Method and system for predicting electricity consumption of user under high photovoltaic permeability

Abstract

The invention provides a user electricity consumption prediction method and system under high photovoltaic permeability, and belongs to the technical field of power grids, and the method comprises the steps: carrying out the online recognition of the identities of all power users, dividing the power users into newly installed photovoltaic users and other users in the current month, and obtaining the electricity consumption data of the power users; determining the distributed photovoltaic installation day of the newly-installed photovoltaic users according to the electric quantity data of the newly-installed photovoltaic users in the current month by using the difference of the electric quantity sequences before and after the distributed photovoltaic installation, and calculating the predicted value of the actual electric quantity of the newly-installed photovoltaic users in the current month and the predicted value of the distributed photovoltaic generating capacity in the next month; calculating the net electricity consumption predicted values of the rest users in the next month; and calculating net electricity consumption predicted values of all users. According to the daily power consumption before and after the distributed photovoltaic is installed and the power consumption curve sequence waveform in the daytime, the newly installed photovoltaic user in the current month is accurately identified, and the prediction accuracy of the power consumption of the user under the condition of high photovoltaic permeability is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of power grids, and in particular relates to a method and system for predicting user power consumption under high photovoltaic penetration. Background Art

[0002] In the context of energy conservation, emission reduction, low carbon and environmental protection, people all over the world are making full use of residential roof resources, laying distributed photovoltaic power generation system equipment, and vigorously developing photovoltaic building integration technology. Distributed photovoltaic power generation is to lay photovoltaic panels on the roof of your own home, and then generate current through photovoltaic equipment such as inverters and connect to the public power grid, so that electricity can be generated and used by yourself, and the surplus can be connected to the Internet. After installing a distributed photovoltaic system, you can use the electricity generated by the system during the day, and sell the excess electricity to the power grid company. It doesn't matter if the power generation cannot meet the load of the home appliances at night or on rainy days, you can get electricity from the power grid.

[0003] The extensive use of distributed photovoltaic systems has made it difficult for power grid companies to accurately predict user electricity consumption. The Chinese invention patent with application publication number CN113780660A discloses a method, system and storage medium for predicting residential electricity consumption. The prediction method includes obtaining monthly historical electricity consumption and photovoltaic power generation historical data of a preset area; establishing a time series-based electricity consumption prediction model and photovoltaic power generation prediction model based on the monthly historical electricity consumption and photovoltaic power generation historical data; predicting the monthly electricity consumption and photovoltaic power generation of residents in the preset area based on the electricity consumption prediction model and photovoltaic power generation prediction model; and calculating the actual electricity consumption of residents in the preset area based on the predicted monthly electricity consumption and photovoltaic power generation of residents. At present, the ARIMA model is usually used to predict residential electricity consumption. For example, a Chinese invention patent with application publication number CN112488427A discloses a method for predicting the available power of photovoltaic power generation in the medium and long term based on the ARIMA model, which includes the following steps: obtaining the equivalent installed capacity of the reporting period; obtaining the effective light resource hours; the effective light resource hours are obtained by using a statistical method through historical data to obtain a higher prediction value, and are obtained through a prediction method based on the ARIMA model; in the process of predicting the medium and long term available power of photovoltaic power generation, the ARIMA model only models the light resource hours, first extracts the medium and long term trend and fixed period information through detrending and differential operations, converts the non-stationary original time series into a stationary series, and then performs time series ARMA analysis; predicts the medium and long term available power of photovoltaic power generation.

[0004] The ARIMA model can make accurate predictions when there are few photovoltaic users, but with the increase in the penetration rate of distributed photovoltaics, the accuracy of the ARIMA model is becoming increasingly unsatisfactory, especially when there are a large number of newly installed photovoltaic users per month, which brings unprecedented challenges to the monthly electricity consumption forecasting of residential users. How to identify newly installed photovoltaic users and accurately calculate the photovoltaic power generation of the current month to improve the forecast accuracy of electricity consumption for the next month when the penetration rate of distributed photovoltaics is constantly changing is a key issue. Summary of the invention

[0005] In order to solve the technical problem in the prior art that it is difficult to accurately identify users who have newly installed photovoltaic power and calculate their photovoltaic power generation for the current month, the present invention provides a method and system for predicting user electricity consumption under a high photovoltaic penetration rate. The method accurately identifies the newly installed photovoltaic users this month based on the difference in electricity consumption sequences before and after the distributed photovoltaic installation. The actual electricity consumption this month and the distributed photovoltaic power generation this month are calculated based on the electricity consumption data of the newly installed photovoltaic users this month. The actual electricity consumption forecast value of next month and the distributed photovoltaic power generation forecast value of next month are then calculated respectively. Compared with directly predicting the net electricity consumption of next month by using the electricity consumption data of the newly installed photovoltaic users this month, the method improves the accuracy of predicting user electricity consumption under a high photovoltaic penetration rate.

[0006] The technical solution adopted by the present invention is: a method for predicting user electricity consumption under high photovoltaic penetration rate, comprising the following steps:

[0007] Online identification of all power users, dividing the power users into new photovoltaic users and other users this month, and obtaining power consumption data of all the power users, including hourly, daily and monthly power consumption;

[0008] By utilizing the difference in electricity consumption sequence before and after the installation of distributed photovoltaics, the distributed photovoltaic installation date of the newly installed photovoltaic users this month is determined according to the electricity consumption data of the newly installed photovoltaic users this month. Based on the distributed photovoltaic installation date and the electricity consumption data of the newly installed photovoltaic users this month, the actual electricity consumption of this month and the distributed photovoltaic power generation of this month are calculated; the actual electricity consumption forecast value of next month and the distributed photovoltaic power generation forecast value of next month are calculated respectively according to the actual electricity consumption of this month and the distributed photovoltaic power generation of this month;

[0009] Calculate the predicted net electricity consumption value of the users for next month based on the electricity consumption data of the remaining users;

[0010] The predicted value of net electricity consumption of all users = the predicted value of net electricity consumption of users next month + the predicted value of actual electricity consumption next month - the predicted value of distributed photovoltaic power generation next month.

[0011] Furthermore, the process of determining the distributed photovoltaic installation day according to the electricity consumption data of the newly installed photovoltaic users this month by using the difference in electricity consumption sequence before and after the distributed photovoltaic installation is as follows:

[0012] Calculate the hourly predicted electricity consumption for the next day based on the daily hourly electricity consumption of the newly installed photovoltaic users this month, and sum up the hourly predicted electricity consumption for the next day to obtain the daily predicted electricity consumption for the next day;

[0013] If the difference between the daily predicted power consumption of the next day and the daily power consumption of the next day is greater than the average error ε av , then compare the changing trends of the hourly electricity consumption sequence waveforms during the daytime period of the next day with that of the previous day. If the changing trends are different, the next day is determined to be a distributed photovoltaic installation day;

[0014] The newly installed photovoltaic users this month whose distributed photovoltaic installation date cannot be determined will be changed to the remaining users.

