Method and system for carrying out load management on electric equipment based on electric energy quality
By using four-scale wavelet energy entropy algorithm and FFT fast Fourier transform in load management, combined with accuracy threshold, the problem of low load management prediction accuracy under the complex changes in electrical data is solved, and higher data accuracy and automation are achieved.
Patent Information
- Application Number
- CN202411965993.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-30
- Publication Date
- 2025-04-25
AI Technical Summary
The prior art is difficult to effectively process electricity consumption data that has both periodic and irregular changes, resulting in low accuracy in load management prediction and insufficient automation.
The four-scale wavelet energy entropy algorithm and FFT fast Fourier transform are used, combined with the accuracy threshold, and the transient change position of the voltage is determined, the first function prediction image is established to predict the probability of power quality, and a power supply load plan is formulated.
It improves the accuracy and reliability of data, effectively shows the complex changes in electric usage scenarios, improves the accuracy and automation of load management, and reduces the cumbersome process of manpower monitoring and adjustment.
Smart Images
Figure CN119891176A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of load control of electrical equipment, and in particular to a method and system for load management of electrical equipment based on power quality. Background Art
[0002] With the growth of electricity demand and technological advancement in modern society, the complexity and diversity of power systems are increasing. Traditional load management methods are becoming less and less effective, and it is difficult to cope with the challenges brought by power-consuming equipment and power consumption scenarios that have both periodic and irregular changes. For example, long-term regional power consumption at the provincial, municipal and town levels often involves a large number of power-consuming equipment and multiple power consumption areas. Power consumption will to a certain extent show annual cyclicality, and also have certain irregularities. Especially in regional power consumption scenarios at the smallest level, power consumption time and power consumption demand often change.
[0003] The integrity of power quality refers to the ability of the power system to continuously provide power that meets standards and is suitable for the normal operation of end-user equipment under steady-state and dynamic conditions. It covers multiple aspects to ensure the stability of power supply in terms of voltage, frequency, waveform and other characteristics. When providing long-term power supply to multiple areas and a large number of electrical equipment, it is often necessary to ensure the integrity of the power quality of the power supply, thereby ensuring the quality of power supply.
[0004] At present, when power quality is used to control equipment load, a deep neural learning network is used to analyze historical power quality data, analyze the distribution and characteristics of power quality, and find out key influencing factors, such as three-phase balance, harmonics, power and current, so as to give optimized planning decisions for power quality management of each distribution network node and assist the power quality management of the distribution network.
[0005] However, this measure involves at least the following two problems: 1) Deep neural learning networks generally have difficulty processing data that has both fixed cycles and irregular changes. Instead, they are good at processing changing data that has a certain specific and fixed internal logic, which is different from the actual changes in electricity consumption. As a result, when using deep neural learning networks to make predictions for load management, the accuracy is often low; 2) Lack of predictive capabilities, and due to low accuracy, relevant personnel are often required to conduct testing and verification, the degree of automation is insufficient, and it is not simple enough. Summary of the invention
[0006] In view of the above two problems, the purpose of the present invention is to propose a method and system for load management of electrical equipment based on power quality. The four-scale wavelet energy entropy algorithm can accurately know each location where voltage transient changes occur, that is, the location where power quality problems exist. Combined with the accuracy comparison brought by FFT fast Fourier transform, the accuracy and reliability of the data are greatly improved; then, a first function prediction image is established to predict the daily probability, monthly probability and annual probability of complete power quality in the future, which conforms to the actual scenario of power consumption, and can effectively show the power data with both fixed period and irregular change characteristics under the power consumption scenario, so as to effectively formulate a power supply load plan; in addition, a second function prediction image is also established for prediction based on the accuracy threshold, which greatly alleviates the complicated process of long-term monitoring and adjustment by manpower, and only occasional manual inspection and adjustment are required.
