Single-household probability net load prediction method, device, equipment and storage medium

Abstract

This invention discloses a method, apparatus, equipment, and storage medium for predicting the probabilistic net load of a single household. The method includes the following steps: (1) clustering the historical net load curves of a single household based on Gaussian mixture model clustering to extract the daily electricity consumption pattern features; (2) clustering the net load data for each time period in the historical net load data of a single household based on Gaussian mixture model clustering to extract the electricity consumption scenario features for each time period; (3) establishing a multi-class logistic regression model based on the obtained daily electricity consumption pattern features and electricity consumption scenario features for each time period to study the probability of occurrence of each electricity consumption scenario in each future time period; (4) establishing a Gaussian process regression model for each scenario to study the probability distribution of net load size under each electricity consumption scenario; and (5) characterizing the final probabilistic net load prediction result of a single household based on Gaussian mixture distribution. The method of this invention is effective and reasonable, providing a reference for accurately predicting the net load of a single household.

Description

Technical Field

[0001] This invention relates to the field of power technology, and in particular to a method, apparatus, equipment and storage medium for predicting probabilistic net load for a single household. Background Technology

[0002] Because the output of distributed renewable energy is highly dependent on weather conditions, its power generation fluctuates greatly and is highly uncertain depending on factors such as irradiance levels and weather types. Simultaneously, the electricity consumption behavior of individual households is also highly unpredictable. Electricity load is closely related to consumer behavior; even under very similar weather and date conditions, the electricity load of a single household may be at completely different levels, exhibiting significant randomness and making accurate load prediction for individual households extremely difficult. For distributed rooftop photovoltaic systems installed after the meter, distribution system operators and retailers can only obtain the net load of users (i.e., user load minus photovoltaic power generation output), as the photovoltaic power generation itself is not visible. This invisibility greatly increases the difficulty of predicting the net load of the after-meter system.

[0003] Therefore, in order to accurately predict the net load of individual users, it is necessary to study a reasonable method for predicting the probabilistic net load of individual users. Summary of the Invention

[0004] Purpose of the invention: This invention addresses the problems existing in the prior art by providing a more reasonable method, apparatus, equipment, and storage medium for predicting the probabilistic net load of a single household.

[0005] To address the aforementioned technical problems, the present invention proposes the following technical solutions:

[0006] A method for predicting net load per household probabilistically, characterized by comprising the following steps:

[0007] Step 1: Based on Gaussian mixture model clustering, cluster the historical net load curves of individual users to extract electricity consumption pattern features;

[0008] Step 2: Based on Gaussian mixture model clustering, cluster the net load data for each time period in the historical net load data of a single user, and extract the electricity consumption scenario features for each time period.

[0009] Step 3: Based on the electricity consumption pattern characteristics and the electricity consumption scenario characteristics of each time period, establish a multi-class logistic regression model to calculate the probability of occurrence of each electricity consumption scenario in each future time period.

[0010] Step 4: Establish Gaussian process regression models for different electricity consumption scenarios and calculate the probability distribution of net load size for each scenario.

[0011] Step 5: Calculate the mixed Gaussian distribution result of the single-household probability net load prediction based on the probability of the electricity consumption scenario and the net load size under the electricity consumption scenario.

[0012] Optionally, step 1 includes the following steps:

[0013] Historical net load data was horizontally normalized.

[0014] A Gaussian mixture model is established to cluster the historical net load data after horizontal normalization. The electricity consumption pattern features are extracted based on the clustering results of the Gaussian mixture model.

[0015] Optionally, step 2 includes the following steps:

[0016] Historical net load data is subjected to longitudinal normalization by time period:

[0017] Gaussian mixture model clustering is performed on the longitudinally normalized historical net load data to obtain a set of net load clustering results for different time periods. The clustering results are used to characterize the electricity consumption scenario features of each time period in the aforementioned segment.

[0018] Optionally, step 3 includes the following steps:

[0019] Establish a multi-class logistic regression model to study the relationship between the characteristics of electricity consumption scenarios and patterns during this period, the load of the previous few periods, the load of the same period of the previous day, and the real-time temperature.