[0015] Furthermore, the daytime period is from 11:00 to 16:00, and the change trend of the hourly power consumption sequence waveform is different, showing that the change trend changes from an up and down fluctuation trend to a downward trend.

[0016] Furthermore, the average error ε av The calculation formula is:

[0017]

[0018] The hourly electricity consumption data of non-PV users is predicted, and the error between the prediction result and the actual electricity consumption data collected from the smart meter is ε k , m is the number of predictions.

[0019] Furthermore, the process of calculating the distributed photovoltaic power generation this month based on the electricity consumption data of the newly installed photovoltaic users this month is as follows:

[0020] The daily power generation on the distributed photovoltaic installation day is obtained by subtracting the daily power consumption on the distributed photovoltaic installation day from the daily predicted power consumption on the distributed photovoltaic installation day;

[0021] The daily power generation is used as input data and the data is normalized:

[0022]

[0023] Where p1 is the processed input data; p o is the input data; p min 、p max are the maximum and minimum values ​​in the data set, respectively, and ε is the proportion of photovoltaic users;

[0024] Will Bring in Find the optimal solution among them;

[0025] Where ω2 is the weight vector of users who install photovoltaics, b is a constant, ξ i is the slack variable, ξ i ≥0; c is the penalty factor, v i is the power generation of the month, IMF is Gaussian white noise;

[0026] Select the kernel function and find the estimation function, then bring the input data into the estimation function to get the distributed photovoltaic power generation this month.

[0027] Furthermore, based on the distributed photovoltaic power generation this month, the peak hour estimation method is used to calculate the forecast value of distributed photovoltaic power generation next month. The process is as follows:

[0028] Assume that the number of peak hours of power generation this month is Z m , t is the number of days of photovoltaic installation this month, then the daily peak hours are:

[0029] Z=Z m / t

[0030] Where Z is the number of peak hours per day,

[0031] Correspondingly, the daily peak load period of users is t1, t2, …, t Z , the corresponding load is recorded as: The forecast value of distributed photovoltaic power generation next month is:

[0032]

[0033] Where λ is the number of hours of sunshine, N sum It represents the effective statistical load days, and K represents the forecasted day.

[0034] Furthermore, the actual power consumption of this month is calculated based on the power consumption data of the newly installed photovoltaic users this month, and then the actual power consumption forecast value of next month is calculated based on the actual power consumption of this month:

[0035] According to the daily electricity consumption before the distributed photovoltaic installation day, the net daily electricity consumption of the rest of the month is calculated using the monthly electricity consumption forecasting model (ARIMA) based on the time series method, and the daily electricity consumption before the distributed photovoltaic installation day and the net daily electricity consumption of the rest of the month are summed to obtain the net electricity consumption of this month; according to the net electricity consumption of this month, the monthly electricity consumption forecasting model based on the time series method is used to calculate the predicted value of actual electricity consumption for next month.

[0036] Furthermore, a monthly electricity consumption prediction model based on a time series method is used to calculate the predicted value of the net electricity consumption of the users next month according to the monthly electricity consumption of the remaining users this month.

[0037] The technical solution adopted by the present invention is: a prediction system for user power consumption under high photovoltaic penetration rate, comprising:

[0038] A data collection unit is used to obtain the identities of all power users and the power consumption data of all the power users; the identities of the power users are divided into new photovoltaic users installed this month and other users; the power consumption data includes hourly power consumption, daily power consumption and monthly power consumption;

[0039] A data storage unit, used to store the distributed photovoltaic installation date of a newly installed photovoltaic user this month, the daily predicted power consumption of the distributed photovoltaic installation date, the identity of the power user and the power consumption data of the power user;

[0040] A distributed photovoltaic installation day determination unit is used to determine the distributed photovoltaic installation day of the newly installed photovoltaic user this month according to the electricity data of the newly installed photovoltaic user this month by utilizing the difference in electricity consumption sequence before and after the distributed photovoltaic installation, calculate the daily predicted electricity consumption of the distributed photovoltaic installation day, and change the newly installed photovoltaic user this month whose distributed photovoltaic installation day cannot be determined to the remaining users;

[0041] The unit for calculating the electricity consumption of the new photovoltaic users installed this month for the next month is used to calculate the actual electricity consumption of this month and the distributed photovoltaic power generation of this month based on the distributed photovoltaic installation date and the electricity consumption data of the new photovoltaic users installed this month; and to calculate the predicted value of the actual electricity consumption of the next month and the predicted value of the distributed photovoltaic power generation of the next month respectively based on the actual electricity consumption of this month and the distributed photovoltaic power generation of this month;

[0042] A unit for calculating the power consumption of other users next month, used to calculate the predicted value of the net power consumption of the users next month according to the power consumption data of the other users;

[0043] A user total power consumption calculation unit is used to calculate the predicted value of net power consumption of all users based on the predicted value of net power consumption of users next month, the predicted value of actual power consumption next month and the predicted value of distributed photovoltaic power generation next month;

[0044] The data acquisition unit is connected to the data storage unit, and the data storage unit is respectively connected to the distributed photovoltaic installation day determination unit, the next month's electricity consumption calculation unit for newly installed photovoltaic users this month, and the next month's electricity consumption calculation unit for remaining users. The user total electricity consumption calculation unit is respectively connected to the next month's electricity consumption calculation unit for newly installed photovoltaic users this month and the next month's electricity consumption calculation unit for remaining users.

[0045] Furthermore, the distributed photovoltaic installation day determination unit calculates the hourly predicted power consumption for the next day based on the daily hourly power consumption of the newly installed photovoltaic users this month, and the hourly predicted power consumption for the next day is summed up to obtain the daily predicted power consumption for the next day;

[0046] If the difference between the daily predicted power consumption of the next day and the daily power consumption of the next day is higher than the average error εav, the change trend of the hourly power consumption sequence waveform during the daytime period of the next day and the previous day is compared. If the change trend is different, the next day is determined to be a distributed photovoltaic installation day.

[0047] The daytime period is from 11:00 to 16:00, and the change trend of the hourly power consumption sequence waveform is different, showing that the change trend changes from an up and down trend to a downward trend.

