Construction of power load control and management terminal operation system based on machine learning technology

Abstract

With the increasing Power Load (PL), the operation of the power system is facing increasingly severe challenges. PL control is an important means to ensure the stability of power system operation and power supply quality. However, traditional PL control methods have limitations and cannot meet the requirements of load control in the new era of power systems. This is because with the development of modern industry and commerce, the demand for electricity is gradually increasing. This article constructed a PL control and management terminal operating system based on machine learning technology to achieve intelligent management of PL, so as to improve the operational efficiency and power supply quality of the power system. This article identified the design concept of a PL control management terminal operating system based on machine learning technology by reviewing the current research status of PL control technology. Based on the operational characteristics and data characteristics of the power system, this article selected suitable machine learning algorithms to process and analyze load data, and established a prototype of a PL control and management terminal operating system based on machine learning technology, so as to realize intelligent processing and analysis of load data and conduct experimental verification. The experimental results show that through the comparative study of 6 sets of data in the tertiary level, the difference between the system and the real tertiary level is 0.079 kw, 0.005 kw and 0.189 kw respectively. Therefore, therefore, the average difference between the predicted value and the measured value of the PL system is about 0.091 kw. This indicated that the system had high accuracy and real-time performance in predicting PL, which could effectively improve the load control efficiency and power supply quality of the power system. The PL control management terminal operating system based on machine learning technology constructed in this article provided new ideas and methods for the development of PL control technology. In the future, system algorithms can be further optimized and a more intelligent PL control and management terminal operating system can be constructed to cope with the growing PL and increasingly complex power system operating environment.

Keywords

Machine learning technology power load management terminal operation systemï¼ power load forecasting model

1. Introduction

The PL control and management terminal is a very important component in the power system, which plays a crucial role in ensuring the stability and safe operation of the power system. At present, with the continuous development of the power system and the improvement of intelligence, the operation system of the PL control management terminal is also facing a series of problems and challenges. Therefore, it is meaningful to analyze the construction of a PL control and management terminal operating system based on machine learning technology, so as to solve its existing problems and provide some reference opinions for the development of the power system.

The PL control and management system in the power system plays an important role in the stable transmission of electricity, and many experts have conducted a series of studies on its important role in various fields. Atif Maqsood analyzed the current trends in naval shipborne power system architecture and believed that the electrification of future warships was inevitable. It would be equipped with PL that absorbed periodic pulse currents from the direct-current microgrid or had significant transients when switching states [1]. Tinghui Ouyang believed that load forecasting was crucial for the effective scheduling and operation of power systems, which became increasingly complex and uncertain, especially with the penetration of distributed power. He proposed a data-driven deep learning framework to predict short-term PL [2]. Xianlun Tang believed that accurate PL forecasting was of great significance for ensuring the safety, stability, and economic operation of the power system. Especially, short-term PL forecasting was the foundation of power grid planning and decision-making. In recent years, machine learning algorithms were widely used for short-term PL forecasting [3]. Million Abayneh Mengistu provided a non-invasive method for online household appliance load detection from aggregated energy consumption data. Enabling online load detection was a related research issue as it could unlock new grid services, such as demand side management. It could enhance the interactivity of energy awareness and might lead to more green behavior [4]. Bhandari Binod believed that decisions in the energy sector must be based on accurate predictions of load demand. Short term forecasting helped to make important decisions in areas such as scheduling, emergency analysis, power flow analysis, prevention of generation and load demand imbalance, and load switching strategies, thereby improving power grid reliability and energy quality [5]. Short term forecasting plays a significant role in analyzing and regulating PL, and advanced methods can be used to predict PL.

