System and method for rapid estimation of flexible load adjustable potential of port area

Abstract

The present application discloses a port flexible load adjustable potential rapid calculation system and method, comprising: a data acquisition module mainly used for real-time acquisition of port power load data; a load dynamic baseline module establishes a load dynamic baseline model based on equipment historical operation data, and generates a load dynamic baseline curve influenced by multiple factors; an adjustable potential calculation module compares real-time operation data with the load dynamic baseline curve of the corresponding period to calculate and obtain the load adjustable interval of each load equipment; an edge computing node cluster module is configured with a real-time data processing engine to realize distributed parallel calculation of the load adjustable potential; and a dispatching optimization module generates a load adjustment instruction set influenced by multiple factors based on the power grid demand response instruction and the adjustable potential data reported by each subarea edge node; the present application has the advantages that: through the load dynamic baseline curve and edge computing, the load adjustable potential is accurately calculated, and the flexibility and reliability of the port power load management are improved.

Description

Technical Field

[0001] This invention relates to potential assessment, and particularly to a system and method for rapid assessment of the adjustable potential of flexible loads in port areas. Background Technology

[0002] The background technology for the rapid calculation method of port power load adjustment potential stems from the challenges posed by the ever-increasing power demand and complex power dispatching during the modernization of ports. As important logistics hubs, ports play a crucial role in global trade, thus their power load demand is highly dynamic and variable. Port power load is affected by multiple factors, including the operational status of port facilities, cargo throughput, number of ships berthing, intensity of loading and unloading operations, and the power supply situation in the surrounding area. With the expansion of port scale and the increase in automation, the demand for electricity in ports is growing, leading to increased difficulty in power dispatching, especially during peak load periods, when the power system faces greater pressure in allocation and load balancing.

[0003] Current load adjustability potential calculation systems on the market typically suffer from slow response times, insufficient accuracy, and poor real-time performance. Traditional methods mostly rely on static load models or simplified prediction algorithms, failing to effectively account for the complex multi-factor environment and dynamic changes in equipment within port areas, resulting in low flexibility and accuracy in load regulation. These systems often depend on a single, centralized computing model, which easily encounters computational bottlenecks when processing massive amounts of data, impacting the efficiency and accuracy of real-time scheduling. Furthermore, traditional methods do not adequately consider the impact of equipment health and environmental changes on load fluctuations in load baseline prediction, making prediction errors more likely. Summary of the Invention

[0004] To improve existing systems and methods for calculating load adjustability potential, this paper presents a rapid system and method for calculating the flexible load adjustability potential in port areas. This method utilizes intelligent algorithms and edge computing to accurately calculate load adjustability potential and optimize scheduling schemes. By leveraging dynamic baseline models and machine learning techniques, the system can efficiently predict load fluctuations, enhancing the flexibility and reliability of power load management in port areas.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] The rapid calculation system for the port area's flexible load adjustment potential includes:

[0007] Data acquisition module: The data acquisition module is mainly used to acquire port area power load data in real time, including smart meter data, equipment operating status data, environmental sensor data and port operation scheduling data;

[0008] Load dynamic baseline module: The load dynamic baseline module establishes a load dynamic baseline model based on historical equipment operating data and generates a load dynamic baseline curve influenced by multiple factors;

[0009] Adjustable potential calculation module: The adjustable potential calculation module calculates and obtains the load adjustable range of each load device by comparing real-time operating data with the load dynamic baseline curve of the corresponding time period;

[0010] Edge computing node cluster module: The edge computing node cluster module is deployed in various functional areas of the port terminal operation area, warehousing and logistics area, and office service area. Each edge computing node is equipped with a real-time data processing engine to realize distributed parallel computing with adjustable load potential.

[0011] The scheduling optimization module optimizes scheduling schemes across multiple time scales based on power grid demand response commands and adjustable potential data reported by edge nodes of each zone, generating a load adjustment command set influenced by multiple factors.

[0012] Interactive interface module: The interactive interface module provides a three-dimensional visualization operation interface, supports dynamic parameter adjustment and comparative analysis between various scheduling schemes.

[0013] Preferably, the data acquisition module specifically includes:

[0014] Electrical parameter data and topology information data are obtained through a distributed array of smart meters.

[0015] Data on the status of power equipment, the health of electrical equipment, and operational status parameters are acquired through embedded monitoring terminals.