[0048] The average error ε av The calculation formula is:

[0049]

[0050] The hourly electricity consumption data of non-PV users is predicted, and the error between the prediction result and the actual electricity consumption data collected from the smart meter is ε k , m is the number of predictions;

[0051] The newly installed photovoltaic users this month for whom the distributed photovoltaic installation date cannot be determined will be changed to the remaining users.

[0052] Furthermore, the process of calculating the next month's electricity consumption of the newly installed photovoltaic users this month by the electricity consumption calculation unit for the newly installed photovoltaic users this month to calculate the distributed photovoltaic power generation this month is as follows:

[0053] The daily power generation on the distributed photovoltaic installation day is obtained by subtracting the daily power consumption on the distributed photovoltaic installation day from the daily predicted power consumption on the distributed photovoltaic installation day;

[0054] The daily power generation is used as input data and the data is normalized:

[0055]

[0056] Where p1 is the processed input data; p o is the input data; p min 、p max are the maximum and minimum values ​​in the data set, respectively, and ε is the proportion of photovoltaic users;

[0057] Will Bring in Find the optimal solution among them;

[0058] Where ω2 is the weight vector of users who install photovoltaics, b is a constant, ξ i is the slack variable, ξ i ≥0; c is the penalty factor, v i is the power generation of the month, IMF is Gaussian white noise;

[0059] Select the kernel function and find the estimation function, then bring the input data into the estimation function to get the distributed photovoltaic power generation for this month;

[0060] The process of calculating the power consumption of the new photovoltaic users installed this month for the next month according to the distributed photovoltaic power generation of this month and using the peak hour estimation method to calculate the predicted value of the distributed photovoltaic power generation for the next month is as follows:

[0061] Assume that the number of peak hours of power generation this month is Z m , t is the number of days of photovoltaic installation this month, then the daily peak hours are:

[0062] Z=Z m / t

[0063] Where Z is the number of peak hours per day,

[0064] Correspondingly, the daily peak load period of users is t1, t2, ..., t Z , the corresponding load is recorded as: The forecast value of distributed photovoltaic power generation next month is:

[0065]

[0066] Where λ is the number of hours of sunshine, N sum represents the number of days with effective load statistics, and K represents the forecasted day;

[0067] The process of calculating the power consumption of the new photovoltaic users installed this month for the next month based on the power consumption data of the new photovoltaic users installed this month, and then calculating the predicted value of the actual power consumption for the next month based on the actual power consumption of this month is as follows:

[0068] According to the daily electricity consumption before the distributed photovoltaic installation day, the net daily electricity consumption of the rest of the month is calculated using the monthly electricity consumption prediction model based on the time series method, and the daily electricity consumption before the distributed photovoltaic installation day and the net daily electricity consumption of the rest of the month are summed to obtain the net electricity consumption of this month; according to the net electricity consumption of this month, the monthly electricity consumption prediction model based on the time series method is used to calculate the predicted value of actual electricity consumption for next month.

[0069] Furthermore, the next month electricity consumption calculation unit for the remaining users utilizes a monthly electricity consumption prediction model based on a time series method to calculate a predicted value of the net electricity consumption of the users next month according to the monthly electricity consumption of the remaining users this month.

[0070] The technical solution adopted by the present invention is also: a computer-readable storage medium, including a program, and the program can be executed by a processor to implement the method for predicting user electricity consumption under high photovoltaic penetration.

[0071] Compared with the prior art, the present invention has the following beneficial effects:

[0072] 1. The present invention uses the difference in electricity consumption sequence before and after the installation of distributed photovoltaics to accurately identify newly installed photovoltaic users, decouples the electricity consumption data of newly installed photovoltaic users, and separately predicts the actual electricity consumption of users and photovoltaic power generation. Since the growth rates and bases of actual electricity consumption and photovoltaic power generation are different, net electricity consumption rises or falls under their joint action. Compared with the method of directly predicting net electricity consumption, indirect prediction can improve the prediction accuracy when there are a large number of newly installed distributed photovoltaics during the continuous change of photovoltaic penetration.

[0073] 2. The present invention combines SVM, ARIMA, and prohet algorithms to accurately calculate the distributed photovoltaic power generation of newly installed photovoltaic users this month, and then uses the peak hour estimation method to calculate the distributed photovoltaic power generation of next month, while the actual power consumption of newly installed photovoltaic users next month is predicted by ARIMA. Different methods are used to predict the actual power consumption and distributed photovoltaic power generation of next month, which improves the prediction accuracy of user power consumption under high photovoltaic penetration.

[0074] 3. The present invention uses the daily power consumption before and after distributed photovoltaics and the power consumption curve sequence waveform during the daytime as the judgment standard to accurately determine the distributed photovoltaic installation day of the newly installed photovoltaic users, provide a guarantee for the decoupling of the power consumption data of the newly installed photovoltaic users, and also identify the users who have registered but not actually used the distributed photovoltaics. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 is a flow chart of a method according to an embodiment of the present invention;

[0076] Figure 2 It is a curve chart of actual power consumption sequence of photovoltaic users according to an embodiment of the present invention;

[0077] Figure 3 A net electricity consumption sequence curve diagram of a photovoltaic user according to an embodiment of the present invention;

[0078] Figure 4 4 is a structural block diagram of a system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0079] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0080] The embodiment of the present invention provides a method for predicting user power consumption under high photovoltaic penetration rate, such as Figure 1 As shown, it includes the following steps:

[0081] Step 1: Identify the identities of all power users online, divide the power users into new photovoltaic users installed this month and other users, and obtain the power consumption data of all the power users.

[0082] Before distributed photovoltaic power is connected to the grid, electricity users need to register with the relevant departments of the power grid. Through the power grid user database, it can be queried whether the electricity user is a photovoltaic user or a non-photovoltaic user; for photovoltaic users, it can also be queried whether they are old photovoltaic users or newly installed photovoltaic users this month.

[0083] The electricity consumption data includes hourly electricity consumption, daily electricity consumption and monthly electricity consumption. Other users need to obtain their monthly electricity consumption this month. New photovoltaic users this month need to obtain their hourly electricity consumption and daily electricity consumption.

[0084] Step 2: Utilize the difference in electricity consumption sequence before and after the distributed photovoltaic installation, and determine the distributed photovoltaic installation date of the newly installed photovoltaic user this month according to the electricity consumption data of the newly installed photovoltaic user this month; based on the distributed photovoltaic installation date and the electricity consumption data of the newly installed photovoltaic user this month, calculate the actual electricity consumption this month and the distributed photovoltaic power generation this month; calculate the actual electricity consumption forecast value for next month and the distributed photovoltaic power generation forecast value for next month according to the actual electricity consumption this month and the distributed photovoltaic power generation this month.

