Bidirectional gating network short-term load prediction method and device based on Copula orthogonal analysis

Abstract

The invention relates to the technical field of load prediction, and discloses a bidirectional gating network short-term load prediction method and device based on Copula orthogonal analysis, and the method comprises the steps: carrying out the analysis of the correlation between the weather characteristics of a to-be-predicted building and the load based on an orthogonal maximum correlation coefficient, screening strong correlation weather features from the weather features according to an analysis result; taking historical weather data corresponding to the strong correlation weather characteristics and historical load data of the building to be predicted as training data, and training a prediction model based on the training data; and inputting real-time weather data corresponding to the strong correlation weather characteristics into the trained prediction model to obtain a short-term load prediction result of the to-be-predicted building. The method has the beneficial effects that strong correlation weather characteristics highly related to load change can be screened out, so that the input quality of a prediction model is effectively improved, and the accuracy of a load prediction result is improved.

Description

Technical Field

[0001] The present application relates to the technical field of load forecasting, and in particular to a method and device for short-term load forecasting of a bidirectional gated network based on Copula orthogonal analysis. Background Art

[0002] With the increasing demand for electricity in various office and commercial buildings, the stable operation of the power system has become particularly important. Accurate power load forecasting plays an important role in the balance of power demand and power supply scheduling of the power system. According to the length of the forecast time, power load forecasting can usually be divided into short-term load forecasting, medium-term load forecasting and long-term load forecasting.

[0003] For short-term load forecasting, electricity load demand is affected by a variety of external factors related to the weather, resulting in strong randomness and volatility in the electricity load in the short term. In related load forecasting technologies, if these external factors are ignored, the load forecast results will be inaccurate. However, if all external factors are taken into account, the complexity of the prediction algorithm dimension will be greatly increased, the load forecasting efficiency will be reduced, and the short-term load forecasting needs cannot be met. Summary of the invention

[0004] The present application provides a short-term load forecasting method for a bidirectional gated network based on Copula orthogonal analysis, which solves the technical problem that the current short-term load forecasting needs to consider the influence of a large number of external factors, resulting in the inability to achieve both dimensionality reduction of the prediction algorithm and accuracy of the prediction results, and achieves the technical effect of reducing the dimension of the prediction algorithm while ensuring the accuracy of the load forecast.

[0005] In order to achieve the above objectives, the main technical solutions adopted in this application include:

[0006] In a first aspect, an embodiment of the present application provides a short-term load forecasting method for a bidirectional gated network based on Copula orthogonal analysis, the short-term load forecasting method for a bidirectional gated network comprising:

[0007] Analyze the correlation between the weather characteristics of the building to be predicted and the load based on the orthogonal maximum correlation coefficient, and select strongly correlated weather characteristics from the weather characteristics according to the analysis results;

[0008] Using the historical weather data corresponding to the strongly correlated weather features and the historical load data of the building to be predicted as training data, and training the prediction model based on the training data;

[0009] The real-time weather data corresponding to the strongly correlated weather features are input into the trained prediction model to obtain a short-term load prediction result of the building to be predicted.

[0010] The bidirectional gated network short-term load forecasting method proposed in the embodiment of the present application uses the orthogonal maximum correlation coefficient to screen out the parts that are highly correlated with the load changes of the building to be predicted from a large number of weather characteristics as strongly correlated weather characteristics, and uses the strongly correlated weather characteristics as the input of the prediction model, thereby improving the accuracy of the load forecasting results without increasing the complexity of the variable dimension.

[0011] In addition, this weather feature screening strategy enables the prediction model to maintain high prediction accuracy and stability when dealing with load forecasting under different weather conditions, thereby improving the generalization ability of the prediction model.

[0012] Optionally, the strongly correlated weather characteristics include at least one of temperature, dew point, humidity, air pressure and wind speed.

[0013] Optionally, the analysis of the correlation between the weather characteristics of the building to be predicted and the load based on the orthogonal maximum correlation coefficient, and screening out strongly correlated weather characteristics from the weather characteristics according to the analysis results, includes:

[0014] Step 1: determining the correlation coefficient between each weather feature of the building to be predicted and the load based on the t-Copula function, and selecting the weather feature corresponding to the largest correlation coefficient to add to the selected feature set;

[0015] Step 2, obtaining orthogonal vectors of the weather features other than the selected feature set relative to the selected feature set, and calculating orthogonal correlation coefficients of the orthogonal vectors;

[0016] Step 3, selecting the weather feature corresponding to the largest orthogonal correlation coefficient and adding it to the selected feature set, and repeating the above step 2 until the number of repetitions reaches a preset value;

[0017] Step 4: Using the weather features in the selected feature set as the strongly correlated weather features.