[0007] This is achieved through the following technical solutions: Firstly, a method for load management of electrical equipment based on power quality is proposed, which is used to manage the load of multiple electrical equipment with both periodic changes and irregular changes. The method comprises the following steps: S1. Obtain source data of multiple monitoring points, calculate each source data using a four-scale wavelet energy entropy algorithm, obtain the corresponding signal value of each wavelet energy entropy, and determine each first position where a voltage transient change occurs from the distance of each signal value to the corresponding source data; S2, performing FFT fast Fourier transform on the source data of multiple monitoring points in step S1 respectively, and obtaining each frequency domain graph corresponding to each source data; based on each frequency domain graph, determining each second position where the voltage transient change occurs; setting an accuracy threshold, comparing each first position with each second position, and when the comparison result does not meet the accuracy threshold, returning to the step of calculating each source data using the four-scale wavelet energy entropy algorithm in step S1 and looping; when the comparison result meets the accuracy threshold, proceeding to step S3; S3, based on each first position in step S1, determining the daily probability, monthly probability and annual probability of complete power quality, and establishing a first function prediction image based on the daily probability, monthly probability and annual probability of complete power quality, for predicting the daily probability, monthly probability and annual probability of complete power quality in the future; S4. Based on the daily probability, monthly probability and annual probability of complete future power quality predicted in step S3, formulate a power supply load plan for multiple power-consuming devices.
[0008] The four-scale wavelet energy entropy algorithm can accurately know each location where voltage transient changes occur, that is, the location where power quality problems exist. Combined with the accuracy comparison brought by FFT fast Fourier transform, the accuracy and reliability of the data are greatly improved. Then, the first function prediction image is established to predict the daily probability, monthly probability and annual probability of complete power quality in the future, which is in line with the actual power consumption scenario, and can effectively show the power data with both fixed period and irregular change characteristics in the power consumption scenario, and then effectively formulate a power supply load plan.
[0009] Preferably, the process of calculating each source data using the four-scale wavelet energy entropy algorithm in step S1 is: selecting a wavelet function, performing four-scale decomposition, downsampling, and wavelet reconstruction. By selecting a wavelet function, performing four-scale decomposition, downsampling, and wavelet reconstruction, the accuracy and reliability of voltage transient change detection are improved, providing accurate data for subsequent load management and prediction.
[0010] Preferably, when performing four-scale decomposition, each source data when the voltage undergoes transient changes and gradually transitions to a stable state is first used as each basic data, and then each basic data is decomposed into three corresponding high-frequency components and one low-frequency component. Four-scale decomposition can obtain four different frequency components, three high-frequency components corresponding to local detail components, and low-frequency components corresponding to overall approximate components, thereby taking into account both the overall and the local, making the final result more accurate.
[0011] Preferably, the lower limit of the accuracy threshold in step S2 is 70%. The lower limit of the accuracy threshold is closely related to the final predicted daily probability and other results. Setting a higher lower limit can effectively ensure the accuracy of the final result.
[0012] Preferably, when setting the accuracy threshold in step S2, in a specific year, the historical data changes of the accuracy threshold in the previous year are first counted, and then a second function prediction image of the accuracy threshold in the previous year is generated; the prediction data in the second function prediction image is used as the data of the accuracy threshold at the corresponding time in the specific year. The historical data of the previous year is used to determine the accuracy threshold change of the new specific year, which eliminates the tedious process of manual continuous adjustment of the setting, and only requires occasional manual monitoring and adjustment, which greatly improves efficiency and simplicity.
[0013] Preferably, the process of establishing the first function prediction image in step S3 includes: data preparation, preprocessing, feature extraction based on time series, selection of analysis model based on seasonal decomposition, training model and prediction. Feature extraction based on time series and selection of analysis model based on seasonal decomposition can make the final prediction result closely related to time, which conforms to the correlation relationship between electricity consumption and time.
