Construction site temporary power dynamic load regulation method based on internet of things

Abstract

The application belongs to the field of construction site safety management and energy saving technology, and provides a dynamic load regulation and control method for temporary electricity of construction site based on Internet of Things, which mainly comprises the following steps: building an electricity monitoring and sensing network for Internet of Things; collecting multi-dimensional real-time data of temporary electricity of construction site; preprocessing the collected data; establishing an electricity load prediction and regulation model by using the processed data; training, optimizing and verifying the model; realizing dynamic regulation and control of electricity load based on the trained model; adjusting the model parameters according to the actual feedback of on-site electricity; recording the regulation and control data and updating the database; establishing an electricity load continuous monitoring and dynamic regulation and control system for real-time regulation and control on demand. The application can realize accurate and dynamic management and control of temporary electricity load of construction site, avoid problems such as overload electricity and low load operation, and improve the safety and energy saving of electricity of construction site.

Description

Technical Field

[0001] This application belongs to the field of construction site safety management and energy-saving technology, specifically, it relates to a method for dynamic load control of temporary power supply at construction sites based on the Internet of Things. Background Technology

[0002] Construction sites for water conservancy and hydropower projects are generally characterized by dispersed work areas, diverse types of electrical equipment, frequent changes in construction procedures, and rapid changes in working conditions. Temporary power systems on-site cover various electrical loads, including lifting equipment, welding tools, lighting systems, and power machinery. The electrical load exhibits strong fluctuations and nonlinear characteristics depending on the construction progress, work shifts, and equipment start-up and shutdown status. Since temporary power networks are mostly planned and constructed on a temporary basis, the rationality of load allocation directly affects the electrical safety and operating costs of the construction site. If lines or equipment operate under overload for extended periods, it can easily lead to safety hazards such as overheating, tripping, or even electrical fires. Conversely, if some equipment operates under low load for extended periods, it will result in energy waste and low power efficiency, posing significant challenges to power management on the construction site.

[0003] Traditional temporary power management at construction sites often relies on manual, periodic inspections and meter readings. This method suffers from significant time lags in monitoring load status, making it difficult to promptly detect sudden load changes and abnormal operating conditions. Furthermore, load regulation typically depends on the experience and judgment of management personnel, exhibiting strong subjectivity and a lack of quantification and precision based on real-time power consumption. This model is ill-suited to the complex and ever-changing power demands of construction sites, unable to proactively intervene before overload risks arise, and unable to optimize low-load operation, thus failing to balance the dual goals of ensuring power safety and reducing energy consumption. Summary of the Invention

[0004] To address the issues of lag, subjectivity, and inefficiency in traditional temporary power management methods at construction sites, and to achieve precise, quantitative, and dynamic control of temporary power load at construction sites, this application provides an Internet of Things-based dynamic load control method for temporary power at construction sites.

[0005] In one approach, a method for dynamic load control of temporary power supply at construction sites based on the Internet of Things (IoT) includes the following steps: Deploy IoT sensing terminals and transmission modules to build an IoT power consumption monitoring and sensing network covering all scenarios of temporary power consumption at construction sites; The construction site collects multi-dimensional electricity consumption data from all areas of production, living and office using an IoT-based electricity monitoring and sensing network, and then preprocesses the collected electricity consumption data. Using preprocessed electricity consumption data, a training set, a validation set, and a test set are set up. An electricity load prediction and dynamic control model is established and trained by combining an LSTM long short-term memory network with a particle swarm optimization algorithm to predict and control the electricity load at the construction site. Based on the power load prediction and dynamic control model, and combined with the load allocation formula, overload control formula, and load rate priority formula, the system performs dynamic allocation, overload early warning, and intelligent control of temporary power load at the construction site.

[0006] Furthermore, the deployment of IoT sensing terminals and transmission modules includes: IoT sensor devices for monitoring current, voltage, power, and temperature are deployed at the main power distribution cabinet, distribution boxes, and equipment terminals for temporary power supply at the construction site. Electricity metering terminals are deployed in production, living, and office areas to collect electricity data from both point and area perspectives at the construction site.

[0007] Build a cloud platform or local server for construction site electricity data, and use wireless transmission technology in combination with wired transmission equipment to construct an Internet of Things data transmission network, so that the data from the sensing terminals can be transmitted to the cloud platform or local server in real time.

[0008] Furthermore, the collected multi-dimensional electricity consumption data includes real-time electricity operation data, basic data of electrical equipment, construction site operating conditions data, and historical electricity consumption data.