Machine learning methods have been applied in many fields, and the technology is gradually maturing in application. It also has strong advantages in predicting PL, so many experts have analyzed and studied this. Quanbo Ge believed that it was difficult to obtain high-performance industrial PL forecasting due to many complex factors. By integrating some machine learning methods targeting industrial power consumers, he conducted in-depth research on industrial PL forecasting [6]. Muhammad Naveed Akhter believed that accurately predicting photovoltaic power generation was a key requirement to ensure the stability and reliability of the power grid. He conducted a systematic and critical review of methods for predicting photovoltaic power generation, mainly focusing on meta heuristics and machine learning methods [7]. Cheng Lefeng stated that data-driven technology would accelerate the development of intelligent energy and power systems. Machine learning formed a typical representative algorithm category for predicting and making judgments by analyzing and learning large amounts of historical and synthetic data to help people make the best decisions [8]. Bento Murilo EC believed that voltage stability margin was an important load margin measure used by the power system operation center to prevent voltage collapse, and used a supervised machine learning technique to predict the load margin range of the power system. The requirements for voltage stability and small signal stability were considered, and the electrical data of certain buses with phasor measurement units was used [9]. Roel Dobbe believed that machine learning technology played an important role in PL management systems. Through the application of machine learning technology, voltage was stabilized and the carrying capacity of electricity was within the operating load [10]. Bebortta, Sujt proposes a dynamic integer linear programming technique to facilitate optimal task unloading, allocating resources from the fog computing layer to IoT devices, while considering the constraints of timely task execution and resource availability. To minimize the task latency and energy consumption of fog nodes. The system can be optimized in terms of power consumption and delay [11]. In order to solve the black box characteristics of machine learning models, Chen, He classified studies according to the application stage of interpretable machine learning techniques to improve the interpretability of black box models, which is of great significance for accelerating the application of machine learning in building energy management [12]. By accurately predicting the PL, the voltage can be effectively stabilized and the power can be controlled within a stable range.

At present, there are some problems in the operation system of the PL control management terminal, such as inaccurate data collection and processing, insufficient system response speed, and low energy utilization rate. In order to solve these problems, machine learning technology would be applied to promote the performance of the PL control and management terminal operating system, thus accelerating the rapid development of the power system. This article first conducted a demand analysis of the PL system, analyzed and studied the operating system of the PL control and management terminal using machine learning technology, and constructed the operating process of the operating system of the PL control and management terminal and the system for the operation of the new PL control and management terminal. Through a series of analyses, the rapid development of the operation system of the PL control management terminal was promoted, thus providing some reference for the establishment of the power system in terms of PL.

2. Evaluation of PL system

Power usage The amount of electricity used by the electrical devices in the power system is referred to as the power load. The kind, amount, and operational status of the electrical equipment as well as the power system’s power supply capacity determine the size of the power load. Power load control is a technological tool for load management in addition to being a crucial component of distribution automation. In order to ensure the safety, stability, and efficient operation of the power grid, it aims to use a reasonable peak and valley price difference, mobilize the majority of users to participate in the peak regulation of the power system, and use automatic control technology to increase the utilization rate of power generation, supply, and equipment. This has significant economic significance as well as social benefits. loads arc A load curve is a relationship curve that shows how the amount of power consumed by the power system changes over time in a specific time frame. One of the crucial aspects of a power system, which refers to its management and operation, is the load curve. Demand side control of energy By directing and managing the mode and behavior of electricity consumption under the assumption that the demand for electricity would be met, power demand side management refers to the optimization of electricity consumption mode and energy savings. The basic techniques for managing power demand side include peak shifting, distributed energy, and energy efficiency management. Power demand side management has been successfully used in practice to increase the utilization efficiency of power resources, lessen the environmental impact, and is one of the key ways to implement the sustainable development strategy.

2.1 System requirements evaluation

For the construction of the PL control and management terminal operation system, the primary task is to conduct demand analysis. The importance of this process is self-evident. Only by fully understanding all possible implicit requirements can a solid foundation be laid for subsequent system construction, so as to better solve a series of problems that exist in traditional systems.

The following categories can be used to group the demand analysis of power load control and management terminal operating systems: 1. Functional requirements, such as the need for load control and management, load prediction and optimization scheduling, and data collecting for power load monitoring. These features can aid in real-time power load monitoring, forecasting, and control, which will increase grid stability and effectiveness. 2. Security needs: Remote control and management of power grid equipment are involved in power load control and management, therefore security requirements are essential. To guard against potential security risks, operating systems need robust security measures including identity identification, access control, and data encryption. 3. Reliability and real-time performance requirements: controlling and managing the power load is a crucial duty, and the operating system needs to operate well in real-time. The system must be capable of quickly and accurately responding to different load control directives and handling faults and crises.