[0014] Preferably, in step S4, when formulating a power supply load plan for multiple power-consuming devices, the higher the daily probability of complete power quality corresponding to the predicted future power quality of any power-consuming device, the lower the priority of power supply. If the daily probability of complete power quality is high, it means that the operation of the power-consuming device is very stable and it is easy to formulate a power supply strategy; if the daily probability of complete power quality is low, it means that the operation of the power-consuming device is unstable and there are many uncertain factors, so it is necessary to give priority to power supply, determine the situation of these power-consuming devices with unstable factors after power supply, and then quickly formulate other power-consuming devices corresponding to the high daily probability of complete power quality.
[0015] In addition, a system for load management of electrical equipment based on power quality is proposed, which is operated by the above-mentioned method for load management of electrical equipment based on power quality. The system includes multiple power quality collectors, a first processing unit, a second processing unit, a function prediction unit and a power supply load plan generation unit; wherein the multiple power quality collectors are respectively used to collect each source data of each corresponding monitoring point; the first processing unit has a built-in four-scale wavelet energy entropy algorithm for determining each first position where a voltage transient change occurs from each source data; the second processing unit is used to perform FFT fast Fourier transform on each source data Transformation is performed to obtain each frequency domain graph corresponding to each source data and, based on each frequency domain graph, determine each second position where a voltage transient change occurs; the function prediction unit is used to determine the daily probability, monthly probability and annual probability of complete power quality when each second position and each corresponding first position meet the accuracy threshold, and at the same time, a first function prediction image is established based on the daily probability, monthly probability and annual probability of complete power quality, which is used to predict the daily probability, monthly probability and annual probability of complete power quality in the future; the power supply load plan generation unit is used to formulate a power supply load plan for multiple electrical equipment based on the predicted daily probability, monthly probability and annual probability of complete power quality in the future.
[0016] Compared with the prior art, the present invention has the following beneficial effects: The technical solution of the present invention utilizes a four-scale wavelet energy entropy algorithm to accurately know each location where a voltage transient change occurs, that is, the location where there is a power quality problem, and then combines the accuracy comparison brought by the FFT fast Fourier transform to greatly improve the accuracy and reliability of the data; then, a first function prediction image is established to predict the daily probability, monthly probability and annual probability of complete future power quality, which conforms to the actual scenario of electricity consumption, and can effectively show the power data with both fixed period and irregular change characteristics under the electricity consumption scenario, thereby effectively formulating a power supply load plan; in addition, a second function prediction image is also established for prediction based on the accuracy threshold, which greatly alleviates the complicated process of long-term human monitoring and adjustment, and only occasional manual detection and adjustment are required. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] Figure 1 The present invention is a flow chart of a method for load management of electrical equipment based on power quality. DETAILED DESCRIPTION
[0018] The technical solutions in the embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings in the embodiments of the present invention.
[0019] like Figure 1 As shown, it is a flow chart of a method for load management of electrical equipment based on power quality. The method is used to manage the load of multiple electrical equipment with both periodic changes and irregular changes. First, the probability of complete power quality is obtained, and then the corresponding power supply load plan is determined based on the corresponding probability. The method specifically includes the following steps: S1. Obtain source data of multiple monitoring points through multiple power quality collectors, calculate each source data using the four-scale wavelet energy entropy algorithm, obtain the corresponding signal value of each wavelet energy entropy, and determine each first position where the voltage transient change occurs from the distance of each signal value to the corresponding source data. The power quality collector, also called the power quality analyzer, is an instrument used to collect power quality signals; the source data is the power quality signal, which is used for subsequent processing and calculation.