[0018] By calculating the correlation coefficient between each weather feature and the load, the parts that are highly correlated with the load change are screened out. At the same time, taking into account the redundancy between weather features, the relevant weather features are reduced in dimension according to the principle of highest correlation, which not only ensures that the screened weather features are highly correlated with the load change, but also ensures the independence and importance of each weather feature, eliminates the redundancy problem of multiple weather features, and thus effectively improves the input quality of the prediction model.

[0019] Optionally, the determining the correlation coefficient between each weather feature of the building to be predicted and the load based on the t-Copula function respectively includes:

[0020] The correlation coefficient between each weather feature and the load is calculated based on formula (1):

[0021]

[0022] Wherein, u is the marginal distribution function of the weather characteristic, v is the marginal distribution function of the load, τ(u,v) is the Kendall correlation coefficient between the weather characteristic and the load, and C(u,v) is the t-Copula function with u and v as independent variables.

[0023] Through the above process, the complex nonlinear dependency between weather characteristics and load can be captured, and then the Kendall correlation coefficient can be derived, which helps to solve the shortcomings of traditional correlation analysis in nonlinear problems and can fully reflect the impact of different weather characteristics on load.

[0024] Optionally, the obtaining orthogonal vectors of the weather features other than the selected feature set relative to the selected feature set, and calculating orthogonal correlation coefficients of the orthogonal vectors includes:

[0025] Based on formula (2), the orthogonal vector β of the weather feature outside the selected feature set and the selected feature set is obtained. n :

[0026]

[0027] In the formula, α n is the historical data vector corresponding to the nth weather feature outside the selected feature set, β1, ..., β n-1 is α n Relative to the orthogonal vectors of each weather feature in the selected feature set, e n β n The normalized orthogonal vector of , <·,·> represents the vector inner product operation;

[0028] Based on formula (3), the orthogonal vector β is obtained: n The orthogonal correlation coefficient is:

[0029] OCC=τ(e n ,y) (3)

[0030] Where, OCC is the orthogonal correlation coefficient, y is the load of the building to be predicted, τ(e n ,y) is e n Kendall correlation coefficient between y and y.

[0031] The Gram-Schmidt Process is used to consider the minimum redundancy between weather features, and the weather features corresponding to the largest orthogonal correlation coefficients are gradually selected to ensure that the features selected each time are the weather features most relevant to the load changes in the current stage, and to eliminate the multicollinearity problem between multiple weather features, thus ensuring the independence of the selected features.

[0032] Optionally, the sampling time of the historical load data lags behind the sampling time of the historical weather data.

[0033] Optionally, the training the prediction model based on the training data includes:

[0034] Inputting the historical weather data into the prediction model and obtaining load prediction data output by the prediction model;

[0035] Obtaining a mean absolute percentage error and a root mean square error of the load forecast data relative to the historical load data;

[0036] When both the mean absolute percentage error and the root mean square error meet preset conditions, training of the prediction model is stopped.

[0037] By inputting historical weather data into the prediction model and obtaining the load forecast data output by the prediction model, the prediction performance of the model can be monitored and evaluated in real time. The two indicators of Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE) can be used to measure the prediction performance of the model from different angles, thereby improving the training effect of the prediction model.

[0038] Optionally, the prediction model is a bidirectional gated neural network model.

[0039] In a second aspect, an embodiment of the present application provides a bidirectional gated network short-term load forecasting device based on Copula orthogonal analysis, the bidirectional gated network short-term load forecasting device comprising:

[0040] A correlation analysis module, used for analyzing the correlation between the weather characteristics of the building to be predicted and the load based on the orthogonal maximum correlation coefficient, and screening out strongly correlated weather characteristics from the weather characteristics according to the analysis results;

[0041] A model training module, used to use the historical weather data corresponding to the strongly correlated weather features and the historical load data of the building to be predicted as training data, and train the prediction model based on the training data;

[0042] The prediction module is used to input the real-time weather data corresponding to the strongly correlated weather characteristics into the trained prediction model to obtain the short-term load prediction result of the building to be predicted.

[0043] The short-term load forecasting device proposed in the embodiment of the present application uses the orthogonal maximum correlation coefficient to screen out the parts that are highly correlated with the load changes of the building to be predicted from numerous weather characteristics as strongly correlated weather characteristics, and uses the strongly correlated weather characteristics as the input of the prediction model, thereby improving the accuracy of the load forecasting results without increasing the complexity of the variable dimension.

[0044] In addition, this weather feature screening strategy enables the prediction model to maintain high prediction accuracy and stability when dealing with load forecasting under different weather conditions, thereby improving the generalization ability of the prediction model.