The following restrictions may apply to the operating system of power load control and management terminals: 1. Hardware resource restrictions: The operating system may be constrained by the real hardware resources – processors, memory, storage, etc. – that are actually accessible. The system’s functionality and processing power may be affected by this in some way. 2. Communication network restrictions: The operating system may be constrained by the capacity, reliability, and latency of the available communication networks because it must communicate with the devices on the power grid. 3. Cost and resource limitations: The development and maintenance expenses, human resources, and time constraints may all have an impact on the operating system’s design and execution.

In conclusion, the function, security, scalability, reliability, and real-time categories can be used to categorize the demand analysis of power load control and management terminal operating systems. Hardware resources, communication networks, cost, and resource limitations may also be considered.

System requirement analysis requires comprehensive collection and organization of relevant requirements from multiple perspectives such as technology, application, and users. Firstly, technically, it is necessary to consider factors such as system stability, scalability, and maintainability, so as to create an efficient and stable operating environment. Secondly, from an application perspective, it is necessary to optimize the security and operability of the system to meet the requirements of management personnel and provide more options and functions. Finally, users need to provide a clear description of the details, features, and personalized requirements of the developed system to determine whether the system can meet its personalized requirements. In this way, a better systematic analysis of user needs can be conducted to create an excellent system [13].

Therefore, in order to meet these different levels of requirements, it is necessary to conduct in-depth exploration and analysis of the system’s requirements, so as to provide good support for subsequent data collection and processing technologies, as well as the selection and optimization of machine learning algorithms. In the process of analyzing these requirements, it is also necessary to pay attention to factors such as cost, time, and resources to ensure the practicality and feasibility of system development.

In summary, conducting a system requirement analysis requires a comprehensive and detailed review of all relevant requirements in order to ensure high-quality terminal operation system design and better serve related work and management applications.

2.2 Data collection and processing technology

In the operation system of PL control management terminal, data collection and processing is a very important link [14, 15]. Firstly, the purpose of data collection is to obtain data for further processing and analysis. The amount of data, real-time performance, accuracy and so on are the key factors affecting the data acquisition effect. Therefore, when collecting data, these factors should be taken into account in order to reduce errors in data collection. For example, multiple sensors are used for data collection to ensure the comprehensiveness and accuracy of the data [16]. In addition to multiple sensors for data collection, appropriate data collection methods, such as questionnaires, field observation, experimental design, can be selected to ensure accurate and complete data acquisition; standardized data collection tools, such as validated questionnaires, checklists or observation record tables, can reduce subjective factors and inconsistency in data collection and ensure the consistency and comparability of data; for large-scale data collection, sampling methods can be used to obtain representative samples. Then, statistical analysis is used to infer the overall situation, reduce the cost and workload of data collection, and improve the accuracy of data collection.

Secondly, the collected data needs to be processed for better analysis. There are many methods for data processing, such as data cleaning, data normalization, data transformation, and so on. Among them, data cleaning is a very important step. This is because during the data collection process, there are often cases of missing or abnormal data, which can have an impact on subsequent analysis. Therefore, after data collection, data should be cleaned to ensure its integrity and accuracy.

The selection of data collection and processing technology is also a key factor affecting the effectiveness of system operation. When selecting these technologies, consideration should be given to system requirements, such as the frequency of data collection, the efficiency of data processing, and so on. At the same time, advanced and reliable technologies should be selected to ensure the stability of the system in long-term operation.

In summary, data acquisition and processing technology plays an important role in the operation system of PL control and management terminals. It can effectively improve the efficiency of PL control and management terminal operation, and save a lot of time [17]. By selecting appropriate technologies and conducting effective processing, data can be made more accurate and comprehensive, thus providing a strong foundation for the selection and optimization of machine learning algorithms.

2.3 Selection of machine learning algorithms

In the operation system of PL control management terminal, the selection of machine learning algorithms is one of the key steps to achieve system performance improvement [18, 19]. At present, commonly used machine learning algorithms include support vector machines, neural networks, decision trees, etc. Different algorithms have different roles and functions. Choosing properly can effectively leverage its advantages. When making specific choices, factors such as system performance requirements, data characteristics, and algorithm performance need to be considered.

Firstly, it is necessary to consider the performance requirements of the system. Different machine learning algorithms have different performance performances, such as training time, recognition accuracy, and generalization ability. Therefore, when selecting algorithms, it is necessary to determine performance indicators based on actual needs and evaluate the performance indicators of each algorithm using experimental data.

