Power consumption value prediction system for power consumption network
By introducing a deep convolutional network model, using basic data of electricity load and rated power for multiple training, the accuracy of future electricity numerical prediction of electricity network is solved, and reliable prediction and power scheduling of electricity network are realized. Key information is provided.
Patent Information
- Application Number
- CN202411140328.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-20
- Publication Date
- 2024-11-12
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The prior art is difficult to accurately predict the electricity consumption value of the power consumption network in the future, resulting in insufficient prediction accuracy, especially due to the diversified power consumption equipment and complex scenes, there are large deviations in manual empirical judgments.
Using an intelligent prediction mechanism based on deep convolutional networks, the deep convolutional network model after multiple trainings is introduced, and the number of training times is positively correlated with the power span. Achieve accurate prediction of the power consumption network.
It realizes reliable prediction of electricity consumption values of the electricity consumption network at the future moments, provides key information for power scheduling and distribution, and improves the stability and accuracy of prediction.
Smart Images

Figure CN118940909A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of power consumption networks, and in particular to a prediction system for power consumption values of power consumption networks. Background Art
[0002] In the field of machine learning and cognitive science, artificial neural network (ANN), also known as neural network (NN) or quasi-neural network, is a mathematical model or computational model that imitates the structure and function of biological neural networks (the central nervous system of animals, especially the brain) and is used to estimate or approximate functions. Neural networks are calculated by connecting a large number of artificial neurons. The deep convolutional network, a network structure that includes many hidden layers, has an expression ability and feature learning level that cannot be matched by traditional machine applications. Therefore, since its application in deep learning algorithm training, it has achieved good application results in many large-scale recognition studies. At present, for each power network, due to the diversity of its power equipment and the complexity of power usage scenarios, it is difficult to accurately predict the power consumption values of the power network at future moments. Instead, it can only be judged based on the power consumption values of the power network at past moments by manual experience. Obviously, the accuracy of such judgment data is insufficient, and even has large deviations. Since artificial neural networks, especially deep convolutional networks, can achieve a reliable numerical conversion bridge between input content and output content after learning, it is expected that an intelligent prediction mechanism based on deep convolutional networks can accurately predict the power consumption values of the power network at set future moments. Summary of the invention
[0003] In order to solve the above problems, the present invention provides a prediction system for the power consumption value of a power network, which can introduce multiple basic data to participate in the intelligent identification of the power consumption value corresponding to the current power network at a set future time. The multiple basic data are various operating parameters of the current power network and multiple power consumption values corresponding to the current power network at multiple times before the set future time. A deep convolutional network model with a customized structure design is also introduced to perform intelligent identification of the power consumption value corresponding to the current power network at the set future time. The customization of the deep convolutional network model structure lies in that, in the deep convolutional network model, the deep convolutional network is continuously trained for each time to obtain a deep convolutional network after each training, and the deep convolutional network after each training is used as the deep convolutional network model. The number of times the deep convolutional network is trained is positively correlated with the power span values in each share of the rated power corresponding to each power load of the current power network, thereby completing the reliable prediction of the power consumption value corresponding to the current power network at the set future time based on the deep convolutional network, and providing key information for the scheduling and allocation of electricity between various power networks.