[0045] In a third aspect, an embodiment of the present application provides a computer device, including:

[0046] A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the bidirectional gated network short-term load forecasting method based on Copula orthogonal analysis described in the above embodiment by executing the computer instructions. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the specific implementation methods of the present application or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0048] Figure 1 A schematic diagram of a flow chart of a short-term load forecasting method for a bidirectional gated network based on Copula orthogonal analysis proposed in an embodiment of the present application;

[0049] Figure 2 A schematic diagram of a process for filtering weather features into a selected feature set proposed in an embodiment of the present application;

[0050] Figure 3 A schematic diagram of the structure of a bidirectional gated network short-term load forecasting device based on Copula orthogonal analysis proposed in an embodiment of the present application;

[0051] Figure 4 The present invention is a schematic diagram of the structure of a computer device for executing the short-term load forecasting method for a bidirectional gated network proposed in an embodiment of the present application. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0053] Electricity load demand is affected by many external factors such as weather, electricity prices, holidays, etc., which leads to randomness and volatility of electricity load. Therefore, electricity load forecasting plays an important role in the balance of electricity demand and power supply scheduling of power systems. For short-term load forecasting, its load fluctuation is more significantly affected by weather changes, which increases the difficulty of forecasting. In order to improve the forecasting accuracy, it is necessary to accurately extract the key factors affecting load changes.

[0054] In traditional correlation analysis, commonly used methods include Pearson correlation coefficient, Spearman rank correlation coefficient and chi-square test. These methods are mainly used to evaluate the linear correlation between variables. However, the relationship between power load and weather characteristics is often nonlinear, and traditional correlation analysis methods are difficult to effectively capture and reflect this complex dependency. Therefore, in practical applications, traditional correlation analysis methods have certain deficiencies in screening key weather characteristics that affect power load, resulting in limited performance of the prediction model.

[0055] In summary, the existing short-term load forecasting methods have certain shortcomings in dealing with the nonlinear relationship and multicollinearity problems of power load data. There is an urgent need for a new method that can more accurately analyze the correlation between power load and weather characteristics, and screen out strongly correlated weather characteristics based on this, thereby improving the accuracy of load forecasting.

[0056] The embodiment of the present application proposes a short-term load forecasting method based on the orthogonal maximum correlation coefficient for two types of building loads, commercial and office. The OMCC method is used to analyze the correlation between weather characteristics and load, and strongly correlated weather characteristics are screened out, thereby improving the accuracy of the load forecasting model and providing strong support for the stable operation and efficient management of the power system.

[0057] According to an embodiment of the present application, an embodiment of short-term load forecasting of a bidirectional gated network based on Copula orthogonal analysis is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0058] Figure 1 is a flow chart of a method for short-term load forecasting of a bidirectional gating network according to an embodiment of the present application. Figure 1 As shown, the process includes the following steps:

[0059] Step S1, analyzing the correlation between the weather characteristics of the building to be predicted and the load based on the orthogonal maximum correlation coefficient, and screening out strongly correlated weather characteristics from the weather characteristics according to the analysis results.

[0060] Step S3, using the historical weather data corresponding to the strongly correlated weather features and the historical load data of the building to be predicted as training data, and training the prediction model based on the training data.

[0061] Step S5, inputting the real-time weather data corresponding to the strongly correlated weather features into the trained prediction model to obtain the short-term load prediction result of the building to be predicted.

[0062] The bidirectional gated network short-term load forecasting method provided in this embodiment uses the orthogonal maximum correlation coefficient to screen out the parts that are highly correlated with the load changes of the building to be predicted from a large number of weather features as strongly correlated weather features, and uses the strongly correlated weather features as the input of the prediction model, thereby improving the accuracy of the load forecasting results without increasing the complexity of the variable dimension.

[0063] In an embodiment of the present application, the strongly correlated weather characteristics include at least one of temperature, dew point, humidity, air pressure and wind speed.

[0064] In the embodiment of the present application, the correlation between the weather characteristics of the building to be predicted and the load is analyzed based on the orthogonal maximum correlation coefficient, and the strongly correlated weather characteristics are screened out from the weather characteristics according to the analysis results, including:

[0065] Step 1: determining the correlation coefficient between each weather feature of the building to be predicted and the load based on the t-Copula function, and selecting the weather feature corresponding to the largest correlation coefficient to add to the selected feature set;

[0066] Step 2, obtaining orthogonal vectors of the weather features other than the selected feature set relative to the selected feature set, and calculating orthogonal correlation coefficients of the orthogonal vectors;

[0067] Step 3, selecting the weather feature corresponding to the largest orthogonal correlation coefficient and adding it to the selected feature set, and repeating the above step 2 until the number of repetitions reaches a preset value;

[0068] Step 4: Using the weather features in the selected feature set as the strongly correlated weather features.

