Construction site off-grid power supply optimization system based on big data fusion analysis

Abstract

The invention discloses a construction site off-grid power supply optimization system based on big data fusion analysis, and belongs to the field of construction site power supply management. Comprising an acquisition preprocessing module, an arrangement fusion module, a graph construction module, a demand prediction module, a resource optimization module, a simulation evaluation module, a decision support module, a real-time monitoring module, a feedback adjustment module and a report analysis module. According to the invention, under the condition of different environments and demand changes, rapid response can be ensured, the power supply stability can be maintained, the overall energy utilization efficiency can be improved, a good operation state can be maintained in case of emergencies, the robustness and reliability of a power supply system can be enhanced, prediction errors can be reduced, and the overall prediction precision of the system can be improved. The method helps to better understand the relationship between past and future demands, enhances the prediction effectiveness, reduces the demands of manual intervention, and improves the intelligent level of the system.

Description

Technical Field

[0001] The present invention relates to the field of construction site power supply management, and in particular to a construction site off-grid power supply optimization system based on big data fusion analysis. Background Art

[0002] In modern construction, the power supply of the construction site is an important factor in ensuring the progress and safety of the construction. The traditional power supply mode often relies on a fixed power source, lacks flexibility, and is difficult to meet the ever-changing power demand on the construction site. In addition, with the improvement of environmental protection requirements and the promotion of renewable energy, off-grid power supply solutions have gradually become a trend in the construction industry. In addition, with the widespread application of renewable energy, the accelerated urbanization process and the rapid development of the construction industry, the power demand on the construction site has gradually increased, especially in the off-grid power supply scenario. How to efficiently and reliably meet this demand has become an urgent problem to be solved. How to optimize resource allocation under different environmental conditions and ensure the stability and economy of power supply has become the key to designing an efficient off-grid power supply system.

[0003] The existing off-grid power supply optimization system for construction sites cannot guarantee a rapid response under different environments and changing demands, which reduces the overall energy utilization efficiency and cannot maintain a good operating state in the face of emergencies; in addition, the overall prediction accuracy of the existing off-grid power supply optimization system for construction sites is low, which reduces the effectiveness of the prediction, increases the demand for manual intervention, and reduces the intelligence level of the system. For this reason, we propose an off-grid power supply optimization system for construction sites based on big data fusion analysis. Summary of the invention

[0004] The purpose of the present invention is to solve the defects in the prior art and to propose a construction site off-grid power supply optimization system based on big data fusion analysis.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] The construction site off-grid power supply optimization system based on big data fusion analysis includes collection and preprocessing module, sorting and fusion module, graph construction module, demand prediction module, resource optimization module, simulation evaluation module, decision support module, real-time monitoring module, feedback adjustment module and report analysis module;

[0007] The acquisition preprocessing module is used to collect power data from the construction site in real time, and to clean, denoise and standardize the collected data;

[0008] The collating and fusion module is used to integrate data from different sources and generate a power demand view;

[0009] The graph construction module is used to structure and process various knowledge about power supply of construction sites;

[0010] The demand forecasting module is used to make short-term and long-term forecasts of the power demand of the construction site based on historical data and real-time data;

[0011] The resource optimization module is used to optimize the resource configuration of the off-grid power supply system according to the prediction results and the currently available resources;

[0012] The simulation evaluation module is used to analyze the performance of different optimization schemes in different scenarios;

[0013] The decision support module is used to combine the knowledge graph and simulation results to recommend the best power supply solution for the user;

[0014] The real-time monitoring module is used to monitor the power supply situation of the construction site in real time;

[0015] The feedback adjustment module is used to dynamically adjust the operating parameters of the demand forecasting module based on real-time monitoring and user feedback;

[0016] The report analysis module is used to generate an operation report and analysis conclusion of the power supply system.