[0009] Furthermore, the logic for establishing the electricity load forecasting and dynamic control model is as follows: An LSTM (Long Short-Term Memory) network combined with a particle swarm optimization algorithm is used to construct a sub-model for electricity load prediction. At the same time, a load regulation sub-model is constructed based on a multi-objective optimization algorithm, forming a composite model that integrates time series prediction and intelligent regulation.

[0010] Furthermore, the training logic for the electricity load forecasting and dynamic control model is as follows: The preprocessed electricity consumption data is divided into training set, validation set and test set according to the proportion. The training set data is input into the composite model. The load forecast value and control strategy are calculated through forward propagation. The prediction error and control deviation are calculated through back propagation. At the same time, regularization and hyperparameter optimization are used to prevent overfitting.

[0011] Furthermore, the verification and testing logic for the electricity load forecasting and dynamic control model is as follows: The test set data is input into the model to calculate the load forecast accuracy of the model. The mean absolute error, root mean square error, and mean absolute percentage error are used as evaluation indicators.

[0012] The model simulates power consumption conditions at a construction site, including full load, peak load, low load, and frequent equipment start-stop, to test the effectiveness of the model's control strategy. By continuously inputting real-time power consumption data, the model's response speed, command accuracy, and fault-free operation time under continuous operation are tested to verify the model's stability.

[0013] The model's load prediction accuracy, regulation effectiveness, and stability are used as comprehensive evaluation indicators for model performance. If the model performance meets the preset target, the model training is deemed qualified. If the model performance does not meet the preset target, the model parameters are readjusted and a second training is conducted.

[0014] Furthermore, the calculation logic of the load allocation formula is as follows: when the load reaches the warning threshold, the model generates a load allocation strategy based on the power consumption priority of the construction process through the load dynamic allocation formula, and performs frequency conversion speed regulation and time-sharing start and stop on low-priority electrical equipment to control the load within a safe range.

[0015] Furthermore, the calculation logic of the overload control formula is as follows: when the load reaches the over-limit threshold, the model automatically calculates the load of unnecessary electrical equipment that needs to be cut off through the overload control formula, under the premise of ensuring the power supply of the core construction process, in order to prevent safety accidents such as circuit overload and short circuit.

[0016] Furthermore, the calculation logic of the load rate priority formula is as follows: the model monitors the load rate of each electrical device in real time, determines the operating status of the device through the load rate optimization formula, issues adjustment prompts to devices whose rated power is greater than the used power and which maintain low load continuous operation, proposes equipment usage optimization strategies, and guides on-site staff to optimize equipment usage methods and improve power efficiency.

[0017] The beneficial effects of this application are: This application achieves full-scenario coverage of temporary power consumption at construction sites by establishing an IoT-based power consumption monitoring and sensing network. Combined with real-time total load calculation logic, it completes real-time quantitative collection and calculation of multi-dimensional power consumption data from the main distribution cabinet to individual devices, enabling timely and accurate identification of power consumption anomalies at construction sites and improving the real-time performance and precision of load monitoring. A composite model using LSTM combined with particle swarm optimization algorithm is employed to achieve forward-looking load prediction. By combining multi-objective optimization algorithms and quantitative control formulas to generate scientific control strategies, it can achieve dynamic allocation of power load and precise overload shedding, effectively mitigating power safety risks. Furthermore, it can continuously adjust model parameters and core parameters of the control formula based on actual power consumption feedback from the construction site, adapting the control system to the dynamic changes in construction procedures and power equipment, enhancing the long-term application value of the control system, and achieving precise, quantitative, and dynamic control of temporary power load at construction sites. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the method flow of a method for dynamic load control of temporary power supply at construction sites based on the Internet of Things in one embodiment of this application; Figure 2 This is a flowchart illustrating the construction of an electricity load prediction and dynamic control model in one embodiment of this application. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0021] Unless otherwise defined, all technical and scientific terms used in this application have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application.

[0022] In one embodiment, please refer to Figure 1 A method for dynamic load control of temporary power supply at construction sites based on the Internet of Things includes the following steps: Step 1: Build an IoT power consumption monitoring and sensing network. By deploying IoT sensing terminals and transmission modules, construct an IoT power consumption monitoring and sensing network that covers all scenarios of temporary power consumption at the construction site.