Secondly, data characteristics need to be considered. The performance of machine learning algorithms is closely related to the characteristics of the training dataset. Therefore, sufficient analysis of the data is necessary before algorithm selection. For example, in the PL management terminal system, some indicators may have missing or outlier, and these characteristics need to be taken into account when selecting algorithms. At the same time, data preprocessing techniques also need to be considered before algorithm selection in order to improve the performance and accuracy of the algorithm.

Finally, the optimization of machine learning algorithms is also an important part of improving system performance. Most machine learning algorithms need to be set with certain “hyperparameter” when they are used. These parameters can directly affect the performance of the algorithm. Therefore, the specific use of the algorithm needs to be set according to the actual situation, and the parameters need to be adjusted to improve the performance.

In summary, in the operation system of PL control management terminal, the selection of machine learning algorithms would have a significant impact on the performance and efficiency of the system. Choosing algorithms based on actual data characteristics, performance indicators, and actual needs can help improve the performance of the system.

2.4 Necessity of applying machine learning technology

Machine learning technology, as an important component of this system, has excellent performance and broad application prospects [20]. At present, machine learning technology has been applied in multiple fields such as PL forecasting, anomaly detection, and power grid optimization, and has achieved good results. This indicates that machine learning technology is gradually maturing and improving in practice in multiple fields.

However, there are still some problems and challenges to be faced in the process of promoting and applying the system. First of all, in the actual operation process, due to the complexity of the power system and the diversity of data, machine learning models often face overfitting and under fitting problems. Secondly, in practical applications, machine learning technology needs to be finely tuned and optimized in conjunction with actual scenarios in order to fully play its role. In addition, the cost and risk of promoting applications are also issues that need to be considered.

To address the above issues, a series of strategies and measures need to be taken to further promote the application of the system. Firstly, targeted research and optimization can be strengthened to create machine learning models with better adaptability and stability. Secondly, by improving the data collection and processing process, data quality and accuracy can be improved, and the error rate of model training can be reduced. In addition, cooperation with various power enterprises and research institutions can be strengthened to promote technology sharing and innovation.

In summary, the promotion and application of the PL control and management terminal operating system based on machine learning technology has broad prospects and potential. At the same time, efforts need to be made to solve a series of problems and challenges, strengthen the combination of technology and practice, and promote the long-term stable development of the system.

2.5 Machine learning based algorithms

Using the support vector regression algorithm in machine learning algorithms, a database regression model is established in the PL control management system to detect the data. The calculation process is as follows:

$\displaystyle f(x_{e})=h\bullet\varphi(x_{e})+n$ (1)

Among them, $\varphi$ represents the weight vector, and $n$ belongs to the bias vector.

After functional analysis, it is used to determine the target value:

$\displaystyle\min K=\frac{1}{2}\left\|\alpha\right\|^{2}+V\sum\limits_{o=1}^{m% }{\left({\beta_{o}^{\bullet}+\beta_{o}}\right)}$ (2)

Where represents the minimum value of the target value $K$ , and represents the data collected in the PL control management system.

Equation (2) is constrained:

$\displaystyle\left\{{{\begin{array}[]{l}{u_{o}-e\partial(x)-d\leqslant% \varepsilon+\beta_{o}}\\ {e\partial(x)+d-u_{o}\leqslant\varepsilon+\beta_{o}^{*}}\\ {\beta_{o},\beta_{o}^{*}\geqslant 0,o=1,2,...,n}\\ \end{array}}}\right.$ (3)

In Eq. (3), are relaxation factors.

When encountering problems with linear regression, the specific functions of support vector machines are as follows:

$\displaystyle f(x)=\sum\limits_{e=1}^{n_{b}}{(s_{e}-s_{e}^{\bullet})}[Q(x_{e})% ,Q(x)]+d$ (4)

Based on the above algorithm, $Q\left({x_{e},x}\right)$ is used to replace $[Q(x_{e}),Q(x)]$ and transformed into a dual problem. The data detection results are obtained by solving the dual problem. The specific process is as follows:

$\displaystyle f(x)=\sum\limits_{e=1}^{n_{b}}{(s_{e}-s_{e}^{\bullet})}Q(x_{e},x% )+d$ (5)

3. Construction of PL system

3.1 System framework

In system applications, the system framework is an indispensable and important step. This step aims to conduct a comprehensive and in-depth analysis of all aspects of the PL control management terminal operating system, so as to further improve the system’s performance and efficiency, and continuously improve the system’s operational quality and service level. Therefore, in the analysis practice of this article, the following strategies would be adopted to build the system framework, as shown in Fig. 1.