[0017] As a further solution of the present invention, the specific steps of the structured processing of the graph construction module are as follows:

[0018] S1.1: Collect various sets of information corresponding to the power supply of the construction site from various sources such as literature, industry reports, expert interviews and databases. Use natural language processing technology to extract key entities of equipment, energy type and supplier from the collected text data;

[0019] S1.2: Use the pre-trained relationship extraction model to identify the relationship between entities, and transform the identified entities and their relationships into the structure of a knowledge graph in the form of each entity as a node and the relationship as an edge;

[0020] S1.3: Collect and integrate information from different sources through the sorting and fusion module, and assign attribute information to each node and edge based on the collected sets of information. Then traverse the constructed knowledge graph, detect and remove redundant and contradictory information, and use graph alignment technology to unify data from different sources, collect new data and user feedback, and continuously iterate and optimize the knowledge graph.

[0021] As a further solution of the present invention, the demand forecasting module performs short-term and long-term forecasting of the power demand of the construction site in the following specific steps:

[0022] S2.1: Collect historical power demand data and real-time monitoring data processed by the acquisition preprocessing module, normalize each group of processed data, divide each group of data into a training set and a validation set, and create a set of power prediction models based on the architecture of the input layer, the bidirectional GRU layer and the output layer;

[0023] S2.2: The patient feature transmission value in the training set is transferred to the prediction model. The prediction model input layer receives the historical power demand data and real-time monitoring data in the training set, and performs forward GRU unit calculation and backward GRU unit calculation on each group of data in the training set through the model Bi-GRU layer;

[0024] S2.3: The output layer activates the hidden states of the forward GRU unit and the backward GRU unit and outputs the final prediction value. The error between the prediction value and the true label is calculated by the mean square error loss function. The model parameters are then optimized using the back propagation algorithm and the gradient descent method. The validation set data is input into the prediction model to verify the performance index of the prediction model. If the model performance index does not meet the preset standard, the model's learning rate, batch size, and number of layers are adjusted, and the training is repeated.

[0025] S2.4: Repeat the training and verification of the prediction model until the preset maximum number of iterations is reached or the model performance indicators reach the preset standards. After the model training is completed, the prediction model is deployed to the construction site power supply management, and the real-time monitoring data is input into the prediction model. The power demand is predicted through the forward propagation of the model to obtain short-term and long-term power demand results, and feedback and adjustments are made based on the prediction results and actual power consumption.

[0026] As a further solution of the present invention, the resource optimization module optimizes the resource configuration of the off-grid power supply system in the following specific steps:

[0027] S3.1: Collect various data that affect resource allocation from the knowledge graph and the prediction results of the prediction model, regard resources as the flow of water, and set the basic parameters of the resource optimization model, such as initial water volume, flow rate and evaporation rate;

[0028] S3.2: Set the initial water volume for each resource based on historical data or real-time monitoring results, according to the power demand and resource consumption model, and through formula E i =d i ·D t The consumption of computing resources to simulate the evaporation of water, where E i represents the evaporation of resource i, d i represents the consumption coefficient of resource i, D t Represents the electricity demand at time t, and then simulates the precipitation of available resources based on weather forecasts and resource generation capabilities;

[0029] S3.3: Based on the resource optimization model, each set of data generated by the resource evaporation and precipitation simulation is calculated through the flow equation Q jh =k jh ·(R j -Rh ) describes the allocation of resources and simulates the transfer of resources between different devices, where Q jh represents the flow from resource j to resource h, k jh represents the flow efficiency coefficient, R j and R h They represent the current water volume of resource j and resource h respectively, and check whether the current resource allocation exceeds the maximum load or storage limit. If so, the flow is adjusted;

[0030] S3.4: Repeat the process of evaporation, precipitation and flow until the set maximum number of iterations or convergence criteria is reached, and generate an optimized resource allocation plan based on the final flow results to clarify the allocation of each resource in the power supply system. At the same time, input the generated resource allocation plan into the simulation evaluation module for analysis and evaluate the effectiveness of the plan.

