Method and system for optimizing power management of temporarily built area based on artificial intelligence

Abstract

The invention discloses a temporary construction area power management optimization method and system based on artificial intelligence, and intelligent optimization of global power management is realized through an integrated device energy consumption management module, a power load management module and a power supply path optimization module. The invention relates to the technical field of temporary construction area power management, in particular to a temporary construction area power management optimization method and system based on artificial intelligence, equipment energy efficiency management adopts an improved dynamic power consumption modeling method, and high-energy-consumption equipment is accurately identified and the energy efficiency evaluation precision is improved in combination with transient characteristics and a hybrid model; according to power load management, the improved weighted clustering and multi-power-supply-node agent reinforcement learning technology is utilized, load distribution is optimized, and the self-adaptive capacity of the system is enhanced; according to power supply path optimization, dynamic adjustment of a power supply path is realized by combining a topological graph structure and a Dijkstra algorithm, line loss is reduced, and stable power supply is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of temporary construction area power management, and specifically refers to an optimization method and system for temporary construction area power management based on artificial intelligence. Background Art

[0002] The optimization method and system for temporary construction area power management based on artificial intelligence is a comprehensive technical solution integrating Internet of Things perception, big data analysis, and intelligent decision-making. The system collects multiple data such as current, voltage, and equipment status in real time by deploying an intelligent sensor network, and uses machine learning algorithms to achieve dynamic optimization management. This technology is particularly suitable for temporary power usage scenarios such as construction sites and post-disaster resettlement areas, effectively solving the problems of slow response and low energy efficiency in traditional manual management.

[0003] However, in the existing temporary construction area power management methods, there are technical problems that the existing temporary construction area power management systems usually adopt a hierarchical management method, in which equipment energy efficiency management, load management, and power supply path optimization are independent of each other, lacking an effective coordination mechanism. This decentralized management method is prone to uneven resource allocation and power system instability. The energy efficiency information of equipment is usually not combined with the load demand prediction in a timely manner, resulting in possible overload of equipment during peak load periods and waste of electricity during low demand periods.

[0004] In the existing equipment energy consumption management methods, there are technical problems that the existing methods mainly rely on traditional regular equipment inspections and static power consumption modeling. This method often ignores the transient behavior of equipment and the impact of complex load fluctuations, resulting in the inability to accurately identify high-energy-consuming equipment and difficult to analyze the abnormal behavior and transient characteristics of equipment.

[0005] In the existing power load management methods, there are technical problems that the existing methods usually based on static load prediction and conventional load distribution algorithms, unable to cope with the complex requirements of sudden changes in power demand or dynamic load distribution. Therefore, the traditional methods cannot handle load fluctuations and lack intelligent scheduling capabilities.

[0006] In the existing power supply path optimization methods, there are technical problems of untimely path optimization and slow response speed. Summary of the Invention

[0007] In view of the above situation, to overcome the defects of the prior art, the present invention provides an optimization method and system for temporary construction area power management based on artificial intelligence. In the existing temporary construction area power management methods, the existing temporary construction area power management system usually adopts a hierarchical management method, in which equipment energy efficiency management, load management, and power supply path optimization are independent of each other, lacking an effective coordination mechanism. This decentralized management method is prone to uneven resource allocation and power system instability. The energy efficiency information of equipment is usually not combined with the load demand prediction in a timely manner, resulting in equipment overload during peak loads and power waste during low demand periods. The present solution creatively adopts an integrated intelligent function of equipment energy consumption management, power load management, and power supply path optimization, realizing the intelligent optimization of global power management and improving the overall coordination efficiency of the power management system. This system can not only monitor the energy efficiency and load status of equipment in real time but also dynamically optimize the power supply path to ensure that power resources can be optimally allocated and scheduled under the conditions of equipment load changes and temporary construction area power demand fluctuations, thereby greatly improving the intelligent level of power management and the stability of system operation. In the existing equipment energy consumption management methods, the existing methods mainly rely on traditional regular equipment inspections and static power consumption modeling. This method often ignores the transient behavior of equipment and the impact of complex load fluctuations, resulting in the inability to accurately identify high-energy-consuming equipment, analyze abnormal behaviors and transient characteristics of equipment. The present solution creatively adopts a dynamic power consumption modeling method improved by combining a hybrid model and transient characteristics for equipment energy consumption management, solving the problem of insufficient accuracy in equipment energy efficiency evaluation. Through the improved transient characteristic loss function, the accuracy is improved in equipment power consumption prediction, making equipment energy efficiency management more accurate and intelligent, thereby reducing energy consumption and failure rates and improving the operation efficiency and stability of equipment. In the existing power load management methods, the existing methods usually rely on static load prediction and conventional load distribution algorithms, unable to cope with the complex demands of sudden power demand changes or dynamic load distribution. Therefore, the traditional methods cannot handle load fluctuations and lack intelligent scheduling capabilities. The present solution creatively adopts a multi-power supply node intelligent agent reinforcement learning method with improved weighted clustering to achieve dynamic balance and intelligent scheduling of power loads. By combining multi-dimensional data such as power load prediction, energy storage status, and equipment priority, power load management has an adaptive ability and can optimize loads according to real-time power demands, equipment status, and environmental changes, improving the self-recovery ability and overall operation efficiency of the power system.In view of the technical problems of untimely path optimization and slow response speed in the existing power supply path optimization methods, this solution creatively adopts an optimization algorithm that combines the standard topology structure for power supply path reconstruction to optimize the power supply path, achieving adaptive adjustment of the dynamic power supply path. In addition, intelligent switches are used to switch the power supply circuit, and visual analysis is carried out in combination with the line loss heat map, further reducing the loss during power transmission and ensuring the efficient and stable operation of the power supply system.

[0008] The technical solution adopted by the present invention is as follows: The present invention provides an optimized method for temporary construction area power management based on artificial intelligence, and this method includes the following steps:

[0009] Step S1: Data scheduling;

[0010] Step S2: Equipment energy consumption management;

[0011] Step S3: Power load management;

[0012] Step S4: Power supply path optimization;

[0013] Step S5: Temporary construction area power management.