[0016] Meteorological parameters and corrosive environmental parameters are obtained through multi-parameter environmental monitoring stations;

[0017] The port operation system acquires real-time data on operation plans, equipment scheduling instructions, and energy constraints.

[0018] Preferably, the load dynamic baseline module specifically includes:

[0019] Feature extraction unit: The feature extraction unit is mainly used to fuse multi-source data such as equipment running sequence, environmental parameters and port operation events, and extract high-dimensional feature vectors containing periodic features, event-driven responses and environmental sensitive factors;

[0020] Hybrid model architecture unit: The hybrid model architecture unit uses a hybrid architecture of GRU neural network and LightGBM, combined with attention mechanism to enhance the impact of key operation events and optimize time series prediction results;

[0021] Dynamic baseline generation unit: The dynamic baseline generation unit is mainly used to calibrate historical similar working conditions based on dynamic time warping, and output adaptive baseline curves and flexible adjustment ranges.

[0022] Preferably, the feature extraction unit specifically includes:

[0023] Based on Fourier transform, daily periodic and weekly harmonic components and port-specific periods are extracted to extract periodic features;

[0024] Based on the load change gradient of the leading time window of loading and unloading operation events, event-driven features are extracted;

[0025] Temperature load response curves are constructed by fitting the slope of load change in different temperature zones using piecewise linear regression.

[0026] Preferably, the hybrid model architecture unit specifically includes:

[0027] Based on a three-layer gated cyclic unit network, 72 hours of historical equipment operation time series data are processed to build a basic time series model. Input features from the feature extraction unit and perform multi-source data fusion through feature splicing and normalization.

[0028] Enhanced modeling is achieved by using a spatiotemporal attention mechanism, which sets attention weight coefficients for the time periods before and after device operation in the time dimension.

[0029] High-order nonlinear modeling of discrete features is performed using LightGBM to construct an interpretive model, and key influencing factors are screened by ranking features by importance.

[0030] Based on the time-series baseline model and the interpretive model, a hybrid model is constructed. Through multi-task joint training, the main task performs load baseline regression prediction, while the auxiliary task completes load fluctuation range estimation and abnormal operating condition binary classification.

[0031] Preferably, the dynamic baseline generation unit specifically includes:

[0032] Based on the acquired hybrid model, the multidimensional distance between the current operating condition data and the historical operating data of the equipment is calculated using the DTW algorithm, and the similarity is also calculated.

[0033] The five sets of time series prediction data with the highest similarity were selected as the baseline to generate the candidate set, and outlier data were removed.

[0034] The current load sequence is mapped to the historical sequence on a nonlinear time axis, the matched historical sequence is dynamically time-calibrated, and the load dynamic baseline curve is synthesized by weighted averaging.

[0035] Output the load dynamic baseline generation results for each time period, and add job event timestamp annotations.

[0036] Preferably, the adjustable potential calculation module specifically includes:

[0037] Based on the real-time collected load data, it is interpolated and aligned with the load dynamic baseline curve within the corresponding time period.

[0038] Based on baseline deviation calculation, the deviation value is obtained. Positive deviation indicates the load can be adjusted downwards, and negative deviation indicates the load can be adjusted upwards.

[0039] Based on the real-time health scores of each device, a dynamic attenuation coefficient is applied to the adjustable potential, and the load adjustable range is dynamically corrected.

[0040] Cross-validate the calculation results of adjacent nodes in the edge computing device to eliminate misjudgments caused by single-point data anomalies.

[0041] Furthermore, a rapid calculation method for the port area's flexible load adjustment potential includes:

[0042] The port area's electrical parameters and topology data are collected through a distributed smart meter array, and the status characteristics of equipment vibration and temperature monitoring terminals are acquired simultaneously.

[0043] Baseline prediction is performed based on a three-layer GRU-LightGBM hybrid model, and optimized by performing dynamic time warping;

[0044] Based on the edge computing cluster, the port area load adjustment potential is calculated, dynamic deviation analysis is performed, and cross-node collaborative verification is conducted.

[0045] Multi-timescale scheduling optimization is carried out based on the adjustable load potential of the port area.

