Method and device for predicting temperature-sensitive electric quantity and medium

Abstract

The invention relates to a method and a device for predicting temperature-sensitive electric power, and a medium. The method comprises the following steps: collecting daily electric power, daily average temperature and analysis delay days of a power system for multiple consecutive days; performing trend decomposition on the daily electric power quantity to obtain temperature-sensitive electric power quantity; setting a critical temperature according to the daily average temperature; dividing training data based on the temperature-sensitive electric power quantity and the daily average temperature of the corresponding day according to the critical temperature; fuzzy reasoning is combined, quantitative characterization parameters are introduced by analyzing the number of hysteresis days and the daily average temperature, and an autoregression kinetic model is adopted to describe the relation between the temperature and the temperature-sensitive electric power quantity; training data is used for regression of the autoregressive dynamic model, the critical temperature is updated in a parameter search mode, the threshold temperature is calculated through an index inversion method, and the autoregressive dynamic model with fitted parameters is obtained and used for predicting the future thermo-sensitive electric quantity. Compared with the prior art, the method has the advantages of high accuracy, small model calculation amount and the like.

Description

Technical Field

[0001] The present invention relates to the field of energy technology, and in particular to a method, device and medium for predicting temperature-sensitive electric power. Background Art

[0002] The existence of the accumulated temperature effect in the power system is widely recognized. The most intuitive manifestation is that continued high temperatures will lead to a further increase in power consumption, while when the temperature drops, the decrease in power consumption will have a certain hysteresis, that is, the power consumption will not drop rapidly, but will slowly fall to a normal level. This effect may be caused by many factors, including the lag in the response of various power consumption groups to temperature changes, especially the sluggishness of people's use of temperature control equipment. In the power system, the accumulated temperature effect is particularly important for the accuracy of prediction of temperature-sensitive power consumption in winter and summer.

[0003] The primary goal of exploring the accumulated temperature effect is to confirm the quantitative relationship between air temperature data and temperature-sensitive power data, so as to build a prediction model. Power data has sparse discrete characteristics and high randomness. It has both endogenous dynamic properties and is significantly affected by the external ambient temperature. Therefore, it is difficult to use traditional polynomial regression to make accurate predictions at the daily granularity data level. At the same time, because the temperature-sensitive power peak is often significantly higher than other data in extremely high temperature weather, simple regression models are generally unable to handle the fitting and prediction of such a small number of outliers; therefore, existing technical solutions are difficult to accurately predict the temperature-sensitive power peak, and accurate prediction of power peak is crucial to the operation and dispatch of power systems. Summary of the invention

[0004] The purpose of the present invention is to provide a method, device and medium for predicting temperature-sensitive electric power in order to overcome the defect of low accuracy in the above-mentioned prior art.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] A method for predicting temperature-sensitive electrical power comprises the following steps:

[0007] S1: Collect the daily power consumption, daily average temperature and analysis delay days of the power system for multiple consecutive days;

[0008] S2: Decompose the trend of daily power and define the non-trend component as temperature-sensitive power;

[0009] S3: Set critical temperature based on daily average temperature;

[0010] S4: dividing the temperature-sensitive power and the daily average temperature of the corresponding day according to the critical temperature to obtain training data;

[0011] S5: Combined with fuzzy reasoning, quantitative characterization parameters of the accumulated temperature effect are introduced by analyzing the number of hysteresis days and daily average temperature;

[0012] S6: Combined with quantitative characterization parameters, an autoregressive dynamic model is designed to describe the relationship between temperature and temperature-sensitive power;

[0013] S7: Use the training data to regress the autoregressive kinetic model, and update the critical temperature through parameter search. Use the index inversion method to calculate the threshold temperature of the accumulated temperature effect, and obtain the autoregressive kinetic model with good fitting parameters;

[0014] S8: Use the autoregressive dynamic model with fitted parameters, input the prediction time and the temperature data within the corresponding time, and predict the future temperature-sensitive power consumption.

[0015] Further, step S5 includes the following steps:

[0016] S51: Set the accumulated temperature covariance threshold T * is a parameter to be determined;

[0017] S52: Calculate the accumulated temperature covariance factor s on day i (i>k) i , the calculation expression is:

[0018]

[0019] Where, T i-k represents the average temperature of the ikth day, k is the number of days of analysis hysteresis, and the temperature weights are b0, b1…, b k satisfy

[0020] S53: Calculate the intensity of the accumulated temperature effect σ on the i-th day (i>k) i , the calculation expression is:

[0021]

[0022] In the formula, σ i is the intensity of accumulated temperature effect on the i-th day.

