Remote intelligent shore power management system

Abstract

The invention relates to a remote intelligent shore power management system, and belongs to the technical field of shore power management, and the system comprises a data collection module which is used for obtaining the initial data of a shore power system in real time; the edge calculation module is used for cleaning and standardizing the initial data on an edge calculation node to obtain target data; the cloud decision module is used for performing power load prediction on the cloud based on the completely trained long-short term memory network model and the target data to obtain power load prediction data, and generating a regulation and control strategy based on the power load prediction data; and the execution module is used for regulating and controlling the shore power system based on the regulation and control strategy. The demand for transmitting data to the cloud is reduced through the edge computing module, network delay is reduced, parameter prediction and comprehensive analysis of the shore power system are performed through an intelligent algorithm trained in the cloud decision module, real-time data are converted into decision information to regulate and control the system, and efficient, safe and intelligent management is achieved.

Description

Technical Field

[0001] The present application relates to the technical field of shore power management, and in particular to a shore power remote intelligent management system. Background Art

[0002] Shore power systems can replace ships’ own power generation when they are at anchor, significantly reducing environmental pollution and noise, and are crucial to promoting the vision of “green ports”.

[0003] In the existing technology, the traditional shore power management system relies on manual operation to confirm parameters and operate equipment. Operators need to check instrument readings regularly to ensure normal power supply. If an abnormality is found, manual adjustment or corresponding measures need to be taken, which is slow to respond and inefficient. Summary of the invention

[0004] In view of this, it is necessary to provide a shore power remote intelligent management system to solve the problems of slow response and low efficiency in the prior art.

[0005] In order to solve the above problems, the present application provides a shore power remote intelligent management system, including: Data collection module, used to obtain the initial data of the shore power system in real time; An edge computing module, used for cleaning and standardizing the initial data on an edge computing node to obtain target data; A cloud decision module is used to perform power load forecasting in the cloud based on a well-trained long short-term memory network model and the target data, obtain power load forecasting data, and generate a control strategy based on the power load forecasting data; An execution module is used to regulate the shore power system based on the regulation strategy.

[0006] In some possible implementations, the edge computing module includes a data cleaning unit, a data aggregation unit, and a data standardization unit, and the target data includes target time series data; The data cleaning unit is used to filter noise and remove invalid data from the initial data to obtain cleaned data; The data aggregation unit is used to aggregate and integrate the cleaned data from different sensors and edge computing nodes to form unified time series data to obtain initial time series data; The data standardization unit is used to perform standardization processing on the initial time series data to obtain target time series data.

[0007] In some possible implementations, the data cleaning unit further includes a feature extraction unit, and the target data further includes feature data; The feature extraction unit is used to extract features from the target time series data to obtain feature data.

[0008] In some possible implementations, the data cleaning unit further includes a preliminary analysis unit, and the target data further includes local operation status data; The preliminary analysis unit is used to perform inductive analysis on the target time series data and the characteristic data to obtain local operation status data.

[0009] In some possible implementations, the long short-term memory network model includes an input layer, a first LSTM layer, a second LSTM layer, and an output layer.

[0010] In some possible implementations, the fully trained long short-term memory network model is obtained through the following steps: Constructing the hierarchical structures of the long short-term memory network model, defining the hyperparameter space and obtaining a data set, the data set including a training set and a validation set; Based on the cane rat optimization algorithm and the data set, the hyperparameters of the long short-term memory network model are optimized with the goal of maximizing the accuracy of the long short-term memory network model with hyperparameter configuration and training on the validation set, and the optimal hyperparameters and their corresponding fully trained long short-term memory network model are obtained.

