimension features during the training process; the key dimensions include decision feasibility, decision cost, decision effect, etc.

[0071] The CNN network extracts the data features of engineering construction decision-making data through a deep network structure, providing a basis for subsequent analysis and prediction; the Attention mechanism splices data features and attention dimension features through preset attention weights, improving the ability of features to represent key information in the data; the RF structure uses the internal Classification and Regression Tree (CART) to screen key features of the input spliced ​​features, eliminating the interference of irrelevant features on the prediction results; the MLP network, as a fully connected network, fuses several features to improve prediction accuracy and efficiency;

[0072] S1-2-4: Optimize and train the initial engineering construction decision analysis model based on the model training set to obtain an optimized engineering construction decision analysis model;

[0073] S1-2-5: Based on the model test set, the optimized engineering construction decision analysis model is tested to obtain several historical prediction analysis labels;

[0074] S1-2-6: Analyze and compare a number of historical prediction analysis labels with a number of real analysis labels to obtain the test accuracy of the optimized engineering construction decision analysis model;

[0075] S1-2-7: If the test accuracy is greater than the accuracy threshold, the final engineering construction decision analysis model is output and several historical engineering construction decision analysis results are generated; otherwise, training optimization is continued;

[0076] S1-3: Based on some pre-processed historical engineering construction decision data and historical engineering construction decision analysis results, a deep Q network (DQN) algorithm is used to construct an engineering construction decision optimization model, including the following steps:

[0077] S1-3-1: Based on the engineering construction decision optimization problem, define the simulation environment of the reinforcement learning algorithm, and build the intelligent agent and experience replay pool of the reinforcement learning algorithm;

[0078] S1-3-2: Define the state space and action space of the reinforcement learning algorithm based on the pre-processed historical engineering construction decision data and historical engineering construction decision analysis results;

[0079] S1-3-3: Based on the simulation environment, analyze the impact of each execution action in the action space on the pre-processed historical engineering construction decision data, and define the reward function of the reinforcement learning algorithm;

[0080] S1-3-4: Construct the input layer, several hidden layers, and output layer of the deep Q network, connect the state space to the input layer, and connect the action space and reward function to the output layer;

[0081] S1-3-5: Integrate the deep Q network, intelligent agent and experience replay pool to build the initial engineering construction decision optimization model;

[0082] S1-3-6: Based on a number of pre-processed historical engineering construction decision data and historical engineering construction decision analysis results, the initial engineering construction decision optimization model is optimized and trained to obtain an optimized engineering construction decision optimization model, and a number of historical engineering construction decision optimization experiences are generated;

[0083] S1-3-7: Store several historical engineering construction decision optimization experiences in the experience playback pool, and evaluate the optimized engineering construction decision optimization model. If several evaluations meet the requirements, output the final engineering construction decision optimization model;

[0084] S2: collecting real-time engineering construction decision data, and using the engineering construction decision analysis model to perform engineering construction decision analysis based on the real-time engineering construction decision data to obtain real-time engineering construction decision analysis results, including the following steps:

[0085] S2-1: collecting real-time engineering construction decision data, performing preprocessing to obtain preprocessed real-time engineering construction decision data, and inputting the preprocessed real-time engineering construction decision data into the engineering construction decision analysis model;

[0086] S2-2: Use the attention dimension feature extraction module to extract N real-time attention dimension features of the pre-processed real-time engineering construction decision data;

[0087] S2-3: Using the data feature extraction module, extract the real-time data features of the pre-processed real-time engineering construction decision data;

[0088] S2-4: Use the feature splicing module to perform feature splicing on N real-time attention dimension features and real-time data features according to the preset attention weights to obtain real-time splicing features;

[0089] S2-5: Use the key feature screening module to screen the real-time splicing features for key features to obtain M real-time key features, including the following steps:

[0090] S2-5-1: Use the RF structure trained in the key feature screening module to extract the feature contribution of several real-time candidate features in the real-time splicing feature;