[0046] Compared with the prior art, the advantages of the present invention are:

[0047] Through multi-level data acquisition, dynamic baseline generation, adjustable potential calculation, and scheduling optimization, the system effectively improves the accuracy and response speed of power load control in the port area. By collecting real-time power load data, equipment operating status, environmental sensor data, and port operation scheduling data, and combining advanced machine learning and data analysis technologies, the system establishes a dynamic load baseline model to accurately predict load change trends. The adjustable potential calculation module quickly calculates the adjustable range of load equipment by comparing real-time load data with the dynamic baseline curve, and performs parallel processing through an edge computing node cluster, ensuring the efficiency and real-time performance of large-scale data processing. Furthermore, the system corrects the load adjustable potential through a dynamic attenuation coefficient, eliminating the impact of varying equipment health conditions and ensuring the stability and reliability of the scheduling scheme. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of the system proposed in this invention;

[0049] Figure 2 This is a schematic diagram of the method proposed in this invention;

[0050] Figure 3 This is a flowchart of the data acquisition module proposed in this invention;

[0051] Figure 4 This is a schematic diagram of the load dynamic baseline module proposed in this invention;

[0052] Figure 5 This is a flowchart of the feature extraction unit proposed in this invention;

[0053] Figure 6 This is a flowchart of the hybrid model architecture unit proposed in this invention;

[0054] Figure 7 This is a flowchart of the dynamic baseline generation unit proposed in this invention;

[0055] Figure 8 This is a flowchart of the adjustable potential calculation module proposed in this invention;

[0056] Figure 9 This is an architecture diagram of the electronic devices in this solution;

[0057] Figure 10 This is a schematic diagram of the computer-readable storage medium structure in this scheme. Detailed Implementation

[0058] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0059] See Figure 1 As shown, the rapid calculation system for the port area's flexible load adjustment potential includes:

[0060] Data acquisition module: The data acquisition module is mainly used to acquire port area power load data in real time, including smart meter data, equipment operating status data, environmental sensor data and port operation scheduling data;

[0061] Load dynamic baseline module: The load dynamic baseline module establishes a load dynamic baseline model based on historical equipment operating data and generates a load dynamic baseline curve influenced by multiple factors;

[0062] Adjustable potential calculation module: The adjustable potential calculation module calculates and obtains the load adjustable range of each load device by comparing real-time operating data with the load dynamic baseline curve of the corresponding time period;

[0063] Edge computing node cluster module: The edge computing node cluster module is deployed in various functional areas of the port terminal operation area, warehousing and logistics area, and office service area. Each edge computing node is equipped with a real-time data processing engine to realize distributed parallel computing with adjustable load potential.

[0064] The scheduling optimization module optimizes scheduling schemes across multiple time scales based on power grid demand response commands and adjustable potential data reported by edge nodes of each zone, generating a load adjustment command set influenced by multiple factors.

[0065] Interactive interface module: The interactive interface module provides a three-dimensional visualization operation interface, supports dynamic parameter adjustment and comparative analysis between various scheduling schemes.

[0066] See Figure 2 As shown, the rapid calculation method for the adjustable potential of flexible load in the port area includes:

[0067] The port area's electrical parameters and topology data are collected through a distributed smart meter array, and the status characteristics of equipment vibration and temperature monitoring terminals are acquired simultaneously.

[0068] Baseline prediction is performed based on a three-layer GRU-LightGBM hybrid model, and optimized by performing dynamic time warping;

[0069] Based on the edge computing cluster, the port area load adjustment potential is calculated, dynamic deviation analysis is performed, and cross-node collaborative verification is conducted.

[0070] Multi-timescale scheduling optimization is carried out based on the adjustable load potential of the port area.

[0071] See Figure 3 As shown, the data acquisition module specifically includes:

[0072] Electrical parameter data and topology information data are obtained through a distributed array of smart meters.

[0073] Data on the status of power equipment, the health of electrical equipment, and operational status parameters are acquired through embedded monitoring terminals.

[0074] Meteorological parameters and corrosive environmental parameters are obtained through multi-parameter environmental monitoring stations;

[0075] The port operation system acquires real-time data on operation plans, equipment scheduling instructions, and energy constraints.