[0020] The multi-class logistic regression model is trained in stages, and the model is shown after training.

[0021] Optionally, step 4 includes the following steps:

[0022] The Gaussian process regression model is trained for each electricity consumption scenario. After training, the independent variables in the test set are input, and the Gaussian process regression model outputs the predicted net load of the electricity consumption scenario in the future, and obtains the probability distribution of the net load size under each electricity consumption scenario.

[0023] Optionally, step 3 specifically includes the following steps:

[0024] Step 31: Establish a multi-class logistic regression model, as shown in equations (7)-(8):

[0025]

[0026] z = a0 + a1x1 + a2x2 + ... + a n x n (8)

[0027] In the formula, e is the natural logarithm, z is the input to the multi-class logistic regression model, and σ zx represents the probability value of belonging to a certain category after inputting z. i α is the independent variable input to the multi-class logistic regression model. i These are the prediction coefficients corresponding to each independent variable, where n is the total number of independent variables;

[0028] Step 32: Train a multi-class logistic regression model to study the relationship between the electricity consumption scenario and daily electricity consumption pattern characteristics during this period, the load of the previous few periods, the load of the same period of the previous day, and the real-time temperature.

[0029] When training a multi-class logistic regression model in different time periods, the input independent variables and input dependent variables of the multi-class logistic regression model are shown in equations (9)-(10):

[0030] X MLR =[m d ,n″ d,t-1 ,n″ d,t-2 ,…,n″ d,t-j ,n″ d-1,t ,T d,t (9)

[0031] Y MLR =[s d,t (10)

[0032] In the formula, X MLR For a multi-class logistic regression model, the set of independent variables, m d For the electricity consumption pattern characteristics on day d, n' d ' ,t The net load value after longitudinal normalization is given for time period t on day d, where j is the maximum number of time periods selected forward, and T is the maximum number of time periods selected forward. d,t Y represents the actual temperature at time t on day d; MLR To input the dependent variable set into a multi-class logistic regression model, s d,t This is the electricity consumption scenario number for the t-th time period on day d;

[0033] Step 33: After the multi-class logistic regression model is trained, input the independent variables in the test set. The multi-class logistic regression model outputs the probability of each electricity consumption scenario occurring in the future during this period. The output result is shown in Equation (11):

[0034]

[0035] In the formula, β s To predict the probability of scenario s occurring in the future during this period, satisfying β s ≥0,

[0036] Optionally, step 4 specifically includes the following steps:

[0037] Step 41, establish a Gaussian process regression model:

[0038] Define a Gaussian process in the space of function f:

[0039] f(x)~GP(m(x),k(x,x')) (12)

[0040] In the formula, GP is a Gaussian process, m(x) is the expectation, and setting m(x) = 0, we can directly deduce the covariance function; k(x,x') is the covariance function, and the covariance functions form the covariance matrix. Where X = [x1, ..., x] n ] T ;

[0041] Given the test input vector x * Training data y and predicted output f * Satisfies a multivariate joint Gaussian distribution:

[0042]

[0043] f * The probability density function is shown in equation (14):

[0044]

[0045] Where N represents a Gaussian distribution, The mean, The expression is shown in equation (15):

[0046]

[0047] cov(f * Let ) be the variance, expressed as in equation (16):

[0048]

[0049] Step 42: Train the Gaussian process regression model for each electricity consumption scenario. The input independent variables and input dependent variables of the Gaussian process regression model are shown in equations (17)-(18):

[0050] X GPR,s =[m d ,n″ d,t-1 ,n″ d,t-2 ,…,n″ d,t-j ,n″ d-1,t ,T d,t ],d∈D s (17)

[0051] Y GPR,s =[n″ d,t ],d∈Ds (18)

[0052] In the formula, X GPR,s Input the set of independent variables D for the Gaussian process regression model under the s-th electricity consumption scenario during this period. s Y represents the set of natural days belonging to the s-th electricity consumption scenario within this time period; GPR,s Input the set of dependent variables for the Gaussian process regression model for the s-th electricity consumption scenario during this period;

[0053] Step 43: After the Gaussian process regression model is trained, input the independent variables in the test set. The Gaussian process regression model outputs the predicted net load of the electricity consumption scenario s in the future period. The output result is shown in Equation (19):

[0054]

[0055] In the formula, E s This represents the predicted net load expectation for electricity consumption scenario s during this period. Let be the predicted net load variance for electricity consumption scenario s during this period.