[0031] As a further solution of the present invention, the simulation evaluation module analyzes the performance of different optimization solutions in different scenarios in the following specific steps:

[0032] S4.1: Based on the resource allocation scheme generated by the resource optimization module, construct the initial state of the evaluation, including the current resource configuration, the status of the equipment and the expected power demand, and use the resource allocation scheme as the root node, create child nodes according to the feasible operations of the root node to construct a set of search trees, and set the initial visit times and reward values ​​of each node to zero;

[0033] S4.2: Calculate the UCB value of each child node, and based on the calculated UCB values ​​of each group, start from the root node and select the child node with the highest UCB value layer by layer through the upper confidence selection strategy until the selected child node is a node that has not been fully expanded;

[0034] S4.3: randomly select executable actions in the current child node state to generate subsequent states, that is, generate new child nodes to add to the search tree, generate subsequent states according to the selected actions, record state information, and determine whether the preset maximum number of simulation steps has been reached;

[0035] S4.4: At the termination state of the simulation, the performance index is calculated, and the simulation results are recorded in the total reward of the current node, and the number of visits to the node is increased. The results of each simulation are back-propagated, and the number of visits and reward values ​​of the unified path nodes are updated;

[0036] S4.5: Repeat the selection, expansion, simulation and backtracking until the performance indicator change value converges to the preset threshold, stop the iteration, perform statistical analysis on the simulation results, compare the performance of the scheme in various scenarios, and feedback to the resource configuration module based on the results of the simulation evaluation to adjust the configuration plan.

[0037] As a further solution of the present invention, the specific steps of dynamic adjustment of the feedback adjustment module are as follows:

[0038] S5.1: Collect real-time monitoring data on power demand, equipment status and resource usage from the system, and collect users’ satisfaction with the current power demand forecast results of the construction site and their suggestions on resource allocation through questionnaire surveys and system feedback functions;

[0039] S5.2: Generate a group of populations and define the number, initial position and speed parameters of the individuals in the population, where each group of individuals represents a parameter configuration scheme, calculate the fitness value of each group of individuals, evaluate its performance under the current parameter configuration state, collect the position with the highest fitness of each group of individuals as the best position, and take the position with the highest fitness value among all individuals as the global best position;

[0040] S5.3: Update the speed of each group of individuals based on the calculated fitness value, optimal position and global optimal position, and then calculate the current position of the individual based on the updated speed, check whether the new position meets the preset constraints, and if not, adjust the new position, update the status of each group of individuals, and record the new position and new speed;

[0041] S5.4: Repeat fitness evaluation and position update until the fitness value change is less than the preset threshold. According to the final fitness evaluation, select the individual with the highest fitness value as the best parameter configuration scheme, combine the real-time monitoring data with user feedback, evaluate the effect of the new configuration and make adjustments.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] 1. The present invention collects various data that affect resource allocation from the knowledge graph and the prediction results of the prediction model, regards resources as the flow of water, and sets the initial water volume, flow rate and evaporation rate of the resource optimization model. The initial water volume of each resource is set based on historical data or real-time monitoring results. According to the power demand and resource consumption model, the resource consumption is calculated to simulate the evaporation of water, and then the precipitation simulation is performed on the available resources according to the weather forecast and resource generation capacity. Based on the resource optimization model, each set of data generated by the resource evaporation and precipitation simulation is described by the flow equation to simulate the transfer of resources between different devices. Check whether the current resource allocation exceeds the maximum load or energy storage limit. If so, adjust the flow and repeat the process of evaporation, precipitation and flow until the set maximum number of iterations or convergence standard is reached. Generate an optimized resource allocation plan based on the final flow result to clarify the allocation of each resource in the power supply system. At the same time, input the generated resource allocation plan into the simulation evaluation module for analysis and evaluate the effectiveness of the plan to ensure that it can respond quickly and maintain power supply stability under different environments and demand changes, improve overall energy utilization efficiency, maintain a good operating state in the face of emergencies, and enhance the robustness and reliability of the power supply system.