[0023] Furthermore, the critical temperature includes a high temperature critical temperature and a low temperature critical temperature, and the training data includes high temperature data above the high temperature critical temperature and low temperature data below the low temperature critical temperature. The high temperature data and the low temperature data are used to regress the autoregressive kinetic model respectively, and the high temperature critical temperature and the low temperature critical temperature are updated through parameter search.

[0024] Further, step S3 includes the following steps:

[0025] S31: sort the daily average temperatures to obtain the minimum temperature and the maximum temperature;

[0026] S32: Divide the temperature between the lowest temperature and the highest temperature into m temperature intervals, so that each interval contains a certain amount of data;

[0027] S33: Calculate the average temperature data T in each temperature interval i And the correlation coefficient r between temperature and temperature-sensitive power i , where i = 1,…,m;

[0028] S34: The |r corresponding to each interval i |, i = 1, ..., m, make a line graph in order, and use the elbow method to find the temperature range where the absolute value of the correlation coefficient begins to decrease and increase;

[0029] S35: The average temperature data corresponding to the interval where the absolute value of the correlation coefficient begins to decrease significantly is taken as T low Initial value, the average temperature data corresponding to the interval where the absolute value of the correlation coefficient begins to rise significantly is taken as T high Initial value.

[0030] Furthermore, the expression of the autoregressive dynamics model is:

[0031]

[0032] Where, L i is the actual value of the temperature-sensitive power on the i-th day, is the predicted value of temperature-sensitive power consumption on the i+1th day, is the rate of change of the temperature-sensitive power with temperature on the i-th day, a1 is the rate of change of the load relative temperature when the accumulated temperature effect fully appears, σ i is the intensity of the accumulated temperature effect on the i-th day, and a2 is the rate of change of the load relative temperature when the accumulated temperature effect does not appear at all.

[0033] Furthermore, in step S7, calculating the threshold temperature of the accumulated temperature effect includes the following steps:

[0034] S71: Establish the autoregressive dynamic model from temperature to temperature-sensitive power data in high temperature data and low temperature data respectively, and obtain a1, a2, b0, b1…, b k And the accumulated temperature covariance threshold T * The optimal parameters are obtained, and the R-square of the high temperature data and the low temperature data is calculated respectively;

[0035] S72: Adjust parameter T high and T low , re-execute S6 after each adjustment, by high and T lowThe initial value neighborhood traversal search maximizes the R square of high temperature data and low temperature data fitting, and fits T high and T low Optimal parameters;

[0036] S73: At this time, parameter T * That is the threshold temperature at which the accumulated temperature effect occurs.

[0037] Furthermore, the autoregressive dynamics model uses the sigmoid activation function to model the magnitude of the accumulated temperature effect.

[0038] Furthermore, the autoregressive kinetic model calculates the threshold temperature of the accumulated temperature effect by calculating the fitting R-square of the training data.

[0039] A second aspect of the present invention is a device for predicting temperature-sensitive electric power, comprising a memory, a processor, and a program stored in the memory, wherein when the processor executes the program, any of the above methods for predicting temperature-sensitive electric power is implemented.

[0040] A third aspect of the present invention is a storage medium having a program stored thereon, which, when executed, implements any of the above methods for predicting temperature-sensitive electric power.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] 1) The present invention calculates the quantitative characterization of the accumulated temperature effect through historical data, and conducts quantitative analysis on the influence of environmental temperature factors on the temperature-sensitive electric power in the power system. It can accurately describe the influence of the accumulated temperature effect on the temperature-sensitive electric power at different temperatures, and optimize the threshold value through the autoregressive dynamic model to make the prediction result more accurate.

[0043] 2) The present invention adopts an autoregressive dynamic model for prediction and uses the activation function Sigmoid to model the intensity of the accumulated temperature effect, which can reduce the difficulty of peak prediction, reduce errors to a greater extent, increase the fitting R-square of the extreme temperature data, and better meet the high-precision requirements of actual temperature-sensitive power peak prediction scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a structural schematic diagram of the present invention. DETAILED DESCRIPTION

[0045] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0046] Example 1

[0047] The present invention is a method, device and medium for predicting temperature-sensitive electric power, the method comprising the following steps:

[0048] S1: Collect the daily power consumption, daily average temperature and analysis delay days of the power system for multiple consecutive days;

[0049] S2: Decompose the trend of daily power and define the non-trend component as temperature-sensitive power;

[0050] S3: Set critical temperature:

[0051] Specifically, the high temperature critical temperature T high and low temperature critical temperature T low ;

[0052] S31: sort the daily average temperatures to obtain the minimum temperature and the maximum temperature;

[0053] S32: Divide the temperature between the lowest temperature and the highest temperature into m temperature intervals, so that each interval contains a certain amount of data;

[0054] S33: Calculate the average temperature data T in each temperature interval i And the correlation coefficient r between temperature and temperature-sensitive power i , where i = 1,…,m;

[0055] S34: The |r corresponding to each interval i |, i = 1, ..., m, make a line graph in order, and use the elbow method to find the temperature range where the absolute value of the correlation coefficient begins to significantly decrease and increase;

[0056] S35: The average temperature data corresponding to the interval where the absolute value of the correlation coefficient begins to decrease significantly is taken as T low Initial value, the average temperature data corresponding to the interval where the absolute value of the correlation coefficient begins to rise significantly is taken as T high Initial value.