[0056] Optionally, step 5 specifically includes the following steps:

[0057] Based on the probability prediction results of electricity consumption scenarios and the probability net load prediction results of different electricity consumption scenarios, the final Gaussian distribution result of the probability net load prediction for a single household is calculated using equation (20):

[0058]

[0059] Where, σ z β represents the probability that the input z belongs to a certain classification result; s To predict the probability of scenario s occurring in the future during this period; E s This represents the predicted net load expectation for electricity consumption scenario s during this period.

[0060] A second aspect of the present invention relates to a single-household probabilistic net load forecasting device, characterized in that it comprises:

[0061] The user net load data feature extraction module is used to extract the electricity consumption pattern features and electricity consumption scenario features of historical net load data.

[0062] The electricity consumption scenario occurrence probability prediction module is used to predict the probability of each electricity consumption scenario occurring in each time period in the future; and

[0063] The probabilistic net load prediction module is used to predict the probability distribution of net load size under each electricity consumption scenario.

[0064] A third aspect of the present invention relates to a single-household probabilistic net load forecasting device, characterized in that it comprises:

[0065] One or more processors;

[0066] Memory, used to store one or more programs;

[0067] When the one or more programs are executed by the one or more processors, the one or more processors implement the above-described method for predicting the probabilistic net load of a single household.

[0068] Beneficial effects: Compared with the prior art, the significant advantage of this invention is that it proposes a probabilistic net load prediction method for single households, providing a certain approach to improve the accuracy of net load prediction for single households. Attached Figure Description

[0069] Figure 1 This is a flowchart illustrating a single-household probabilistic net load prediction method provided in Embodiment 1 of the present invention.

[0070] Figure 2 The above are the probability prediction results of a single-household probabilistic net load prediction method under different confidence levels provided in Embodiment 1 of the present invention.

[0071] Figure 3 This is a schematic diagram of a single-household probabilistic net load prediction device provided in Embodiment 2 of the present invention;

[0072] Figure 4 This is a schematic diagram of the structure of a single-household probability net load prediction device provided in Embodiment 3 of the present invention. Detailed Implementation

[0073] This invention provides a method, apparatus, device, and storage medium for predicting the probabilistic net load of a single household. First, based on Gaussian mixture model clustering, the daily net load curves in the training dataset are clustered to extract the daily electricity consumption pattern features. Then, based on Gaussian mixture model clustering, the net load data for each time period in the training dataset is clustered to extract the electricity consumption scenario features for each time period. Next, based on the obtained daily electricity consumption pattern features and time-period electricity consumption scenario features, a multi-class logistic regression model is established to study the probability of occurrence of each electricity consumption scenario in each future time period. Then, a Gaussian process regression model is established for each scenario to study the probability distribution of net load size under each electricity consumption scenario. Finally, the final probabilistic net load prediction result for a single household is represented by a Gaussian mixture distribution.

[0074] Example 1

[0075] The following is a detailed description of the single-household probabilistic net load prediction method provided by the embodiments of the present invention. Figure 1 See the flowchart for the single-household probabilistic net load forecasting method. Figure 1 As shown, the method may include the following steps:

[0076] Step 1: Based on Gaussian mixture model clustering, cluster the daily net load curves in the training dataset to extract the daily electricity consumption pattern features.

[0077] This step specifically includes:

[0078] Step 11: Perform horizontal normalization on a daily basis according to equation (1):

[0079]

[0080] In the formula, n d,t Let n' be the original net load value for time period t on day d, where T is the maximum number of time periods per day. d,t This represents the net load value after horizontal normalization during time period t on day d.