[0076] See Figure 4 As shown, the load dynamic baseline module specifically includes:

[0077] Feature extraction unit: The feature extraction unit is mainly used to fuse multi-source data such as equipment running sequence, environmental parameters and port operation events, and extract high-dimensional feature vectors containing periodic features, event-driven responses and environmental sensitive factors;

[0078] Hybrid model architecture unit: The hybrid model architecture unit uses a hybrid architecture of GRU neural network and LightGBM, combined with attention mechanism to enhance the impact of key operation events and optimize time series prediction results;

[0079] Dynamic baseline generation unit: The dynamic baseline generation unit is mainly used to calibrate historical similar working conditions based on dynamic time warping, and output adaptive baseline curves and flexible adjustment ranges.

[0080] See Figure 5 As shown, the feature extraction unit specifically includes:

[0081] Based on Fourier transform, daily periodic and weekly harmonic components and port-specific periods are extracted to extract periodic features;

[0082] Based on the load change gradient of the leading time window of loading and unloading operation events, event-driven features are extracted;

[0083] Temperature load response curves are constructed by fitting the slope of load change in different temperature zones using piecewise linear regression.

[0084] Specifically, based on the acquired time series data x(t), the frequency domain signal data is obtained through Fast Fourier Transform, as shown in the formula:

[0085]

[0086] Where f is the frequency and j is the imaginary unit;

[0087] From X(f) obtained by Fourier transform, the amplitude |X(f)| and phase arg(X(f)) information at different frequencies are extracted. The frequencies corresponding to the daily cycle (24 hours) and the weekly cycle (7 days) are as follows:

[0088]

[0089] If a port has special periodicity (such as each shift, working day, etc.), specific periodic components can be extracted by observing the frequency components related to these periods in the spectrum, extracting the harmonic components of the daily and weekly periods, such as the first, second, and third harmonics, and observing their amplitude and phase changes to determine the significance of the periodicity characteristics.

[0090] The load change gradient refers to the rate of change of the load value within a time window. It is obtained by calculating the difference in load within the time window, using the following formula:

[0091]

[0092] Where Δx is the size of the time window. Observe the gradient change pattern within the time window. If the gradient change is large, it can be considered that the load fluctuation may be driven by an upcoming work event.

[0093] Based on the distribution of temperature data T(t), different temperature ranges are divided. For each temperature range, a linear regression model is used to fit the relationship between load and temperature, and the slope of load change is calculated, which represents the rate of change of load with temperature within that temperature range.

[0094] See Figure 6 As shown, the hybrid model architecture unit specifically includes:

[0095] Based on a three-layer gated cyclic unit network, 72 hours of historical equipment operation time series data are processed to build a basic time series model. Input features from the feature extraction unit and perform multi-source data fusion through feature splicing and normalization.

[0096] Enhanced modeling is achieved by using a spatiotemporal attention mechanism, which sets attention weight coefficients for the time periods before and after device operation in the time dimension.

[0097] High-order nonlinear modeling of discrete features is performed using LightGBM to construct an interpretive model, and key influencing factors are screened by ranking features by importance.

[0098] Based on the time-series baseline model and the interpretive model, a hybrid model is constructed. Through multi-task joint training, the main task performs load baseline regression prediction, while the auxiliary task completes load fluctuation range estimation and abnormal operating condition binary classification.

[0099] Specifically, GRU is a variant of gated recurrent unit network that can effectively handle long-term dependencies in time-series data. The input layer receives 72 hours of processed historical runtime data, forming a network of shape (n... s amples, 72, n f The three-dimensional tensor of time series is used to learn long-term dependencies of time series using a three-layer GRU network:

[0100] h t =GRU(h t-1 x t )

[0101] Among them, h t In the hidden state, x t Input for the current time step;

[0102] The main task of the output layer is to perform load baseline regression prediction using the time-series features extracted by the GRU layer, while the auxiliary tasks are load fluctuation range estimation and abnormal operating condition binary classification.

[0103] Discrete features (such as equipment type, working status, etc.) are converted into a format suitable for the LightGBM model. LightGBM is used to model discrete features. The model is trained by selecting an appropriate loss function. The features are ranked by the feature importance method, and the features that have the most impact on load forecasting are selected.

[0104] Based on the combination of time series model and interpretive model, stronger prediction ability is achieved through multi-task joint training. Its main task is to perform load baseline regression prediction, and obtain the load baseline prediction through GRU network and spatiotemporal attention mechanism. The auxiliary task is to estimate the load fluctuation range, use regression method to predict the upper and lower limits of load fluctuation, and perform binary classification of abnormal operating conditions, and use classification method to predict whether the equipment is in abnormal operating conditions.