[0014] Furthermore, in step S1, the data scheduling is used to collect, optimize, and schedule the data required for temporary construction area power management. Specifically, through data acquisition, the original power data is obtained, and through data processing and partition transmission, the intelligent functions of equipment energy consumption management, power load management, and power supply path optimization are supported, and an optimized dataset for temporary construction area power management is obtained;

[0015] The data acquisition specifically reads the original data from sensors and device controllers through industrial communication protocols, and collects power data by performing high-frequency sampling on power parameters;

[0016] The original power data includes power sensing data, device control data, environmental monitoring data, and management system data;

[0017] The data processing includes the following steps: anomaly detection, time synchronization, missing value repair, and feature extraction; the feature extraction includes harmonic component extraction and transient feature calculation;

[0018] By performing data processing operations on the original power data, an optimized dataset for temporary construction area power management is obtained;

[0019] The optimized dataset for temporary construction area power management includes time-series power parameter data, real-time data stream data, anomaly record data, current harmonic data, load prediction data, and cable impedance data;

[0020] By performing a partition transfer operation on the optimized dataset of the temporary construction area power management, the time-series power parameter data, abnormal record data, and current harmonic data are used for the equipment energy consumption management task, the time-series power parameter data and system load prediction data are used for the power load management task, and the time-series power parameter data, real-time data stream data, and cable impedance data are used for the power supply path optimization task.

[0021] Further, in step S2, the equipment energy consumption management is used to identify high-energy-consuming equipment and analyze the abnormal behavior of the temporary construction area power equipment. Specifically, based on the optimized dataset of the temporary construction area power management, a dynamic power consumption modeling method combining a hybrid model and transient characteristics is adopted to perform equipment energy consumption management, and equipment energy efficiency evaluation data is obtained, including the following steps:

[0022] Step S21: Enhanced preprocessing, specifically extracting the current waveform, voltage waveform, and equipment metadata from the time-series power parameter data, abnormal record data, and current harmonic data in the optimized dataset of the temporary construction area power management, and extracting harmonic decomposition features and transient feature data, and performing data reconstruction to obtain equipment energy consumption management input data;

[0023] Step S22: Hybrid extraction of equipment features, specifically constructing a hybrid convolutional long short-term model and optimizing the model structure to obtain a network for hybrid extraction of equipment features, and constructing an improved transient feature loss function to perform hybrid extraction of equipment features on the equipment energy consumption management input data to obtain equipment hybrid feature data;

[0024] The optimization of the model structure specifically constructs a waveform-improved input layer, a convolutional module, a bidirectional long short-term module, and a fully connected layer in sequence;

[0025] Step S23: Dynamic power consumption modeling, specifically constructing a dynamic power consumption function to perform dynamic modeling and calculation of the power consumption of each power equipment to obtain dynamic power consumption calculation data;

[0026] Step S24: Improved equipment anomaly detection, specifically performing improved equipment anomaly detection based on the equipment hybrid feature data and the dynamic power consumption calculation data through feature reconstruction and anomaly score calculation to obtain equipment anomaly estimates, and constructing a threshold trigger mechanism to perform anomaly marking when the equipment anomaly estimate is greater than the set threshold;

[0027] The feature reconstruction calculates the current harmonic distortion rate, power factor, and transient energy every 10 seconds to perform frequency feature reconstruction to obtain reconstructed feature data;

[0028] The anomaly score calculation specifically constructs an improved isolation forest algorithm and adds transient feature weights on the basis of the standard algorithm to calculate the equipment anomaly estimate;

[0029] Step S25: Energy efficiency score calculation, specifically, based on the device hybrid characteristic data and the dynamic power consumption calculation data, comprehensively calculate the device energy consumption score to obtain the device energy consumption and energy efficiency evaluation data;

[0030] Step S26: Device energy consumption management, specifically, based on the dynamic power consumption calculation data, the device anomaly estimation value, and the device energy consumption and energy efficiency evaluation data, perform device energy consumption management to obtain the device energy efficiency evaluation data.

[0031] Furthermore, in step S3, the power load management is used to dynamically balance the supply and demand of power load resources. Specifically, based on the temporary construction area power management optimization data set, an intelligent reinforcement learning method for multiple power supply nodes with improved weighted clustering is adopted to perform power load management to obtain the power load distribution plan reference data, including the following steps:

[0032] Step S31: Input data modeling, specifically, extract the load prediction data, energy storage state data, and generator fuel quantity data from the temporary construction area power management optimization data set, and define the state space parameters by constructing device priority tags to obtain the state space parameter data;

[0033] Step S32: Load improved weighted clustering, specifically, based on the temporary construction area power management optimization data set and the state space parameter data, reconstruct the feature vector to obtain the device reconstructed feature vector data, calculate the clustering center through improved weighted calculation, and perform dynamic clustering adjustment in combination with the elbow method to obtain the load clustering data;

[0034] Step S33: Multi-agent reinforcement learning, specifically, define the main grid agent, generator agent, and energy storage agent to perform multi-agent definition, construct a standard deep Q network as the agent decision network, define the action space parameters and the power load improvement reward function, and perform multi-agent reinforcement learning based on the state space parameter data to obtain the power load optimization strategy data;

[0035] The action space parameters include power-off action, frequency reduction action, and full-power operation action;

[0036] The calculation formula of the power load improvement reward function is:

[0037] ;

[0038] where r t is the power load improvement reward function, is the critical load power supply indication function, used to represent the power supply stability reward, IC is the critical load power supply judgment identifier, F diesel(t) is the generator fuel quantity data, which is used to represent the fuel consumption reward, and SOC(t) is the state of charge data, which is used to represent the energy storage balance reward;

[0039] Step S34: Dynamically adjust the load priority. Specifically, by constructing an elastic priority adjustment algorithm, dynamically adjust the load priority, and through the dynamic adjustment of the load priority, optimize the resource reallocation after the access of new devices to obtain dynamic load priority data;

[0040] The elastic priority adjustment algorithm specifically performs dynamic priority adjustment by calculating the priority change amount;

[0041] Step S35: Power load management. Specifically, based on the power load optimization strategy data and the dynamic load priority data, by constructing an agent power constraint condition, perform comprehensive power load management to obtain reference data for the power load distribution plan.