[0091] The formula is:

[0092]

[0093] In the formula, is the feature contribution of the jth real-time candidate feature; is the feature contribution of the jth real-time candidate feature in the i-th tree of the random forest; i is the CART tree indicator; j is the real-time candidate feature indicator; n is the total number of CARTs;

[0094]

[0095] In the formula, GI m GI l GI r is the Gini index of CART tree node m, node l and node r of random forest; p mk is the proportion of category k in CART tree node m; K is the total number of categories; m, l, r are node indicators; k is the category indicator;

[0096] S2-5-2: normalizing the feature contributions of several real-time candidate features to obtain corresponding normalized feature contributions;

[0097] The formula is:

[0098]

[0099] Where VIM j is the contribution of the feature after normalization; J is the total number of real-time candidate features;

[0100] S2-5-3: Generate feature selection standard values ​​for several real-time candidate features based on the normalized feature contribution;

[0101] The formula is:

[0102]

[0103] In the formula, CFC j VIM is the feature selection standard value for the jth real-time candidate feature; j' is the normalized feature contribution of the j'th real-time candidate feature; j' is the indicator of the candidate feature;

[0104] S2-5-4: sort the real-time candidate features in descending order according to the feature selection standard value, and select the first M real-time candidate features as real-time key features to obtain M real-time key features, where M is the total number of real-time key features;

[0105] S2-6: Use the feature fusion analysis module to fuse the M real-time key features to obtain real-time fusion features;

[0106] S2-7: Based on the feature fusion analysis module, the engineering construction decision analysis is performed according to the real-time fusion features to obtain the real-time engineering construction decision analysis results;

[0107] S3: According to the real-time engineering construction decision analysis results, the engineering construction decision optimization model is used to optimize the real-time engineering construction decision data to obtain the optimized real-time engineering construction decision data, including the following steps:

[0108] S3-1: According to the real-time engineering construction decision analysis results, the corresponding several target historical engineering construction decision optimization experiences are matched in the experience playback pool of the engineering construction decision optimization model;

[0109] S3-2: Based on several historical engineering construction decision optimization experiences, the action space of the engineering construction decision optimization model is updated to obtain the updated action space A' = [a' 1 ,...,a' j" ,...,a' I], where a' j" is the updated j-th "action value", j" is the action indicator;

[0110] S3-3: Update the state space of the engineering construction decision optimization model according to the real-time engineering construction decision data, and obtain the updated state space S'=[s' 1 ,...,s' i" ,...,s' I ], where s' i" is the updated i-th state value, i is the state indicator;

[0111] S3-4: input the updated state space into the input layer of the deep Q network of the engineering construction decision optimization model, and input the updated action space into the output layer of the deep Q network;

[0112] S3-5: Use the agent of the engineering construction decision optimization model to control the deep Q network to generate updated Q values ​​for each possible action in the action space;

[0113] S3-6: Use the reward function to obtain the reward value of each possible action in the updated action space, and update the Q value of the possible action according to the reward value to obtain the updated Q value of the possible action;

[0114] The formula is:

[0115] Q(s' p' ,a' p' )=(1-α")·Q(s p' ,a p' )+α"·(R(s p' ,a p' ,s' p' )+γ·Q max (s p' ,a p' ))

[0116] In the formula, Q(s' p' ,a' p' ) is the updated state value s' p' and the updated action value a' p' The corresponding updated Q value; Q(s p' ,a p' ) is the state value s p' and action value a p' The corresponding predicted Q value; α" is the learning rate; Q max (s p' ,a p' ) is the highest predicted Q value;

[0117] S3-7: Repeat the above steps until the number of iterations reaches the iteration number threshold, and use the possible action corresponding to the highest updated Q value as the execution action of the engineering construction decision optimization;

[0118] S3-8: According to the execution action, the real-time engineering construction decision data is optimized to obtain the optimized real-time engineering construction decision data;

