Power grid demand side management method and system based on multi-source data analysis

Abstract

The invention relates to the technical field of power grid management, and discloses a power grid demand side management method and system based on multi-source data analysis, and the method comprises the steps: inputting predictive load characteristic data and non-predictive load characteristic data into a pre-configured second deep learning model, and obtaining a power demand load capacity; judging whether the power grid is in an overload state in a future time period or not; acquiring utilization state data of K load type energy storage devices in the location of the target area, and screening out the optimal load type energy storage device based on the utilization state data; in the overload operation period, the optimal load type energy storage equipment is controlled according to the temperature influence load capacity, and power is supplied to the target area; the method can reveal the potential mechanism behind the temperature change and the power grid load change, is beneficial to relieving the power load pressure caused by the temperature change, and reduces the operation risk of the power grid.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid management, and more specifically, to a power grid demand side management method and system based on multi-source data analysis. Background Art

[0002] With the acceleration of industrialization and urbanization, people's demand for electricity continues to grow, which has caused the power system to bear huge power supply pressure; traditional power grid management is mainly concentrated on the supply side, that is, to cope with the growth of electricity demand by improving power generation capacity, improving transmission facilities, etc.; however, compared with the supply-side management of the power grid, the demand-side management of the power grid (Demand Side Management, DSM) has higher flexibility, higher stability and reliability; in addition, with the rapid development of load-type virtual power plants, it provides more flexible and sustainable solutions for the demand-side management of the power grid, while promoting the consumption of renewable energy and reducing environmental impact.

[0003] At present, most of the existing power grid demand side management methods or systems use flexible electricity price policies or reward and punishment mechanisms to guide users to reduce electricity consumption during peak hours, thereby balancing electricity supply and demand; however, solving the load pressure of the power system by influencing user behavior often increases the uncertainty of user electricity consumption, which is easy to bring users a poor electricity consumption experience and reduce the revenue of power companies; of course, there are also some demand side management technology documents in other directions, such as the patent with authorization announcement number CN112086978B, which discloses an energy dispatching and control system based on demand side response. Although the above method can use energy storage equipment to adjust the power grid load, it is found through research and practical application of the above method and the existing technology that the above method and the existing technology have at least the following defects:

[0004] (1) The data source is single and lacks deep data mining, which makes it impossible to accurately reveal the underlying mechanism behind temperature changes and grid load changes, that is, it is impossible to understand the power load pressure caused by temperature changes;

[0005] (2) It is impossible to determine the impact of temperature on load and select the best load-type energy storage device based on the underlying mechanism. Furthermore, it is impossible to use load-type virtual power plants to flexibly adjust power supply, alleviate the pressure on power load caused by temperature changes, and reduce the operating risks of the power grid and improve the safety and stability of the power grid. Summary of the invention

[0006] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a power grid demand side management method and system based on multi-source data analysis.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A power grid demand side management method based on multi-source data analysis, the method comprising:

[0009] S101: Obtain predictable load characteristic data and non-predictable load characteristic data of the target area in the future period [T+N, T+2N], input the predictable load characteristic data and the non-predictable load characteristic data into a pre-configured second deep learning model, and obtain the power demand load in the future period, where T and N are both integers greater than zero;

[0010] S102: judging whether the power grid is in an overload state in the future time period according to the power demand load; if so, marking the future time period as an overload operation period, and determining the temperature impact load in the overload operation period based on the pre-configured third deep learning model; if not, setting T=T+M, and returning to step S101, where M is an integer greater than zero;

[0011] S103: Obtaining utilization status data of K load-type energy storage devices in the target area, and selecting the best load-type energy storage device based on the utilization status data, where K is an integer greater than zero;

[0012] S104: When in the overload operation period, the optimal load-type energy storage device is controlled according to the temperature-affected load amount to supply power to the target area.

[0013] Further, the predictive load characteristic data includes weather data, electricity price and population in the future period [T+N, T+2N]; wherein the weather data includes rainfall / snowfall, temperature, humidity and wind speed;

[0014] The acquisition logic of the non-predictable load characteristic data is as follows:

[0015] Obtaining utilization coefficient characteristic data of the target area during the period from the past time TN to the current time T;

[0016] The utilization coefficient characteristic data includes the number of electricity consumption units, the number of electrical equipment in each electricity consumption unit, the rotation speed of the electricity meter of each electricity consumption unit at each moment, and the operation time of the electricity meter of each electricity consumption unit at each moment;

[0017] Input the utilization coefficient feature data from the past time TN to the current time T into the pre-configured first neural network to obtain the utilization coefficient in the future time period [T+N, T+2N];

[0018] The utilization coefficient is used as the unpredictable load characteristic data to obtain the unpredictable load characteristic data in the future period [T+N, T+2N].