[0042] Furthermore, in step S4, the power supply path optimization is used to adaptively adjust the power supply path strategy of the temporary construction area. Specifically, based on the power management optimization data set of the temporary construction area, adopt an optimization algorithm that combines the reconstruction of the power supply path of the standard topological graph structure to perform power supply path optimization to obtain reference data for the optimal power supply path, including the following steps:

[0043] Step S41: Input data processing. Specifically, construct the power supply nodes and edge data of the temporary construction area to obtain the power supply topological graph data of the temporary construction area; the power supply nodes of the temporary construction area specifically include power supply points and load points; the edge data specifically uses the cable impedance data as the edge data;

[0044] Step S42: Reconstruct the standard topological graph. Specifically, based on the power supply topological graph data of the temporary construction area, obtain the optimized power supply topological graph data through constructing an adjacency matrix and dynamic weight calculation;

[0045] Step S43: Construct an optimization algorithm. Specifically, starting from the substation, use the standard Dijkstra algorithm to calculate the shortest path to all load nodes, and perform dynamic adjustment by updating the weight every 5 minutes to obtain the optimized path data;

[0046] Step S44: Power supply path optimization. Specifically, based on the optimized path data, use intelligent switches to automatically switch the power supply circuit, and through visual analysis of the line loss thermal map, obtain reference data for the power load distribution plan.

[0047] Further, in step S5, the temporary construction area power management is used for global monitoring and systematic optimization. Specifically, through data aggregation, combining the equipment energy efficiency evaluation data, the power load distribution scheme reference data, and the optimal power supply path reference data, a temporary construction area power comprehensive management system based on a rule engine is constructed. According to the anomaly detection situation in the equipment energy efficiency evaluation data, automatic control and manual confirmation are executed, and by relying on the power load distribution scheme reference data and the optimal power supply path reference data, the power equipment in the temporary construction area is controlled to obtain the temporary construction area power comprehensive management reference data;

[0048] The rule engine adopts the Drools engine and matches predefined rules;

[0049] The predefined rules include overload protection and energy storage scheduling.

[0050] The temporary construction area power management optimization system based on artificial intelligence provided by the present invention includes a data scheduling module, an equipment energy consumption management module, a power load management module, a power supply path optimization module, and a temporary construction area power management module;

[0051] The data scheduling module is used for data scheduling. Through data scheduling, a temporary construction area power management optimization data set is obtained, and the temporary construction area power management optimization data set is sent to the equipment energy consumption management module, the power load management module, and the power supply path optimization module;

[0052] The equipment energy consumption management module is used for equipment energy consumption management. Through equipment energy consumption management, equipment energy efficiency evaluation data is obtained, and the equipment energy efficiency evaluation data is sent to the temporary construction area power management module;

[0053] The power load management module is used for power load management. Through power load management, power load distribution scheme reference data is obtained, and the power load distribution scheme reference data is sent to the temporary construction area power management module;

[0054] The power supply path optimization module is used for power supply path optimization. Through power supply path optimization, optimal power supply path reference data is obtained, and the optimal power supply path reference data is sent to the temporary construction area power management module;

[0055] The temporary construction area power management module is used for temporary construction area power management. Through temporary construction area power management, temporary construction area power comprehensive management reference data is obtained.

[0056] The beneficial effects achieved by the present invention by adopting the above scheme are as follows:

[0057] (1)In the existing temporary construction area power management methods, the existing temporary construction area power management system usually adopts a hierarchical management method. Among them, equipment energy efficiency management, load management, and power supply path optimization are independent of each other, lacking an effective coordination mechanism. This decentralized management method easily leads to uneven resource allocation and instability of the power system. The energy efficiency information of equipment is usually not combined with the load demand prediction in a timely manner, resulting in possible overload of equipment during peak load periods and waste of electricity during low demand periods. This solution creatively adopts an integrated intelligent function of equipment energy consumption management, power load management, and power supply path optimization, realizing the intelligent optimization of global power management, improving the overall coordination efficiency of the power management system. This system can not only monitor the energy efficiency and load status of equipment in real time, but also dynamically optimize the power supply path to ensure that under the conditions of equipment load changes and power demand fluctuations in the temporary construction area, power resources can be optimally configured and dispatched, thus greatly improving the intelligent level of power management and the stability of system operation;

[0058] (2)In the existing equipment energy consumption management methods, the existing methods mainly rely on traditional regular equipment inspections and static power consumption modeling. This method often ignores the impact of the transient behavior of equipment and complex load fluctuations, thus leading to the technical problems of being unable to accurately identify high-energy-consuming equipment, difficult to analyze the abnormal behavior and transient characteristics of equipment. This solution creatively adopts a dynamic power consumption modeling method improved by combining a hybrid model and transient characteristics for equipment energy consumption management, solving the problem of insufficient accuracy in equipment energy efficiency evaluation. Through the improved transient characteristic loss function, the accuracy is improved in equipment power consumption prediction, making equipment energy efficiency management more accurate and intelligent, thereby reducing energy consumption and failure rates and improving the operation efficiency and stability of equipment;

[0059] (3)In the existing power load management methods, the existing methods usually rely on static load prediction and conventional load distribution algorithms, unable to cope with the complex demands of sudden changes in power demand or dynamic load distribution. Therefore, the traditional methods cannot handle load fluctuations and lack intelligent scheduling capabilities. This solution creatively adopts a multi-power supply node agent reinforcement learning method with improved weighted clustering to achieve dynamic balance and intelligent scheduling of power loads. By combining multi-dimensional data such as power load prediction, energy storage status, and equipment priority, power load management has an adaptive ability and can optimize loads according to real-time power demands, equipment status, and environmental changes, improving the self-recovery ability and overall operation efficiency of the power system;

[0060] (4) In view of the technical problems of untimely path optimization and slow response speed in the existing power supply path optimization methods, this solution creatively adopts an optimization algorithm that combines the reconstruction of the power supply path with the standard topology structure to optimize the power supply path, realizing the adaptive adjustment of the dynamic power supply path. In addition, intelligent switches are used to switch the power supply circuit, and visual analysis is carried out in combination with the line loss heat map to further reduce the loss during the power transmission process, ensuring the efficient and stable operation of the power supply system. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 FIG. is a schematic flow chart of the temporary construction area power management optimization method based on artificial intelligence provided by the present invention;

[0062] Figure 2 FIG. is a schematic diagram of the temporary construction area power management optimization system based on artificial intelligence provided by the present invention;

[0063] Figure 3 FIG. is a schematic flow chart of equipment energy consumption management in step S2;

[0064] Figure 4 FIG. is a schematic flow chart of power load management in step S3;

[0065] Figure 5 FIG. is a schematic flow chart of power supply path optimization in step S4.