[0020] In this embodiment, the process of calculating each source data using the four-scale wavelet energy entropy algorithm in step S1 is: selecting a wavelet function, performing four-scale decomposition, performing downsampling, and performing wavelet reconstruction. When selecting a wavelet function, the db4 wavelet function can be selected, which has good time and frequency resolution and moderate computational complexity, and is suitable for the time variation characteristics and computational requirements of the power quality signal in this scheme. Then, four-scale decomposition is performed under the selected wavelet function, and each source data when the voltage undergoes transient changes and gradually transitions to a stable state is first taken as each basic data, and the object to be calculated is determined; each basic data is then decomposed into three corresponding high-frequency components and one low-frequency component, and the three high-frequency components correspond to the local detail information in the horizontal, vertical and diagonal directions, respectively, and the low-frequency component corresponds to the overall approximate component, so that the subsequent results can take into account both the whole and the local, making the final result more accurate. Then, at each decomposed scale, downsampling, also known as subsampling or interlaced sampling, is used to reduce the redundancy of the data while maintaining the main features of each sub-signal. Downsampling is achieved by selecting a sample every certain number of samples, which helps to reduce the dimension of the data while retaining key information. Finally, through the inverse operation, namely wavelet reconstruction, a signal closely related to the source data can be reconstructed from each of these decomposed frequency components, and the corresponding signal value of each wavelet energy entropy can be obtained. By selecting wavelet functions, performing four-scale decomposition, downsampling and wavelet reconstruction, the accuracy and reliability of voltage transient change detection are improved, providing accurate data for subsequent load management and prediction.
[0021] S2. Perform FFT fast Fourier transform on the source data of multiple monitoring points in step S1 respectively, and obtain each frequency domain diagram corresponding to each source data. The corresponding amplitude and phase information will be displayed on each frequency domain diagram; based on each frequency domain diagram, find the abnormal spectrum part, such as the sudden appearance of new frequency components or the sudden and significant increase of existing frequency components, which corresponds to the voltage transient change, so that each second position where the voltage transient change occurs can be determined. Then, set the accuracy threshold, compare each first position with each second position, and when the comparison result does not meet the accuracy threshold, return to the step of calculating each source data using the four-scale wavelet energy entropy algorithm in step S1 and loop until the comparison result meets the accuracy threshold; and when the comparison result meets the accuracy threshold, go to step S3.
[0022] In this embodiment, the lower limit of the accuracy threshold in step S2 can be set to 70%. The lower limit of the accuracy threshold is closely related to the final predicted daily probability and other results. Setting a higher lower limit can effectively ensure the accuracy of the final result.
[0023] In the process of load management, the accuracy threshold will inevitably fluctuate according to the time of year. It can be set by manual adjustment, but relying solely on manual adjustment is too cumbersome and will undoubtedly increase the work pressure of the staff. Therefore, the image of the accuracy threshold of the new year can be predicted based on the relevant data of the accuracy threshold of the previous year, and it can be used as the default value. At the same time, only occasional monitoring or a small amount of correction is required, which can greatly improve the simplicity. Specifically, when setting the accuracy threshold in step S2, in a specific year, the historical data changes of the accuracy threshold in the previous year are first counted, and then the second function prediction image of the accuracy threshold of the previous year is generated; the prediction data in the second function prediction image is used as the data of the accuracy threshold at the corresponding time in the specific year.
[0024] S3. Based on each first position in step S1, determine the daily probability of complete power quality, and then based on each daily probability, calculate the corresponding monthly probability and annual probability; at the same time, based on the daily probability, monthly probability and annual probability of complete power quality, establish a first function prediction image for predicting the daily probability, monthly probability and annual probability of complete power quality in the future.
[0025] In this embodiment, the process of establishing the first function prediction image and the process of establishing the second function prediction image both include: data preparation, preprocessing, feature extraction based on time series, selection of analysis model based on seasonal decomposition, model training and prediction. Taking the establishment of the first function prediction image as an example, data preparation is to prepare the input data, that is, each first position and the corresponding time; preprocessing is to clean the data and exclude abnormal values that are obviously problematic; feature extraction based on time series is to extract time-related features from the input data after preprocessing, such as seasonal components and lag features; then, in order to match the characteristics of the time series, the seasonal decomposition model is selected for training, and finally the trained seasonal decomposition model is used to predict future time points and generate the corresponding function image.