[0019] Furthermore, the generation logic of the preconfigured first neural network is as follows:

[0020] Acquire historical utilization coefficient training data, and divide the historical utilization coefficient training data into a utilization coefficient training set and a utilization coefficient test set; the historical utilization coefficient training data includes utilization coefficient feature data and its corresponding utilization coefficient;

[0021] Constructing a first regression network, using the utilization coefficient feature data in the utilization coefficient training set as input data of the first regression network, and using the utilization coefficient in the utilization coefficient training set as output data of the first regression network, training the first regression network, and obtaining an initial first regression network;

[0022] The initial first regression network is model verified by utilizing a coefficient test set, and an initial first regression network that is less than or equal to a preset first test error threshold is output as a preconfigured first neural network.

[0023] Furthermore, the logic for obtaining the utilization coefficient in the historical utilization coefficient training data is:

[0024] Obtain the total number of electrical equipment in the target area, the actual power consumption of each electrical equipment, the rated power capacity of each electrical equipment, and the power consumption time of each electrical equipment;

[0025] The total number of electrical devices, the actual power consumption of each electrical device, the rated power capacity of each electrical device, and the power consumption time of each electrical device are input into a pre-built mathematical calculation model to obtain a utilization coefficient; wherein the formula of the mathematical calculation model is: Where: α is the utilization coefficient, E i is the actual power consumption of the ith electrical equipment, in kilowatt-hours (kWh); P i is the rated power capacity of the ith electrical equipment, in kilowatts (kW); C i is the usage time of the ith electrical equipment, and Q is the usage time of the ith electrical equipment.

[0026] Furthermore, the generation logic of the preconfigured second deep learning model is as follows:

[0027] Acquire historical power load training data, and divide the historical power load training data into a power load training set and a power load test set; the historical power load training data includes power load characteristic data and its corresponding power demand load; wherein the power load characteristic data includes predictive load characteristic data and non-predictive load characteristic data;

[0028] Constructing a second regression network, using the power load characteristic data in the power load training set as input data of the second regression network, and using the power demand load in the power load training set as output data of the second regression network, training the second regression network, and obtaining an initial second regression network;

[0029] The initial second regression network is model verified using the power load test set, and an initial second regression network that is less than or equal to a preset second test error threshold is output as a preconfigured second neural network.

[0030] Further, the determining whether the power grid is in an overload state in the future period includes:

[0031] Comparing the power demand load with the preset standard power supply load threshold;

[0032] If the power demand load is greater than or equal to the preset standard power supply load threshold, it is determined that the power grid is in an overload state in the future period;

[0033] If the power demand load is less than the preset standard power supply load threshold, it is determined that the power grid is not in an overload state in the future period.

[0034] Furthermore, the process of determining the temperature-affected load is as follows:

[0035] Extracting the power demand load during the overload operation period, and extracting the temperature difference during the overload operation period;

[0036] Inputting the power demand load and the temperature difference into a preconfigured third deep learning model to obtain the temperature-affected load;

[0037] The generation logic of the preconfigured third deep learning model is as follows:

[0038] Acquire historical temperature impact load training data, and divide the historical temperature impact load training data into a temperature impact load training set and a temperature impact load test set; the historical temperature impact load training data includes a temperature difference and its corresponding temperature impact load amount;

[0039] Constructing a third regression network, taking the temperature difference in the temperature impact load training set as input data of the third regression network, taking the temperature impact load amount in the temperature impact load training set as output data of the third regression network, training the third regression network, and obtaining an initial third regression network;

[0040] The temperature impact load test set is used to perform model verification on the initial third regression network, and an initial third regression network that is less than or equal to a preset third test error threshold is output as a preconfigured third neural network.

[0041] Furthermore, the logic for obtaining the temperature impact load amount in the historical temperature impact load training data is as follows:

[0042] A1: In a test scenario, obtain test data of S test periods, where the test data includes actual mean temperature in the S test periods and power demand load in the S test periods, where S is an integer greater than zero;

[0043] A2: Extract the actual average temperature of the target area in the vth test period, and obtain the power demand load in the vth test period;

[0044] A3: Obtain the standard temperature value of the target area and the standard power demand load corresponding to the standard temperature value, calculate the difference between the actual average temperature and the standard temperature value in the vth test period, and obtain the temperature difference; and calculate the difference between the power demand load in the vth test period and the standard power demand load, mark the difference between the power demand load in the vth test period and the standard power demand load as the temperature impact load, associate and bind the temperature difference with the temperature impact load, set v=v+1, and return to step A2, where v is an integer greater than zero;

[0045] A4: Repeat the above steps A2 to A3 until v=S and the cycle ends, and the temperature influence load associated with each temperature difference is obtained.