[0023] Specifically, the deployment of IoT sensing terminals and transmission modules includes: deploying IoT sensor devices for monitoring current, voltage, power, and temperature at the main power distribution cabinet, distribution boxes, and electrical equipment terminals for temporary power supply at the construction site; deploying electricity metering terminals in production, living, and office areas to collect electricity data from the construction site using a point-to-area approach. A cloud platform or local server for the construction site's electricity data is built, configured with data storage modules, and employing wireless transmission technologies such as 5G / 4G, LoRa, and Bluetooth, along with wired transmission equipment such as industrial switches, to construct a low-latency, high-stability IoT data transmission network, enabling real-time transmission of data from sensing terminals to the cloud platform or local server.

[0024] In particular, smart switches, frequency converters and other IoT control terminals can be deployed in power distribution cabinets and electrical equipment to support remote and automatic power control based on cloud platform commands.

[0025] Step 2: Multi-dimensional electricity consumption data collection and preprocessing. Collect multi-dimensional electricity consumption data of the entire construction site, including production, living and office areas, through the Internet of Things electricity consumption monitoring and sensing network, and preprocess the collected electricity consumption data.

[0026] Specifically, the collected multi-dimensional electricity consumption data includes real-time electricity operation data, basic data of electrical equipment, construction site operating condition data, and historical electricity consumption data. Real-time electricity operation data includes the total load of the construction site, the load distribution in each area, and core electricity consumption data such as power, current, voltage, power factor, and duration of electricity consumption for individual equipment. Basic data of electrical equipment includes basic information such as rated power, rated voltage, load rate, equipment type, and operating conditions for construction machinery, office and living equipment, etc. Construction site operating condition data includes operating condition data affecting electricity load, such as construction progress, work area distribution, equipment start-up and shutdown plans, and on-site personnel. In addition, external data such as ambient temperature, humidity, weather conditions, and grid voltage fluctuations can be collected, as well as historical electricity load data and energy-saving control case data from similar construction projects, forming a multi-dimensional electricity consumption dataset that provides a multi-source data foundation for subsequent model prediction and control strategy production.

[0027] In this embodiment, the formula for calculating the real-time total load at the construction site is: in, for The total effective power load at the construction site at all times. Number of areas requiring electricity For the first The number of electrical devices in each area for Time of the first Region 1 Real-time power of the device for Time of the first Region 1 The power factor of the equipment.

[0028] In this embodiment, the collected electricity consumption data undergoes preprocessing including data cleaning, standardization, normalization, and feature extraction. Missing, outlier, and duplicate values ​​are removed from the collected data; data interrupted due to monitoring and sensing equipment malfunctions is interpolated to eliminate noise; the collected electricity consumption data is scaled proportionally to fall within a defined small range, removing the influence of data units and eliminating magnitude differences between data points; Z-score standardization is used; then, MinMax is applied to transform all electricity consumption data to the range [0,1], completing data normalization. Finally, feature values ​​such as load change rate, peak occurrence time, equipment load rate, and regional electricity consumption ratio are extracted from the original electricity consumption data to construct the model input feature set.

[0029] Step 3: Establish an electricity load prediction and dynamic control model. Using the preprocessed electricity data, set up training, validation and test sets, and establish and train an electricity load prediction and dynamic control model constructed by LSTM long short-term memory network combined with particle swarm optimization algorithm to predict and control the electricity load at the construction site.

[0030] In this embodiment, please refer to Figure 2 The power load forecasting and dynamic control model is established by using an LSTM long short-term memory network combined with a particle swarm optimization algorithm to construct a power load forecasting sub-model, and at the same time, a load control sub-model is constructed based on a multi-objective optimization algorithm, forming a composite model that integrates time series forecasting and intelligent control.

[0031] Specifically, the parameters of the LSTM network, such as the number of hidden layer nodes, learning rate, number of iterations, and batch size, are first initialized to generate a weight matrix of the same dimension as the input feature set. The LSTM model is then used as the sub-model for electricity load prediction. The key parameters of the LSTM model are treated as particles in a particle swarm optimization (PSO) system, with the minimum load prediction error as the fitness function. The optimal parameter combination is found through the global search capability of the PSO system to optimize the prediction accuracy of the LSTM model. Then, using multi-objective optimization functions such as total load not exceeding limits, matching equipment load rates, and low energy consumption, constraints such as construction process power priority, equipment start-up and shutdown constraints, and load allocation thresholds are set to construct a load regulation sub-model based on linear programming. A core regulation formula is embedded into the load regulation sub-model to provide mathematical support for the model's regulation logic.