[0081] Step 12: Establish a Gaussian mixture model to cluster the horizontally normalized net load data on a daily basis:

[0082] (1) Start iterating by taking the initial values ​​of the parameters;

[0083] (2) E-Step: Based on the current parameters, calculate the response γ of the k-th sub-model to the sampled data using equation (2). jk :

[0084]

[0085] In the formula, α k α is the proportionality coefficient of the k-th Gaussian distribution function, representing the probability that the data belongs to the k-th Gaussian distribution function, satisfying α. k ≥0, It is a Gaussian distribution function. Let μ represent the k-th sub-model. k Let σ be the mean of the k-th Gaussian distribution function. k Let be the standard deviation of the k-th Gaussian distribution function. Let be the variance of the k-th Gaussian distribution function.

[0086] (3) M-Step: Calculate the model parameters for the new iteration according to equations (3)-(5), and adjust the model parameters to maximize the probability of the model generating these parameters:

[0087]

[0088]

[0089]

[0090] (4) Repeat E-Step and M-Step until convergence.

[0091] Step 13: Perform Gaussian mixture model clustering on the daily net load data to obtain the clustering result set of the net load curve for each day. The clustering results depend on the electricity consumption patterns and dynamics reflected in the net load curve of that day, and are used to characterize the macroscopic features of the net load curve of that day.

[0092] Step 2: Based on Gaussian mixture model clustering, cluster the net load data for each time period in the training dataset and extract the electricity consumption scenario features for each time period.

[0093] This step includes the following steps:

[0094] Step 21: Perform longitudinal normalization on the historical net load data according to the time period based on equation (6):

[0095]

[0096] In the formula, D is the maximum number of days in each time period, and n″ d,t This represents the net load value after longitudinal normalization for the t-th time period on day d.

[0097] Step 22: Perform Gaussian mixture model clustering on the longitudinally normalized net load data to obtain the time-segmented net load clustering result set. The clustering results are used to characterize the electricity consumption scenario to which the net load belongs in each time period.

[0098] Step 3: Based on the obtained daily electricity consumption pattern characteristics and electricity consumption scenario characteristics for each time period, establish a multi-class logistic regression model to study the probability of occurrence of each electricity consumption scenario in each time period in the future.

[0099] Step 3 includes the following steps:

[0100] Step 31: Establish a multi-class logistic regression model, as shown in equations (7)-(8):

[0101]

[0102] z = a0 + a1x1 + a2x2 + ... + a n x n (8)

[0103] In the formula, e is the natural logarithm, z is the input to the multi-class logistic regression model, and σ z x represents the probability value of belonging to a certain category after inputting z. i α is the independent variable input to the multi-class logistic regression model. i Here are the prediction coefficients for each independent variable, and n is the total number of independent variables.

[0104] Step 32: Train a multi-class logistic regression model to study the relationship between electricity consumption scenarios and input features (daily electricity consumption pattern features, load of the previous few periods, load of the same period of the previous day, and real-time temperature) during this period:

[0105] When training a multi-class logistic regression model in different time periods, the input independent variables and input dependent variables of the multi-class logistic regression model are shown in equations (9)-(10):

[0106] X MLR =[m d ,n″ d,t-1 ,n″ d,t-2 ,…,n″ d,t-j ,n″ d-1,t ,T d,t (9)

[0107] Y MLR =[s d,t (10)

[0108] In the formula, X MLR For a multi-class logistic regression model, the set of independent variables, m d For the electricity consumption pattern characteristics on day d, n' d ' ,t The net load value after longitudinal normalization is given for time period t on day d, where j is the maximum number of time periods selected forward, and T is the maximum number of time periods selected forward. d,t Y represents the actual temperature at time t on day d; MLR To input the dependent variable set into a multi-class logistic regression model, s d,t Let be the electricity consumption scenario number for time period t on day d. The set of independent variables mentioned above may include: electricity consumption pattern characteristics, net load value after longitudinal normalization, actual temperature, etc.