[0011] In some possible implementations, based on the cane rat optimization algorithm and the data set, the hyperparameters of the long short-term memory network model are optimized with the goal of maximizing the accuracy of the long short-term memory network model with hyperparameter configuration and training on the validation set, and the optimal hyperparameters and the corresponding fully trained long short-term memory network model are obtained, including: S401, determining the population size, maximum number of iterations and exploration and exploitation balance parameters in the sugarcane rat optimization algorithm, and using the hyperparameter combination as the position of the sugarcane rat, using the accuracy of the long short-term memory network model trained by each sugarcane rat corresponding to the hyperparameter configuration evaluated on the verification set as the fitness, and initializing the population; S402, calculating the fitness of each cane rat in the population, selecting the cane rat with the highest fitness from the population as the dominant male, and updating the positions of other cane rats based on the position of the dominant male; S403, determining whether the current number of iterations is less than a preset value; S404, if the current number of iterations is less than a preset value, then, according to the current number of iterations and the fitness of the dominant male, the influence factor for simulating abundant food sources, the first coefficient for simulating changes in food sources, and the second coefficient for guiding the cane rats to move to other available food sources are calculated and updated, and based on the exploration and exploitation balance parameter, the influence factor, the first coefficient, and the second coefficient, the cane rats in the population are caused to enter the exploration stage or the exploitation stage; S405, ensuring that the hyperparameters corresponding to all cane rats are within the preset range, increasing the number of iterations by one, and returning to step S402; S406. If the current number of iterations is not less than the preset value, the iteration is stopped to obtain the optimal hyperparameter combination and its corresponding fully trained long short-term memory network model.

[0012] In some possible implementations, the formula for updating the position of the cane rat during the exploration phase is:

[0013] In the formula, Indicates i A new location for cane rats, Indicates i The previous position of the cane rat, Indicates the position of the dominant male, represents the random number of food sources and shelters used to simulate dispersion, represents the influencing factors used to simulate abundant food sources; The formula for updating the position of cane rats in the development stage is:

[0014] In the formula, represents the position of a random female, represents the random number used to simulate the number of pups produced by each female cane rat each year.

[0015] In some possible implementations, the loss function for training the long short-term memory network model uses an adaptive weighted loss function, and the calculation formula is:

[0016]

[0017]

[0018] In the formula, represents the adaptive weighted loss, Indicates the actual power load, is the predicted power load, Indicates at time The adaptive weight of represents the adaptive weight for the next time step, represents the error feedback function, represents the adjustment factor, Indicates i The prediction error of each sample point.

[0019] In some possible implementations, the system further includes a feedback and adaptive control module; The adaptive control module is used to dynamically adjust the long short-term memory network model in the cloud decision module according to the operation results and environmental feedback of the edge computing module and the cloud decision module.

[0020] The beneficial effect of the present application is that the shore power remote intelligent management system provided by the present application, when faced with massive real-time data under various changes in working conditions, first reduces the need for data transmission to the cloud through the edge computing module, reduces network latency, and then uses the trained intelligent algorithm in the cloud decision module to predict and comprehensively analyze the parameters of the shore power system, converts real-time data into decision-making information, and finally uses the execution module to timely regulate and optimize the shore power system, with fast response and high efficiency, to achieve efficient, safe and intelligent management. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A schematic diagram of the structure of an embodiment of the shore power remote intelligent management system provided by the present application; Figure 2 A schematic diagram of the structure of an embodiment of the edge computing module provided in this application; Figure 3 A schematic flow chart of an embodiment of the steps for obtaining a fully trained long short-term memory network model provided in the present application; Figure 4 For this application Figure 3 A flowchart of an embodiment of step S302 in FIG. DETAILED DESCRIPTION

[0022] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0023] It should be understood that the schematic drawings are not drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of the present application. It should be understood that the operations of the flowcharts may be implemented out of order, and steps without logical contextual relationships may be reversed in order or implemented simultaneously. In addition, those skilled in the art may add one or more other operations to the flowcharts, or may remove one or more operations from the flowcharts, under the guidance of the content of this application. Some of the block diagrams shown in the accompanying drawings are functional entities, and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor systems and / or microcontroller systems.

[0024] The descriptions of "first", "second", etc. involved in the embodiments of the present application are only used to describe the implicit purpose, and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the technical features defined as "first" and "second" may explicitly or implicitly include at least one of the features. "And / or" describes the association relationship of the associated objects, indicating that three relationships may exist, for example: A and / or B, which can represent: A exists alone, A and B exist at the same time, and B exists alone.

[0025] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0026] The present application provides a shore power remote intelligent management system, which is described below.