[0046] Furthermore, the utilization status data includes the utilization cost, current charge and failure rate of each load-type energy storage device;

[0047] The screening of the best load-type energy storage device comprises:

[0048] Extract the utilization status data of K load-type energy storage devices, and extract the load affected by temperature;

[0049] Combine the utilization status data of each load-type energy storage device with the temperature-affected load in turn to calculate the matching coefficient of K load-type energy storage devices; the calculation formula is: μ j =exp[(F j +(γ′ j -γ″) 2 )×φ j ]; where: μ j is the matching coefficient of the jth load-type energy storage device, F j is the utilization cost of the jth load-type energy storage device, γ j ′ is the current charge of the jth load-type energy storage device, γ″ is the temperature-affected load, φ j is the failure rate of the jth load-type energy storage device;

[0050] The matching coefficients of the K load-type energy storage devices are sorted from small to large, and the load-type energy storage device corresponding to the first matching coefficient is taken as the optimal load-type energy storage device.

[0051] A power grid demand side management system based on multi-source data analysis, comprising:

[0052] A data acquisition module is used to obtain the predictable load characteristic data and the non-predictable load characteristic data of the target area in the future time period [T+N, T+2N], input the predictable load characteristic data and the non-predictable load characteristic data into the pre-configured second deep learning model, and obtain the power demand load in the future time period, where T and N are both integers greater than zero;

[0053] A load judgment module is used to judge whether the power grid is in an overload state in a future period according to the power demand load. If so, the future period is marked as an overload operation period, and the temperature-affected load in the overload operation period is determined based on a preconfigured third deep learning model; if not, T=T+M is set, and the data acquisition module is triggered, where M is an integer greater than zero;

[0054] An equipment screening module, used to obtain the utilization status data of K load-type energy storage devices in the target area, and screen out the best load-type energy storage device based on the utilization status data, where K is an integer greater than zero;

[0055] The power dispatching module is used to control the optimal load-type energy storage device according to the temperature impact load when in the overload operation period, and supply power to the target area.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] The present application discloses a method and system for power grid demand side management based on multi-source data analysis, including: inputting predictive load characteristic data and non-predictive load characteristic data into a preconfigured second deep learning model to obtain the power demand load; judging whether the power grid is in an overload state in the future period; obtaining the utilization status data of K load-type energy storage devices in the target area, and screening out the best load-type energy storage device based on the utilization status data; when in the overload operation period, controlling the best load-type energy storage device according to the temperature-affected load to supply power to the target area; based on the above process, the present invention can perform deep data mining and accurately reveal the potential mechanism behind the temperature change and the power grid load change, which is conducive to understanding the power load pressure caused by the temperature change; in addition, by determining the temperature-affected load and screening out the best load-type energy storage device, the present invention can use the load-type virtual power plant to flexibly adjust the power supply, which is conducive to alleviating the power load pressure caused by temperature changes, and further, it is conducive to reducing the operation risk of the power grid and improving the safety and stability of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 A flowchart of a power grid demand side management method based on multi-source data analysis provided by the present invention;

[0059] Figure 2 A module structure diagram of a power grid demand side management system based on multi-source data analysis provided by the present invention. DETAILED DESCRIPTION

[0060] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0061] Example 1

[0062] See also Figure 1 As shown, this embodiment discloses a power grid demand side management method based on multi-source data analysis, the method comprising:

[0063] S101: Obtain predictable load characteristic data and non-predictable load characteristic data of the target area in the future period [T+N, T+2N], input the predictable load characteristic data and the non-predictable load characteristic data into a pre-configured second deep learning model, and obtain the power demand load in the future period, where T and N are both integers greater than zero;

[0064] It should be noted that: the target area refers to an area with power demand, such as a building;

[0065] Those skilled in the art will appreciate that, in other alternative implementations, the target area may also be a residential area or a town, etc., which may be selected according to actual needs, and the present invention does not make too many limitations on this;