[0057] S4: dividing the temperature-sensitive power and the daily average temperature of the corresponding day according to the critical temperature to obtain training data;

[0058] Based on the high temperature critical temperature and the low temperature critical temperature, high temperature data with a daily average temperature higher than the high temperature critical temperature and low temperature data with a daily average temperature lower than the low temperature critical temperature are obtained;

[0059] S5: The quantitative characterization of the accumulated temperature effect is introduced in combination with fuzzy reasoning; the quantitative characterization of the accumulated temperature effect includes the accumulated temperature covariance threshold, the accumulated temperature covariance factor and the accumulated temperature effect intensity. The specific calculation process is as follows:

[0060] S51: Set the accumulated temperature covariance threshold T* is a parameter to be determined;

[0061] S52: Calculate the accumulated temperature covariance factor s on day i (i>k) i , the calculation expression is:

[0062]

[0063] Among them, T i-k represents the temperature data of the ikth day, b0, b1…, b k and

[0064] S53: Calculate the intensity of the accumulated temperature effect σ on the i-th day (i>k) i , the calculation expression is:

[0065]

[0066] S6: Combined with the accumulated temperature effect, an autoregressive dynamic model is designed to describe the relationship between temperature and temperature-sensitive power;

[0067] The expression of the autoregressive dynamics model is:

[0068]

[0069] Where, L i is the actual value of the temperature-sensitive power on the i-th day, is the predicted value of temperature-sensitive power consumption on the i+1th day, is the rate of change of the temperature-sensitive power with temperature on the i-th day.

[0070] S7: Use the training data to regress the autoregressive kinetic model, and update the critical temperature through parameter search. Use the index inversion method to calculate the threshold temperature of the accumulated temperature effect, and obtain the autoregressive kinetic model with good fitting parameters;

[0071] Specifically, the autoregressive kinetic model is regressed using high temperature data and low temperature data respectively, and the high temperature critical temperature and the low temperature critical temperature are updated by parameter search. The threshold temperature of the accumulated temperature effect is calculated by the index inversion method, and the autoregressive kinetic model with good fitting parameters is obtained.

[0072] S71: Establish the autoregressive dynamic model from temperature to temperature-sensitive power data in high temperature data and low temperature data respectively, and obtain a1, a2, b0, b1…, b k And the accumulated temperature covariance threshold T * The optimal parameters are obtained, and the R-square of the high temperature data and the low temperature data is calculated respectively;

[0073] S72: Adjust parameter T high and Tlow , re-execute S6 after each adjustment, by high and T low The initial value neighborhood traversal search maximizes the R square of high temperature data and low temperature data fitting, and fits T high and T low Optimal parameters;

[0074] S73: At this time, parameter T * That is the threshold temperature at which the accumulated temperature effect occurs.

[0075] S8: Use the autoregressive dynamic model with fitted parameters, input the prediction time and the temperature data within the corresponding time, and predict the future temperature-sensitive power consumption.

[0076] The temperature data within the corresponding time used in the prediction process may be derived from the prediction data of the weather forecast, and the specific method is not limited in the present invention.

[0077] The autoregressive dynamic model uses the Sigmoid activation function to model the intensity of the accumulated temperature effect, which is smoother and more in line with the actual scene. It is more accurate than simply making a simple "0-1" binary judgment on whether there is an accumulated temperature effect on a certain day. At the same time, the autoregressive dynamic model makes full use of past data to reduce the difficulty of peak prediction, reduce errors to a greater extent, and increase the fitting R-square of the extreme temperature data, which better meets the high-precision requirements of actual temperature-sensitive power peak prediction scenarios.