[0066] The drawings are used to provide a further understanding of the present invention and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention and do not constitute a limitation to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0067] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0068] In the description of the present invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc. indicating the orientation or position relationship are based on the orientation or position relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation to the present invention.

[0069] Embodiment 1, refer to Figure 1, the power management optimization method for temporary construction areas based on artificial intelligence provided by the present invention includes the following steps:

[0070] Step S1: Data scheduling;

[0071] Step S2: Equipment energy consumption management;

[0072] Step S3: Power load management;

[0073] Step S4: Power supply path optimization;

[0074] Step S5: Power management for temporary construction areas.

[0075] By performing the above operations, in the existing power management method for temporary construction areas, there is a problem that the existing power management system for temporary construction areas usually adopts a hierarchical management method, in which equipment energy efficiency management, load management, and power supply path optimization are independent of each other, lacking an effective coordination mechanism. This decentralized management method is prone to uneven resource allocation and instability of the power system. The energy efficiency information of equipment is usually not combined with the load demand prediction in a timely manner, resulting in possible overload of equipment during peak load periods and waste of electricity during low-demand periods. The present solution creatively adopts an integrated intelligent function of equipment energy consumption management, power load management, and power supply path optimization, realizing the intelligent optimization of global power management and improving the overall coordination efficiency of the power management system. This system can not only monitor the energy efficiency of equipment and the load status in real time, but also dynamically optimize the power supply path to ensure that under the conditions of changes in equipment load and fluctuations in power demand in temporary construction areas, power resources can be optimally allocated and scheduled, thus greatly improving the intelligent level of power management and the stability of system operation.

[0076] Example 2, refer to Figure 1 and Figure 2 , based on the above embodiment, in step S1, the data scheduling is used to collect, optimize, and schedule the data required for power management in temporary construction areas. Specifically, through data acquisition, the original power data is obtained, and through data processing and partition transmission, the intelligent functions of equipment energy consumption management, power load management, and power supply path optimization are supported to obtain the optimized data set for power management in temporary construction areas;

[0077] The data acquisition specifically reads the original data from sensors and device controllers through industrial communication protocols and collects power data by performing high-frequency sampling on power parameters;

[0078] The original power data includes power sensing data, device control data, environmental monitoring data, and management system data;

[0079] The power sensing data includes current waveform data, voltage waveform data, power factor data, and harmonic distortion rate data;

[0080] The device control data includes device metadata and real-time operation parameter data. The device metadata includes device type, rated power, and working state parameters. The real-time operation parameter data includes air conditioner set temperature and motor load rate data;

[0081] The environmental monitoring data includes temperature, humidity, and light intensity parameters;

[0082] The management system data includes power load prediction values, energy storage status, and power transmission topology information;

[0083] The data processing includes the following steps: anomaly detection, time synchronization, missing value repair, and feature extraction; the feature extraction includes harmonic component extraction and transient feature calculation;

[0084] By performing data processing operations on the raw power data, an optimized dataset for temporary construction area power management is obtained;

[0085] The optimized dataset for temporary construction area power management includes time-series power parameter data, real-time data stream data, anomaly record data, current harmonic data, load prediction data, and cable impedance data;

[0086] By performing partition transmission operations on the optimized dataset for temporary construction area power management, the time-series power parameter data, anomaly record data, and current harmonic data are used for device energy consumption management tasks, the time-series power parameter data and system load prediction data are used for power load management tasks, and the time-series power parameter data, real-time data stream data, and cable impedance data are used for power supply path optimization tasks.

[0087] Embodiment 3, refer to Figure 1 、 Figure 2 and Figure 3 Based on the above embodiment, in step S2, the device energy consumption management is used to identify high-energy-consuming devices and analyze the abnormal behaviors of temporary construction area power equipment. Specifically, based on the optimized dataset for temporary construction area power management, a dynamic power consumption modeling method combining a hybrid model and transient features is adopted for device energy consumption management to obtain device energy efficiency evaluation data, including the following steps:

[0088] Step S21: Enhanced preprocessing, specifically, current waveforms, voltage waveforms, and device metadata are extracted from the time-series power parameter data, anomaly record data, and current harmonic data in the optimized dataset for temporary construction area power management, and harmonic decomposition features and transient feature data are extracted for data reconstruction to obtain input data for device energy consumption management;

[0089] Step S22: Hybrid extraction of device features. Specifically, by constructing a hybrid convolutional long short-term model and optimizing the model structure, a network for hybrid extraction of device features is obtained, and an improved transient feature loss function is constructed to perform hybrid extraction of device features on the input data of the device energy consumption management, obtaining device hybrid feature data;

[0090] The optimization of the model structure specifically includes successively constructing a waveform-improved input layer, a convolutional module, a bidirectional long short-term module, and a fully connected layer;

[0091] The waveform-improved input layer specifically uses the current waveform with a 200-millisecond window and 10-dimensional harmonic components as the input layer dimension;

[0092] The convolutional module specifically uses a one-dimensional convolutional layer, sets the convolutional model parameters as (kernel = 5, filters = 32, stride = 2), uses the ReLU activation function and sets the pooling dimension as a max pooling layer with a size of 2, and outputs convolutional feature data with a size of 32×49;

[0093] The bidirectional long short-term module specifically uses a 64-dimensional hidden layer and outputs a 128-dimensional time series feature vector;

[0094] The fully connected layer receives and fuses the convolutional feature data and the time series feature vector, outputs 64-dimensional device hybrid feature data, and constructs the network for hybrid extraction of device features;

[0095] The calculation formula of the improved transient feature loss function is:

[0096] ;