[0027] Figure 1 A schematic diagram of an embodiment of the shore power remote intelligent management system provided in this application is as follows: Figure 1 As shown, the shore power remote intelligent management system 1 includes: The data collection module 11 is used to obtain the initial data of the shore power system in real time, and the initial data includes the operation data (including voltage, current and power factor, etc.) and environmental data (including temperature, humidity and wind speed, etc.) of the shore power system; The edge computing module 12 is used to deploy edge computing nodes with computing, storage and network functions at a location close to the data source of the shore power system, and to clean and standardize the initial data on the edge computing nodes to obtain target data; A cloud decision module 13 is used to perform power load forecasting in the cloud based on a well-trained long short-term memory network model and target data, obtain power load forecast data, and generate a control strategy based on the power load forecast data; Specifically, the cloud-based decision-making module 13 performs multi-dimensional decision support based on the power load forecast data and generates a control strategy. The multi-dimensional decision support includes performance evaluation: based on the power load forecast data, the actual operating efficiency of the shore power system is evaluated, and the inefficiencies in the operation of the shore power system are identified to help managers understand the operating status and optimization space of the system; decision support: based on the power load forecast data, considering the power demand in different time periods, providing decision-making recommendations on the energy use of the shore power system, and then optimizing the dispatching strategies of renewable energy, energy storage systems and main power grids within the port. For example, for possible system high-load time periods, resources can be deployed in advance to avoid system overload, so that the port shore power system can operate smoothly and economically.

[0028] The execution module 14 is used to regulate the shore power system based on the regulation strategy.

[0029] Specifically, the execution module 14 will drive various hardware devices in the shore power system (such as inverters, switchgear, load balancers, etc.) to achieve precise regulation of voltage, current, and power factor, as well as rapid response when necessary, such as emergency power-off protection, load transfer, etc., to ensure that the system operates efficiently according to the optimization strategy.

[0030] Compared with the existing technology, when faced with massive real-time data under various working conditions, this application first reduces the need for data transmission to the cloud through the edge computing module, reduces network latency, and then uses the trained intelligent algorithm in the cloud decision module to predict and comprehensively analyze the parameters of the shore power system, converting real-time data into decision-making information. Finally, the execution module is used to timely regulate and optimize the shore power system, with fast response and high efficiency, to achieve efficient, safe and intelligent management.

[0031] In order to obtain better target data, in some embodiments, such as Figure 2 As shown, the edge computing module 12 includes a data cleaning unit 121, a data aggregation unit 122 and a data standardization unit 123, and the target data includes target time series data; The data cleaning unit 121 is used to filter noise and remove invalid data from the initial data to obtain cleaned data. In a specific embodiment, the data cleaning unit enhances the signal and filters out unnecessary frequency components through a signal conditioning circuit (such as a low noise amplifier LNA) at the hardware level, and dynamically adjusts the filtering parameters of the signal conditioning circuit through a filtering algorithm (such as average filtering, median filtering, LMS and RLS algorithms, etc.) at the software level to obtain a better denoising effect; The data aggregation unit 122 is used to aggregate and integrate the cleaned data from different sensors and edge computing nodes to form unified time series data, and obtain initial time series data; The data standardization unit 123 is used to perform standardization processing on the initial time series data, converting the initial time series data of different dimensions to the same scale to obtain the target time series data; It should be noted that, in a specific embodiment, the minimum-maximum normalization method is used to scale the initial time series data to [0,1], and the formula is: ,in is the initial time series data, is the minimum value of a certain type of data in the initial time series data, is the maximum value of the same type of data in the initial time series data, is the target data after normalization.

[0032] In order to more intuitively compare the operating conditions of different time periods or different devices, in some embodiments, such as Figure 2 As shown, the edge computing module also includes a feature extraction unit 124, and the target data also includes feature data; The feature extraction unit 124 is used to extract features from the target time series data to obtain feature data, where the feature data is a feature that has an important impact on the operation of the shore power system, such as voltage fluctuation amplitude, load change trend, etc.; Furthermore, in order to more intuitively compare the operating conditions of different time periods or different devices, in some embodiments, Figure 2 As shown, the edge computing module also includes a preliminary analysis unit 125, and the target data also includes local operating status data; The preliminary analysis unit 123 is used to summarize and analyze the target time series data and feature data to obtain local operation status data, which mainly consists of a real-time operation status comprehensive report, an energy efficiency and power quality report, and an equipment health status report.