[0085] Step 2.1: Using the difference in electricity consumption sequence before and after the distributed photovoltaic installation, the process of determining the distributed photovoltaic installation day according to the electricity consumption data of the newly installed photovoltaic users this month is as follows:

[0086] The monthly electricity consumption forecasting model (ARIMA) based on the time series method is used to calculate the hourly forecast electricity consumption for the next day according to the daily hourly electricity consumption of the newly installed photovoltaic users this month. The daily forecast electricity consumption for the next day is calculated. If the difference between the daily forecast electricity consumption for the next day and the daily electricity consumption for the next day is higher than the average error εav, the change trend of the hourly electricity consumption sequence waveform during the daytime period of the next day and the previous day is compared. If the change trends are different, it is determined that the next day is a distributed photovoltaic installation day.

[0087] The newly installed photovoltaic users this month whose distributed photovoltaic installation date cannot be determined will be changed to the remaining users. For some newly installed photovoltaic users this month, although they have filed with the relevant departments of the power grid and are recorded as photovoltaic users in the power grid user database, their distributed photovoltaic equipment is not actually used, so they should actually be the remaining users. Therefore, the remaining users actually include three parts, namely old photovoltaic users, non-photovoltaic users and newly installed photovoltaic users this month whose distributed photovoltaic equipment is not actually used.

[0088] The daytime period is from 11 to 16 o'clock, and the change trend of the hourly electricity consumption sequence waveform is different, showing that the change trend changes from an up and down trend to a downward trend. During the daytime, the hourly electricity consumption sequence waveform of non-photovoltaic users shows an up and down trend. After the installation of distributed photovoltaics, the output of distributed photovoltaics affects the hourly electricity consumption sequence waveform to a downward trend. During the day, the random fluctuations of the actual electricity consumption and net electricity consumption sequences are different. Based on this, the random fluctuations of the electricity consumption sequences of two consecutive days are compared. The random fluctuations of the actual electricity consumption sequence are obtained by removing the trend in the actual consumption sequence, such as Figure 2 and 3 As shown in Figure 1. Comparison of the fluctuations of a user's electricity consumption sequence before and after the installation of distributed photovoltaics. Although the actual electricity consumption sequence curve and the net electricity consumption sequence curve are very similar due to the small output of distributed photovoltaics, by observing the fluctuations of the two curves, it is found that the sequence waveform has changed during the period of 11-16 o'clock after the installation of distributed photovoltaics. Without the installation of distributed photovoltaics, the user's net electricity consumption curve is a curve with ups and downs, which indicates that the user's electricity consumption is unstable. When the user installs distributed photovoltaics, even if the output of distributed photovoltaics is small, its net electricity consumption curve is on a downward trend.

[0089] The formula for calculating the average error is:

[0090]

[0091] The hourly electricity consumption data of non-PV users is predicted, and the error between the prediction result and the actual electricity consumption data collected from the smart meter is ε k , m is the number of predictions.

[0092] Step 2.2: The process of calculating the distributed photovoltaic power generation this month based on the electricity consumption data of the newly installed photovoltaic users this month is as follows:

[0093] (1) The daily power generation on the distributed photovoltaic installation day is obtained by subtracting the daily power consumption on the distributed photovoltaic installation day from the daily predicted power consumption on the distributed photovoltaic installation day.

[0094] (2) Calculate the distributed photovoltaic power generation for this month based on the daily power generation using the SAP algorithm.

[0095] The SAP algorithm is a combination of SVM, ARIMA, and prohet algorithms. Assume that the data set u(u1, u2, u3, ..., u n ) is the n-dimensional daily power generation, is the input quantity; v(v1, v2, v3, ..., v n ) is the n-dimensional distributed photovoltaic power generation this month, and is the output, which is brought into the SAP algorithm. The training data set is {(u1,v1),(u2,v2),(u3,v3),...,(u n ,v n )}, the training data set is mapped nonlinearly Mapped into n-dimensional space, the following estimation function is constructed Among them: ω2 is the weight vector of users who install photovoltaics, and b is a constant. Find the optimal hyperplane, construct the minimization function, and transform it into solving the optimization problem:

[0096]

[0097] Where: i is the slack variable, ξ i ≥0; c is the penalty factor.

[0098] IMF is Gaussian white noise, adding variables r1(t) is the residual, X(t) is the variable at a certain moment

[0099] Then combine the above formula with the variables r1(t) and Decompose and combine the Lagrange dual problem to obtain the second modal component:

[0100]

[0101] In the formula, α, η, α i ,η i is the Lagrangian function.

[0102] The residual function is the following formula:

[0103]

[0104] ε1(n i (t)) is the residual signal, which satisfies the termination condition of decomposition. The final formula of decomposition is:

[0105]

[0106] Finally, the original signal can be decomposed into: n is the nth decomposition.

[0107] Combining the two equations, we can get the function of t:

[0108]

[0109] Using kernel function The final estimation function expression is:

[0110]

[0111] The process of calculating the distributed photovoltaic power generation of this month based on the daily power generation of the distributed photovoltaic installation day is as follows:

[0112] Take the daily power generation as input data and normalize the data:

[0113]

[0114] Where p1 is the processed input data; p o is the input data; p min 、p max are the maximum and minimum values ​​in the data set respectively, and ε is the proportion of photovoltaic users.

[0115] Will Bring in Find the optimal solution among them;

[0116] Where ω2 is the weight vector of users who install photovoltaics, b is a constant, ξ i is the slack variable, ξ i ≥0; c is the penalty factor, v i is the power generation of the month, IMF is Gaussian white noise;

[0117] Select the kernel function and find the estimation function, then bring the input data into the estimation function to get the distributed photovoltaic power generation this month.

[0118] Step 2.3: Based on the distributed photovoltaic power generation of this month, the peak hour estimation method is used to calculate the forecast value of distributed photovoltaic power generation for next month. The process is as follows:

[0119] Based on the distributed photovoltaic power generation this month and the peak power generation hours during the month when the photovoltaic power generation is installed, the forecast value of distributed photovoltaic power generation for next month is calculated.

[0120] Assume that the number of peak hours of power generation this month is Z m , t is the number of days of photovoltaic installation this month, then the daily peak hours are:

[0121] Z=Z m / t

[0122] Where Z is the number of peak hours per day,

[0123] Correspondingly, the daily peak load period of users is t1, t2, ..., tZ , the corresponding load is recorded as: The forecast value of distributed photovoltaic power generation next month is:

[0124]

[0125] Where λ is the number of hours of sunshine, N sum It represents the effective statistical load days, and K represents the forecasted day.