[0097] In the formula, is the improved transient feature loss function, max(·) is the function to find the maximum value, f(x a ) is the anchor sample feature vector, used to represent the feature representation of the target device to be recognized at the current moment, f(x p ) is the positive sample feature vector, used to represent the feature representations of devices of the same category as the anchor sample at different time periods, f(x n ) is the negative sample feature vector, used to represent the feature representations of devices of different categories from the anchor sample, ||·|| 2 is the L2 norm operator, is the margin parameter, is the transient feature weight coefficient, as a whole is the change rate of the transient feature;

[0098] Preferably, the margin parameter specifically takes the value of 0.5, and the transient feature weight coefficient The value range of [] is set to [0.1, 0.3], and the specific value is selected through cross-validation;

[0099] Step S23: Dynamic power consumption modeling, specifically, by constructing a dynamic power consumption function, dynamic modeling and calculation of the device power consumption are performed for each power device to obtain dynamic power consumption calculation data;

[0100] The calculation formula of the dynamic power consumption function is:

[0101] ;

[0102] In the formula, P dev (t) is the output of the dynamic power consumption function, used to represent the real-time power consumption of the power device at time t, t is the time index, K is the total number of devices, k is the device index, w k is the device weight coefficient, specifically trained by the least squares method, and the calculation formula is: , where, is the update operator, is the learning rate parameter, P meas (t) is the actual measured power of the device, is the non-linear mapping function of the device state characteristics, specifically using the Gaussian radial basis function, S(t) is the device state parameter, used to represent the real-time operation parameter data of the device;

[0103] Step S24: Improve device anomaly detection, specifically, through feature reconstruction and anomaly score calculation, based on the device hybrid feature data and the dynamic power consumption calculation data, perform improved device anomaly detection to obtain device anomaly estimates, and through constructing a threshold trigger mechanism, when the device anomaly estimate is greater than the set threshold, perform anomaly marking;

[0104] For the feature reconstruction, the current harmonic distortion rate, power factor, and transient energy are calculated every 10 seconds to perform frequency feature reconstruction to obtain reconstructed feature data;

[0105] For the anomaly score calculation, specifically, by constructing an improved isolation forest algorithm, adding transient feature weights on the basis of the standard algorithm, calculate the device anomaly estimate, and the calculation formula is:

[0106] ;

[0107] In the formula, Score(x) is the device anomaly estimate, x is the reconstructed feature data, E(·) is the mean function, h(x) is the path length of the input sample in the isolation tree, c(N) is the normalization factor, N is the total number of samples in the reconstructed feature data, is the transient feature weight, x trans is the transient energy in the reconstructed feature data;

[0108] Preferably, the transient feature weight is specifically set to 0.3;

[0109] The set threshold is specifically set to 0.75;

[0110] Step S25: Calculate the energy efficiency score. Specifically, based on the device hybrid feature data and the dynamic power consumption calculation data, comprehensively calculate the device energy consumption score to obtain the device energy consumption and energy efficiency evaluation data;

[0111] The calculation formula for the comprehensive calculation of the device energy consumption score is:

[0112] ;

[0113] In the formula, EfficiencyScore is the device energy consumption and energy efficiency evaluation data, P dev (t) is the output of the dynamic power consumption function, P ideal (t) is the device reference power consumption, P rated is the device rated power consumption;

[0114] Step S26: Device energy consumption management. Specifically, based on the dynamic power consumption calculation data, the device anomaly estimation, and the device energy consumption and energy efficiency evaluation data, perform device energy consumption management to obtain the device energy efficiency evaluation data.

[0115] By performing the above operations, in the existing device energy consumption management method, there are technical problems that the existing method mainly relies on traditional regular device inspections and static power consumption modeling, which often ignores the transient behavior of the device and the impact of complex load fluctuations, resulting in the inability to accurately identify high - energy - consuming devices, difficulty in analyzing the abnormal behavior and transient characteristics of the device. This solution creatively uses a dynamic power consumption modeling method combined with a hybrid model and transient feature improvement for device energy consumption management, solves the problem of insufficient accuracy in device energy efficiency evaluation, improves the accuracy in device power consumption prediction through the improved transient feature loss function, makes device energy efficiency management more accurate and intelligent, thereby reducing energy consumption and the failure rate, and improving the operation efficiency and stability of the device.

[0116] Example 4, refer to Figure 1 、 Figure 2 and Figure 4 , based on the above - mentioned example, in step S3, the power load management is used to dynamically balance the supply and demand of power load resources. Specifically, based on the temporary construction area power management optimization data set, an intelligent reinforcement learning method for multi - power supply nodes with improved weighted clustering is used for power load management to obtain the reference data for the power load distribution plan, including the following steps:

[0117] Step S31: Input data modeling, specifically extracting load prediction data, energy storage state data, and generator fuel quantity data from the prefabricated area power management optimization dataset, and defining state space parameters by constructing device priority tags to obtain state space parameter data;

[0118] The calculation formula for the state space parameter data is:

[0119] ;

[0120] In the formula, s t is the state space parameter, L pred (t) is the load prediction data, SOC(t) is the energy storage state data, F diesel (t) is the generator fuel quantity data, P dev is the device priority tag;

[0121] The device priority tag specifically includes non-critical devices (0), ordinary devices (1), and critical devices (2);

[0122] Step S32: Load improvement weighted clustering, specifically reconstructing the feature vector based on the prefabricated area power management optimization dataset and the state space parameter data to obtain device reconstruction feature vector data, calculating the clustering center through improved weighted calculation, and performing dynamic clustering adjustment in combination with the elbow method to obtain load clustering data;

[0123] The calculation formula for the device reconstruction feature vector data is:

[0124] ;

[0125] In the formula, x i is the load device reconstruction feature vector data, is the average power feature, is the peak power feature, is the device weight feature, and i is the load device index;

[0126] The calculation formula for the improved weighted calculation is:

[0127] ;

[0128] In the formula, is the output of the improved weighted distance calculation, is the clustering center, is the mean clustering center, is the peak clustering center, is the device weight clustering center;

[0129] Step S33: Multi-agent reinforcement learning, specifically by defining a main power grid agent, a generator agent, and an energy storage agent to perform multi-agent definition, and by constructing a standard deep Q-network as the agent decision network, improving the reward function by defining action space parameters and power load, and performing multi-agent reinforcement learning based on the state space parameter data to obtain power load optimization strategy data;