[0105] The total loss function of the hybrid model is obtained by weighted summation of the loss functions of the primary task and the auxiliary task, as shown in the formula:

[0106] L total =λ1L MSE +λ2L CE +λ3L regression

[0107] Among them, L MSE Main task loss function, L CE This is the loss function for auxiliary tasks.

[0108] See Figure 7 As shown, the dynamic baseline generation unit specifically includes:

[0109] Based on the acquired hybrid model, the multidimensional distance between the current operating condition data and the historical operating data of the equipment is calculated using the DTW algorithm, and the similarity is also calculated.

[0110] The five sets of time series prediction data with the highest similarity were selected as the baseline to generate the candidate set, and outlier data were removed.

[0111] The current load sequence is mapped to the historical sequence on a nonlinear time axis, the matched historical sequence is dynamically time-calibrated, and the load dynamic baseline curve is synthesized by weighted averaging.

[0112] Output the load dynamic baseline generation results for each time period, and add job event timestamp annotations.

[0113] Specifically, based on the current sequence x current ={x1, x2, ..., x T} and historical sequence Xi ={x i1 x i2 , ..., x iT Let DTW distance matrix D be defined, where D(t,j) represents the cumulative minimum distance from the current sequence to a subset of historical sequences, and the formula is:

[0114] D(t,j)=dist(x) t x ij )+min(D(t-1,j-1),D(t-1,j),D(t,j-1))

[0115] Where, dist(x) t x ij () is the distance metric between the current time t and the historical time j, usually using Euclidean distance;

[0116] Outlier removal is performed on the five historical time series data with the smallest DTW distance between the current load series and the historical series. Then, the time axis is nonlinearly mapped using radial basis functions to transform the time coordinates of the historical series into the time coordinates of the current load data.

[0117] Dynamic time calibration is performed on the matched historical time series data. Interpolation methods are used to adjust the historical data so that the data points on the time axis correspond one-to-one with the current operating condition data points. The calibrated historical data is then aligned with the time steps of the current load sequence.

[0118] Based on the calibrated historical sequences, a weight is assigned to each historical sequence, and the final load dynamic baseline curve is synthesized by weighted averaging, as shown in the formula:

[0119]

[0120] in, For the calibrated historical sequence, the weight is proportional to the similarity between the current load sequence and the historical sequence, calculated using the similarity formula:

[0121]

[0122] Link the time points in the load dynamic baseline curve with the timestamps of work events to highlight key time points of equipment operation in charts or reports.

[0123] See Figure 8 As shown, the adjustable potential calculation module specifically includes:

[0124] Based on the real-time collected load data, it is interpolated and aligned with the load dynamic baseline curve within the corresponding time period.

[0125] Based on baseline deviation calculation, the deviation value is obtained. Positive deviation indicates the load can be adjusted downwards, and negative deviation indicates the load can be adjusted upwards.

[0126] Based on the real-time health scores of each device, a dynamic attenuation coefficient is applied to the adjustable potential, and the load adjustable range is dynamically corrected.

[0127] Cross-validate the calculation results of adjacent nodes in the edge computing device to eliminate misjudgments caused by single-point data anomalies.

[0128] Specifically, the deviation between the real-time load data and the dynamic baseline curve is calculated as follows: δ(t) = x current (t)-y baseline When δ(t) > 0, it indicates that the load exceeds the baseline and there is no adjustment potential; when δ(t) < 0, it indicates that the load is below the baseline and the load can be adjusted upward.

[0129] Based on health score S i (t), calculate the dynamic attenuation coefficient α i (t), this coefficient is used to adjust the load adjustability potential. It is typically calculated using a linear relationship, with the formula:

[0130]

[0131] Among them, S max To obtain the corrected load adjustable range, a dynamic attenuation coefficient is applied to the load adjustable range to the maximum value of the equipment health score.

[0132] In edge computing devices, for any two adjacent nodes j1 and j2, calculate their load deviation difference Δδ(t)=δ j1 (t)-δ j2 (t), if the difference in deviation between two nodes exceeds a certain threshold ∈, it indicates that the calculation result of one of the nodes may be abnormal. For nodes with large deviation differences, the weighted average of the neighboring nodes is used to correct the deviation value of that node. The formula is:

[0133]

[0134] Among them, w1 and w2 are weighting coefficients, which are usually inversely proportional to the distance between nodes.