Figure 1.

Improved PL control management terminal operation flow.

Firstly, the functionality of the system is analyzed, including improving the collection methods of PL data, improving the accuracy of load forecasting, enhancing the applicability, flexibility, and effectiveness of analytical models, and analyzing load control strategies to improve control stability and real-time performance.

Secondly, the reliability of the system is analyzed, including strengthening the security protection measures of the system, ensuring the confidentiality and integrity of system data, and avoiding possible faults and problems by establishing a sound backup mechanism and debugging and testing.

Finally, the scalability of the system is analyzed, including further improving the system’s technical documentation and code annotations, and carrying out standardization activities to achieve green and sustainable development of the system.

For the analysis of the system framework, innovative methods would be adopted in order to achieve more excellent results in the application practice of the PL control management terminal operating system, provide practical and feasible data support and technical support for the sustained and stable development of the power market, and make positive contributions to the sustainable development and beautiful tomorrow of the world power industry.

3.2 Establishment of a new PL system

Based on machine learning technology, an operating system for PL control and management terminal has been constructed. One of the main application scenarios of this system is PL forecasting and control. PL forecasting and control are very important links in the power system, which are of great significance for improving the efficiency, stability, and reliability of the power system. The main purpose of PL forecasting and control is to accurately predict future loads through the analysis and prediction of historical data, and take corresponding control measures based on the prediction results, such as adjusting the output of generators, regulating the voltage of transmission lines, and regulating the electricity consumption of users, so as to achieve the goal of balancing the power supply and demand relationship, as shown in Fig. 2.

Figure 2.

PL control management terminal operation system.

In practical applications, the prediction and control scenarios of PL are relatively complex, including weather factors, holidays, electricity demand, power grid load, and other factors. Therefore, machine learning technology is adopted, mainly including support vector machine, decision tree, random forest and other models, and a large number of historical data are used for training and testing to improve the accuracy of prediction and control. Through experimental verification, it can be seen that using machine learning technology can significantly improve the accuracy and efficiency of PL prediction and control, thereby effectively analyzing the operational efficiency of the power system.

In summary, PL prediction and control play an important role in the power system. Through the application of machine learning technology, the accuracy and efficiency of prediction and control can be significantly improved. In the future, PL forecasting and control would continue to be deeply studied, and the system would be analyzed to further improve the stability and reliability of the power system.

4. Experimental design of PL

4.1 Experimental preparation

Before conducting experimental design and data collection, it is necessary to have a clear and clear plan and determination of the experimental purpose, experimental process, and data collection sources. The purpose of the experiment is to improve the accuracy and efficiency of PL prediction and control by constructing a PL control terminal operating system based on machine learning technology, and to achieve scientific, accurate, and dynamic management of PL. Therefore, the process of experimental design and data collection is crucial.

Firstly, for the design of the experimental process, it is necessary to determine the necessary equipment, data collection plan, experimental variables, and other key elements based on the experimental purpose and research content, and establish a complete experimental plan and execution plan. In determining the data collection plan, a combination of historical and real-time data was used to collect, integrate, and analyze data from multiple directions and dimensions. A data collection plan was constructed with historical load data, weather data, economic data, population data, and other main data sources. In the data collection and processing process, key technical measures such as data cleaning, data preprocessing, feature selection, and model construction were adopted to enhance data quality and experimental reliability. Secondly, for the implementation of data collection, it is necessary to conduct reasonable, detailed, and comprehensive data collection according to the experimental plan and data collection plan.

In summary, in order to verify the effectiveness of PL prediction and control, a series of experiments were designed, and data collection and analysis and evaluation of experimental results were conducted. Among them, various scenarios including weather factors, holidays, working days, etc. were included, and different models were used to evaluate and compare the results. Through the analysis and evaluation of experimental results, it can be seen that using appropriate models and algorithms can achieve high-precision and efficient prediction and control results. A large amount of PL data has been collected, and experimental verification and data analysis have been conducted using a machine learning based PL control and management terminal operating system. In terms of experimental design and data collection, the advantages and characteristics of machine learning technology have been fully utilized. Through reasonable planning, meticulous design, and comprehensive implementation, a solid foundation has been laid for subsequent experimental analysis and result evaluation.