[0044] 2. The present invention collects various real-time monitoring data on power demand, equipment status and resource usage from the system, collects users' satisfaction with the current power demand forecast results of the construction site and suggestions for resource allocation through questionnaires and system feedback functions, generates a group of populations, and defines the number, initial position and speed parameters of the individuals in the population, wherein each group of individuals represents a parameter configuration scheme, calculates the fitness value of each group of individuals, evaluates its performance under the current parameter configuration state, collects the position with the highest fitness of each group of individuals as the optimal position, and the position with the highest fitness value among all individuals as the global optimal position, and updates each group of individuals based on the calculated fitness value, optimal position and global optimal position. The speed of each group of individuals is calculated, and the current position of the individual is calculated based on the updated speed. It is checked whether the new position meets the preset constraints. If not, the new position is adjusted, the status of each group of individuals is updated, the new position and new speed are recorded, and the fitness evaluation and position update are repeated until the fitness value change is less than the preset threshold. According to the final fitness evaluation, the individual with the highest fitness value is selected as the best parameter configuration scheme. The real-time monitoring data is combined with user feedback, the effect of the new configuration is evaluated and adjusted, the prediction error is reduced, and the overall prediction accuracy of the system is improved, which helps to better understand the relationship between past and future demands, enhance the effectiveness of prediction, reduce the need for manual intervention, and improve the intelligence level of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0046] Figure 1 This is a system block diagram of the construction site off-grid power supply optimization system based on big data fusion analysis proposed by the present invention. DETAILED DESCRIPTION

[0047] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0048] Example 1

[0049] Reference Figure 1 The construction site off-grid power supply optimization system based on big data fusion analysis includes acquisition and preprocessing module, sorting and fusion module, graph construction module, demand forecasting module, resource optimization module, simulation and evaluation module, decision support module, real-time monitoring module, feedback adjustment module and report analysis module.

[0050] The collection and preprocessing module is used to collect power data from the construction site in real time, and to clean, denoise and standardize the collected data; the sorting and fusion module is used to integrate data from different sources and generate a power demand view; the graph construction module is used to structure the various knowledge about the power supply of the construction site.

[0051] Specifically, various groups of information corresponding to the power supply of the construction site are collected from literature, industry reports, expert interviews and databases. Through natural language processing technology, key entities of equipment, energy type and supplier are extracted from the collected text data. The pre-trained relationship extraction model is used to identify the relationship between entities. The identified entities and their relationships are converted into the structure of a knowledge graph in the form of each entity as a node and the relationship as an edge. Information from different sources is collected and integrated through the sorting and fusion module, and attribute information is assigned to each node and edge according to the collected groups of information. Then, the constructed knowledge graph is traversed to detect and remove redundant and contradictory information. Graph alignment technology is used to unify data from different sources, collect new data and user feedback, and continuously iterate and optimize the knowledge graph.

[0052] The demand forecasting module is used to make short-term and long-term forecasts of construction site electricity demand based on historical data and real-time data.

[0053] Specifically, the historical power demand data and real-time monitoring data processed by the collection preprocessing module are collected, and each group of processed data is normalized, and each group of data is divided into a training set and a validation set. Based on the architecture of the input layer, the bidirectional GRU layer and the output layer, a set of power prediction models is created, and the patient feature transmission value in the training set is transmitted to the prediction model. The prediction model input layer accepts the historical power demand data and real-time monitoring data in the training set, and performs forward GRU unit calculation and backward GRU unit calculation on each group of data in the training set through the model Bi-GRU layer. The output layer activates the hidden states of the forward GRU unit and the backward GRU unit and then outputs the final prediction value. The prediction value is calculated by the mean square error loss function. The error between the real label and the real label is calculated, and the back propagation algorithm and gradient descent method are used to optimize the model parameters. The validation set data is input into the prediction model to verify the performance indicators of the prediction model. If the model performance indicators do not meet the preset standards, the learning rate, batch size, and number of layers of the model are adjusted, and the training is repeated until the preset maximum number of iterations is reached or the model performance indicators meet the preset standards. After the model training is completed, the prediction model is deployed to the construction site power supply management, and the real-time monitoring data is input into the prediction model. The power demand is predicted through the forward propagation of the model to obtain short-term and long-term power demand results, and feedback and adjustments are made based on the prediction results and actual power consumption.

[0054] The resource optimization module is used to optimize the resource configuration of the off-grid power supply system based on the prediction results and the currently available resources.