[0119] S4: Based on the optimized real-time engineering construction decision data, use the engineering construction decision analysis model to perform post-engineering construction decision analysis to obtain optimized real-time engineering construction decision analysis results;

[0120] S5: Use the preset decision storage blockchain system to store the optimized real-time engineering construction decision data and the optimized real-time engineering construction decision analysis results on the chain;

[0121] The decision storage blockchain system includes a blockchain network consisting of several data servers as nodes, a call interface, smart contracts, and an Inter Planetary File System (IPFS);

[0122] The steps include:

[0123] S5-1: Based on the digital asset blockchain system, the optimized real-time engineering construction decision data and the optimized real-time engineering construction decision analysis results are encrypted using an asymmetric encryption algorithm according to the private key in the pre-generated public-private key pair, to obtain the encrypted optimized real-time engineering construction decision data and the encrypted optimized real-time engineering construction decision analysis results;

[0124] S5-2: Use the IPFS system to obtain the encrypted and optimized real-time engineering construction decision data and the encrypted and optimized real-time engineering construction decision analysis results real-time data hash value, and generate a real-time data storage request;

[0125] S5-3: Based on the calling interface, the smart contract is executed to generate encrypted and optimized real-time engineering construction decision data and encrypted and optimized real-time engineering construction decision analysis results of real-time transaction data, and the real-time transaction data and real-time digital asset hash value are converted into real-time data blocks, and the real-time data storage request and real-time data block are sent to the blockchain network;

[0126] S5-4: Based on the blockchain network, consensus is reached on the real-time data storage request. If the consensus is successful, the node that receives the real-time data block in the blockchain network is used to store the real-time data block on the chain and update the distributed log;

[0127] S5-5: Store the encrypted and optimized real-time engineering construction decision data and the encrypted and optimized real-time engineering construction decision analysis results in the IPFS system.

[0128] Embodiment 2:

[0129] like Figure 2 As shown, this embodiment provides a decision management system based on machine learning, which is used to implement a decision management method. The system is deployed in a cloud data center and includes a model building unit, a decision analysis unit, a decision optimization unit, a post-decision analysis unit, and an on-chain storage unit;

[0130] A model building unit, used to build an engineering construction decision analysis model and an engineering construction decision optimization model based on a number of historical engineering construction decision data using a machine learning algorithm;

[0131] A decision analysis unit is used to collect real-time engineering construction decision data, and to perform engineering construction decision analysis based on the real-time engineering construction decision data using an engineering construction decision analysis model to obtain real-time engineering construction decision analysis results;

[0132] A decision optimization unit is used to optimize the real-time engineering construction decision data according to the real-time engineering construction decision analysis results and use the engineering construction decision optimization model to obtain the optimized real-time engineering construction decision data;

[0133] A post-decision analysis unit is used to perform post-decision analysis on the engineering construction according to the optimized real-time engineering construction decision data and use the engineering construction decision analysis model to obtain the optimized real-time engineering construction decision analysis results;

[0134] The on-chain storage unit is used to use a preset decision storage blockchain system to store the optimized real-time engineering construction decision data and the optimized real-time engineering construction decision analysis results on the chain.

[0135] The present invention discloses a decision management method and system based on machine learning, which adopts cloud storage to avoid data loss caused by local hardware crashes. At the same time, blockchain technology is used to ensure the decentralization and traceability of engineering construction decision data storage, thereby improving data security; deep learning algorithms are used to analyze data, and decision data is accurately and efficiently mined and utilized; reinforcement learning algorithms are used to optimize decisions, thereby improving the quality and effect of decisions and improving practicality.

[0136] The present invention is not limited to the above optional implementations, and anyone can derive other various forms of products under the enlightenment of the present invention. The above specific implementations should not be understood as limiting the scope of protection of the present invention. The scope of protection of the present invention should be based on the definition in the claims, and the description can be used to interpret the claims.