4.2 Experimental sources

The terminal can measure voltage, current, power, and power factor, as well as complete the power quality calculation, which includes harmonic statistics, power factor statistics, and voltage qualified rate statistics. The terminal can also select voltage and current analog acquisition functions based on the use requirements. The phrase “measurement point” refers to the point at which an electrical terminal and a measuring device sequentially link to measure a certain set of electrical quantity values. Each measuring point has a distinct logical positioning code, which is the unique object identification of the device in the terminal and the parameter configuration of the data application, when a physically identical electrical connection point is measured by a number of devices or is measured by one device but transmitted to a terminal in a variety of ways.

After data analysis and preprocessing, normalization and feature selection were carried out to ensure the accuracy and reliability of the data. At the same time, according to the characteristics and change rules of sample data, a variety of machine learning algorithms are used for modeling and analysis, such as support vector machine, random forest, neural network, etc. By combining cross validation and error analysis techniques, various models were evaluated and compared, and the most effective model was ultimately selected for experimental verification.

The PL data and weather data of a large power company in North China were selected as sample data. Firstly, a new prediction system was established to predict the PL value, and then the actual PL value was recorded through experiments. By comparing the predicted and actual values, the accuracy of the prediction, system energy consumption, and system efficiency were analyzed.

4.2.1 Prediction accuracy

The PL of the power company was tested based on three different levels. Each time lasted for 10 minutes. Subsequently, the load forecasting model was used for evaluation, and the data was recorded through real detection. The three levels were divided into three groups. 1 and 2 were the first level, 3 and 4 were the second level, and 5 and 6 were the third level. Among them, 1, 3, and 5 were the prediction groups, and 2, 4, and 6 were the actual measurement groups. By comparing the system prediction values with the actual load values, the accuracy of the prediction could be clearly seen, as shown in Fig. 3.

Figure 3.

Comparison of the predicted value of the forecast model with the actual PL value.

As shown in Fig. 3, there was little difference between the predicted value and the actual PL value. Based on specific data, the average value of 1 in the first level was about 8.053 kw, while the average value of 2 was about 7.974 kw, with a difference of about 0.079 kw; the average value of the second level 3 was about 11.971 kw, while the average value of 4 was about 11.966 kw, with a difference of about 0.005 kw; the average value of the third level 5 was about 15.872 kw, while the average value of 6 was about 16.061 kw, with a difference of about 0.189 kw. The average value of the difference between the three values was about 0.091 kw, indicating that the prediction accuracy was high and the error was controlled within an acceptable range. It played an important role in quickly detecting PL values in power operation. The prediction model could quickly detect changes in PL values, indicating that the prediction model could quickly adapt to load changes and had high practical value.

4.2.2 System energy consumption

The systems of six different regions of the power company were subjected to energy consumption testing. The six regions were divided into two groups, labeled as Group A and Group B. Group A used the traditional PL detection system, and the three regions of Group A were labeled as 1, 2, and 3; Group B used a new PL detection system, with the three areas of Group B marked as 4, 5, and 6. Each detection lasted for 5 minutes, and experimental data was recorded and analyzed, as shown in Fig. 4.

Figure 4.

Comparison of energy consumption between traditional and new PL detection system.

As shown in Fig. 4, the energy consumption of Group A ranged from 30% to 50%, while that of Group B ranged from 5% to 10%. It was evident that using a new PL detection system could effectively reduce the energy consumption of the prediction system. Therefore, it could be concluded that using a machine learning based PL control management terminal operating system for load control could promote the efficiency of power system load dispatch and effectively reduce system energy consumption. By comparing experimental data and simulation data, it could be seen that the control effect was significant, which could effectively reduce system energy consumption indicators and reduce the frequency and degree of peak load occurrence, thus meeting practical operational requirements.

4.2.3 Stability of system operation

The systems of six different regions of the power company were tested for operational stability, with three regions being a group for stability testing. The group using the traditional PL detection system was recorded as Group A, and the group using the new PL detection system was recorded as Group B. In order to better distinguish each area in each group, the three areas in the traditional PL detection system would be marked as 1, 2, and 3; the three areas in the new PL detection system would be recorded as 4, 5, and 6, with each detection lasting for 10 minutes. The experimental data would be recorded and analyzed, as shown in Fig. 5.