[0055] Specifically, various data affecting resource allocation are collected from the knowledge graph and the prediction results of the prediction model. Resources are regarded as the flow of water, and the basic parameters of the resource optimization model, such as the initial water volume, flow rate, and evaporation rate, are set. The initial water volume of each resource is set based on historical data or real-time monitoring results. According to the power demand and resource consumption model, the initial water volume of each resource is set based on the formula E. i =d i ·D t The consumption of computing resources to simulate the evaporation of water, where E i represents the evaporation of resource i, d i represents the consumption coefficient of resource i, D t Represents the electricity demand at time t, and then simulates the precipitation of available resources based on weather forecasts and resource generation capacity. Based on the resource optimization model, each set of data generated by resource evaporation and precipitation simulation is calculated through the flow equation Q jh =k jh ·(R j -R h ) describes the allocation of resources and simulates the transfer of resources between different devices, where Q jhrepresents the flow from resource j to resource h, k jh represents the flow efficiency coefficient, R j and R h They represent the current water volume of resource j and resource h respectively. At the same time, it is checked whether the current resource allocation exceeds the maximum load or energy storage limit. If so, the flow is adjusted and the process of evaporation, precipitation and flow is repeated until the set maximum number of iterations or convergence criteria is reached. The optimized resource allocation plan is generated according to the final flow result to clarify the allocation of each resource in the power supply system. At the same time, the generated resource allocation plan is input into the simulation evaluation module for analysis and the effectiveness of the plan is evaluated.

[0056] Example 2

[0057] Reference Figure 1 The construction site off-grid power supply optimization system based on big data fusion analysis includes acquisition and preprocessing module, sorting and fusion module, graph construction module, demand forecasting module, resource optimization module, simulation and evaluation module, decision support module, real-time monitoring module, feedback adjustment module and report analysis module.

[0058] The simulation evaluation module is used to analyze the performance of different optimization solutions in different scenarios.

[0059] Specifically, according to the resource allocation scheme generated by the resource optimization module, the initial state of the evaluation is constructed, including the current resource configuration, the status of the equipment and the expected power demand, and the resource allocation scheme is used as the root node. According to the feasible operations of the root node, child nodes are created to construct a set of search trees. The initial visit times and reward values ​​of each node are set to zero, and the UCB value of each child node is calculated. Based on the calculated UCB values ​​of each group, starting from the root node, the child node with the highest UCB value is selected layer by layer through the upper confidence selection strategy until the selected child node is a node that has not been fully expanded. In the current child node state, an executable action is randomly selected to generate a subsequent state, that is, a new The child node is added to the search tree, and the subsequent state is generated according to the selected action. The state information is recorded, and it is determined whether the preset maximum number of simulation steps has been reached. At the termination state of the simulation, the performance index is calculated, and the simulation result is recorded in the total reward of the current node, and the number of visits to the node is increased. The result of each simulation is back-propagated, and the number of visits and reward values ​​of the unified path nodes are updated. The selection, expansion, simulation and backtracking are repeated until the change value of the performance index converges to the preset threshold. The iteration is stopped, the simulation results are statistically analyzed, and the performance of the schemes in various scenarios is compared. According to the results of the simulation evaluation, feedback is given to the resource configuration module to adjust the configuration scheme.

[0060] The decision support module is used to combine the knowledge graph and simulation results to recommend the best power supply plan for users; the real-time monitoring module is used to monitor the power supply situation of the construction site in real time; the feedback adjustment module is used to dynamically adjust the operating parameters of the demand prediction module based on real-time monitoring and user feedback; the report analysis module is used to generate operation reports and analysis conclusions of the power supply system.

[0061] Specifically, the real-time monitoring data of power demand, equipment status and resource usage are collected from the system. The user's satisfaction with the current power demand forecast results of the construction site and the suggestions for resource allocation are collected through questionnaires and system feedback functions. A group of populations is generated, and the number, initial position and speed parameters of the individuals in the population are defined, where each group of individuals represents a parameter configuration scheme. The fitness value of each group of individuals is calculated, and its performance under the current parameter configuration state is evaluated. The position with the highest fitness of each group of individuals is collected as the optimal position, and the position with the highest fitness value among all individuals is taken as the global optimal position. The speed of each group of individuals is updated based on the calculated fitness value, optimal position and global optimal position. The current position of the individual is calculated based on the updated speed, and the new position is checked to see if it meets the preset constraints. If not, the new position is adjusted, the status of each group of individuals is updated, the new position and the new speed are recorded, and the fitness evaluation and position update are repeated until the fitness value change is less than the preset threshold. According to the final fitness evaluation, the individual with the highest fitness value is selected as the optimal parameter configuration scheme. The real-time monitoring data is combined with user feedback to evaluate the effect of the new configuration and make adjustments.