[0130] The action space parameters include power-off actions, frequency reduction actions, and full-power operation actions;

[0131] The calculation formula of the improved power load reward function is:

[0132] ;

[0133] In the formula, r t is the improved power load reward function, is the critical load power supply indication function, used to represent the power supply stability reward, IC is the critical load power supply judgment identifier, F diesel (t) is the generator fuel quantity data, used to represent the fuel consumption reward, and SOC(t) is the energy storage state data, used to represent the energy storage balance reward;

[0134] Step S34: Dynamic adjustment of load priority, specifically by constructing an elastic priority adjustment algorithm to perform dynamic adjustment of load priority, and through the dynamic adjustment of load priority, optimizing the resource reallocation after the access of new devices to obtain dynamic load priority data;

[0135] The elastic priority adjustment algorithm specifically performs priority dynamic adjustment by calculating the priority change amount, and the calculation formula is:

[0136] ;

[0137] In the formula, is the priority change amount, is the power demand value of the newly accessed device, is the normalized value of the current system load state, is the enhancement coefficient;

[0138] Preferably, the specific value range of the enhancement coefficient is set to [1.0, 3.0];

[0139] Step S35: Power load management, specifically based on the power load optimization strategy data and dynamic load priority data, by constructing agent power constraint conditions to perform comprehensive power load management to obtain power load distribution plan reference data;

[0140] The calculation formula of the agent power constraint condition is:

[0141] ;

[0142] In the formula, The whole is the sum of the total power of all devices, which is used to represent the power constraint condition of the agent. is the total number of load devices, i is the load device index, and P i (·) is the device power, and a i is the action decision output after multi-agent reinforcement learning, and P gridmax is the maximum power supply value of the power supply grid, and P diesel is the maximum power value of the generator, and P battery is the maximum power value of the energy storage system.

[0143] By performing the above operations, in the existing power load management methods, there are technical problems that the existing methods usually rely on static load prediction and conventional load distribution algorithms, and cannot cope with sudden changes in power demand or complex requirements of dynamic load distribution. Therefore, the traditional methods cannot handle load fluctuations and lack intelligent scheduling capabilities. This solution creatively adopts a multi-power supply node agent reinforcement learning method with improved weighted clustering to achieve dynamic balance and intelligent scheduling of power loads. By combining multi-dimensional data such as power load prediction, energy storage status, and device priority, the power load management has adaptability and can optimize the load according to real-time power demand, device status, and environmental changes, improving the self-recovery ability and overall operation efficiency of the power system.

[0144] Example 5, refer to Figure 1 、 Figure 2 and Figure 5 , based on the above example, in step S4, the power supply path optimization is used to adaptively adjust the power supply path strategy of the temporary construction area. Specifically, according to the power management optimization data set of the temporary construction area, an optimization algorithm combining the reconstruction of the power supply path with the standard topological graph structure is adopted to optimize the power supply path and obtain the optimal power supply path reference data, including the following steps:

[0145] Step S41: Input data processing, specifically constructing the power supply nodes and edge data of the temporary construction area to obtain the power supply topology graph data of the temporary construction area; the power supply nodes of the temporary construction area specifically include power supply points and load points; the edge data specifically uses the cable impedance data as the edge data;

[0146] Step S42: Standard topological graph reconstruction, specifically obtaining the optimized power supply topology graph data by constructing an adjacency matrix and calculating dynamic weights based on the power supply topology graph data of the temporary construction area;

[0147] Step S43: Construct an optimization algorithm. Specifically, starting from the substation, use the standard Dijkstra algorithm to calculate the shortest paths to all load nodes, and dynamically adjust by updating the weights every 5 minutes to obtain optimized path data;

[0148] Step S44: Optimize the power supply path. Specifically, based on the optimized path data, use intelligent switches to automatically switch the power supply circuits, and through visual analysis of the line loss heat map, obtain reference data for the power load distribution plan.

[0149] By performing the above operations, in view of the technical problems of untimely path optimization and slow response speed existing in the existing power supply path optimization methods, this solution creatively adopts an optimization algorithm that combines the reconstruction of the power supply path with the standard topology structure to optimize the power supply path, realizes the adaptive adjustment of the dynamic power supply path. In addition, intelligent switches are used to switch the power supply circuits, and visual analysis is combined with the line loss heat map to further reduce the losses in the power transmission process and ensure the efficient and stable operation of the power supply system.

[0150] Example 6, refer to Figure 1 and Figure 2 In this example, based on the above example, in step S5, the power management of the temporary construction area is used for global monitoring and systematic optimization. Specifically, through data aggregation, combining the equipment energy efficiency evaluation data, the reference data for the power load distribution plan, and the reference data for the optimal power supply path, a power comprehensive management system for the temporary construction area based on a rule engine is constructed. According to the abnormal detection situation in the equipment energy efficiency evaluation data, automatic control and manual confirmation are executed, and the power equipment in the temporary construction area is controlled based on the reference data for the power load distribution plan and the optimal power supply path to obtain reference data for the power comprehensive management of the temporary construction area;

[0151] Preferably, for the data aggregation, it is set to collect the equipment energy efficiency evaluation data, the reference data for the power load distribution plan, and the reference data for the optimal power supply path every 30 seconds and store them in the database;

[0152] The rule engine uses the Drools engine and matches predefined rules;

[0153] The predefined rules include overload protection and energy storage scheduling;

[0154] Table 1 is an example table of Drools syntax code for the predefined rules. As shown in the table, the rule parameter represents the defined rule. Node(load > 0.9*capacity) is the data model of the power node, specifically indicating that when the load exceeds 90% of the capacity, the overload protection condition is triggered. triggerAlarm(·) is the function to trigger an alarm, and shedNonCriticalLoad($node) is the function to cut off non-critical loads. $battery : Battery(SOC < 0.2) is the data model of the energy storage battery, specifically indicating that when the battery charge is lower than 20%, the energy storage scheduling condition is triggered. Time(hour >= 18 && hour <= 6) is the system time object, specifically indicating that the time range is from 18:00 in the evening to 6:00 in the morning the next day. startGenerator() is the power supply scheduling function.