Figure 5.

Comparison of stability between traditional and new PL systems.

As shown in Fig. 5, the stability wave values of Group A system were between 10 and 100, while the stability wave values of Group B system were between 20 and 30. It was evident that using the new PL detection system could effectively stabilize the daily operation of the system. Therefore, it could be concluded that using the PL control management terminal operating system of machine learning for load control could enhance the stability of power supply in the power system. By comparing the stability of traditional PL systems with new PL systems, it could be seen that the operation effect was significant and could effectively meet the actual operational needs.

In summary, the system was analyzed, its intelligence and reliability were strengthened, and its efficiency and operational stability were promoted. In practical use, the system achieved good results, thus achieving intelligent control and refined management of PL, with high practical value and economic benefits. The application scenarios and experimental verification results of the PL control management terminal operating system based on machine learning technology indicated that the system had good application prospects and promotion value.

5. System promotion strategies and measures

In order to promote the widespread application of machine learning technology based PL control management terminal operating system in the power industry, the following system promotion strategies and measures have been emphasized and formulated.

Firstly, through communication and negotiation with technical personnel and enterprises related to the power industry, the work of technology demonstration and promotion should be strengthened to present the high quality and efficiency of the system to the vast number of users. During the negotiation process, the technical features and advantages of the system can be introduced in detail. At the same time, users would be invited to participate in the testing and optimization of the system, so as to further promote its practicality.

Secondly, for individual investors and small and medium-sized power enterprises, the promotion rate of the system can be further improved by issuing free trial periods. For individual investors and small enterprises, due to limited investment and technological strength, more meticulous services and refined promotion strategies are often needed. Users with special needs can apply for the opportunity to use the system for free for a certain period of time, so as to have a comprehensive understanding and understanding of its functions and performance, and provide users with more reference value and word-of-mouth effects based on user experience.

Before implementing the system promotion, it is necessary to carry out various related publicity, strengthen the attention and promotion efforts of various information media, and enable enterprises and users to fully understand the excellent performance and broad application prospects of the system through various news media such as television stations, Weibo, WeChat, and newspapers.

Finally, a detailed system promotion plan should be developed, and feedback and suggestions from all parties should be continuously collected during the promotion process to enhance user experience and satisfaction, and strive to promote the stability and execution efficiency of the system. At the same time, comprehensive evaluation and statistics are conducted on the implementation effects of various promotion strategies and measures, and timely adjustments are made to promotion plans and strategies to achieve maximum promotion and application.

6. Conclusions

With the rapid development of smart grid technology, PL control management terminal operating system has been widely applied and promoted. At present, machine learning technology has been applied in many fields such as PL prediction, abnormality detection, power grid optimization and other fields, and has achieved good results. This paper first investigate the requirements of the system to emphasize the importance of constructing the system, collect the power load data with the collected power load data. Then select the machine learning algorithm and study the factors affecting the machine learning algorithm to build the system. Finally, the experiment found that the load prediction model can effectively handle the nonlinear relationship in load data with high prediction accuracy. Based on the load prediction model, the PL control management terminal operating system is tested and verified. The experimental results show that the system can effectively reduce the energy consumption in the PL control process, and has good stability and operation efficiency in the operation process. The power load control management terminal operating system is explored in this work utilizing machine learning technology; nevertheless, due to time constraints, this paper also has several limitations: Resilient for the algorithm: The load control management system must be able to adjust to the diverse circumstances of many scenarios. The black box nature of machine learning algorithms, for instance, makes it occasionally challenging to explain the prediction or decision process of the algorithms, leading to uncertainty about the system results, which may lead to trust issues in scenarios involving power load control management. On the other hand, machine learning algorithms may be less responsive to abnormalities and uncertainty.

The sole way for the operating system for the power load control and management terminal is as follows: Deep learning and reinforcement learning combined:

Combining deep learning and reinforcement learning techniques can improve the control system’s capacity to model and decide in difficult scenarios.

Intelligent edge computing and the Internet of Things: These two technologies enable distributed load control management systems, lessen reliance on a centralized server, and enable more flexible and real-time load management.

Footnotes

Acknowledgments

State Grid Sichuan Electric Power Company Marketing Management Control System Expansion, No.: CDSM [2020] No. 41 (Approval Document).