Claims

1. A decision management method based on machine learning, characterized in that: The steps include: Based on the cloud data center, according to a number of historical engineering construction decision data, using machine learning algorithms, we build engineering construction decision analysis models and engineering construction decision optimization models; Collect real-time engineering construction decision data, and use the engineering construction decision analysis model to perform engineering construction decision analysis based on the real-time engineering construction decision data to obtain real-time engineering construction decision analysis results; According to the real-time engineering construction decision analysis results, the engineering construction decision optimization model is used to optimize the real-time engineering construction decision data to obtain the optimized real-time engineering construction decision data; According to the optimized real-time engineering construction decision data, the engineering construction decision analysis model is used to perform post-engineering construction decision analysis to obtain the optimized real-time engineering construction decision analysis results; Use the preset decision storage blockchain system to store the optimized real-time engineering construction decision data and the optimized real-time engineering construction decision analysis results on the chain.

2. A decision management method based on machine learning according to claim 1, characterized in that: Based on the cloud data center and some historical engineering construction decision data, the machine learning algorithm is used to build an engineering construction decision analysis model and an engineering construction decision optimization model, including the following steps: Based on the cloud data center, historical engineering construction decision data of several historical engineering construction projects are collected and preprocessed to obtain several preprocessed historical engineering construction decision data; Based on a number of pre-processed historical engineering construction decision data, a deep learning algorithm is used to build an engineering construction decision analysis model and generate a number of historical engineering construction decision analysis results; Based on several pre-processed historical engineering construction decision data and historical engineering construction decision analysis results, a reinforcement learning algorithm is used to construct an engineering construction decision optimization model.

3. A decision management method based on machine learning according to claim 2, characterized in that: The engineering construction decision analysis model is constructed based on the N-GAN-CNN-Attention-RF-MLP algorithm, and the engineering construction decision analysis model includes N attention dimension feature extraction modules constructed based on the GAN algorithm, a data feature extraction module constructed based on the CNN algorithm, a feature splicing module constructed based on the Attention mechanism, a key feature screening module constructed based on the RF algorithm, and a feature fusion analysis module constructed based on the MLP algorithm.

4. The decision management method based on machine learning according to claim 3, characterized in that: Based on several pre-processed historical engineering construction decision data, a deep learning algorithm is used to construct an engineering construction decision analysis model, and several historical engineering construction decision analysis results are generated, including the following steps: Clustering is performed on some pre-processed historical engineering construction decision data to obtain some cluster centers and cluster clusters; A true analysis label is set for each cluster center, and the true analysis label of the cluster center is diffused to the corresponding cluster clusters, so as to obtain a number of historical engineering construction decision analysis samples with true analysis labels; Several historical engineering construction decision analysis samples were divided into model training set and model test set in a ratio of 7:3, and the initial engineering construction decision analysis model was constructed using a deep learning algorithm. According to the model training set, the initial engineering construction decision analysis model is optimized and trained to obtain an optimized engineering construction decision analysis model; According to the model test set, the optimized engineering construction decision analysis model is tested to obtain several historical prediction analysis labels; Analyze and compare several historical prediction analysis labels with several real analysis labels to obtain the test accuracy of the optimized engineering construction decision analysis model; If the test accuracy is greater than the accuracy threshold, the final engineering construction decision analysis model is output and several historical engineering construction decision analysis results are generated. Otherwise, training optimization continues.

5. The decision management method based on machine learning according to claim 2, characterized in that: The engineering construction decision optimization model is constructed based on the DQN algorithm.