[0155] Table 1 Example table of Drools syntax code for predefined rules

[0156]

[0157] Example 7, refer to Figure 1 and Figure 2 , this example is based on the above example. The optimized power management system for temporary construction areas provided by the present invention based on artificial intelligence includes a data scheduling module, a device energy consumption management module, a power load management module, a power supply path optimization module, and a temporary construction area power management module;

[0158] The data scheduling module is used for data scheduling. Through data scheduling, an optimized data set for power management in the temporary construction area is obtained, and the optimized data set for power management in the temporary construction area is sent to the device energy consumption management module, the power load management module, and the power supply path optimization module;

[0159] The device energy consumption management module is used for device energy consumption management. Through device energy consumption management, device energy efficiency evaluation data is obtained, and the device energy efficiency evaluation data is sent to the temporary construction area power management module;

[0160] The power load management module is used for power load management. Through power load management, reference data for power load distribution schemes is obtained, and the reference data for power load distribution schemes is sent to the temporary construction area power management module;

[0161] The power supply path optimization module is used for power supply path optimization. Through power supply path optimization, reference data for the optimal power supply path is obtained, and the reference data for the optimal power supply path is sent to the temporary construction area power management module;

[0162] The temporary construction area power management module is used for power management in the temporary construction area. Through the power management in the temporary construction area, comprehensive power management reference data for the temporary construction area is obtained.

[0163] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device.

[0164] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention.

[0165] The above description of the present invention and its embodiments is not restrictive. What is shown in the drawings is only one of the embodiments of the present invention, and the actual structure is not limited thereto. Generally speaking, if those of ordinary skill in the art are inspired by it and design similar structural modes and embodiments without creative efforts without departing from the purpose of the present invention, they shall fall within the protection scope of the present invention.

Claims

1. The artificial intelligence-based temporary construction area power management optimization method is characterized by: The method comprises the following steps: Step S1: Data scheduling, obtaining the temporary construction area power management optimization data set; Step S2: Equipment energy consumption management, using a dynamic power consumption modeling method that combines a hybrid model and transient feature improvements to manage equipment energy consumption and obtain equipment energy efficiency evaluation data, including the following steps: Step S21: Enhanced preprocessing; Step S22: Equipment feature hybrid extraction, by constructing a hybrid convolutional long-term and short-term model and an improved transient feature loss function, to perform equipment feature hybrid extraction; Step S23: Dynamic power consumption modeling; Step S24: Improved equipment anomaly detection, by feature reconstruction and anomaly score calculation, to improve equipment anomaly detection, and by constructing a threshold trigger mechanism for anomaly marking; Step S25: Energy efficiency score calculation; Step S26: Equipment energy consumption management; Step S3: Power load management, adopting the multi-power supply node agent reinforcement learning method with improved weighted clustering to perform power load management and obtain reference data for power load distribution scheme, including the following steps: Step S31: input data modeling; Step S32: load improvement weighted clustering; Step S33: multi-agent reinforcement learning, defining the main grid agent, generator agent and energy storage agent, and performing multi-agent reinforcement learning by defining action space parameters and power load improvement reward function; Step S34: dynamic adjustment of load priority; Step S35: power load management; Step S4: power supply path optimization, using an optimization algorithm combined with standard topology structure power supply path reconstruction to optimize the power supply path and obtain optimal power supply path reference data; Step S5: temporary construction area power management, obtain temporary construction area power comprehensive management reference data.

2. The artificial intelligence-based temporary construction area power management optimization method according to claim 1 is characterized by: In step S1, the data scheduling is used to collect, optimize and schedule the data required for power management in temporary construction areas, specifically, to obtain the original power data through data collection, and to support the intelligent functions of equipment energy consumption management, power load management and power supply path optimization through data processing and partition transmission, so as to obtain the power management optimization data set of temporary construction areas; The data collection specifically reads raw data from sensors and device controllers through industrial communication protocols, and collects power data by high-frequency sampling of power parameters; The raw power data includes power sensor data, equipment control data, environmental monitoring data and management system data; The temporary construction area power management optimization data set includes time-series power parameter data, real-time data stream data, abnormal record data, current harmonic data, load prediction data and cable impedance data; By performing partition transmission operations on the temporary construction area power management optimization data set, the time series power parameter data, abnormal record data and current harmonic data are used for equipment energy consumption management tasks, the time series power parameter data and system load prediction data are used for power load management tasks, and the time series power parameter data, real-time data stream data and cable impedance data are used for power supply path optimization tasks.

3. The artificial intelligence-based temporary construction area power management optimization method according to claim 2 is characterized by: In step S2, the equipment energy consumption management is used to identify high-energy-consuming equipment and analyze abnormal behaviors of power equipment in temporary construction areas. Specifically, based on the temporary construction area power management optimization data set, a dynamic power consumption modeling method combining a hybrid model and transient characteristics is used to perform equipment energy consumption management to obtain equipment energy efficiency evaluation data, including the following steps: Step S21: Enhanced preprocessing, specifically, extracting current waveform, voltage waveform and equipment metadata from the time-series power parameter data, abnormal record data and current harmonic data in the temporary construction area power management optimization data set, extracting harmonic decomposition characteristics and transient characteristic data, performing data reconstruction, and obtaining equipment energy consumption management input data; Step S22: extracting mixed device features, specifically by constructing a mixed convolutional long-term and short-term model and optimizing the model structure to obtain a mixed device feature extraction network, and constructing an improved transient feature loss function to perform mixed device feature extraction on the device energy consumption management input data to obtain mixed device feature data; The model structure optimization specifically includes sequentially constructing a waveform improvement input layer, a convolutional module, a bidirectional long-term and short-term module, and a fully connected layer; Step S23: Dynamic power consumption modeling, specifically, by constructing a dynamic power consumption function, dynamically modeling and calculating the power consumption of each power device, and obtaining dynamic power consumption calculation data; Step S24: improving device anomaly detection, specifically, performing improved device anomaly detection according to the device mixed feature data and the dynamic power consumption calculation data through feature reconstruction and anomaly score calculation, obtaining device anomaly estimation, and constructing a threshold trigger mechanism, when the device anomaly estimation is greater than a set threshold, marking the anomaly; The feature reconstruction is performed by calculating the current harmonic distortion rate, power factor and transient energy every 10 seconds, performing frequency feature reconstruction, and obtaining reconstructed feature data; The anomaly score calculation is specifically performed by constructing an improved isolation forest algorithm, adding transient feature weights on the basis of the standard algorithm, and calculating the device anomaly estimation; Step S25: energy efficiency score calculation, specifically, performing comprehensive calculation of the equipment energy consumption score based on the equipment mixed characteristic data and the dynamic power consumption calculation data to obtain equipment energy consumption and energy efficiency evaluation data; Step S26: equipment energy consumption management, specifically, performing equipment energy consumption management based on the dynamic power consumption calculation data, the equipment abnormality estimation data and the equipment energy consumption and energy efficiency evaluation data to obtain equipment energy efficiency evaluation data.