[0066] Specifically, the predictive load characteristic data includes weather data, electricity prices and population in the future period [T+N, T+2N]; wherein the weather data includes rainfall / snowfall, temperature, humidity and wind speed;

[0067] It should be noted that: the predictive load characteristic data is obtained by accessing various platforms or through sensor monitoring; for example, the weather data in the future period [T+N, T+2N] can be obtained through various weather forecast platforms, or through various sensors, including but not limited to rainfall / snowfall sensors, temperature and humidity sensors, and wind speed sensors; the population is obtained by accessing various population census platforms, and the electricity price is obtained by accessing the electricity market platform or the electric energy management platform;

[0068] In implementation, the acquisition logic of the non-predictable load characteristic data is as follows:

[0069] Obtaining utilization coefficient characteristic data of the target area during the period from the past time TN to the current time T;

[0070] The utilization coefficient characteristic data includes the number of electricity consumption units, the number of electrical equipment in each electricity consumption unit, the rotation speed of the electricity meter of each electricity consumption unit at each moment, and the operation time of the electricity meter of each electricity consumption unit at each moment;

[0071] It should be noted that: the number of the electricity-consuming units and the number of the electricity-consuming devices in each electricity-consuming unit are pre-stored in the system database and are collected and recorded in advance by the technical personnel; the speed of the meter of each electricity-consuming unit at each moment and the operation time of the meter of each electricity-consuming unit at each moment are obtained by the sensors installed on each meter, and the sensors on the meter include but are not limited to speed sensors and timers, etc.;

[0072] Input the utilization coefficient feature data from the past time TN to the current time T into the pre-configured first neural network to obtain the utilization coefficient in the future time period [T+N, T+2N];

[0073] Specifically, the generation logic of the preconfigured first neural network is as follows:

[0074] Acquire historical utilization coefficient training data, and divide the historical utilization coefficient training data into a utilization coefficient training set and a utilization coefficient test set; the historical utilization coefficient training data includes utilization coefficient feature data and its corresponding utilization coefficient;

[0075] Among them, the acquisition of the utilization coefficient characteristic data in the historical utilization coefficient training data is introduced by referring to the above content, and no further description is given here.

[0076] The logic for obtaining the utilization coefficient in the historical utilization coefficient training data is as follows:

[0077] Obtain the total number of electrical equipment in the target area, the actual power consumption of each electrical equipment, the rated power capacity of each electrical equipment, and the power consumption time of each electrical equipment;

[0078] It should be noted that the total number of electrical devices in the target area, the actual power consumption of each electrical device and the rated power capacity of each electrical device are obtained by offline collection and recording in advance by technical personnel; the power consumption time of each electrical device is obtained by pre-installing sensor equipment or by offline collection and recording in advance by technical personnel;

[0079] The total number of electrical devices, the actual power consumption of each electrical device, the rated power capacity of each electrical device, and the power consumption time of each electrical device are input into a pre-built mathematical calculation model to obtain a utilization coefficient; wherein the formula of the mathematical calculation model is: Where: α is the utilization coefficient, E i is the actual power consumption of the ith electrical equipment, in kilowatt-hours (kWh); P i is the rated power capacity of the ith electrical equipment, in kilowatts (kW); C i is the usage time of the ith electrical equipment, Q is the usage time of the ith electrical equipment;

[0080] Constructing a first regression network, using the utilization coefficient feature data in the utilization coefficient training set as input data of the first regression network, and using the utilization coefficient in the utilization coefficient training set as output data of the first regression network, training the first regression network, and obtaining an initial first regression network;

[0081] The initial first regression network is model-verified by utilizing a coefficient test set, and an initial first regression network having a value less than or equal to a preset first test error threshold is output as a preconfigured first neural network;

[0082] It should be noted that: the first regression network is specifically one of deep neural network algorithms such as convolutional neural network (CNN), recurrent neural network (RNN) or long short-term memory network (LSTM);

[0083] Taking the utilization factor as the unpredictable load characteristic data, the unpredictable load characteristic data in the future period [T+N, T+2N] is obtained;

[0084] In implementation, the generation logic of the preconfigured second deep learning model is as follows:

[0085] Acquire historical power load training data, and divide the historical power load training data into a power load training set and a power load test set; the historical power load training data includes power load characteristic data and its corresponding power demand load; wherein the power load characteristic data includes predictive load characteristic data and non-predictive load characteristic data;

[0086] It should be noted that: the acquisition logic of the power load characteristic data in the historical power load training data refers to the above description and will not be repeated here; the power demand load in the historical power load training data is obtained by actual records of technicians;

[0087] Constructing a second regression network, using the power load characteristic data in the power load training set as input data of the second regression network, and using the power demand load in the power load training set as output data of the second regression network, training the second regression network, and obtaining an initial second regression network;

[0088] Using the power load test set to perform model verification on the initial second regression network, outputting an initial second regression network that is less than or equal to a preset second test error threshold as a preconfigured second neural network;

[0089] It should be noted that: like the first regression network, the second regression network is specifically one of deep neural network algorithms such as convolutional neural network (CNN), recurrent neural network (RNN) or long short-term memory network (LSTM).