[0109] Step 33: After the multi-class logistic regression model is trained, input the independent variables in the test set. The multi-class logistic regression model outputs the probability of each electricity consumption scenario occurring in the future during this period. The output result is shown in Equation (11):

[0110]

[0111] In the formula, β s To predict the probability of scenario s occurring in the future during this period, satisfying β s ≥0,

[0112] Step 4: Establish Gaussian process regression models for different scenarios to study the probability distribution of net load size under each electricity consumption scenario.

[0113] Step 4 includes the following steps:

[0114] Step 41, establish a Gaussian process regression model:

[0115] A Gaussian process is defined on the space of any finite number of random variables with a joint Gaussian distribution:

[0116] f(x)~GP(m(x),k(x,x')) (12)

[0117] In the formula, GP is a Gaussian process, m(x) is the expectation, usually set m(x) = 0, and the covariance function is directly inferred; k(x,x') is the covariance function, and the covariance functions form the covariance matrix. Where X = [x1, ..., x] n ] T .

[0118] Given the test input vector x * Training data y and predicted output f * Satisfies a multivariate joint Gaussian distribution:

[0119]

[0120] f * The probability density function is shown in equation (14):

[0121]

[0122] Where N represents a Gaussian distribution, The mean, The expression is shown in equation (15):

[0123]

[0124] cov(f * Let ) be the variance, expressed as in equation (16):

[0125]

[0126] Step 42: Train the Gaussian process regression model for each electricity consumption scenario. The input independent variables and input dependent variables of the Gaussian process regression model are shown in equations (17)-(18):

[0127] X GPR,s =[m d ,n″ d,t-1 ,n″ d,t-2 ,…,n″ d,t-j ,n″ d-1,t ,T d,t ],d∈D s (17)

[0128] Y GPR,s =[n″ d,t],d∈D s (18)

[0129] In the formula, X GPR,s Input the set of independent variables D for the Gaussian process regression model under the s-th electricity consumption scenario during this period. s Y represents the set of natural days belonging to the s-th electricity consumption scenario within this time period; GPR,s Input the set of dependent variables for the Gaussian process regression model under the s-th electricity consumption scenario in this period.

[0130] Step 43: After the Gaussian process regression model is trained, input the independent variables in the test set. The Gaussian process regression model outputs the predicted net load of the electricity consumption scenario s in the future period. The output result is shown in Equation (19):

[0131]

[0132] In the formula, E s This represents the predicted net load expectation for electricity consumption scenario s during this period. Let be the predicted net load variance for electricity consumption scenario s during this period.

[0133] Step 5: Characterize the final single-household probabilistic net load prediction results based on Gaussian mixture distribution.

[0134] Step 5 includes the following steps:

[0135] Step 51: Based on the probability prediction results of the electricity consumption scenario and the probability net load prediction results of each electricity consumption scenario, the final Gaussian distribution result of the probability net load prediction for a single household is calculated using equation (20):

[0136]

[0137] The dataset uses net load data from a residential user, with 96 sampling points per day. It is divided into training, validation, and test sets in an 8:1:1 ratio. The dataset undergoes both daily horizontal normalization and time-period vertical normalization. Gaussian mixture model clustering is then applied to the normalized data to extract electricity consumption pattern features from the net load data. Each time period is divided into four electricity consumption scenarios, which are then input into a multi-class logistic regression model and a Gaussian process regression model for training, validation, and testing.

[0138] Based on the prediction results of the multi-class logistic regression model and the Gaussian process regression model, the complete prediction results of a residential user's net load on a certain day at different confidence levels are as follows: Figure 2 As shown.

[0139] Example 2

[0140] Figure 3This is a schematic diagram of a single-household probabilistic net load forecasting device provided in an embodiment of the present invention. The device can be implemented using software and / or hardware, and can be configured in a terminal device. The device includes:

[0141] The user net load data feature extraction module is used to extract the daily electricity consumption pattern features and the electricity consumption scenario features for each time period from the historical net load data.

[0142] The electricity consumption scenario occurrence probability prediction module is used to predict the probability of occurrence of each electricity consumption scenario in each time period in the future.