[0026] S4. Based on the daily probability, monthly probability and annual probability of complete future power quality predicted in step S3, formulate a power supply load plan for multiple power-consuming devices.
[0027] In this embodiment, in step S4, when formulating power supply load plans for multiple power-consuming devices, the higher the daily probability of complete power quality corresponding to the predicted future power quality of any power-consuming device, the lower the priority of power supply. If the daily probability of complete power quality is high, it means that the operation of the power-consuming device is very stable and it is easy to formulate a power supply strategy; if the daily probability of complete power quality is low, it means that the operation of the power-consuming device is unstable and there are many uncertain factors, so it is necessary to give priority to power supply to ensure the basic power demand of users, determine the specific situation of these power-consuming devices with unstable factors, whether they need overload power supply or can reduce load power supply, and then quickly formulate load strategies for other power-consuming devices corresponding to the high daily probability of complete power quality.
[0028] In this embodiment, different power-consuming equipment can also be prioritized. When the daily probability of complete power quality is the same, the power supply of each power-consuming equipment with a high priority is guaranteed first, and the power supply of power-consuming equipment with a low priority is reduced. For example, the power-consuming equipment in the military area is set to the highest level one, the power-consuming equipment in the hospital is set to the second highest level two, the power-consuming equipment in the residential area is set to the third level, the power-consuming equipment in the commercial area is set to the fourth level, and the power-consuming equipment in the residential activity area is set to the fifth level. In this way, the operation of power-consuming equipment in higher priority areas can be guaranteed as much as possible to meet the power supply load requirements in emergency situations.
[0029] In addition, the present application also proposes a system for load management of electrical equipment based on power quality, which is operated by adopting the above-mentioned method for load management of electrical equipment based on power quality. The system includes multiple power quality collectors, a first processing unit, a second processing unit, a function prediction unit and a power supply load plan generation unit; wherein the multiple power quality collectors are respectively used to collect each source data of each corresponding monitoring point; the first processing unit has a built-in four-scale wavelet energy entropy algorithm for determining each first position where a voltage transient change occurs from each source data; the second processing unit is used to perform FFT fast Fourier transform on each source data Leaf transformation, obtain each frequency domain graph corresponding to each source data and determine each second position where voltage transient changes occur based on each frequency domain graph; the function prediction unit is used to determine the daily probability, monthly probability and annual probability of complete power quality when each second position and each corresponding first position meet the accuracy threshold, and at the same time establish a first function prediction image based on the daily probability, monthly probability and annual probability of complete power quality, which is used to predict the daily probability, monthly probability and annual probability of complete power quality in the future; the power supply load plan generation unit is based on the predicted daily probability, monthly probability and annual probability of complete power quality in the future, which is used to formulate power supply load plans for multiple electrical equipment. The detailed beneficial effects of the system can be referred to the content of the above-mentioned method for load management of electrical equipment based on power quality, which will not be repeated here.
[0030] In summary, the present invention utilizes a four-scale wavelet energy entropy algorithm to accurately know each location where voltage transient changes occur, that is, the location where there is a power quality problem, and combines it with the accuracy comparison brought by the FFT fast Fourier transform, which greatly improves the accuracy and reliability of the data; then, a first function prediction image is established to predict the daily probability, monthly probability and annual probability of complete future power quality, which conforms to the actual power consumption scenario, and can effectively show the power data with both fixed period and irregular change characteristics under the power consumption scenario, thereby effectively formulating a power supply load plan; in addition, a second function prediction image is also established for the accuracy threshold for prediction, which greatly alleviates the complicated process of long-term human monitoring and adjustment, and only occasional manual detection and adjustment are required, which is significantly progressive.
[0031] The above embodiments are only for illustrating the technical idea of the present invention, and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the present invention.