[0090] S102: judging whether the power grid is in an overload state in the future time period according to the power demand load; if so, marking the future time period as an overload operation period, and determining the temperature impact load in the overload operation period based on the pre-configured third deep learning model; if not, setting T=T+M, and returning to step S101, where M is an integer greater than zero;

[0091] In implementation, the determining whether the power grid is in an overload state in the future period includes:

[0092] Comparing the power demand load with the preset standard power supply load threshold;

[0093] It should be noted that: the preset standard power supply load threshold is pre-stored in the system database, and the preset standard power supply load threshold reveals the safe power supply load of the power grid. When the supply is less than the demand, it means that the power grid is overloaded;

[0094] If the power demand load is greater than or equal to the preset standard power supply load threshold, it is determined that the power grid is in an overload state in the future period;

[0095] If the power demand load is less than the preset standard power supply load threshold, it is determined that the power grid is not in an overload state in the future period;

[0096] In practice, the process of determining the temperature-affected load is as follows:

[0097] Extracting the power demand load during the overload operation period, and extracting the temperature difference during the overload operation period;

[0098] Inputting the power demand load and the temperature difference into a preconfigured third deep learning model to obtain the temperature-affected load;

[0099] Specifically, the generation logic of the preconfigured third deep learning model is as follows:

[0100] Acquire historical temperature impact load training data, and divide the historical temperature impact load training data into a temperature impact load training set and a temperature impact load test set; the historical temperature impact load training data includes a temperature difference and its corresponding temperature impact load amount;

[0101] It should be noted that the logic for obtaining the temperature difference in the historical temperature impact load training data is as follows: the actual temperature value at each moment in the test period is collected by temperature sensing, and then the actual temperature value is averaged according to the test period, and the difference between the actual average temperature obtained and the standard temperature value is calculated to obtain the temperature difference;

[0102] It should also be noted that the standard temperature value is set in advance by the technicians, and the standard temperature value reveals the temperature value that has the smallest impact on the power demand load; for example, assuming that when the temperature is 25 degrees Celsius, the power demand load of the target area is the lowest at this time, in other words, the power demand load of the target area is least affected by the temperature at this time, and this temperature is used as the standard temperature value;

[0103] The logic for obtaining the temperature impact load amount in the historical temperature impact load training data is as follows:

[0104] A1: In a test scenario, obtain test data of S test periods, where the test data includes actual mean temperature in the S test periods and power demand load in the S test periods, where S is an integer greater than zero;

[0105] A2: Extract the actual average temperature of the target area in the vth test period, and obtain the power demand load in the vth test period;

[0106] A3: Obtain the standard temperature value of the target area and the standard power demand load corresponding to the standard temperature value, calculate the difference between the actual average temperature and the standard temperature value in the vth test period, and obtain the temperature difference; and calculate the difference between the power demand load in the vth test period and the standard power demand load, mark the difference between the power demand load in the vth test period and the standard power demand load as the temperature impact load, associate and bind the temperature difference with the temperature impact load, set v=v+1, and return to step A2, where v is an integer greater than zero;

[0107] It should be noted that the standard temperature value of the target area and the standard power demand load corresponding to the standard temperature value are obtained by actual experiments of technicians and pre-stored in the system database, and each standard temperature value of the target area corresponds to a corresponding standard power demand load, which is determined according to specific experimental results;

[0108] A4: Repeat the above steps A2 to A3 until v=S, and end the cycle to obtain the temperature impact load associated with each temperature difference;

[0109] It can be understood that: in the test phase, each temperature difference is associated with the corresponding temperature-affected load, and then by measuring the temperature change, the power load generated by the temperature change in the power demand load, that is, the temperature-affected load, is obtained;

[0110] Constructing a third regression network, taking the temperature difference in the temperature impact load training set as input data of the third regression network, taking the temperature impact load amount in the temperature impact load training set as output data of the third regression network, training the third regression network, and obtaining an initial third regression network;

[0111] Using the temperature impact load test set to perform model verification on the initial third regression network, outputting an initial third regression network that is less than or equal to a preset third test error threshold as a preconfigured third neural network;

[0112] It should be noted that: like the above-mentioned first regression network or second regression network, the second regression network is specifically one of deep neural network algorithms such as convolutional neural network (CNN), recurrent neural network (RNN) or long short-term memory network (LSTM).