Claims

1. The construction site off-grid power supply optimization system based on big data fusion analysis is characterized by: It includes collection and preprocessing module, sorting and fusion module, graph construction module, demand forecasting module, resource optimization module, simulation evaluation module, decision support module, real-time monitoring module, feedback adjustment module and report analysis module; The acquisition preprocessing module is used to collect power data from the construction site in real time, and to clean, denoise and standardize the collected data; The collating and fusion module is used to integrate data from different sources and generate a power demand view; The graph construction module is used to structure and process various knowledge about power supply of construction sites; The demand forecasting module is used to make short-term and long-term forecasts of the power demand of the construction site based on historical data and real-time data; The resource optimization module is used to optimize the resource configuration of the off-grid power supply system according to the prediction results and the currently available resources; The simulation evaluation module is used to analyze the performance of different optimization schemes in different scenarios; The decision support module is used to combine the knowledge graph and simulation results to recommend the best power supply solution for the user; The real-time monitoring module is used to monitor the power supply situation of the construction site in real time; The feedback adjustment module is used to dynamically adjust the operating parameters of the demand forecasting module based on real-time monitoring and user feedback; The report analysis module is used to generate an operation report and analysis conclusion of the power supply system.

2. The construction site off-grid power supply optimization system based on big data fusion analysis according to claim 1 is characterized in that: The specific steps of the structured processing of the graph construction module are as follows: S1.1: Collect various sets of information corresponding to the power supply of the construction site from various sources such as literature, industry reports, expert interviews and databases. Use natural language processing technology to extract key entities of equipment, energy type and supplier from the collected text data; S1.2: Use the pre-trained relationship extraction model to identify the relationship between entities, and transform the identified entities and their relationships into the structure of a knowledge graph in the form of each entity as a node and the relationship as an edge; S1.3: Collect and integrate information from different sources through the sorting and fusion module, and assign attribute information to each node and edge based on the collected sets of information. Then traverse the constructed knowledge graph, detect and remove redundant and contradictory information, and use graph alignment technology to unify data from different sources, collect new data and user feedback, and continuously iterate and optimize the knowledge graph.

3. The construction site off-grid power supply optimization system based on big data fusion analysis according to claim 2 is characterized in that: The specific steps of the demand forecasting module for short-term and long-term forecasting of the power demand of the construction site are as follows: S2.1: Collect historical power demand data and real-time monitoring data processed by the acquisition preprocessing module, normalize each group of processed data, divide each group of data into a training set and a validation set, and create a set of power prediction models based on the architecture of the input layer, the bidirectional GRU layer and the output layer; S2.2: The patient feature transmission value in the training set is transferred to the prediction model. The prediction model input layer receives the historical power demand data and real-time monitoring data in the training set, and performs forward GRU unit calculation and backward GRU unit calculation on each group of data in the training set through the model Bi-GRU layer; S2.3: The output layer activates the hidden states of the forward GRU unit and the backward GRU unit and outputs the final prediction value. The error between the prediction value and the true label is calculated by the mean square error loss function. The model parameters are then optimized using the back propagation algorithm and the gradient descent method. The validation set data is input into the prediction model to verify the performance index of the prediction model. If the model performance index does not meet the preset standard, the model's learning rate, batch size, and number of layers are adjusted, and the training is repeated. S2.4: Repeat the training and verification of the prediction model until the preset maximum number of iterations is reached or the model performance indicators reach the preset standards. After the model training is completed, the prediction model is deployed to the construction site power supply management, and the real-time monitoring data is input into the prediction model. The power demand is predicted through the forward propagation of the model to obtain short-term and long-term power demand results, and feedback and adjustments are made based on the prediction results and actual power consumption.