[0135] In the edge computing node cluster module, the node cluster deployment is as follows:

[0136] Port area terminal operation area: The load mainly comes from loading and unloading equipment, transportation vehicles, etc., and the load is monitored and scheduled in real time.

[0137] Warehousing and logistics area: The load comes from warehousing systems, cargo storage and retrieval equipment, automated transportation systems, etc.

[0138] Office service area: The load mainly consists of office equipment, air conditioning, lighting, etc.

[0139] The scheduling optimization module makes scheduling decisions at multiple time scales (short-term, medium-term, and long-term).

[0140] Short-term dispatch optimization primarily addresses the matching between grid demand response commands and the load regulation capabilities of each node. The goal is to ensure load balance within a region under grid demand response commands, while also considering the health status of equipment and the impact of external factors.

[0141] Medium-term dispatch optimization typically focuses on the effectiveness of grid demand response commands, as well as long-term influencing factors such as equipment health status and load forecasting errors. The goal is to smooth load dispatch within a single day, avoiding excessive load fluctuations.

[0142] Long-term dispatch optimization considers more factors, including equipment maintenance cycles, grid load growth trends, and the volatility of renewable energy generation. The goal is to ensure load balance and avoid long-term load accumulation.

[0143] Based on the optimization results, load adjustment instructions are generated for each time period. For each load adjustment instruction, the corresponding adjustment cost and efficiency are given.

[0144] Furthermore, the method according to the embodiments of this application can also be achieved by means of... Figure 9 The architecture of the electronic device shown is used to implement this. For example... Figure 9 As shown, the electronic device 500 may include a bus 501, one or more CPUs 502, a read-only memory (ROM) 503, a random access memory (RAM) 504, a communication port 505 connected to a network, an input / output component 506, a hard disk 507, etc. The storage device in the electronic device 500, such as the ROM 503 or the hard disk 507, may store the rapid calculation system and method for the port area flexible load adjustable potential provided in this application. The electronic device 500 may also include a terminal interface 508. Of course, Figure 9 The architecture shown is merely exemplary and can be omitted as needed when implementing different devices. Figure 9 One or more components in the illustrated electronic device.