Claims
1. A method for load management of electrical equipment based on power quality, used for load management of multiple electrical equipment with both periodic changes and irregular changes, characterized in that: The steps include: S1. Obtain source data of multiple monitoring points, calculate each source data using a four-scale wavelet energy entropy algorithm, obtain the corresponding signal value of each wavelet energy entropy, and determine each first position where a voltage transient change occurs from the distance of each signal value to the corresponding source data; S2, performing FFT fast Fourier transform on the source data of the multiple monitoring points in step S1 respectively to obtain each frequency domain graph corresponding to each source data; based on each frequency domain graph, determining each second position where the voltage transient change occurs; Set an accuracy threshold, compare each first position with each second position, and when the comparison result does not meet the accuracy threshold, return to step S1 to calculate each source data using the four-scale wavelet energy entropy algorithm and loop; when the comparison result meets the accuracy threshold, go to step S3; S3, based on each first position in step S1, determining the daily probability, monthly probability and annual probability of complete power quality, and establishing a first function prediction image based on the daily probability, monthly probability and annual probability of complete power quality, for predicting the daily probability, monthly probability and annual probability of complete power quality in the future; S4. Based on the daily probability, monthly probability and annual probability of complete future power quality predicted in step S3, formulate a power supply load plan for multiple power-consuming devices.
2. A method for load management of electrical equipment based on power quality according to claim 1, characterized in that: The process of calculating each source data using the four-scale wavelet energy entropy algorithm in step S1 is: selecting a wavelet function, performing four-scale decomposition, performing downsampling, and performing wavelet reconstruction.
3. The method for load management of electrical equipment based on power quality according to claim 1, characterized in that: When performing four-scale decomposition, each source data when the voltage undergoes transient changes and gradually transitions to a stable state is first taken as each basic data, and then each basic data is decomposed into three corresponding high-frequency components and one low-frequency component.
4. The method for load management of electrical equipment based on power quality according to claim 1, characterized in that: The lower limit value of the accuracy threshold in step S2 is 70%.
5. The method for load management of electrical equipment based on power quality according to claim 1, characterized in that: When setting the accuracy threshold in step S2, in a specific year, firstly count the historical data changes of the accuracy threshold in the previous year, and then generate a second function prediction image of the accuracy threshold in the previous year; and use the predicted data in the second function prediction image as the data of the accuracy threshold at the corresponding time in the specific year.
6. The method for load management of electrical equipment based on power quality according to claim 1, characterized in that: The process of establishing the first function prediction image in step S3 includes: data preparation, preprocessing, feature extraction based on time series, analysis model selection based on seasonal decomposition, model training and prediction.
7. The method for load management of electrical equipment based on power quality according to claim 1, characterized in that: In step S4, when formulating a power supply load plan for multiple power-consuming devices, the higher the daily probability that the predicted future power quality of any power-consuming device is complete, the lower the priority of power supply.
8. A system for load management of electrical equipment based on power quality, operating by using the method for load management of electrical equipment based on power quality as claimed in any one of claims 1 to 7, characterized in that: The system includes multiple power quality collectors, a first processing unit, a second processing unit, a function prediction unit and a power supply load plan generation unit; wherein the multiple power quality collectors are respectively used to collect each source data of each corresponding monitoring point; the first processing unit is equipped with a four-scale wavelet energy entropy algorithm, which is used to determine each first position where a voltage transient change occurs from each source data; the second processing unit is used to perform FFT fast Fourier transform on each source data, obtain each frequency domain graph corresponding to each source data, and determine each second position where a voltage transient change occurs based on each frequency domain graph; the function prediction unit is used to determine the daily probability, monthly probability and annual probability of complete power quality when each second position meets the accuracy threshold with each corresponding first position, and at the same time establish a first function prediction image based on the daily probability, monthly probability and annual probability of complete power quality, which is used to predict the daily probability, monthly probability and annual probability of complete power quality in the future; the power supply load plan generation unit is used to formulate a power supply load plan for multiple electrical equipment based on the predicted daily probability, monthly probability and annual probability of complete power quality in the future.