[0033] In order to better obtain a well-trained long short-term memory network model, in some embodiments, such as Figure 3 As shown in Figure 1, the fully trained long short-term memory network model is obtained through the following steps: S301, constructing the hierarchical structure of the long short-term memory network model, defining the hyperparameter space and obtaining the data set, the data set includes a training set and a validation set; It should be noted that in this application, the long short-term memory network model includes an input layer, a first LSTM layer (returning each output in the sequence to provide a complete sequence for the subsequent LSTM layer), a second LSTM layer (not returning the sequence, but only returning the last output, i.e., the predicted data), and an output layer, which are used to capture the temporal relationship in the target time series data and perform load value prediction. Hyperparameters include the number of LSTM units, learning rate, batch size, and number of training rounds.

[0034] S302. Based on the cane rat optimization algorithm, the hyperparameters of the long short-term memory network model are optimized with the goal of maximizing the accuracy of the trained long short-term memory network model with well-configured hyperparameters on the validation set, and the optimal hyperparameters and their corresponding well-trained long short-term memory network model are obtained.

[0035] In order to better apply the Greater Cane Rat Algorithm (GCRA), in some embodiments, such as Figure 4 , step S302 includes: S401, determining the population size, maximum number of iterations and exploration and exploitation balance parameters in the sugarcane rat optimization algorithm, and using the hyperparameter combination as the position of the sugarcane rat, using the accuracy of the long short-term memory network model trained by each sugarcane rat corresponding to the hyperparameter configuration evaluated on the validation set as the fitness, and initializing the population; It should be noted that the population size ( ), maximum number of iterations ( ) and the exploration and development balance parameter ( It is usually set to 0.5 to maintain a balance between exploration and development).

[0036] It should also be noted that the randomly generated There are groups of hyperparameter combinations, each group represents the location of a cane rat. Vidi Cane rats ( ) is randomly generated using the following formula: , where rand Represents a random number between 0 and 1. and are the upper and lower bounds respectively.

[0037] S402, calculating the fitness of each cane rat in the population, selecting the cane rat with the highest fitness from the population as the dominant male, and updating the positions of other cane rats based on the position of the dominant male; It should be noted that the positions of other cane rats in the population are updated according to the following formula:

[0038] In the formula, Indicates i The location of the cane rats, Indicates the middle population i The previous position of the cane rat, Indicates the position of the dominant male.

[0039] S403, determining whether the current number of iterations is less than a preset value; S404, if the current number of iterations is less than a preset value, then according to the current number of iterations and the fitness of the dominant male, the influence factor for simulating the abundance of food sources, the first coefficient for simulating the change of food sources, and the second coefficient for guiding the cane rats to move to other available food sources are calculated and updated, and based on the exploration and exploitation balance parameter, the influence factor, the first coefficient, and the second coefficient, the cane rats in the population are caused to enter the exploration stage or the exploitation stage; It should be noted that the impact factor The calculation formula is: , where is the objective function value of the dominant male, is the current iteration number, Indicates the maximum number of iterations, the first coefficient and the second coefficient The calculation formula is: , , where rand is a random number between 0 and 1, represents the number of pups produced by each female cane rat each year (a random value between 1 and 4).

[0040] It should also be noted that each cane rat decides whether to enter the exploration stage or the development stage based on the relationship between the random number and the exploration and exploitation balance parameter; at the same time, the long short-term memory network model is trained using the hyperparameter configuration corresponding to the new position, and the accuracy is evaluated. If the accuracy of the new position is higher, the new position is accepted, otherwise the original position is retained.

[0041] The formula for updating the position of cane rats during the exploration phase is:

[0042] In the formula, Indicates i A new location for cane rats, Indicates i The previous position of the cane rat, Indicates the position of the dominant male, represents the first random number used to simulate scattered food sources and shelters, defining random values ​​within the boundary of the problem space, represents the impact factor; The formula for updating the position of cane rats in the development stage is:

[0043] In the formula, represents the position of a random female (selected randomly among cane rats excluding the dominant male), The second random number used to simulate the number of pups produced by each female cane rat each year, usually randomly taking a value of 1 to 4.