[0145] Figure 10 This is a schematic diagram of a computer-readable storage medium structure provided in one embodiment of this application. Figure 10The diagram illustrates a computer-readable storage medium 600 according to one embodiment of this application. The computer-readable storage medium 600 stores computer-readable instructions. When executed by a processor, the computer-readable instructions can perform the rapid calculation system and method for the adjustable potential of flexible loads in port areas according to an embodiment of this application, as described with reference to the above figures. The storage medium 600 includes, but is not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0146] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0147] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0148] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A rapid calculation system for the adjustable potential of flexible load in port areas, characterized by: include: Data acquisition module: The data acquisition module is used to acquire port area power load data in real time, including smart meter data, equipment operating status data, environmental sensor data and port operation scheduling data; Load dynamic baseline module: The load dynamic baseline module establishes a load dynamic baseline model based on the historical operating data of the equipment and generates a load dynamic baseline curve affected by multiple factors; The load dynamic baseline module includes: Feature extraction unit: The feature extraction unit is used to fuse multi-source data of equipment operation sequence, environmental parameters and port operation events to extract a high-dimensional feature vector containing periodic features, event-driven response and environmental sensitivity factors; Hybrid model architecture unit: The hybrid model architecture unit uses a hybrid architecture of GRU neural network and LightGBM, combined with attention mechanism to enhance the impact of key operation events and optimize time series prediction results; Dynamic baseline generation unit: The dynamic baseline generation unit is used to calibrate historical similar working conditions based on dynamic time warping, and output an adaptive baseline curve and elastic adjustment range; The feature extraction unit includes: Based on Fourier transform, daily periodic and weekly harmonic components and port-specific periods are extracted to extract periodic features; Based on the load change gradient of the leading time window of loading and unloading operation events, event-driven features are extracted; Temperature load response curves are constructed by fitting the slope of load change in different temperature zones using piecewise linear regression. The hybrid model architecture unit includes: Based on a three-layer gated cyclic unit network, 72 hours of historical equipment operation time series data are processed to build a basic time series model. Input features from the feature extraction unit and perform multi-source data fusion through feature splicing and normalization. Enhanced modeling is achieved by using a spatiotemporal attention mechanism, which sets attention weight coefficients for the time periods before and after device operation in the time dimension. High-order nonlinear modeling of discrete features is performed using LightGBM to construct an interpretive model, and key influencing factors are screened by ranking features by importance. Based on the time series base model and the interpretive model, a hybrid model is constructed. Through multi-task joint training, the main task performs load baseline regression prediction, and the auxiliary task completes load fluctuation range estimation and abnormal operating condition binary classification. The dynamic baseline generation unit includes: Based on the acquired hybrid model, the multidimensional distance between the current operating condition data and the historical operating data of the equipment is calculated using the DTW algorithm, and the similarity is also calculated. The five sets of time series prediction data with the highest similarity were selected as the baseline to generate the candidate set, and outlier data were removed. The current load sequence is mapped to the historical sequence on a nonlinear time axis, the matched historical sequence is dynamically time-calibrated, and the load dynamic baseline curve is synthesized by weighted averaging. Output the load dynamic baseline generation results for each time period, and add job event timestamp annotations; Adjustable potential calculation module: The adjustable potential calculation module calculates and obtains the load adjustable range of each load device by comparing real-time operating data with the load dynamic baseline curve of the corresponding time period; Edge computing node cluster module: The edge computing node cluster module is deployed in various functional areas of the port terminal operation area, warehousing and logistics area, and office service area. Each edge computing node is equipped with a real-time data processing engine to realize distributed parallel computing with adjustable load potential. The scheduling optimization module optimizes scheduling schemes across multiple time scales based on power grid demand response commands and adjustable potential data reported by edge nodes of each zone, generating a load adjustment command set influenced by multiple factors. Interactive interface module: The interactive interface module provides a three-dimensional visualization operation interface, supports dynamic parameter adjustment and comparative analysis between various scheduling schemes.

2. The rapid calculation system for the adjustable potential of flexible load in port areas according to claim 1, characterized in that: The data acquisition module specifically includes: Electrical parameter data and topology information data are obtained through a distributed array of smart meters. Data on the status of power equipment, the health of electrical equipment, and operational status parameters are acquired through embedded monitoring terminals. Meteorological parameters and corrosive environmental parameters are obtained through multi-parameter environmental monitoring stations; The port operation system acquires real-time data on operation plans, equipment scheduling instructions, and energy constraints.

3. The rapid calculation system for the adjustable potential of flexible load in port areas according to claim 1, characterized in that: The adjustable potential calculation module includes: Based on the real-time collected load data, it is interpolated and aligned with the load dynamic baseline curve within the corresponding time period. Based on baseline deviation calculation, the deviation value is obtained. Positive deviation indicates the load can be adjusted downwards, and negative deviation indicates the load can be adjusted upwards. Based on the real-time health scores of each device, a dynamic attenuation coefficient is applied to the adjustable potential, and the load adjustable range is dynamically corrected. Cross-validate the calculation results of adjacent nodes in the edge computing device to eliminate misjudgments caused by single-point data anomalies.

4. A method for rapid calculation of the adjustable potential of flexible load in port areas, using the rapid calculation system for the adjustable potential of flexible load in port areas as described in any one of claims 1-3, characterized in that: include: The port area's electrical parameters and topology data are collected through a distributed smart meter array, and the status characteristics of equipment vibration and temperature monitoring terminals are acquired simultaneously. Baseline prediction is performed based on a three-layer GRU-LightGBM hybrid model, and optimized by performing dynamic time warping; Based on the edge computing cluster, the port area load adjustment potential is calculated, dynamic deviation analysis is performed, and cross-node collaborative verification is conducted. Multi-timescale scheduling optimization is carried out based on the adjustable load potential of the port area.

5. An electronic device, characterized in that: include: At least one processor; and a memory communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform the rapid calculation method for the port area flexible load adjustment potential as described in claim 4.

6. A computer-readable storage medium storing computer-readable instructions, characterized in that: When the computer-readable instructions are executed by the processor, the method for rapid calculation of the port area's flexible load adjustment potential as described in claim 4 is implemented.

Images