[0032] Furthermore, the output of the electricity load forecasting sub-model is used as the input condition for the load control sub-model, realizing a composite model of "load forecasting - strategy generation - command issuance" to ensure the foresight and accuracy of the control strategy. The fused composite model also requires parameter initialization. Based on the specific requirements of the construction site, parameters such as the core parameters of the control formula, the threshold for issuing control commands, execution delay, and feedback verification time are set to complete the composite model initialization.

[0033] In this embodiment, the preprocessed electricity consumption data is divided into a training set, a validation set, and a test set in a ratio of 7:2:1. The training set is used for training the electricity load prediction and dynamic control model, the validation set is used for parameter adjustment, and the test set is used for model performance testing.

[0034] Specifically, the training set data is input into the composite model. Forward propagation is used to calculate the predicted load and control strategy, while backpropagation is used to calculate the prediction error and control deviation. The model weights and control formula parameters are continuously adjusted. The learning rate and number of iterations are dynamically adjusted using an exponential decay method, with the following adjustment formula: in, for Learning rate at any given moment The initial learning rate, The attenuation coefficient is... The number of iterations is set; a model convergence threshold is set, and training stops when the prediction error is less than the convergence threshold or when the maximum number of iterations is reached.

[0035] Furthermore, a hybrid L1+L2 regularization approach is used to optimize the composite model. A weight penalty term is added to the loss function to prevent overfitting. The formula for the hybrid regularization loss function is as follows: in, The loss value. The original loss function, , For regularization parameters, The L1 norm of the weight parameters, Let L2 be the norm of the weight parameters.

[0036] Simultaneously, a combination of grid search and cross-validation is used to optimize hyperparameters, namely, optimizing hyperparameters such as the number of hidden layer nodes, regularization parameters, particle swarm population size, and correction coefficients of the control formula, and selecting the hyperparameter combination with the best performance on the validation set.

[0037] Furthermore, based on the electrical specifications at the construction site, the rationality of the control strategies generated by the composite model based on the control formula can be verified, strategies that do not meet the requirements of construction safety and procedures can be eliminated, and the strategy generation logic can be optimized.

[0038] In this embodiment, the verification and testing of the electricity load forecasting and dynamic control model are carried out by inputting test set data into the model, calculating the load forecasting accuracy of the model, and using mean absolute error (MAE), root mean square error (RMSE), and mean absolute percentage error (MAPE) as evaluation indicators for verification testing.

[0039] Specifically, the effectiveness test simulates different power consumption conditions at a construction site, including full load, peak load, low load, and frequent equipment start-ups and shutdowns, to evaluate the model's control strategy's performance in achieving reasonable load allocation, overload warnings, and energy consumption reduction. The stability test continuously inputs real-time power consumption data to test the model's response speed, command accuracy, and fault-free operating time under long-term continuous operation. Finally, the model's load prediction accuracy, control effectiveness, and stability are used as comprehensive evaluation indicators of model performance. If the model performance meets the preset target, the model training is deemed successful; if the model performance does not meet the preset target, the model parameters are readjusted for a second training.

[0040] Step 4: Implementation of dynamic load control for temporary power consumption. Based on the trained and verified power load prediction and dynamic control model, combined with control formulas such as load allocation formula, overload control formula, and load rate priority formula, dynamic allocation, overload early warning, and intelligent control of temporary power load at the construction site are carried out.

[0041] In this embodiment, the load forecasting and early warning can be generated by a composite model based on real-time electricity consumption data to predict the trend of electricity load changes within the target forecast period in the future. When the predicted load is about to reach the early warning threshold of the rated load of the distribution cabinet, the platform will issue an early warning. When the load is about to reach the over-limit threshold of the rated load, an emergency overload warning will be issued.

[0042] As a preferred implementation, the target prediction period is 15 minutes to 2 hours in the future, the warning threshold is 80% of the rated load of the distribution cabinet, and the over-limit threshold is 95% of the rated load of the distribution cabinet.

[0043] In this embodiment, the formula for calculating load allocation is: in, For the first time after regulation Region 1 Distribute the load of the equipment. For the first Electricity priority coefficient for the region, core construction area Non-core areas ; For the first Load rate correction factor for each device. Adapt to the rated load rate of the equipment; for System load correction factor at time. The load is dynamically calculated based on the difference between the real-time total load and the warning threshold. That is, when the load reaches the warning threshold, the model generates a load allocation strategy based on the power consumption priority of the construction process through a dynamic load allocation formula, and performs variable frequency speed regulation and time-sharing start-stop on low-priority electrical equipment to control the load within a safe range.