4. The artificial intelligence-based temporary construction area power management optimization method according to claim 3 is characterized by: In step S3, the power load management is used to dynamically balance the supply and demand of power load resources. Specifically, based on the temporary construction area power management optimization data set, a multi-power supply node intelligent agent reinforcement learning method with improved weighted clustering is used to perform power load management to obtain power load distribution scheme reference data, including the following steps: Step S31: input data modeling, specifically extracting load forecast data, energy storage status data and generator fuel quantity data from the temporary construction area power management optimization data set, and defining state space parameters by constructing equipment priority tags to obtain state space parameter data; Step S32: load improved weighted clustering, specifically, reconstructing feature vectors based on the temporary construction area power management optimization data set and the state space parameter data to obtain equipment reconstructed feature vector data, and performing cluster center calculation through improved weighted calculation, and performing dynamic clustering adjustment in combination with the elbow method to obtain load clustering data; Step S33: multi-agent reinforcement learning, specifically, by defining the main grid agent, the generator agent and the energy storage agent, performing multi-agent definition, and by constructing a standard deep Q network as the agent decision network, by defining the action space parameters and the power load improvement reward function, and performing multi-agent reinforcement learning according to the state space parameter data to obtain the power load optimization strategy data; The action space parameters include power-off action, frequency reduction action and full-power operation action; The calculation formula of the power load improvement reward function is: ； In the formula, r t is the power load improvement reward function, is the key load power supply indication function, used to indicate the power supply stability reward, IC is the key load power supply judgment identifier, F diesel (t) is the generator fuel quantity data, which is used to represent the fuel consumption reward, and SOC(t) is the energy storage state data, which is used to represent the energy storage balance reward; Step S34: Dynamically adjust the load priority, specifically by constructing an elastic priority adjustment algorithm to dynamically adjust the load priority, and optimize the resource reallocation after the new device is connected through the dynamic load priority adjustment to obtain dynamic load priority data; The elastic priority adjustment algorithm specifically performs dynamic priority adjustment by calculating the priority change amount; Step S35: Power load management, specifically, based on the power load optimization strategy data and dynamic load priority data, by constructing intelligent body power constraints, comprehensive power load management is performed to obtain power load distribution plan reference data.

5. The artificial intelligence-based temporary construction area power management optimization method according to claim 4 is characterized by: In step S4, the power supply path optimization is used to adaptively adjust the power supply path strategy of the temporary construction area. Specifically, based on the temporary construction area power management optimization data set, an optimization algorithm combined with the power supply path reconstruction of the standard topology graph structure is used to optimize the power supply path to obtain the optimal power supply path reference data, including the following steps: Step S41: input data processing, specifically constructing temporary construction area power supply nodes and edge data to obtain temporary construction area power supply topology data; the temporary construction area power supply nodes specifically include power supply points and load points; the edge data specifically uses cable impedance data as edge data; Step S42: standard topology reconstruction, specifically, obtaining optimized power supply topology data by constructing an adjacency matrix and performing dynamic weight calculation based on the temporary construction area power supply topology data; Step S43: construct an optimization algorithm, specifically, taking the substation as the starting point, using the standard Dijkstra algorithm to calculate the shortest path to all load nodes, and dynamically adjusting by updating the weight every 5 minutes to obtain optimized path data; Step S44: Optimizing the power supply path, specifically, using an intelligent switch to automatically switch the power supply circuit according to the optimized path data, and obtaining reference data for the power load distribution plan through visual analysis of the line loss heat map.

6. The artificial intelligence-based temporary construction area power management optimization method according to claim 5 is characterized by: In step S5, the temporary construction area power management is used for global monitoring and systematic optimization. Specifically, through data aggregation, the equipment energy efficiency evaluation data, the power load distribution scheme reference data and the optimal power supply path reference data are combined to build a temporary construction area power comprehensive management system based on a rule engine. According to the abnormal detection in the equipment energy efficiency evaluation data, automatic control and manual confirmation are performed, and the temporary construction area power equipment is controlled according to the power load distribution scheme reference data and the optimal power supply path reference data to obtain the temporary construction area power comprehensive management reference data.

7. An artificial intelligence-based temporary construction area power management optimization system, used to implement the artificial intelligence-based temporary construction area power management optimization method as described in any one of claims 1 to 6, characterized in that: It includes data scheduling module, equipment energy consumption management module, power load management module, power supply path optimization module and temporary construction area power management module.

8. The artificial intelligence-based temporary construction area power management optimization system according to claim 7 is characterized by: The data scheduling module is used for data scheduling, and obtains a temporary construction area power management optimization data set through data scheduling, and sends the temporary construction area power management optimization data set to the equipment energy consumption management module, the power load management module and the power supply path optimization module; The equipment energy consumption management module is used for equipment energy consumption management, obtains equipment energy efficiency evaluation data through equipment energy consumption management, and sends the equipment energy efficiency evaluation data to the temporary construction area power management module; The power load management module is used for power load management, obtains power load distribution scheme reference data through power load management, and sends the power load distribution scheme reference data to the temporary construction area power management module; The power supply path optimization module is used for power supply path optimization, obtains optimal power supply path reference data through power supply path optimization, and sends the optimal power supply path reference data to the temporary construction area power management module; The temporary construction area power management module is used for temporary construction area power management, and obtains temporary construction area power comprehensive management reference data through temporary construction area power management.

Images