4. The construction site off-grid power supply optimization system based on big data fusion analysis according to claim 3 is characterized in that: The specific steps of optimizing the resource configuration of the off-grid power supply system by the resource optimization module are as follows: S3.1: Collect various data that affect resource allocation from the knowledge graph and the prediction results of the prediction model, regard resources as the flow of water, and set the basic parameters of the resource optimization model, such as initial water volume, flow rate and evaporation rate; S3.2: Set the initial water volume for each resource based on historical data or real-time monitoring results, according to the power demand and resource consumption model, and through formula E i =d i ·D t The consumption of computing resources to simulate the evaporation of water, where E i represents the evaporation of resource i, d i represents the consumption coefficient of resource i, D t Represents the electricity demand at time t, and then simulates the precipitation of available resources based on weather forecasts and resource generation capabilities; S3.3: Based on the resource optimization model, each set of data generated by the resource evaporation and precipitation simulation is calculated through the flow equation Q jh =k jh ·(R j -R h ) describes the allocation of resources and simulates the transfer of resources between different devices, where Q jh represents the flow from resource j to resource h, k jh represents the flow efficiency coefficient, R j and R h They represent the current water volume of resource j and resource h respectively, and check whether the current resource allocation exceeds the maximum load or storage limit. If so, the flow is adjusted; S3.4: Repeat the process of evaporation, precipitation and flow until the set maximum number of iterations or convergence criteria is reached, and generate an optimized resource allocation plan based on the final flow results to clarify the allocation of each resource in the power supply system. At the same time, input the generated resource allocation plan into the simulation evaluation module for analysis and evaluate the effectiveness of the plan.

5. The construction site off-grid power supply optimization system based on big data fusion analysis according to claim 4 is characterized in that: The specific steps of the simulation evaluation module to analyze the performance of different optimization solutions in different scenarios are as follows: S4.1: Based on the resource allocation scheme generated by the resource optimization module, construct the initial state of the evaluation, including the current resource configuration, the status of the equipment and the expected power demand, and use the resource allocation scheme as the root node, create child nodes according to the feasible operations of the root node to construct a set of search trees, and set the initial visit times and reward values ​​of each node to zero; S4.2: Calculate the UCB value of each child node, and based on the calculated UCB values ​​of each group, start from the root node and select the child node with the highest UCB value layer by layer through the upper confidence selection strategy until the selected child node is a node that has not been fully expanded; S4.3: randomly select executable actions in the current child node state to generate subsequent states, that is, generate new child nodes to add to the search tree, generate subsequent states according to the selected actions, record state information, and determine whether the preset maximum number of simulation steps has been reached; S4.4: At the termination state of the simulation, the performance index is calculated, and the simulation results are recorded in the total reward of the current node, and the number of visits to the node is increased. The results of each simulation are back-propagated, and the number of visits and reward values ​​of the unified path nodes are updated; S4.5: Repeat the selection, expansion, simulation and backtracking until the performance indicator change value converges to the preset threshold, stop the iteration, perform statistical analysis on the simulation results, compare the performance of the scheme in various scenarios, and feedback to the resource configuration module based on the results of the simulation evaluation to adjust the configuration plan.

6. The construction site off-grid power supply optimization system based on big data fusion analysis according to claim 3 is characterized in that: The specific steps of dynamic adjustment of the feedback adjustment module are as follows: S5.1: Collect real-time monitoring data on power demand, equipment status and resource usage from the system, and collect users’ satisfaction with the current power demand forecast results of the construction site and their suggestions on resource allocation through questionnaire surveys and system feedback functions; S5.2: Generate a group of populations and define the number, initial position and speed parameters of the individuals in the population, where each group of individuals represents a parameter configuration scheme, calculate the fitness value of each group of individuals, evaluate its performance under the current parameter configuration state, collect the position with the highest fitness of each group of individuals as the best position, and take the position with the highest fitness value among all individuals as the global best position; S5.3: Update the speed of each group of individuals based on the calculated fitness value, optimal position and global optimal position, and then calculate the current position of the individual based on the updated speed, check whether the new position meets the preset constraints, and if not, adjust the new position, update the status of each group of individuals, and record the new position and new speed; S5.4: Repeat fitness evaluation and position update until the fitness value change is less than the preset threshold. According to the final fitness evaluation, select the individual with the highest fitness value as the best parameter configuration scheme, combine the real-time monitoring data with user feedback, evaluate the effect of the new configuration and make adjustments.

Images