[0126] Step 2.4: The process of calculating the actual power consumption of this month based on the power consumption data of the newly installed photovoltaic users this month, and then calculating the predicted value of the actual power consumption of next month based on the actual power consumption of this month is as follows:

[0127] According to the daily electricity consumption before the distributed photovoltaic installation day, ARIMA is used to calculate the net daily electricity consumption of the rest of the month. The daily electricity consumption before the distributed photovoltaic installation day and the net daily electricity consumption of the rest of the month are summed to obtain the net electricity consumption of this month; based on the net electricity consumption of this month, ARIMA is used to calculate the predicted value of actual electricity consumption for next month.

[0128] Step 3: Calculate the predicted value of the net electricity consumption of the users in the next month based on the electricity consumption data of the remaining users.

[0129] ARIMA is used to calculate the predicted value of the net electricity consumption of the users next month based on the monthly electricity consumption of the remaining users this month. The monthly electricity consumption prediction model based on the time series method is based on the online time series method, which uses differential operation to normalize the electricity consumption, lists the objective function, and brings the monthly electricity consumption of this month into the objective function to obtain the predicted value of the net electricity consumption of the users next month. The specific calculation process is:

[0130] (1) Get the monthly electricity consumption of this month,

[0131] (2) Perform a stationary test on the monthly electricity consumption of this month: If it is a non-stationary sequence, use the difference operation to convert the monthly electricity consumption of this month into a stationary sequence, and then perform a stationary test again; if it is a stationary sequence, fit the difference stationary model and list the objective function;

[0132] (3) Substitute this month’s electricity consumption into the objective function to obtain the predicted value of the user’s net electricity consumption for next month.

[0133] For non-stationary series, the difference operation can make a type of non-stationary series (i.e., series with trends) stationary. If the first-order difference cannot make the time series stationary, the second-order difference, the third-order difference, and the d-th-order difference can be performed, and finally the series can be turned into a stationary series.

[0134] The essence of differential operation is to extract deterministic information by autoregression:

[0135]

[0136] ▽ d x t is the d-order variance, (1-B) = x t -x t-1 , where x t is the value that needs to be predicted at the current moment.

[0137] Choice of differential method:

[0138] (1) The non-stationary series contains a significant linear trend, and the first-order difference can achieve trend stability;

[0139] (2) Non-stationary sequences contain curvilinear trends, and low-order (second or third order) differences can usually extract the influence of curvilinear trends;

[0140] (3) For non-stationary sequences with a fixed period, performing a difference operation with a step length equal to the period length can usually better extract the period information.

[0141] The difference stationary model is used for fitting difference stationary series. The difference stationary model refers to a model in which the highest autocorrelation order after d-order difference is p and the highest moving average order is q. It usually contains p+q independent unknown coefficients.

[0142] The prediction principle of the differential stabilization model is the minimum mean square error prediction principle, and the objective function formula is:

[0143]

[0144] Where: x t is the value to be predicted at the current moment, is the regression coefficient calculated using the regression algorithm, p is the P time points before the current time point selected for calculation, θ1...,θ q is the autoregressive coefficient, ε t is the current random interference, which is a zero-mean white noise sequence.

[0145] The solution for the regression coefficient is: using the formula for the sum of squared errors, Using partial derivatives to find the minimum value of J(a,b), we get and They represent the average value of the predicted net electricity consumption of users next month and the average value of the monthly electricity consumption this month respectively.

[0146] Step 4: Calculate the predicted net electricity consumption of all users based on the predicted net electricity consumption of users next month, the predicted actual electricity consumption of next month and the predicted distributed photovoltaic power generation of next month.

[0147] The predicted value of net electricity consumption of all users = the predicted value of net electricity consumption of users next month + the predicted value of actual electricity consumption next month - the predicted value of distributed photovoltaic power generation next month.

[0148] In order to verify the effectiveness of the method of the present invention, two indicators are used for evaluation:

[0149] (1) Root Mean Squared Error (RMSE). Root Mean Squared Error can be used as a basis for estimating the size of the forecast error in advance. The difference between Root Mean Squared Error and Mean Absolute Error is mainly manifested in the following aspects: First, because each error is squared during the calculation of Root Mean Squared Error, the influence factor of large errors in the indicator is enhanced, and the sensitivity of this indicator is improved. Second, Root Mean Squared Error can provide research error distribution and calculate forecast confidence intervals when making further predictions. Its calculation formula is:

[0150]

[0151] y pre is the predicted value, y ac is the actual value.

[0152] (2) Accuracy of identifying photovoltaic users α update :

[0153]

[0154] Where n cor is the number of correctly identified PV users, n toa is the total number of users identified.

[0155] The prediction errors of the direct prediction method A based on ARIMA and the method B proposed in the present invention in three scenarios are shown in Table 1: a small number of DPV (distributed photovoltaics) in the table represent that less than 20% of users install distributed photovoltaics, some DPV represent that 20%-50% of users install distributed photovoltaics, and most DPV represent that more than 50% of users install distributed photovoltaics.

[0156] Table 1 Prediction error comparison table

[0157]

[0158] In the case of a small number of photovoltaic users, direct prediction is better than the method proposed in the present invention, but the photovoltaic penetration rate is generally proportional to the number of newly installed photovoltaic users. As the photovoltaic penetration rate increases, the number of newly added photovoltaic users will also increase rapidly. In the case of high penetration rates, the root mean square error and mean absolute percentage error of the method of the present invention are both reduced compared to the direct prediction method. The present invention can accurately and effectively predict the photovoltaic electricity consumption of users in the case of high penetration rates. The method proposed in the present invention has the advantages of detecting distributed photovoltaic installations and updating the distributed photovoltaic customer database, making the indirect monthly net electricity consumption forecast better than the direct forecast.