[0113] S103: Obtaining utilization status data of K load-type energy storage devices in the target area, and selecting the best load-type energy storage device based on the utilization status data, where K is an integer greater than zero;

[0114] It should be understood that a load-based virtual power plant is a system that uses renewable energy, energy storage and other resources to flexibly dispatch and optimize the power system through intelligent control technology and market mechanisms. On the demand side, various load-based energy storage devices (such as solar energy storage devices and wind energy storage devices) are used to regulate and dispatch electricity, which is conducive to greatly alleviating the load pressure of the centralized power grid.

[0115] Specifically, the utilization status data includes the utilization cost, current charge and failure rate of each load-type energy storage device;

[0116] It should be noted that the utilization cost and failure rate of the load-type energy storage device are obtained by experimental calculation in advance by technical personnel, and the current charge is collected by the monitoring system or sensor of the load-type energy storage device;

[0117] In implementation, the screening of the best load-type energy storage device includes:

[0118] Extract the utilization status data of K load-type energy storage devices, and extract the load affected by temperature;

[0119] Combine the utilization status data of each load-type energy storage device with the temperature-affected load in turn to calculate the matching coefficient of K load-type energy storage devices; the calculation formula is: μ j =exp[(F j +(γ′ j -γ″) 2 )×φ j ]; where: μ j is the matching coefficient of the jth load-type energy storage device, F j is the utilization cost of the jth load-type energy storage device, γ′ j is the current charge of the jth load-type energy storage device, γ″ is the load affected by temperature, φ j is the failure rate of the jth load-type energy storage device;

[0120] The matching coefficients of the K load-type energy storage devices are sorted from small to large, and the load-type energy storage device corresponding to the first matching coefficient is taken as the optimal load-type energy storage device.

[0121] S104: When in the overload operation period, the optimal load-type energy storage device is controlled according to the temperature-affected load amount to supply power to the target area;

[0122] It should be noted that: when the power demand load and the temperature-affected load are known, the difference between the power demand load and the temperature-affected load is calculated to obtain the power demand caused by non-temperature changes; therefore, when in the overload operation period, the power grid is first used for power supply to meet the power demand caused by non-temperature changes; then, the power grid is interrupted, the line between the optimal load-type energy storage device and the target area is connected, and in the overload operation period, the optimal load-type energy storage device is controlled according to the temperature-affected load to supply power to the target area, which can alleviate the power load pressure caused by temperature changes;

[0123] By revealing the potential mechanism behind temperature changes and grid load changes, and screening out the best load-type energy storage device based on the potential mechanism, and using the best load-type energy storage device to switch power supply, the present invention is conducive to using load-type virtual power plants to flexibly adjust power supply. Further, it is conducive to alleviating the power load pressure caused by temperature changes, reducing the operating risks of the power grid, and improving the safety and stability of the power grid.

[0124] Example 2

[0125] See also Figure 2 As shown, based on the same inventive concept, this embodiment discloses a power grid demand side management system based on multi-source data analysis. For details of this embodiment, please refer to the above embodiment 1. The system includes:

[0126] The data acquisition module 210 is used to obtain the predictable load characteristic data and the non-predictable load characteristic data of the target area in the future time period [T+N, T+2N], input the predictable load characteristic data and the non-predictable load characteristic data into the pre-configured second deep learning model, and obtain the power demand load in the future time period, where T and N are both integers greater than zero;

[0127] A load judgment module 220 is used to judge whether the power grid is in an overload state in a future period according to the power demand load. If so, the future period is marked as an overload operation period, and the temperature impact load in the overload operation period is determined based on the pre-configured third deep learning model; if not, T=T+M is set, and the data acquisition module is triggered, where M is an integer greater than zero 210;

[0128] The device screening module 230 is used to obtain the utilization status data of K load-type energy storage devices in the target area, and screen out the best load-type energy storage device based on the utilization status data, where K is an integer greater than zero;

[0129] The power dispatching module 240 is used to control the optimal load-type energy storage device according to the temperature-affected load amount to supply power to the target area when the overload operation period occurs.