[0143] The probabilistic net load prediction module is used to predict the probability distribution of net load size under each electricity consumption scenario.

[0144] The single-household probabilistic net load forecasting device provided in this embodiment of the invention can be used to perform a single-household probabilistic net load forecasting method provided in Embodiment 2, and has the corresponding functions and beneficial effects of the execution method.

[0145] It is worth noting that in the embodiments of the above-mentioned device, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of the present invention.

[0146] Example 3

[0147] Figure 4 This is a schematic diagram of the device provided in Embodiment 3 of the present invention. The present invention provides services for the implementation of the single-household probabilistic net load prediction method in Embodiment 1, and can be configured with the single-household probabilistic net load prediction device in Embodiment 2 above. Figure 4 A block diagram of an exemplary device 12 suitable for implementing embodiments of the present invention is shown. Figure 4 The device 12 shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0148] like Figure 4 As shown, device 12 is represented as a general-purpose computing device. Components of device 12 may include, but are not limited to: one or more processors or processing units 16, system memory 28, and a bus 18 connecting different system components, including system memory 28 and processing unit 16.

[0149] Bus 18 represents one or more of several bus architectures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of the various bus architectures. For example, these architectures include, but are not limited to, the Industry Standard Architecture (I-Step A) bus, the Micro Channel Architecture (MAC) bus, the Enhanced I-Step A bus, the Video Electronics Standards Association (VE-Step A) local bus, and the Peripheral Component Interconnect (PCI) bus.

[0150] Device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by device 12, including volatile and non-volatile media, removable and non-removable media.

[0151] System memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. Device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 34 may be used to read and write non-removable, non-volatile magnetic media (…). Figure 3 Not shown; usually referred to as a "hard drive"). Although Figure 4 Not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk") and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 18 via one or more data media interfaces. Memory 28 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the embodiments of the present invention.

[0152] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in memory 28. Such program modules 42 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. Program modules 42 typically perform the functions and / or methods described in the embodiments of the present invention.

[0153] Device 12 can also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, etc.), and with one or more devices that enable a user to interact with device 12, and / or with any device that enables device 12 to communicate with one or more other computing devices (e.g., network card, modem, etc.). This communication can be performed via input / output (I / O) interface 22. Furthermore, device 12 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 20. Figure 3 As shown, network adapter 20 communicates with other modules of device 12 via bus 18. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0154] The processing unit 16 executes various functional applications and data processing by running programs stored in the system memory 28, such as implementing the single-household probabilistic net load prediction method provided in Embodiment 1 of the present invention.

[0155] Example 4

[0156] Embodiment 4 of the present invention also provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform the method described in Embodiment 1.

[0157] The computer storage medium of this invention can be any combination of one or more computer-readable media. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0158] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0159] Program code contained on a computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0160] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​such as Java, procedural malltalk, and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0161] Of course, the computer-executable instructions provided in the embodiments of the present invention are not limited to the above-described method operations.