[0004] According to the present invention, a system for predicting power consumption values of a power network is provided, the system comprising: A first analysis mechanism is used to obtain various operating parameters of the current power network, wherein the various operating parameters of the current power network are the number of power loads in the current power network and the respective share rated powers corresponding to the respective power loads; The second analysis mechanism is used to obtain a plurality of power consumption values corresponding to a plurality of moments before a set future moment in the current power consumption network, wherein the number of moments before the set future moment is proportional to the power load quantity of the current power consumption network; A sequential training mechanism, used for obtaining a deep convolutional network model, in which the deep convolutional network is trained successively for multiple times to obtain a deep convolutional network after each training, and the deep convolutional network after each training is used as the deep convolutional network model; An intelligent identification device, connected to the first analysis mechanism, the second analysis mechanism and the sequential training mechanism, respectively, for inputting a plurality of power consumption values corresponding to a plurality of moments before a set future moment of the current power consumption network, the number of power loads of the current power consumption network and the respective shares of the constant power corresponding to each power load in parallel into the deep convolutional network model, and executing the deep convolutional network model to obtain the power consumption value corresponding to the current power consumption network at the set future moment output by the deep convolutional network model; A real-time notification device, connected to the intelligent identification device, for displaying in real time the power consumption value corresponding to the current power consumption network at the set future time upon receiving the power consumption value corresponding to the current power consumption network at the set future time; The sequential training mechanism is used to obtain a deep convolutional network model, in which the deep convolutional network is trained continuously for each time to obtain a deep convolutional network after each training, and the deep convolutional network after each training is used as the deep convolutional network model, including: the number of times the deep convolutional network is trained is positively correlated with the power span value in each share of the rated power corresponding to each power load; The number of times the deep convolutional network is trained is positively correlated with the power span values in each share of the constant power corresponding to each power load, including: subtracting the minimum rated power value in each share of the constant power corresponding to each power load from the maximum rated power value in each share of the constant power corresponding to each power load to obtain the power span value in each share of the constant power corresponding to each power load; Among them, the second analysis mechanism is used to obtain multiple power consumption values corresponding to multiple moments before the set future moment of the current power network, and the number of moments before the set future moment is proportional to the power load quantity of the current power network, including: the multiple moments before the set future moment include the current moment, and the multiple moments before the set future moment and the set future moment are evenly spaced on the time axis.
[0005] It can be seen that the present invention mainly has the following four significant technical effects: Technical effect 1: Introducing multiple basic data to participate in the intelligent identification of the power consumption value corresponding to the current power consumption network at a set future time, wherein the multiple basic data are various operating parameters of the current power consumption network and multiple power consumption values corresponding to multiple times before the set future time of the current power consumption network; Technical effect 2: Specifically, the various operating parameters of the current power consumption network are the number of power loads of the current power consumption network and the respective fixed powers corresponding to the respective power loads, and the number of the multiple moments before the future moment is set to be proportional to the number of power loads of the current power consumption network; Technical effect three: A deep convolutional network model with a customized structure design is introduced to perform intelligent identification of the power consumption value corresponding to the current power network at a set future moment. The customization of the deep convolutional network model structure lies in that, in the deep convolutional network model, the deep convolutional network is trained continuously for each time to obtain a deep convolutional network after each training, and the deep convolutional network after each training is used as the deep convolutional network model. What is particularly critical is that the number of times the deep convolutional network is trained is positively correlated with the power span values in each share of the rated power corresponding to each power load of the current power network; Technical effect four: subtract the minimum rated power value in each share of the constant power corresponding to each power load in the current power network from the maximum rated power value in each share of the constant power corresponding to each power load in the current power network to obtain the power span value in each share of the constant power corresponding to each power load in the current power network.
[0006] The power consumption value prediction system of the power consumption network of the present invention is stable in operation and intelligent in design. Since a deep convolutional network model can be introduced to use multiple basic data to perform intelligent identification of the power consumption value corresponding to the current power consumption network at a set future time, a reliable prediction of the power consumption value of the current power consumption network at a set future time based on the deep convolutional network can be completed. BRIEF DESCRIPTION OF THE DRAWINGS
[0007] The embodiments of the present invention will be described below with reference to the accompanying drawings, wherein:
[0008] Figure 1 Schematic diagram of the internal structure of a system for predicting power consumption values of a power network according to a first embodiment of the present invention.
[0009] Figure 2 Schematic diagram of the internal structure of a system for predicting power consumption values of a power network according to a second embodiment of the present invention.
[0010] Figure 3 Schematic diagram of the internal structure of a system for predicting power consumption values of a power network according to a third embodiment of the present invention. DETAILED DESCRIPTION
[0011] The following will describe in detail an embodiment of a system for predicting power consumption values of a power consumption network according to the present invention with reference to the accompanying drawings.