[0044] In this embodiment, the calculation logic for overload control is as follows: When the load reaches the over-limit threshold, the model automatically calculates the load of unnecessary electrical equipment that needs to be cut off through the overload control formula, while ensuring the power supply for the core construction process, to prevent safety accidents such as circuit overload and short circuit. The overload control formula is: in, The amount of overload that needs to be reduced. The number of non-core electrical equipment. for Time of the first The real-time load of non-core equipment in Taiwan is selected by the model based on the load value. Matching non-core equipment is cut off to achieve precise overload shearing.

[0045] In this embodiment, the calculation logic for prioritizing equipment load rate control is based on a composite model that monitors the load rate of each electrical device in real time. The operating status of the equipment is determined through a load rate optimization formula. Adjustment prompts are issued to devices whose rated power exceeds their operating power and which maintain low-load continuous operation. Equipment usage optimization strategies are proposed to guide on-site personnel in optimizing equipment usage and improving power efficiency. The priority load rate calculation formula is as follows: like If so, the equipment is determined to be operating under low load. for Time of the first Region 1 The actual load rate of the equipment. This is the rated effective power of the equipment. For the standard load rate threshold of the equipment, construction machinery Office and living equipment .

[0046] Furthermore, after the control terminal executes the control instructions corresponding to each control formula, it feeds back the equipment operating status and load change data to the cloud platform in real time. The real-time total load of the construction site is recalculated using the real-time total load calculation formula to verify the control effect and realize closed-loop management of the control process.

[0047] Furthermore, based on the actual feedback results of electricity consumption at the construction site, the model parameters and control formula parameters can be periodically adjusted to dynamically iterate and optimize the composite electricity load prediction and dynamic control model, thereby improving the model's adaptability.

[0048] Specifically, this can be achieved by collecting actual electricity load data, equipment operation data, energy consumption data, and adjustment suggestions from on-site personnel after the model has adjusted the control formula, thus forming a model optimization feedback dataset. Based on this feedback dataset, the model's prediction threshold and control priority coefficients can be adjusted periodically. Load rate correction factor System load correction factor and equipment standard load rate threshold The core parameters of the control formula are optimized to enable the model to adapt to the dynamic changes in construction procedures, electrical equipment, and work surfaces at the construction site. For new types of electrical equipment and special construction conditions encountered at the construction site, the model's algorithm logic can be iteratively upgraded simultaneously, adding new feature values ​​and constraints, while the control formula is expanded and modified to adapt to new power consumption scenarios. The control strategies can be updated in the load control strategy library based on the energy-saving effect and safe operation of power consumption at the construction site, matching different parameter combinations of the control formula with specific power consumption conditions, adding personalized and scenario-based control strategies, and enhancing the model's practical application value.

[0049] In summary, this application achieves full coverage of temporary power consumption at construction sites by building an IoT-based power monitoring and sensing network. Combined with real-time total load calculation logic, it completes real-time quantitative collection and calculation of multi-dimensional power consumption data from the main distribution cabinet to individual devices, overcoming the lag of manual inspection. This enables timely and accurate identification of power consumption anomalies, significantly improving the real-time performance and precision of load monitoring. Furthermore, it employs a composite model combining LSTM and particle swarm optimization to achieve forward-looking load prediction. By combining multi-objective optimization algorithms and quantitative control formulas to generate scientific control strategies, it can achieve dynamic load allocation and precise overload shedding, effectively mitigating power safety risks. Finally, through load rate optimization logic, it can promptly identify low-load operating equipment, improving on-site power efficiency and energy saving levels. This application can also continuously adjust the model parameters and core parameters of the control formula based on the actual feedback of electricity consumption at the construction site, so that the control system can adapt to the dynamic changes of construction procedures and electrical equipment, and enhance the long-term application value of the control system. At the same time, the automated and intelligent operation of the monitoring and control process greatly reduces the workload of manual inspection and metering, reduces the manpower and material costs of on-site electricity management, and avoids the subjective errors caused by manual judgment, thus comprehensively improving the overall benefits of temporary electricity management at the construction site.