[0044] S405, ensuring that all hyperparameter values ​​are within a predefined range, increasing the number of iterations by one, and returning to step S402; It should be noted that after each iteration, the boundaries are checked to ensure that all hyperparameter values ​​are within the predefined range; then the fitness of each cane rat in this generation is calculated, and the position is updated according to the following formula to obtain the cane rat with the highest fitness in this generation as the dominant male, , and then enter the next iteration.

[0045] S406. If the current number of iterations is not less than the preset value, the iteration is stopped to obtain the optimal hyperparameter combination and its corresponding fully trained long short-term memory network model.

[0046] In order to better train the LSTM network model, in some embodiments, the loss function for training the LSTM network model adopts an adaptive weighted loss function, and the calculation formula is:

[0047]

[0048]

[0049] In the formula, represents the adaptive weighted loss, Indicates the actual power load, is the predicted power load, Indicates at time The adaptive weight of represents the adaptive weight for the next time step, represents the error feedback function, represents the adjustment factor, Indicates i The prediction error of each sample point can be calculated by aggregating the power load data by time period to identify peaks and troughs, and assigning initial weights with higher peaks and lower troughs. The weights can then be dynamically adjusted based on the performance of the model on the validation set. If the prediction error of the model in a certain period of time is continuously high, the weight of that period can be increased.

[0050] It should be noted that in order to prevent overfitting, the cutoff condition for LSTM model training is whether the error is improving.

[0051] In order to improve the operating efficiency and reliability of the system, in some embodiments, such as Figure 1 As shown, the shore power remote intelligent management system 1 also includes a feedback and adaptive control module 15; The adaptive control module 15 is used to dynamically adjust the long short-term memory network model in the cloud decision module 13 according to the operation results and environmental feedback of the edge computing module 12 and the cloud decision module 13, form a self-learning mechanism, improve the operation efficiency and reliability of the system, and ensure that the system operation strategy matches the actual working conditions.

[0052] To facilitate management decision making, in some embodiments, Figure 1 As shown, the shore power remote intelligent management system also includes a user interface module 16, which is used to display the target data of the edge computing module 12 and the control strategy of the cloud decision module 13, and can adjust the shore power system through manual control functions.

[0053] It should be noted that the user interface module 16 provides an intuitive operation interface, displaying real-time monitoring data, system health status display, operation suggestions, manual control options and report generation, so as to facilitate management personnel to make decisions; the real-time data monitoring interface provides real-time data monitoring function, and users can check the operating status and environmental parameters of the shore power system at any time; the system health status display interface displays the health status of the system as a whole and each module through a graphical interface; the manual control function allows users to manually adjust system parameters as needed, perform emergency operations or special settings; the report generation module can automatically generate operation reports, analysis reports and forecast reports, and supports export and printing.

[0054] Finally, the shore power remote intelligent management system 1 also includes a communication module, which adopts the safe and efficient MQTT lightweight data transmission protocol, and has data encapsulation and decapsulation, error retransmission and security authentication functions. With its small size, low power consumption and support for reliable communication under unstable networks, it ensures the integrity and security of data transmission between the data collection module 11, the edge computing module 12 and the cloud decision module 13, the execution module 14, the feedback and adaptive control module 15 and the user interface module 16, while effectively reducing the amount of data transmission and optimizing bandwidth usage.

[0055] The above is a detailed introduction to a shore power remote intelligent management system provided by the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for technical personnel in this field, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

[0056] The above is only a preferred specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed in the present application should be covered within the protection scope of the present application.

Claims

1. A shore power remote intelligent management system, characterized in that: include: Data collection module, used to obtain the initial data of the shore power system in real time; An edge computing module, used for cleaning and standardizing the initial data on an edge computing node to obtain target data; A cloud decision module is used to perform power load forecasting in the cloud based on a well-trained long short-term memory network model and the target data, obtain power load forecasting data, and generate a control strategy based on the power load forecasting data; An execution module is used to regulate the shore power system based on the regulation strategy.

2. The shore power remote intelligent management system according to claim 1 is characterized in that: The edge computing module includes a data cleaning unit, a data aggregation unit and a data standardization unit, and the target data includes target time series data; The data cleaning unit is used to filter noise and remove invalid data from the initial data to obtain cleaned data; The data aggregation unit is used to aggregate and integrate the cleaned data from different sensors and edge computing nodes to form unified time series data to obtain initial time series data; The data standardization unit is used to perform standardization processing on the initial time series data to obtain target time series data.