[0012] Figure 1 This is a schematic diagram of the internal structure of a system for predicting power consumption values of a power network according to a first embodiment of the present invention, wherein the system comprises: A first analysis mechanism is used to obtain various operating parameters of the current power network, wherein the various operating parameters of the current power network are the number of power loads in the current power network and the respective share rated powers corresponding to the respective power loads; For example, the first analysis mechanism is used to obtain various operating parameters of the current power network, and the various operating parameters of the current power network are the number of power loads in the current power network and the respective share constant powers corresponding to each power load, including: an FPGA chip can be selected to implement the first analysis mechanism, which is used to obtain various operating parameters of the current power network, and the various operating parameters of the current power network are the number of power loads in the current power network and the respective share constant powers corresponding to each power load; The second analysis mechanism is used to obtain a plurality of power consumption values corresponding to a plurality of moments before a set future moment in the current power consumption network, wherein the number of moments before the set future moment is proportional to the power load quantity of the current power consumption network; A sequential training mechanism, used for obtaining a deep convolutional network model, in which the deep convolutional network is trained successively for multiple times to obtain a deep convolutional network after each training, and the deep convolutional network after each training is used as the deep convolutional network model; An intelligent identification device, connected to the first analysis mechanism, the second analysis mechanism and the sequential training mechanism, respectively, for inputting a plurality of power consumption values corresponding to a plurality of moments before a set future moment of the current power consumption network, the number of power loads of the current power consumption network and the respective shares of the constant power corresponding to each power load in parallel into the deep convolutional network model, and executing the deep convolutional network model to obtain the power consumption value corresponding to the current power consumption network at the set future moment output by the deep convolutional network model; A real-time notification device, connected to the intelligent identification device, for displaying in real time the power consumption value corresponding to the current power consumption network at the set future time upon receiving the power consumption value corresponding to the current power consumption network at the set future time; The sequential training mechanism is used to obtain a deep convolutional network model, in which the deep convolutional network is trained continuously for each time to obtain a deep convolutional network after each training, and the deep convolutional network after each training is used as the deep convolutional network model, including: the number of times the deep convolutional network is trained is positively correlated with the power span value in each share of the rated power corresponding to each power load; The number of times the deep convolutional network is trained is positively correlated with the power span values in each share of the constant power corresponding to each power load, including: subtracting the minimum rated power value in each share of the constant power corresponding to each power load from the maximum rated power value in each share of the constant power corresponding to each power load to obtain the power span value in each share of the constant power corresponding to each power load; The second analysis mechanism is used to obtain a plurality of power consumption values corresponding to a plurality of moments before a set future moment in the current power network, wherein the number of moments before the set future moment is proportional to the power load quantity of the current power network, including: the plurality of moments before the set future moment include the current moment, and the plurality of moments before the set future moment and the set future moment are evenly spaced on the time axis; Among them, the multiple power consumption values corresponding to the current power network at multiple moments before the set future moment, the number of power loads of the current power network, and the respective shares of constant power corresponding to each power load are input into the deep convolutional network model in parallel, and the deep convolutional network model is executed to obtain the power consumption value corresponding to the current power network at the set future moment output by the deep convolutional network model, including: before inputting the multiple power consumption values corresponding to the current power network at multiple moments before the set future moment, the number of power loads of the current power network, and the respective shares of constant power corresponding to each power load in parallel into the deep convolutional network model, the multiple power consumption values corresponding to the current power network at multiple moments before the set future moment, the number of power loads of the current power network, and the respective shares of constant power corresponding to each power load are converted into octal values.
[0013] Figure 2 Schematic diagram of the internal structure of a system for predicting power consumption values of a power network according to a second embodiment of the present invention.