6. The decision management method based on machine learning according to claim 5, characterized in that: Based on some pre-processed historical engineering construction decision data and historical engineering construction decision analysis results, a reinforcement learning algorithm is used to build an engineering construction decision optimization model, which includes the following steps: Based on the engineering construction decision optimization problem, define the simulation environment of the reinforcement learning algorithm, and build the intelligent agent and experience replay pool of the reinforcement learning algorithm; Based on the pre-processed historical engineering construction decision data and historical engineering construction decision analysis results, define the state space and action space of the reinforcement learning algorithm; Based on the simulation environment, the impact of each execution action in the action space on the pre-processed historical engineering construction decision data is analyzed, and the reward function of the reinforcement learning algorithm is defined; Construct the input layer, several hidden layers, and output layer of the deep Q network, connect the state space to the input layer, and connect the action space and reward function to the output layer; Integrate deep Q networks, intelligent agents, and experience replay pools to build an initial engineering construction decision optimization model; According to a number of pre-processed historical engineering construction decision data and historical engineering construction decision analysis results, the initial engineering construction decision optimization model is optimized and trained to obtain an optimized engineering construction decision optimization model, and a number of historical engineering construction decision optimization experiences are generated; Several historical engineering construction decision optimization experiences are stored in the experience playback pool, and the optimized engineering construction decision optimization model is evaluated. If several evaluations meet the requirements, the final engineering construction decision optimization model is output.

7. The decision management method based on machine learning according to claim 4, characterized in that: Collecting real-time engineering construction decision data, and using an engineering construction decision analysis model to perform engineering construction decision analysis based on the real-time engineering construction decision data to obtain real-time engineering construction decision analysis results, including the following steps: Collecting real-time engineering construction decision data, preprocessing it, obtaining preprocessed real-time engineering construction decision data, and inputting the preprocessed real-time engineering construction decision data into the engineering construction decision analysis model; Use the attention dimension feature extraction module to extract N real-time attention dimension features of the pre-processed real-time engineering construction decision data; Use the data feature extraction module to extract the real-time data features of the pre-processed real-time engineering construction decision data; Use the feature splicing module to perform feature splicing on N real-time attention dimension features and real-time data features according to the preset attention weights to obtain real-time splicing features; Use the key feature screening module to screen the real-time splicing features for key features and obtain M real-time key features; Use the feature fusion analysis module to fuse M real-time key features to obtain real-time fusion features; Based on the feature fusion analysis module, engineering construction decision analysis is performed according to real-time fusion features to obtain real-time engineering construction decision analysis results.

8. The decision management method based on machine learning according to claim 6, characterized in that: According to the real-time engineering construction decision analysis results, the engineering construction decision optimization model is used to optimize the real-time engineering construction decision data to obtain the optimized real-time engineering construction decision data, including the following steps: According to the real-time engineering construction decision analysis results, a number of corresponding target historical engineering construction decision optimization experiences are matched in the experience playback pool of the engineering construction decision optimization model; According to several target historical engineering construction decision optimization experiences, the action space of the engineering construction decision optimization model is updated to obtain an updated action space; According to the real-time engineering construction decision data, the state space of the engineering construction decision optimization model is updated to obtain an updated state space; Input the updated state space into the input layer of the deep Q network of the engineering construction decision optimization model, and input the updated action space into the output layer of the deep Q network; Using the intelligent agent of the engineering construction decision optimization model, the deep Q network is controlled to generate updated Q values ​​for each possible action in the action space; Use the reward function to obtain the reward value of each possible action in the updated action space, and update the Q value of the possible action according to the reward value to obtain the updated Q value of the possible action; Repeat the above steps until the number of iterations reaches the iteration threshold, and take the possible action corresponding to the highest updated Q value as the execution action for engineering construction decision optimization; According to the execution action, the real-time engineering construction decision data is optimized to obtain the optimized real-time engineering construction decision data.

9. The decision management method based on machine learning according to claim 1, characterized in that: The decision storage blockchain system includes a blockchain network consisting of several data servers as nodes, a calling interface, a smart contract and an IFPS system.

10. A decision management system based on machine learning, used to implement the decision management method according to any one of claims 1 to 9, characterized in that: The system is deployed in a cloud data center and includes a model building unit, a decision analysis unit, a decision optimization unit, a post-decision analysis unit and an on-chain storage unit.

Images