3. The shore power remote intelligent management system according to claim 2 is characterized in that: The data cleaning unit further includes a feature extraction unit, and the target data further includes feature data; The feature extraction unit is used to extract features from the target time series data to obtain feature data.

4. The shore power remote intelligent management system according to claim 2 is characterized in that: The data cleaning unit further includes a preliminary analysis unit, and the target data further includes local operation status data; The preliminary analysis unit is used to perform inductive analysis on the target time series data and the characteristic data to obtain local operation status data.

5. The shore power remote intelligent management system according to claim 1 is characterized in that: The long short-term memory network model includes an input layer, a first LSTM layer, a second LSTM layer and an output layer.

6. The shore power remote intelligent management system according to claim 1 is characterized in that: The fully trained long short-term memory network model is obtained through the following steps: Constructing the hierarchical structures of the long short-term memory network model, defining the hyperparameter space and obtaining a data set, the data set including a training set and a validation set; Based on the cane rat optimization algorithm and the data set, the hyperparameters of the long short-term memory network model are optimized with the goal of maximizing the accuracy of the long short-term memory network model with hyperparameter configuration and training on the validation set, and the optimal hyperparameters and their corresponding fully trained long short-term memory network model are obtained.

7. The shore power remote intelligent management system according to claim 6 is characterized in that: Based on the cane rat optimization algorithm and the data set, the hyperparameters of the long short-term memory network model are optimized with the goal of maximizing the accuracy of the long short-term memory network model with hyperparameter configuration and training on the validation set, and the optimal hyperparameters and the corresponding fully trained long short-term memory network model are obtained, including: S401, determining the population size, maximum number of iterations and exploration and exploitation balance parameters in the sugarcane rat optimization algorithm, and using the hyperparameter combination as the position of the sugarcane rat, using the accuracy of the long short-term memory network model trained by each sugarcane rat corresponding to the hyperparameter configuration evaluated on the verification set as the fitness, and initializing the population; S402, calculating the fitness of each cane rat in the population, selecting the cane rat with the highest fitness from the population as the dominant male, and updating the positions of other cane rats based on the position of the dominant male; S403, determining whether the current number of iterations is less than a preset value; S404, if the current number of iterations is less than a preset value, then, according to the current number of iterations and the fitness of the dominant male, the influence factor for simulating abundant food sources, the first coefficient for simulating changes in food sources, and the second coefficient for guiding the cane rats to move to other available food sources are calculated and updated, and based on the exploration and exploitation balance parameter, the influence factor, the first coefficient, and the second coefficient, the cane rats in the population are caused to enter the exploration stage or the exploitation stage; S405, ensuring that the hyperparameters corresponding to all cane rats are within the preset range, increasing the number of iterations by one, and returning to step S402; S406. If the current number of iterations is not less than the preset value, the iteration is stopped to obtain the optimal hyperparameter combination and its corresponding fully trained long short-term memory network model.

8. The shore power remote intelligent management system according to claim 7 is characterized in that: The formula for updating the position of cane rats during the exploration phase is: In the formula, Indicates i A new location for cane rats, Indicates i The previous position of the cane rat, Indicates the position of the dominant male, represents the random number of food sources and shelters used to simulate dispersion, represents the influencing factors used to simulate abundant food sources; The formula for updating the position of cane rats in the development stage is: In the formula, represents the position of a random female, represents the random number used to simulate the number of pups produced by each female cane rat each year.

9. The shore power remote intelligent management system according to claim 6, characterized in that: The loss function for training the long short-term memory network model uses an adaptive weighted loss function, and the calculation formula is: In the formula, represents the adaptive weighted loss, Indicates the actual power load, is the predicted power load, Indicates at time The adaptive weight of represents the adaptive weight for the next time step, represents the error feedback function, represents the adjustment factor, Indicates i The prediction error of each sample point.

10. The shore power remote intelligent management system according to claim 1, characterized in that: The system also includes a feedback and adaptive control module; The adaptive control module is used to dynamically adjust the long short-term memory network model in the cloud decision module according to the operation results and environmental feedback of the edge computing module and the cloud decision module.

Images