[0159] The embodiment of the present invention also provides a system for predicting user power consumption under high photovoltaic penetration rate, such as Figure 4 Shown, including

[0160] The data collection unit 1 is used to obtain the identities of all power users and the power consumption data of all the power users; the identities of the power users are divided into the users who newly installed photovoltaic power this month and the remaining users; the power consumption data includes hourly power consumption, daily power consumption and monthly power consumption;

[0161] Data storage unit 2, used to store the distributed photovoltaic installation date of the newly installed photovoltaic user this month, the daily predicted power consumption of the distributed photovoltaic installation date, the identity of the power user and the power consumption data of the power user;

[0162] The distributed photovoltaic installation day determination unit 3 is used to determine the distributed photovoltaic installation day of the newly installed photovoltaic user this month according to the electricity data of the newly installed photovoltaic user this month by using the difference in electricity consumption sequence before and after the distributed photovoltaic installation, calculate the daily predicted electricity consumption of the distributed photovoltaic installation day, and change the newly installed photovoltaic user this month whose distributed photovoltaic installation day cannot be determined to the remaining users;

[0163] The electricity consumption calculation unit 4 for the new photovoltaic users installed this month is used to calculate the actual electricity consumption of this month and the distributed photovoltaic power generation of this month based on the distributed photovoltaic installation date and the electricity consumption data of the new photovoltaic users installed this month; and calculate the predicted value of the actual electricity consumption of next month and the predicted value of the distributed photovoltaic power generation of next month respectively based on the actual electricity consumption of this month and the distributed photovoltaic power generation of this month;

[0164] The power consumption calculation unit 5 for the remaining users in the next month is used to calculate the predicted value of the net power consumption of the users in the next month according to the power consumption data of the remaining users;

[0165] The user total power consumption calculation unit 6 is used to calculate the net power consumption prediction value of all users according to the user net power consumption prediction value for next month, the actual power consumption prediction value for next month and the distributed photovoltaic power generation prediction value for next month;

[0166] The data acquisition unit 1 is connected to the data storage unit 2, and the data storage unit 2 is respectively connected to the distributed photovoltaic installation day determination unit 3, the next month's electricity consumption calculation unit 4 for newly installed photovoltaic users this month, and the next month's electricity consumption calculation unit 5 for other users, and the user total electricity consumption calculation unit 6 is respectively connected to the next month's electricity consumption calculation unit 4 for newly installed photovoltaic users this month, and the next month's electricity consumption calculation unit 5 for other users.

[0167] When the prediction system is working, the data acquisition unit 1 obtains the identities and power consumption data of all power users from the power grid user database, and then stores them in the data storage unit 2. The identities of power users are divided into new photovoltaic users and other users this month; the power consumption data includes hourly power consumption, daily power consumption and monthly power consumption.

[0168] The distributed photovoltaic installation day determination unit 3 reads the electricity consumption data of the newly installed photovoltaic users this month from the data storage unit 2; determines the distributed photovoltaic installation day of the newly installed photovoltaic users this month according to the electricity consumption data of the newly installed photovoltaic users this month by using the difference in electricity consumption sequence before and after the distributed photovoltaic installation, and calculates the daily predicted electricity consumption of the distributed photovoltaic installation day; changes the newly installed photovoltaic users this month whose distributed photovoltaic installation day cannot be determined to the remaining users. The above results are stored in the data storage unit 2.

[0169] The specific calculation process is: according to the daily hourly electricity consumption of the newly installed photovoltaic users this month, the hourly predicted electricity consumption of the next day is calculated, and the hourly predicted electricity consumption of the next day is summed up to obtain the daily predicted electricity consumption of the next day;

[0170] If the difference between the daily predicted power consumption of the next day and the daily power consumption of the next day is greater than the average error ε av , then compare the changing trend of the hourly electricity consumption sequence waveform during the daytime period of the next day with that of the previous day. If the changing trend is different, it is determined that the next day is the distributed photovoltaic installation day.

[0171] The daytime period is from 11:00 to 16:00, and the change trend of the hourly power consumption sequence waveform is different, showing that the change trend changes from an up and down trend to a downward trend.

[0172] The average error ε av The calculation formula is:

[0173]

[0174] The hourly electricity consumption data of non-PV users is predicted, and the error between the prediction result and the actual electricity consumption data collected from the smart meter is ε k , m is the number of predictions;

[0175] The newly installed photovoltaic users this month for whom the distributed photovoltaic installation date cannot be determined will be changed to the remaining users.

[0176] The next month's electricity consumption calculation unit 4 for the new photovoltaic users installed this month and the next month's electricity consumption calculation unit 5 for the remaining users read data from the data storage unit 2 respectively.

[0177] The electricity consumption calculation unit 4 for the new photovoltaic users installed this month for the next month reads the electricity consumption data of the new photovoltaic users installed this month, the distributed photovoltaic installation day and the daily predicted electricity consumption of the distributed photovoltaic installation day from the data storage unit 2. The actual electricity consumption of each new photovoltaic user installed this month and the distributed photovoltaic power generation of this month are calculated; and then the actual electricity consumption forecast value of the next month and the distributed photovoltaic power generation forecast value of the next month are calculated.

[0178] The specific calculation process is:

[0179] According to the electricity consumption data of the newly installed photovoltaic users this month, the process of calculating the distributed photovoltaic power generation this month is:

[0180] The daily power generation on the distributed photovoltaic installation day is obtained by subtracting the daily power consumption on the distributed photovoltaic installation day from the daily predicted power consumption on the distributed photovoltaic installation day;

[0181] The daily power generation is used as input data and the data is normalized:

[0182]

[0183] Where p1 is the processed input data; p o is the input data; p min 、p max are the maximum and minimum values ​​in the data set, respectively, and ε is the proportion of photovoltaic users;

[0184] Will Bring in Find the optimal solution among them;

[0185] Where ω2 is the weight vector of users who install photovoltaics, b is a constant, ξ i is the slack variable, ξ i ≥0; c is the penalty factor, v i is the power generation of the month, IMF is Gaussian white noise;

[0186] Select the kernel function and find the estimation function, then bring the input data into the estimation function to get the distributed photovoltaic power generation this month.

[0187] According to the distributed photovoltaic power generation of this month, the process of calculating the forecast value of distributed photovoltaic power generation for next month using the peak hour estimation method is as follows:

[0188] Assume that the number of peak hours of power generation this month is Z m , t is the number of days of photovoltaic installation this month, then the daily peak hours are:

[0189] Z=Z m / t

[0190] Where Z is the number of peak hours per day,

[0191] Correspondingly, the daily peak load period of users is t1, t2, …, t Z , the corresponding load is recorded as: The forecast value of distributed photovoltaic power generation next month is:

[0192]

[0193] Where λ is the number of hours of sunshine, N sum It represents the effective statistical load days, and K represents the forecasted day.

[0194] The process of calculating the actual power consumption of this month based on the power consumption data of the newly installed photovoltaic users this month, and then calculating the predicted value of the actual power consumption of next month based on the actual power consumption of this month is as follows:

[0195] According to the daily electricity consumption before the distributed photovoltaic installation day, the net daily electricity consumption of the rest of the month is calculated using the monthly electricity consumption prediction model based on the time series method, and the daily electricity consumption before the distributed photovoltaic installation day and the net daily electricity consumption of the rest of the month are summed to obtain the net electricity consumption of this month; according to the net electricity consumption of this month, the monthly electricity consumption prediction model based on the time series method is used to calculate the predicted value of actual electricity consumption for next month.