[0162] The above description is merely a preferred embodiment of the present invention and should not be construed as limiting the scope of the invention. Therefore, any equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A single household probabilistic net load forecasting method, characterized by, Includes the following steps: Step 1: Based on Gaussian mixture model clustering, cluster the historical net load data curves of individual users to extract electricity consumption pattern features; Step 2: Based on Gaussian mixture model clustering, cluster the net load data for each time period in the historical net load data of a single user, and extract the electricity consumption scenario features for each time period. Step 3: Based on the electricity consumption pattern characteristics and the electricity consumption scenario characteristics of each time period, establish a multi-class logistic regression model to calculate the probability of occurrence of each electricity consumption scenario in each future time period. Step 4: Establish Gaussian process regression models for different electricity consumption scenarios and calculate the probability distribution of net load size for each scenario. Step 5: Calculate the mixed Gaussian distribution result of the single-household probability net load prediction based on the probability of the electricity consumption scenario and the net load size under the electricity consumption scenario; Step 3 specifically includes the following steps: Step 31, establish a multi-class logistic regression model, as shown in equations (7)-(8): （7） （8） wherein, ln is the natural logarithm, z X is the input of the multinomial logistic regression model, X is the input z P is the probability value of the posterior belonging to a certain classification result, X is the independent variable input of the multinomial logistic regression model, β is the predicted coefficient corresponding to each independent variable, n N is the total number of independent variables; Step 32: Train a multi-class logistic regression model to study the relationship between the electricity consumption scenario and daily electricity consumption pattern characteristics during this period, the load of the previous few periods, the load of the same period of the previous day, and the real-time temperature. When training a multi-class logistic regression model in different time periods, the input independent variables and input dependent variables of the multi-class logistic regression model are shown in equations (9)-(10): （9） （10） In the formula, is a set of independent variables for the multi-classification logistic regression model, is the daily electricity mode feature, d is the actual temperature of the d t is the normalized net load value of the j is the maximum number of time periods selected forward, is the actual temperature of the d t is a set of dependent variables for the multi-classification logistic regression model, is the electricity scene number of the d t ​​​​​​ Step 33: After the multi-class logistic regression model is trained, input the independent variables in the test set. The multi-class logistic regression model outputs the probability of each electricity consumption scenario occurring in the future during this period. The output result is shown in Equation (11): （11） In the formula, To predict the probability of occurrence of the future scene of this period s , meet , ; Step 4 specifically includes the following steps: Step 41, establish a Gaussian process regression model: In function f Spatially define a Gaussian process: （12） wherein GP is a Gaussian process, is the expectation, let make inferences directly on the covariance function; is a covariance function, the covariance function constitutes a covariance matrix wherein ; Given a test input vector Training data y and predicted output Satisfies a multivariate joint Gaussian distribution: （13） The probability density function is shown in equation (14): （14） in, N Indicates a Gaussian distribution. The mean, The expression is shown in equation (15): （15） The variance is expressed as shown in equation (16): （16） Step 42: Train the Gaussian process regression model for each electricity consumption scenario. The input independent variables and input dependent variables of the Gaussian process regression model are shown in equations (17)-(18): （17） （18） In the formula, For this period of time s The set of input independent variables for a Gaussian process regression model under a given electricity consumption scenario For this period of time, belonging to the first s A collection of natural days for each electricity consumption scenario; Input the set of dependent variables for the Gaussian process regression model under the s-th electricity consumption scenario in this time period; Step 43: After the Gaussian process regression model is trained, input the independent variables from the test set, and the Gaussian process regression model will output the future electricity consumption scenario for that period. s The predicted net load is output as shown in equation (19): （19） In the formula, Electricity consumption scenario during this period s The projected net load expectation, Electricity consumption scenario during this period s The predicted net load variance.

2. The method for predicting probabilistic net load for a single household according to claim 1, characterized in that, Step 1 includes the following steps: Historical net load data was horizontally normalized. A Gaussian mixture model is established to cluster the historical net load data after horizontal normalization. The electricity consumption pattern features are extracted based on the clustering results of the Gaussian mixture model.

3. The method for predicting probabilistic net load for a single household according to claim 1, characterized in that, Step 2 includes the following steps: Historical net load data is subjected to longitudinal normalization by time period: Gaussian mixture model clustering is performed on the longitudinally normalized historical net load data to obtain a set of net load clustering results for different time periods. The clustering results are used to characterize the electricity consumption scenario features of each time period in the aforementioned segment.

4. The method for predicting probabilistic net load for a single household according to claim 1, characterized in that, Step 3 includes the following steps: Establish a multi-class logistic regression model to study the relationship between the characteristics of electricity consumption scenarios and patterns during this period, the load of the previous few periods, the load of the same period of the previous day, and the real-time temperature. The multi-class logistic regression model is trained in stages, and the model is shown after training.

5. The method for predicting probabilistic net load for a single household according to claim 1, characterized in that, Step 4 includes the following steps: The Gaussian process regression model is trained for each electricity consumption scenario. After training, the independent variables in the test set are input, and the Gaussian process regression model outputs the predicted net load of the electricity consumption scenario in the future, and obtains the probability distribution of the net load size under each electricity consumption scenario.