[0078] The program code for implementing the method of the present invention can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0079] In the context of the present invention, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0080] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A method for predicting temperature-sensitive electric power, characterized in that: The following steps are involved: S1: Collect the daily power consumption, daily average temperature and analysis delay days of the power system for multiple consecutive days; S2: Decompose the trend of daily power and define the non-trend component as temperature-sensitive power; S3: Set critical temperature based on daily average temperature; S4: dividing the temperature-sensitive power and the daily average temperature of the corresponding day according to the critical temperature to obtain training data; S5: Combined with fuzzy reasoning, quantitative characterization parameters of the accumulated temperature effect are introduced by analyzing the number of hysteresis days and daily average temperature; S6: Combined with quantitative characterization parameters, an autoregressive dynamic model is designed to describe the relationship between temperature and temperature-sensitive power; S7: Use the training data to regress the autoregressive kinetic model, and update the critical temperature through parameter search. Use the index inversion method to calculate the threshold temperature of the accumulated temperature effect, and obtain the autoregressive kinetic model with good fitting parameters; S8: Use the autoregressive dynamic model with fitted parameters, input the prediction time and the temperature data within the corresponding time, and predict the future temperature-sensitive power consumption.

2. A method for predicting temperature-sensitive electric power according to claim 1, characterized in that: Step S5 includes the following steps: S51: Set the accumulated temperature covariance threshold T * is a parameter to be determined; S52: Calculate the accumulated temperature covariance factor s on day i (i>k) i , the calculation expression is: Where, T i-k represents the average temperature of the ikth day, k is the number of days of analysis hysteresis, and the temperature weights are b0, b1…, b k satisfy S53: Calculate the intensity of the accumulated temperature effect σ on the i-th day (i>k) i , the calculation expression is: In the formula, σ i is the intensity of accumulated temperature effect on the i-th day.

3. A method for predicting temperature-sensitive electric power according to claim 1, characterized in that: The critical temperature includes a high temperature critical temperature and a low temperature critical temperature. The training data includes high temperature data higher than the high temperature critical temperature and low temperature data lower than the low temperature critical temperature. The high temperature data and the low temperature data are used to regress the autoregressive kinetic model respectively, and the high temperature critical temperature and the low temperature critical temperature are updated through parameter search.

4. A method for predicting temperature-sensitive electric power according to claim 3, characterized in that: Step S3 includes the following steps: S31: sort the daily average temperatures to obtain the minimum temperature and the maximum temperature; S32: Divide the temperature between the lowest temperature and the highest temperature into m temperature intervals, so that each interval contains a certain amount of data; S33: Calculate the average temperature data T in each temperature interval i And the correlation coefficient r between temperature and temperature-sensitive power i , where i = 1,…,m; S34: The |r corresponding to each interval i |, i = 1, ..., m, make a line graph in order, and use the elbow method to find the temperature range where the absolute value of the correlation coefficient begins to decrease and increase; S35: The average temperature data corresponding to the interval where the absolute value of the correlation coefficient begins to decrease significantly is taken as T low Initial value, the average temperature data corresponding to the interval where the absolute value of the correlation coefficient begins to rise significantly is taken as R high Initial value.

5. A method for predicting temperature-sensitive electric power according to claim 4, characterized in that: The expression of the autoregressive dynamics model is: Where, L i is the actual value of the temperature-sensitive power on the i-th day, is the predicted value of temperature-sensitive power consumption on the i+1th day, is the rate of change of the temperature-sensitive power with temperature on the i-th day, a1 is the rate of change of the load relative temperature when the accumulated temperature effect fully appears, σ i is the intensity of the accumulated temperature effect on the i-th day, and a2 is the rate of change of the load relative temperature when the accumulated temperature effect does not appear at all.

6. A method for predicting temperature-sensitive electric power according to claim 5, characterized in that: In step S7, the calculation of the threshold temperature of the accumulated temperature effect includes the following steps: S71: Establish the autoregressive dynamic model from temperature to temperature-sensitive power data in high temperature data and low temperature data respectively, and obtain a1, a2, b0, b1…, b k And the accumulated temperature covariance threshold T * The optimal parameters are obtained, and the R-square of the high temperature data and the low temperature data is calculated respectively; S72: Adjust parameter T high and T low , re-execute S6 after each adjustment, by high and T low The initial value neighborhood traversal search maximizes the R square of high temperature data and low temperature data fitting, and fits T high and T low Optimal parameters; S73: At this time, parameter T * That is the threshold temperature at which the accumulated temperature effect occurs.

7. A method for predicting temperature-sensitive electric power according to claim 1, characterized in that: The autoregressive dynamics model uses a sigmoid activation function to model the magnitude of the accumulated temperature effect.

8. A method for predicting temperature-sensitive electric power according to claim 1, characterized in that: The autoregressive kinetic model calculates the threshold temperature of the accumulated temperature effect by calculating the fitting R-square of the training data.

9. A device for predicting temperature-sensitive electric power, comprising a memory, a processor, and a program stored in the memory, characterized in that: When the processor executes the program, a method for predicting temperature-sensitive electric power as described in any one of claims 1-8 is implemented.

10. A storage medium having a program stored thereon, characterized in that: When the program is executed, a method for predicting temperature-sensitive electric power as described in any one of claims 1-8 is implemented.

Images