[0196] The remaining users' next month's electricity consumption calculation unit 5 reads the remaining users' electricity consumption data from the data storage unit 2, and calculates the predicted value of the net electricity consumption of each remaining user next month. The specific calculation process is: using the monthly electricity consumption prediction model based on the time series method, according to the monthly electricity consumption of the remaining users this month, the predicted value of the net electricity consumption of the user next month is calculated.

[0197] The user total power consumption calculation unit 6 obtains the calculation results of the next month's power consumption calculation unit 4 of the newly installed photovoltaic users this month and the next month's power consumption calculation unit 5 of the remaining users, and calculates the predicted net power consumption of all users.

[0198] The predicted value of net electricity consumption of all users = the predicted value of net electricity consumption of users next month + the predicted value of actual electricity consumption next month - the predicted value of distributed photovoltaic power generation next month.

[0199] An embodiment of the present invention further provides a computer-readable storage medium, comprising a program, wherein the program can be executed by a processor to implement the method for predicting user electricity consumption under high photovoltaic penetration.

[0200] The present invention has been described in detail above through embodiments, but the contents described are only exemplary embodiments of the present invention and cannot be considered to limit the scope of implementation of the present invention. The protection scope of the present invention is defined by the claims. Anyone who utilizes the technical solution described in the present invention, or a technician in the field, inspired by the technical solution of the present invention, designs a similar technical solution within the essence and protection scope of the present invention to achieve the above technical effects, or makes equal changes and improvements to the scope of application, etc., shall still fall within the scope of protection covered by the patent of the present invention.

Claims

1. A method for predicting user electricity consumption under high photovoltaic penetration, characterized in that: The following steps are involved: Online identification of all power users, dividing the power users into new photovoltaic users and other users this month, and obtaining power consumption data of all the power users, including hourly, daily and monthly power consumption; By utilizing the difference in electricity consumption sequence before and after the installation of distributed photovoltaics, the distributed photovoltaic installation date of the newly installed photovoltaic users this month is determined according to the electricity consumption data of the newly installed photovoltaic users this month; based on the distributed photovoltaic installation date, the actual electricity consumption of this month and the distributed photovoltaic power generation of this month are calculated according to the electricity consumption data of the newly installed photovoltaic users this month; the actual electricity consumption forecast value of next month and the distributed photovoltaic power generation forecast value of next month are calculated according to the actual electricity consumption of this month and the distributed photovoltaic power generation of this month; Calculate the predicted net electricity consumption value of the users for next month based on the electricity consumption data of the remaining users; The predicted value of net electricity consumption of all users = the predicted value of net electricity consumption of users next month + the predicted value of actual electricity consumption next month - the predicted value of distributed photovoltaic power generation next month.

2. A method for predicting user electricity consumption under high photovoltaic penetration as claimed in claim 1, characterized in that: The process of determining the distributed photovoltaic installation day based on the electricity consumption data of the newly installed photovoltaic users this month by using the difference in electricity consumption sequence before and after the distributed photovoltaic installation is as follows: Calculate the hourly predicted electricity consumption for the next day based on the daily hourly electricity consumption of the newly installed photovoltaic users this month, and sum up the hourly predicted electricity consumption for the next day to obtain the daily predicted electricity consumption for the next day; If the difference between the daily predicted power consumption of the next day and the daily power consumption of the next day is greater than the average error ε av , then compare the changing trends of the hourly electricity consumption sequence waveforms during the daytime period of the next day with that of the previous day. If the changing trends are different, the next day is determined to be a distributed photovoltaic installation day; The newly installed photovoltaic users this month whose distributed photovoltaic installation date cannot be determined will be changed to the remaining users.

3. A method for predicting user electricity consumption under high photovoltaic penetration rate as claimed in claim 2, characterized in that: The daytime period is from 11:00 to 16:00, and the change trend of the hourly power consumption sequence waveform is different, showing that the change trend changes from an up and down fluctuation trend to a downward trend.

4. A method for predicting user electricity consumption under high photovoltaic penetration as claimed in claim 2, characterized in that: The average error ε av The calculation formula is: The hourly electricity consumption data of non-PV users is predicted, and the error between the prediction result and the actual electricity consumption data collected from the smart meter is ε k , m is the number of predictions.

5. A method for predicting user electricity consumption under high photovoltaic penetration rate according to any one of claims 1 to 4, characterized in that: The process of calculating the distributed photovoltaic power generation this month based on the electricity consumption data of the newly installed photovoltaic users this month is as follows: The daily power generation on the distributed photovoltaic installation day is obtained by subtracting the daily power consumption on the distributed photovoltaic installation day from the daily predicted power consumption on the distributed photovoltaic installation day; The daily power generation is used as input data and the data is normalized: Where p1 is the processed input data; p o is the input data; p min 、p max are the maximum and minimum values ​​in the data set, respectively, and ε is the proportion of photovoltaic users; Will Bring in Find the optimal solution among them; Where ω2 is the weight vector of users who install photovoltaics, b is a constant, ξ i is the slack variable, ξ i ≥0; c is the penalty factor, v i is the power generation of the month, IMF is Gaussian white noise; Select the kernel function and find the estimation function, then bring the input data into the estimation function to get the distributed photovoltaic power generation this month.

6. A method for predicting user electricity consumption under high photovoltaic penetration rate according to any one of claims 1 to 4, characterized in that: According to the distributed photovoltaic power generation this month, the peak hour estimation method is used to calculate the forecast value of distributed photovoltaic power generation next month. The process is as follows: Assume that the number of peak hours of power generation this month is Z m , t is the number of days of photovoltaic installation this month, then the daily peak hours are: Z=Z m / t Where Z is the number of peak hours per day, Correspondingly, the user's daily peak load period is t1, t2,,t Z , the corresponding load is recorded as: The forecast value of distributed photovoltaic power generation next month is: Where λ is the number of hours of sunshine, N sum It represents the effective statistical load days, and K represents the forecasted day.

7. A method for predicting user electricity consumption under high photovoltaic penetration rate according to any one of claims 1 to 4, characterized in that: The process of calculating the actual power consumption of this month based on the power consumption data of the newly installed photovoltaic users this month, and then calculating the predicted value of the actual power consumption of next month based on the actual power consumption of this month is as follows: According to the daily electricity consumption before the distributed photovoltaic installation day, the net daily electricity consumption of the rest of the month is calculated using the monthly electricity consumption prediction model based on the time series method, and the daily electricity consumption before the distributed photovoltaic installation day and the net daily electricity consumption of the rest of the month are summed to obtain the net electricity consumption of this month; according to the net electricity consumption of this month, the monthly electricity consumption prediction model based on the time series method is used to calculate the predicted value of actual electricity consumption for next month.