References

Atif

Damian

Keith

Leila

Yue

. STFT cluster analysis for DC pulsed load monitoring and fault detection on naval shipboard power systems. IEEE Transactions on Transportation Electrification.2020; 6(2): 821-831.

Tinghui

Yusen

Huajin

Zhiyu

Stephen

. Modeling and forecasting short-term power load with copula model and deep belief network. IEEE Transactions on Emerging Topics in Computational Intelligence.2019; 3(2): 127-136.

Xianlun

Yuyan

Ting

Yingjie

. Short-term power load forecasting based on multi-layer bidirectional recurrent neural network. IET Generation, Transmission & Distribution.2019; 13(17): 3847-3854.

Million

Awet

Chirstian

Andrea

Edoardo

. A cloud-based on-line disaggregation algorithm for home appliance loads. IEEE Transactions on Smart Grid.2018; 10.3: 3430-3439.

Bhandari

Shree

Ajay

. Short term electric load forecasting of kathmandu valley of nepal using artificial neural network. Kathford Journal of Engineering and Management.2018; 1(1): 43-48.

Quanbo

Chen

Haoyu

Zhenyu

Gang

Jianmin

Qiang

. Industrial power load forecasting method based on reinforcement learning and PSO-LSSVM. IEEE Transactions on Cybernetics. 2020; 52(2): 1112-1124.

Muhammad

Saad

Hazlie

Noraisyah

. Review on forecasting of photovoltaic power generation based on machine learning and metaheuristic techniques. IET Renewable Power Generation.2019; 13(7): 1009-1023.

Cheng

Tao

. A new generation of AI: A review and perspective on machine learning technologies applied to smart energy and electric power systems. International Journal of Energy Research.2019; 43(6): 1928-1973.

Bento Murilo

. Monitoring of the power system load margin based on a machine learning technique. Electrical Engineering.2020; 104(1): 249-258.

10.

Roel

Oscar

David

Daniel

Duncan

Claire

. Toward distributed energy services: Decentralizing optimal power flow with machine learning. IEEE Transactions on Smart Grid.2019; 11(2): 1296-1306.

11.

Bebortta

Tripathy

Modibbo

Ali

. An optimal fog-cloud offloading framework for big data optimization in heterogeneous IoT networks. Decision Analytics Journal.2023; 100295.

12.

Chen

Xiao

Guo,

Yan,

. Interpretable machine learning for building energy management: A state-of-the-art review. Advances in Applied Energy.2023; 100123.

13.

Zhao

Kok

. Predicting electrical power output of combined cycle power plants using a novel artificial neural network optimized by electrostatic discharge algorithm. Measurement.2022; 198: 111405. doi: 10.1016/j.measurement.2022.111405.

14.

Cui

Jianhui

Meng

. Machine learning-based anomaly detection for load forecasting under cyberattacks. IEEE Transactions on Smart Grid.2019; 10(5): 5724-5734.

15.

Yue

Atif

Damian

Keith

. Wavelet Transform Data-Driven Machine Learning-Based Real-Time Fault Detection for Naval DC Pulsating Loads. IEEE Transactions on Transportation Electrification.2021; 8(2): 1956-1965.

16.

Migdal

. Construction of Environmental Economic Dispatching Model Based on Electricity Market Environment. Academic Journal of Environmental Biology.2021; 2(4): 1-9. doi: 10.38007/AJEB.2021.020401.

17.

Sitton

. Reactive Power Optimization of Power System Based on Distributed Cooperative Particle Swarm Optimization Algorithm. Distributed Processing System.2020; 1(2): 46-53. doi: 10.38007/DPS.2020.010206.

18.

Attique

Tek

Brice

Shafiqur

. Comparative evaluation of machine learning models and input feature space for non-intrusive load monitoring. Journal of Modern Power Systems and Clean Energy.2021; 9(5): 1161-1171.

19.

Madam

Alok

Xiaochun

Anuj

Manoj

Achyut

. A Non-invasive Approach to Identify Insulin Resistance with Triglycerides and HDL-c Ratio Using Machine learning. Neural Process. Lett.2023; 55(1): 93-113.

20.

Rai

Mala

. Analysis of classical and machine learning based short-term and mid-term load forecasting for smart grid. International Journal of Sustainable Energy. 2021; 40(9): 821-839.