[0050] The above are merely optional embodiments of this application and are not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for dynamic load control of temporary power supply at construction sites based on the Internet of Things, characterized in that, The method includes the following steps: Deploy IoT sensing terminals and transmission modules to build an IoT power consumption monitoring and sensing network covering all scenarios of temporary power consumption at construction sites; The IoT power consumption monitoring and sensing network collects multi-dimensional power consumption data from the entire construction site, including production, living, and office areas, and preprocesses the collected power consumption data. Using preprocessed electricity consumption data, a training set, a validation set, and a test set are set up. An electricity load prediction and dynamic control model is established and trained by combining an LSTM long short-term memory network with a particle swarm optimization algorithm to predict and control the electricity load at the construction site. Based on the power load prediction and dynamic control model, and combined with the load allocation formula, overload control formula, and load rate priority formula, the system performs dynamic allocation, overload early warning, and intelligent control of temporary power load at the construction site.

2. The method for dynamic load control of temporary power supply at construction sites based on the Internet of Things as described in claim 1, characterized in that, The deployment of IoT sensing terminals and transmission modules includes: IoT sensor devices for monitoring current, voltage, power, and temperature are deployed at the main power distribution cabinet, distribution boxes, and equipment terminals for temporary power supply at the construction site; electricity metering terminals are deployed in production, living, and office areas to collect electricity data from both point and area perspectives at the construction site. Build a cloud platform or local server for construction site electricity data, and use wireless transmission technology in combination with wired transmission equipment to construct an Internet of Things data transmission network, so that the data from the sensing terminals can be transmitted to the cloud platform or local server in real time.

3. The method for dynamic load control of temporary power supply at construction sites based on the Internet of Things as described in claim 1, characterized in that, The collected multi-dimensional electricity consumption data includes real-time electricity operation data, basic data of electrical equipment, construction site operating conditions data, and historical electricity consumption data.

4. The method for dynamic load control of temporary power supply at construction sites based on the Internet of Things as described in claim 1, characterized in that, The logic for establishing the electricity load forecasting and dynamic control model is as follows: An LSTM (Long Short-Term Memory) network combined with a particle swarm optimization algorithm is used to construct a sub-model for electricity load prediction. At the same time, a load regulation sub-model is constructed based on a multi-objective optimization algorithm, forming a composite model that integrates time series prediction and intelligent regulation.

5. The method for dynamic load control of temporary power supply at construction sites based on the Internet of Things as described in claim 4, characterized in that, The training logic for the electricity load forecasting and dynamic control model is as follows: The preprocessed electricity consumption data is divided into training set, validation set and test set according to the proportion. The training set data is input into the composite model. The load forecast value and control strategy are calculated through forward propagation. The prediction error and control deviation are calculated through back propagation. At the same time, regularization and hyperparameter optimization are used to prevent overfitting.

6. The method for dynamic load control of temporary power supply at construction sites based on the Internet of Things as described in claim 5, characterized in that, The verification and testing logic for the electricity load forecasting and dynamic control model is as follows: The test set data is input into the model to calculate the load forecast accuracy of the model. The mean absolute error, root mean square error, and mean absolute percentage error are used as evaluation indicators. The model simulates power consumption conditions at a construction site, including full load, peak load, low load, and frequent equipment start-stop, to test the effectiveness of the model's control strategy. By continuously inputting real-time power consumption data, the model's response speed, command accuracy, and fault-free operation time under continuous operation are tested to verify the model's stability. The model's load prediction accuracy, regulation effectiveness, and stability are used as comprehensive evaluation indicators for model performance. If the model performance meets the preset target, the model training is deemed qualified. If the model performance does not meet the preset target, the model parameters are readjusted and a second training is conducted.

7. The method for dynamic load control of temporary power supply at construction sites based on the Internet of Things as described in claim 1, characterized in that, The calculation logic of the load allocation formula is as follows: When the load reaches the warning threshold, the model generates a load allocation strategy based on the power priority of the construction process and the load dynamic allocation formula. It then uses frequency conversion speed regulation and time-sharing start-stop for low-priority electrical equipment to keep the load within a safe range.

8. The method for dynamic load control of temporary power supply at construction sites based on the Internet of Things as described in claim 1, characterized in that, The calculation logic of the overload control formula is as follows: When the load reaches the overload threshold, the model automatically calculates the load of unnecessary electrical equipment that needs to be cut off through the overload control formula, while ensuring the power supply for the core construction process.

9. The method for dynamic load control of temporary power supply at construction sites based on the Internet of Things as described in claim 1, characterized in that, The calculation logic of the load rate priority formula is as follows: The model monitors the load rate of each electrical device in real time, determines the operating status of the device through the load rate optimization formula, issues adjustment prompts for devices whose rated power is greater than the used power and which maintain low load continuous operation, and proposes optimization strategies for device use.

Images