8. The method for predicting user electricity consumption under high photovoltaic penetration rate according to claim 1, characterized in that: The monthly electricity consumption prediction model based on the time series method is used to calculate the predicted value of the net electricity consumption of the users next month according to the monthly electricity consumption of the remaining users this month.

9. A prediction system for user electricity consumption under high photovoltaic penetration, characterized in that: include A data collection unit is used to obtain the identities of all power users and the power consumption data of all the power users; the identities of the power users are divided into new photovoltaic users installed this month and other users; the power consumption data includes hourly power consumption, daily power consumption and monthly power consumption; A data storage unit, used to store the distributed photovoltaic installation date of the newly installed photovoltaic user this month, the daily predicted power consumption of the distributed photovoltaic installation date, the identity of the power user and the power consumption data of the power user; A distributed photovoltaic installation day determination unit is used to determine the distributed photovoltaic installation day of the newly installed photovoltaic user this month according to the electricity data of the newly installed photovoltaic user this month by utilizing the difference in electricity consumption sequence before and after the distributed photovoltaic installation, calculate the daily predicted electricity consumption of the distributed photovoltaic installation day, and change the newly installed photovoltaic user this month whose distributed photovoltaic installation day cannot be determined to the remaining users; The unit for calculating the electricity consumption of the new photovoltaic users installed this month for the next month is used to calculate the actual electricity consumption of this month and the distributed photovoltaic power generation of this month based on the distributed photovoltaic installation date and the electricity consumption data of the new photovoltaic users installed this month; and to calculate the predicted value of the actual electricity consumption of next month and the predicted value of the distributed photovoltaic power generation of next month respectively based on the actual electricity consumption of this month and the distributed photovoltaic power generation of this month; A unit for calculating the power consumption of other users next month, used to calculate the predicted value of the net power consumption of the users next month according to the power consumption data of the other users; A user total power consumption calculation unit is used to calculate the predicted value of net power consumption of all users based on the predicted value of net power consumption of users next month, the predicted value of actual power consumption next month and the predicted value of distributed photovoltaic power generation next month; The data acquisition unit is connected to the data storage unit, and the data storage unit is respectively connected to the distributed photovoltaic installation day determination unit, the next month's electricity consumption calculation unit for newly installed photovoltaic users this month, and the next month's electricity consumption calculation unit for remaining users. The user total electricity consumption calculation unit is respectively connected to the next month's electricity consumption calculation unit for newly installed photovoltaic users this month and the next month's electricity consumption calculation unit for remaining users.

10. A user power consumption prediction system under high photovoltaic penetration rate as claimed in claim 9, characterized in that: The distributed photovoltaic installation day determination unit calculates the hourly predicted power consumption for the next day based on the daily hourly power consumption of the newly installed photovoltaic users this month, and calculates the daily predicted power consumption for the next day by summing up the hourly predicted power consumption for the next day; If the difference between the daily predicted power consumption of the next day and the daily power consumption of the next day is greater than the average error ε av , then compare the changing trend of the hourly electricity consumption sequence waveform during the daytime period of the next day with that of the previous day. If the changing trend is different, it is determined that the next day is the distributed photovoltaic installation day. The daytime period is from 11:00 to 16:00, and the change trend of the hourly power consumption sequence waveform is different, showing that the change trend changes from an up and down trend to a downward trend. The average error ε av The calculation formula is: The hourly electricity consumption data of non-PV users is predicted, and the error between the prediction result and the actual electricity consumption data collected from the smart meter is ε k , m is the number of predictions; The newly installed photovoltaic users this month whose distributed photovoltaic installation date cannot be determined will be changed to the remaining users.

11. A user power consumption prediction system under high photovoltaic penetration rate as claimed in claim 10, characterized in that: The process of calculating the next month's electricity consumption of the newly installed photovoltaic users this month by the electricity consumption calculation unit is as follows: The daily power generation on the distributed photovoltaic installation day is obtained by subtracting the daily power consumption on the distributed photovoltaic installation day from the daily predicted power consumption on the distributed photovoltaic installation day; The daily power generation is used as input data and the data is normalized: Where p1 is the processed input data; p o is the input data; p min 、p max are the maximum and minimum values ​​in the data set, respectively, and ε is the proportion of photovoltaic users; Will Bring in Find the optimal solution among them; Where ω2 is the weight vector of users who install photovoltaics, b is a constant, ξ i is the slack variable, ξ i ≥0; c is the penalty factor, v i is the power generation of the month, IMF is Gaussian white noise; Select the kernel function and find the estimation function, then bring the input data into the estimation function to get the distributed photovoltaic power generation for this month; The process of calculating the power consumption of the new photovoltaic users installed this month for the next month according to the distributed photovoltaic power generation of this month and using the peak hour estimation method to calculate the predicted value of the distributed photovoltaic power generation for the next month is as follows: Assume that the number of peak hours of power generation this month is Z m , t is the number of days of photovoltaic installation this month, then the daily peak hours are: Z=Z m / t Where Z is the number of peak hours per day, Correspondingly, the user's daily peak load period is t1, t2,,t Z , the corresponding load is recorded as: The forecast value of distributed photovoltaic power generation next month is: Where λ is the number of hours of sunshine, N sum represents the number of days with effective load statistics, and K represents the forecasted day; The process of calculating the power consumption of the new photovoltaic users installed this month for the next month based on the power consumption data of the new photovoltaic users installed this month, and then calculating the predicted value of the actual power consumption for the next month based on the actual power consumption of this month is as follows: According to the daily electricity consumption before the distributed photovoltaic installation day, the net daily electricity consumption of the rest of the month is calculated using the monthly electricity consumption prediction model based on the time series method, and the daily electricity consumption before the distributed photovoltaic installation day and the net daily electricity consumption of the rest of the month are summed to obtain the net electricity consumption of this month; according to the net electricity consumption of this month, the monthly electricity consumption prediction model based on the time series method is used to calculate the predicted value of actual electricity consumption for next month.

12. A user power consumption prediction system under high photovoltaic penetration rate as claimed in claim 10, characterized in that: The next month's electricity consumption calculation unit for the remaining users utilizes a monthly electricity consumption prediction model based on a time series method to calculate a predicted value of the net electricity consumption of the users next month according to the monthly electricity consumption of the remaining users this month.

13. A computer-readable storage medium, characterized in that: The method comprises a program which can be executed by a processor to implement a method for predicting user electricity consumption under high photovoltaic penetration rate as described in any one of claims 1 to 8.

Images