[0130] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters, weights and thresholds in the formula are set by technicians in this field according to actual conditions.

[0131] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center through a wired network or a wireless network. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD) or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0132] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0133] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0134] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the units is only one, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0135] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0136] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0137] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0138] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A power grid demand side management method based on multi-source data analysis, characterized in that: The method comprises: S101: Obtain predictable load characteristic data and non-predictable load characteristic data of the target area in the future period [T+N, T+2N], input the predictable load characteristic data and the non-predictable load characteristic data into a pre-configured second deep learning model, and obtain the power demand load in the future period, where T and N are both integers greater than zero; S102: judging whether the power grid is in an overload state in the future time period according to the power demand load; if so, marking the future time period as an overload operation period, and determining the temperature impact load in the overload operation period based on the pre-configured third deep learning model; if not, setting T=T+M, and returning to step S101, where M is an integer greater than zero; S103: Obtaining utilization status data of K load-type energy storage devices in the target area, and selecting the best load-type energy storage device based on the utilization status data, where K is an integer greater than zero; S104: When in the overload operation period, the optimal load-type energy storage device is controlled according to the temperature-affected load amount to supply power to the target area.

2. The power grid demand side management method based on multi-source data analysis according to claim 1 is characterized in that: The predictive load characteristic data includes weather data, electricity price and population in the future period [T+N, T+2N]; wherein the weather data includes rainfall / snowfall, temperature, humidity and wind speed; The acquisition logic of the non-predictable load characteristic data is as follows: Obtaining utilization coefficient characteristic data of the target area during the period from the past time TN to the current time T; The utilization coefficient characteristic data includes the number of electricity consumption units, the number of electrical equipment in each electricity consumption unit, the rotation speed of the electricity meter of each electricity consumption unit at each moment, and the operation time of the electricity meter of each electricity consumption unit at each moment; Input the utilization coefficient feature data from the past time TN to the current time T into the pre-configured first neural network to obtain the utilization coefficient in the future time period [T+N, T+2N]; The utilization coefficient is used as the unpredictable load characteristic data to obtain the unpredictable load characteristic data in the future period [T+N, T+2N].

3. The power grid demand side management method based on multi-source data analysis according to claim 2 is characterized in that: The generation logic of the preconfigured first neural network is as follows: Acquire historical utilization coefficient training data, and divide the historical utilization coefficient training data into a utilization coefficient training set and a utilization coefficient test set; the historical utilization coefficient training data includes utilization coefficient feature data and its corresponding utilization coefficient; Constructing a first regression network, using the utilization coefficient feature data in the utilization coefficient training set as input data of the first regression network, and using the utilization coefficient in the utilization coefficient training set as output data of the first regression network, training the first regression network, and obtaining an initial first regression network; The initial first regression network is model verified by utilizing a coefficient test set, and an initial first regression network that is less than or equal to a preset first test error threshold is output as a preconfigured first neural network.

4. The power grid demand side management method based on multi-source data analysis according to claim 3 is characterized in that: The logic for obtaining the utilization coefficient in the historical utilization coefficient training data is: Obtain the total number of electrical equipment in the target area, the actual power consumption of each electrical equipment, the rated power capacity of each electrical equipment, and the power consumption time of each electrical equipment; The total number of electrical devices, the actual power consumption of each electrical device, the rated power capacity of each electrical device, and the power consumption time of each electrical device are input into a pre-built mathematical calculation model to obtain a utilization coefficient; wherein the formula of the mathematical calculation model is: Where: α is the utilization coefficient, E i is the actual power consumption of the ith electrical equipment, in kilowatt-hours (kWh); P i is the rated power capacity of the ith electrical equipment, in kilowatts (kW); C i is the usage time of the ith electrical equipment, and Q is the usage time of the ith electrical equipment.

5. The power grid demand side management method based on multi-source data analysis according to claim 4 is characterized in that: The generation logic of the preconfigured second deep learning model is as follows: Acquire historical power load training data, and divide the historical power load training data into a power load training set and a power load test set; the historical power load training data includes power load characteristic data and its corresponding power demand load; wherein the power load characteristic data includes predictive load characteristic data and non-predictive load characteristic data; Constructing a second regression network, using the power load characteristic data in the power load training set as input data of the second regression network, and using the power demand load in the power load training set as output data of the second regression network, training the second regression network, and obtaining an initial second regression network; The initial second regression network is model verified using the power load test set, and an initial second regression network that is less than or equal to a preset second test error threshold is output as a preconfigured second neural network.