[0014] Compared to Figure 1 According to the second embodiment of the present invention, the system for predicting the power consumption value of the power network may further include: A vibration measuring mechanism, connected to the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism, respectively, for measuring the current vibration amplitude of each of the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism; Wherein, the vibration measuring mechanism is respectively connected with the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism, and is used to respectively measure the current vibration amplitudes of the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism, including: the vibration measuring mechanism includes a plurality of vibration measuring units, which are respectively connected with the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism, so as to complete the respective measurements of the current vibration amplitudes of the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism; Wherein, the vibration measuring mechanism comprises a plurality of vibration measuring units, which are respectively connected to the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism to complete the respective measurement of the current vibration amplitude of the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism, including: the plurality of vibration measuring units are a plurality of vibration sensing circuits, which are respectively connected to the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism to complete the respective measurement of the current vibration amplitude of the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism; Wherein, the plurality of vibration measuring units are a plurality of vibration sensing circuits, which are used to be connected to the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism respectively, so as to complete the respective measurement of the current vibration amplitude of the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism respectively, including: the structures of the plurality of vibration sensing circuits are the same; Among them, the multiple vibration measurement units are multiple vibration sensing circuits, which are used to be connected to the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism respectively, so as to complete the respective measurements of the current vibration amplitudes of the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism. It also includes: the multiple vibration sensing circuits have the same vibration measurement upper limit value and vibration measurement lower limit value.
[0015] Figure 3 Schematic diagram of the internal structure of a system for predicting power consumption values of a power network according to a third embodiment of the present invention.
[0016] Compared to Figure 1 According to the third embodiment of the present invention, the system for predicting the power consumption value of the power network may further include: a data caching device, arranged near the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequence training mechanism and connected to the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequence training mechanism respectively, for providing data caching services for the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequence training mechanism respectively; Wherein, the data cache device is arranged near the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequence training mechanism and is respectively connected to the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequence training mechanism, and is used to provide data cache services for the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequence training mechanism respectively, including: the data cache device is connected to the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequence training mechanism through different data cache interfaces; And wherein, the data cache device is arranged near the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequential training mechanism and is respectively connected to the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequential training mechanism, and is used to provide data caching services to the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequential training mechanism respectively, and also includes: the data transmission bandwidth of different data cache interfaces adopted by the data cache device is the same.
[0017] In addition, in the prediction system of the electricity consumption value of the power network, multiple electricity consumption values corresponding to the current power network at multiple moments before a set future moment, the number of power loads of the current power network, and the respective shares of constant power corresponding to each power load are input in parallel into the deep convolutional network model, and the deep convolutional network model is executed to obtain the electricity consumption value corresponding to the current power network at the set future moment output by the deep convolutional network model, which also includes: the electricity consumption value corresponding to the current power network at the set future moment output by the deep convolutional network model is in octal value representation.
[0018] Although the embodiments of the present invention have been described to some extent with specific examples, it should be understood that the present invention is not limited thereto. It will also be understood by those skilled in the art that various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention.
Claims
1. A system for predicting power consumption values of a power network, characterized in that: The system comprises: A first analysis mechanism is used to obtain various operating parameters of the current power network, wherein the various operating parameters of the current power network are the number of power loads in the current power network and the respective share rated powers corresponding to the respective power loads; The second analysis mechanism is used to obtain a plurality of power consumption values corresponding to a plurality of moments before a set future moment in the current power consumption network, wherein the number of moments before the set future moment is proportional to the power load quantity of the current power consumption network; A sequential training mechanism, used for obtaining a deep convolutional network model, in which the deep convolutional network is trained successively for multiple times to obtain a deep convolutional network after each training, and the deep convolutional network after each training is used as the deep convolutional network model; An intelligent identification device, connected to the first analysis mechanism, the second analysis mechanism and the sequential training mechanism, respectively, for inputting a plurality of power consumption values corresponding to a plurality of moments before a set future moment of the current power consumption network, the number of power loads of the current power consumption network and the respective shares of the constant power corresponding to each power load in parallel into the deep convolutional network model, and executing the deep convolutional network model to obtain the power consumption value corresponding to the current power consumption network at the set future moment output by the deep convolutional network model; A real-time notification device, connected to the intelligent identification device, for displaying in real time the power consumption value corresponding to the current power consumption network at the set future time upon receiving the power consumption value corresponding to the current power consumption network at the set future time; The sequential training mechanism is used to obtain a deep convolutional network model, in which the deep convolutional network is trained continuously for each time to obtain a deep convolutional network after each training, and the deep convolutional network after each training is used as the deep convolutional network model, including: the number of times the deep convolutional network is trained is positively correlated with the power span value in each share of the rated power corresponding to each power load; The number of times the deep convolutional network is trained is positively correlated with the power span values in each share of the constant power corresponding to each power load, including: subtracting the minimum rated power value in each share of the constant power corresponding to each power load from the maximum rated power value in each share of the constant power corresponding to each power load to obtain the power span value in each share of the constant power corresponding to each power load; Among them, the second analysis mechanism is used to obtain multiple power consumption values corresponding to multiple moments before the set future moment of the current power network, and the number of moments before the set future moment is proportional to the power load quantity of the current power network, including: the multiple moments before the set future moment include the current moment, and the multiple moments before the set future moment and the set future moment are evenly spaced on the time axis.