6. The method for predicting probabilistic net load for a single household according to claim 1, characterized in that, Step 5 specifically includes the following steps: Based on the probability prediction results of electricity consumption scenarios and the probability net load prediction results of different electricity consumption scenarios, the final Gaussian distribution result of the probability net load prediction for a single household is calculated using equation (20): （20） in, For input z The probability value of belonging to a certain classification result; To predict the probability of scenario s occurring in the future during this period; This represents the predicted net load expectation for electricity consumption scenario s during this period.

7. A single-household probabilistic net load forecasting device, characterized in that, include: The user net load data feature extraction module is used to extract the electricity consumption pattern features and electricity consumption scenario features of historical net load data. The electricity consumption scenario occurrence probability prediction module is used to predict the probability of occurrence of each electricity consumption scenario in each time period in the future. as well as The probabilistic net load prediction module is used to predict the probability distribution of net load size under each electricity consumption scenario; The method for predicting the probability of occurrence of each electricity consumption scenario in each future time period specifically includes the following steps: A multi-class logistic regression model is established, and the formulas are shown in equations (7) and (8): （7） （8） In the formula, It is the natural logarithm. z As input to a multi-class logistic regression model, For input z The probability value of belonging to a certain classification result. The independent variables input to the multi-class logistic regression model, These are the prediction coefficients corresponding to each independent variable. n The total number of independent variables; Training a multi-class logistic regression model to study the relationship between electricity consumption scenarios and daily electricity consumption patterns during this period, load in previous periods, load in the same period of the previous day, and real-time temperature: When training a multi-class logistic regression model in different time periods, the input independent variables and input dependent variables of the multi-class logistic regression model are shown in equations (9)-(10): （9） （10） In the formula, To input the set of independent variables into a multi-class logistic regression model, For the first d Characteristics of daily electricity consumption patterns For the first d Heavenly t Net load value after longitudinal normalization over the time period j To select the maximum number of time periods forward, For the first d Heavenly t The actual temperature during the period; Input the dependent variable set to the multi-category logistic regression model. For the first d Heavenly t Electricity consumption scenario number for a given time period; After the multi-class logistic regression model is trained, the independent variables in the test set are input, and the multi-class logistic regression model outputs the probability of each electricity consumption scenario occurring in the future during this period. The output results are shown in Equation (11): （11） In the formula, To predict future scenarios during this period s The probability of occurrence satisfies , ; The probability distribution for predicting the net load size under each electricity consumption scenario specifically includes the following steps: Establish a Gaussian process regression model: In function f Define a Gaussian process in space: （12） In the formula, GP For Gaussian processes, For the expectation, let This allows for direct inferences about the covariance function; The covariance functions form the covariance matrix. ,in ; Given a test input vector Training data y and predicted output Satisfies a multivariate joint Gaussian distribution: （13） The probability density function is shown in equation (14): （14） in, N Indicates a Gaussian distribution. The mean, The expression is shown in equation (15): （15） The variance is expressed as shown in equation (16): （16） The Gaussian process regression model is trained for different electricity consumption scenarios. The input independent variables and input dependent variables of the Gaussian process regression model are shown in equations (17)-(18): （17） （18） In the formula, For this period of time s The set of input independent variables for a Gaussian process regression model under a given electricity consumption scenario For this period of time, belonging to the first s A collection of natural days for each electricity consumption scenario; Input the set of dependent variables for the Gaussian process regression model under the s-th electricity consumption scenario in this time period; After the Gaussian process regression model is trained, input the independent variables from the test set, and the Gaussian process regression model will output the future electricity consumption scenario for that period. s The predicted net load is output as shown in equation (19): （19） In the formula, Electricity consumption scenario during this period s The projected net load expectation, Electricity consumption scenario during this period s The predicted net load variance.

8. A single-household probabilistic net load forecasting device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the single-household probabilistic net load forecasting method according to any one of claims 1-6.

Images