6. The power grid demand side management method based on multi-source data analysis according to claim 5 is characterized in that: The determining whether the power grid is in an overload state in the future period includes: Comparing the power demand load with the preset standard power supply load threshold; If the power demand load is greater than or equal to the preset standard power supply load threshold, it is determined that the power grid is in an overload state in the future period; If the power demand load is less than the preset standard power supply load threshold, it is determined that the power grid is not in an overload state in the future period.

7. The power grid demand side management method based on multi-source data analysis according to claim 6 is characterized in that: The determination process of the temperature-affected load is as follows: Extracting the power demand load during the overload operation period, and extracting the temperature difference during the overload operation period; Inputting the power demand load and the temperature difference into a preconfigured third deep learning model to obtain the temperature-affected load; The generation logic of the preconfigured third deep learning model is as follows: Acquire historical temperature impact load training data, and divide the historical temperature impact load training data into a temperature impact load training set and a temperature impact load test set; the historical temperature impact load training data includes a temperature difference and its corresponding temperature impact load amount; Constructing a third regression network, taking the temperature difference in the temperature impact load training set as input data of the third regression network, taking the temperature impact load amount in the temperature impact load training set as output data of the third regression network, training the third regression network, and obtaining an initial third regression network; The temperature impact load test set is used to perform model verification on the initial third regression network, and an initial third regression network that is less than or equal to a preset third test error threshold is output as a preconfigured third neural network.

8. The power grid demand side management method based on multi-source data analysis according to claim 7 is characterized in that: The logic for obtaining the temperature impact load in the historical temperature impact load training data is as follows: A1: In a test scenario, obtain test data of S test periods, where the test data includes actual mean temperature in the S test periods and power demand load in the S test periods, where S is an integer greater than zero; A2: Extract the actual average temperature of the target area in the vth test period, and obtain the power demand load in the vth test period; A3: Obtain the standard temperature value of the target area and the standard power demand load corresponding to the standard temperature value, calculate the difference between the actual average temperature and the standard temperature value in the vth test period, and obtain the temperature difference; and calculating the difference between the power demand load of the vth test period and the standard power demand load, marking the difference between the power demand load of the vth test period and the standard power demand load as the temperature impact load, and associating and binding the temperature difference with the temperature impact load, and setting v=v+1, and returning to step A2, where v is an integer greater than zero; A4: Repeat the above steps A2 to A3 until v=S and the cycle ends, and the temperature influence load associated with each temperature difference is obtained.

9. The power grid demand side management method based on multi-source data analysis according to claim 8, characterized in that: The utilization status data includes the utilization cost, current charge and failure rate of each load-type energy storage device; The screening of the best load-type energy storage device comprises: Extract the utilization status data of K load-type energy storage devices, and extract the load affected by temperature; Combine the utilization status data of each load-type energy storage device with the temperature-affected load in turn to calculate the matching coefficient of K load-type energy storage devices; the calculation formula is: μ j =exp[(F j +(γ′ j -γ″) 2 )×φ j ]; where: μ j is the matching coefficient of the jth load-type energy storage device, F j is the utilization cost of the jth load-type energy storage device, γ′ j is the current charge of the jth load-type energy storage device, γ″ is the load affected by temperature, φ j is the failure rate of the jth load-type energy storage device; The matching coefficients of the K load-type energy storage devices are sorted from small to large, and the load-type energy storage device corresponding to the first matching coefficient is taken as the optimal load-type energy storage device.

10. A power grid demand side management system based on multi-source data analysis, characterized in that: include: A data acquisition module is used to obtain the predictable load characteristic data and the non-predictable load characteristic data of the target area in the future time period [T+N, T+2N], input the predictable load characteristic data and the non-predictable load characteristic data into the pre-configured second deep learning model, and obtain the power demand load in the future time period, where T and N are both integers greater than zero; A load judgment module is used to judge whether the power grid is in an overload state in a future period according to the power demand load. If so, the future period is marked as an overload operation period, and based on a pre-configured third deep learning model, the temperature influence load in the overload operation period is determined; If not, set T=T+M and trigger the data acquisition module, where M is an integer greater than zero; An equipment screening module, used to obtain the utilization status data of K load-type energy storage devices in the target area, and screen out the best load-type energy storage device based on the utilization status data, where K is an integer greater than zero; The power dispatching module is used to control the optimal load-type energy storage device according to the temperature impact load when in the overload operation period, and supply power to the target area.

Images