2. The power consumption value prediction system of the power network according to claim 1, characterized in that: Inputting multiple power consumption values corresponding to the current power network at multiple moments before a set future moment, the number of power loads of the current power network, and the respective shares of constant power corresponding to each power load in parallel into the deep convolutional network model, and executing the deep convolutional network model to obtain the power consumption value corresponding to the current power network at the set future moment output by the deep convolutional network model includes: before inputting multiple power consumption values corresponding to the current power network at multiple moments before a set future moment, the number of power loads of the current power network, and the respective shares of constant power corresponding to each power load in parallel into the deep convolutional network model, performing octal value conversion processing on the multiple power consumption values corresponding to the current power network at multiple moments before a set future moment, the number of power loads of the current power network, and the respective shares of constant power corresponding to each power load.
3. The system for predicting power consumption values of a power network according to claim 2, characterized in that: The system further comprises: A vibration measuring mechanism, connected to the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism, respectively, for measuring the current vibration amplitude of each of the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism; Among them, the vibration measuring mechanism is respectively connected to the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism, and is used to respectively measure the current vibration amplitudes of the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism. The vibration measuring mechanism includes multiple vibration measuring units, which are respectively connected to the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism to complete the respective measurements of the current vibration amplitudes of the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism.
4. The system for predicting power consumption values of a power network as claimed in claim 3, characterized in that: The vibration measuring mechanism includes a plurality of vibration measuring units, which are used to be respectively connected to the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism to complete the respective measurements of the current vibration amplitudes of the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism. The plurality of vibration measuring units are a plurality of vibration sensing circuits, which are used to be respectively connected to the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism to complete the respective measurements of the current vibration amplitudes of the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism.
5. The system for predicting power consumption values of a power network as claimed in claim 4, characterized in that: The multiple vibration measurement units are multiple vibration sensing circuits, which are used to be connected to the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism respectively, so as to complete the respective measurements of the current vibration amplitudes of the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism, including: the structures of the multiple vibration sensing circuits are the same.
6. The system for predicting power consumption values of a power network according to claim 5, characterized in that: The multiple vibration measurement units are multiple vibration sensing circuits, which are used to be connected to the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism respectively, so as to complete the respective measurements of the current vibration amplitudes of the intelligent recognition device, the first analysis mechanism, the second analysis mechanism and the sequential training mechanism. It also includes: the multiple vibration sensing circuits have the same vibration measurement upper limit value and vibration measurement lower limit value.
7. The system for predicting power consumption values of a power network according to claim 2, characterized in that: The system further comprises: A data caching device is arranged near the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequential training mechanism and is respectively connected to the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequential training mechanism, and is used to provide data caching services for the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequential training mechanism respectively.
8. The power consumption value prediction system of the power network according to claim 7, characterized in that: A data cache device is arranged near the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequential training mechanism and is respectively connected to the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequential training mechanism, and is used to provide data cache services for the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequential training mechanism respectively, including: the data cache device is connected to the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequential training mechanism through different data cache interfaces.
9. The system for predicting power consumption values of a power network as claimed in claim 8, characterized in that: A data cache device is arranged near the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequential training mechanism and is respectively connected to the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequential training mechanism, and is used to provide data cache services to the intelligent recognition device, the first parsing mechanism, the second parsing mechanism and the sequential training mechanism respectively. It also includes: the data transmission bandwidth of different data cache interfaces used by the data cache device is the same.