[0069] By calculating the correlation coefficient between each weather feature and the load, the parts that are highly correlated with the load change are screened out. At the same time, taking into account the redundancy between weather features, the relevant weather features are reduced in dimension according to the principle of highest correlation, which not only ensures that the screened weather features are highly correlated with the load change, but also ensures the independence and importance of each weather feature, eliminates the redundancy problem of multiple weather features, and thus effectively improves the input quality of the prediction model.

[0070] In order to overcome the shortcomings of traditional correlation analysis methods in dealing with nonlinear relationships, the t-Copula function provides a new method. The t-Copula function can effectively separate the marginal distribution and the dependency structure between variables, and flexibly handle the complex nonlinear relationship between multidimensional data. Through the t-Copula function, the nonlinear dependency between power load and weather characteristics can be more accurately captured, improving the accuracy of the prediction model.

[0071] The classic two-variable Copula joint distribution function can be defined as:

[0072] F(u,v)=C(F1(u),F2(v));

[0073] Where C(F1(u),F2(v)) is the selected copula function, F(u,v) is the joint distribution function, F1(u) is the marginal distribution function of variable u, and F2(v)2 is the marginal distribution function of variable v.

[0074] Furthermore, the t-Copula function expression is:

[0075]

[0076] Where ρ is the linear correlation coefficient, which can be estimated by the maximum likelihood method, and k is the degree of freedom.

[0077] The Kendall correlation coefficient can be derived through the Copula function, so the correlation coefficient between each weather feature of the building to be predicted and the load is determined based on the t-Copula function, including:

[0078] The correlation coefficient between each weather feature and the load is calculated based on the following formula:

[0079] τ(u,v)=4∫0 1 ∫0 1 C(u,v)dC(u,v)-1;

[0080] Wherein, u is the marginal distribution function of the weather characteristic, v is the marginal distribution function of the load, τ(u,v) is the Kendall correlation coefficient between the weather characteristic and the load, and C(u,v) is the t-Copula function with u and v as independent variables.

[0081] The t-Copula function can capture the complex nonlinear dependency between weather characteristics and load, overcome the limitations of traditional linear correlation analysis methods, and more accurately reflect the complex nonlinear relationship between variables. For example, a Pearson correlation coefficient of 0 does not necessarily mean that there is no correlation. There may be complex nonlinear relationships, and Copula can capture these relationships.

[0082] Therefore, the correlation analysis method proposed in the embodiment of the present application can select the weather feature that is most relevant to the load of the building to be predicted as the first selected strongly correlated weather feature, which serves as the basis for subsequent feature screening.

[0083] The embodiment of the present application takes into account the redundancy between multiple weather features, so the obtaining of orthogonal vectors of the weather features other than the selected feature set relative to the selected feature set and calculating the orthogonal correlation coefficients of the orthogonal vectors includes:

[0084] The orthogonal vector β between the weather feature outside the selected feature set and the selected feature set is obtained based on the following formula: n :

[0085]

[0086] In the formula, α n is the historical data vector corresponding to the nth weather feature outside the selected feature set, β1, ..., β n-1 is α n Relative to the orthogonal vectors of each weather feature in the selected feature set, e n β n The normalized orthogonal vector of , <·,·> represents the vector inner product operation;

[0087] The orthogonal vector β is obtained based on the following formula n The orthogonal correlation coefficient is:

[0088] OCC=τ(e n ,y);

[0089] Where, OCC is the orthogonal correlation coefficient, y is the load of the building to be predicted, τ(e n ,y) is e n Kendall correlation coefficient between y and y.

[0090] Orthogonal Maximum Correlation Coefficient (OMCC) is an improved correlation analysis method that can eliminate the multicollinearity problem between features through orthogonalization while considering nonlinear relationships. The OMCC method gradually selects the features with the greatest correlation with the target variable and performs orthogonalization to ensure the independence of the selected features. This can more accurately screen out weather features that are highly correlated with power load and provide more effective input data for the prediction model.

[0091] Figure 2 The process of filtering weather features into the selected feature set by the OMCC method is shown, and the following is a detailed description:

[0092] Step S11, determine the candidate feature set S c And the target y, that is, the candidate feature set S c contains all weather characteristics, and the target y here is the load of the building to be predicted.

[0093] Step S12, generating an initial orthogonal vector e1, and setting the initial number of iterations n=1.

[0094] Step S13: when the iteration starts, the number of iterations n is increased by 1.

[0095] Step S14, calculate the current normalized orthogonal vector as:

[0096] e n =e(e n-1 , x i ), {x i}=argmax(OCC=τ(e n ,y));

[0097] Among them, x i ∈S c -S n-1 , indicating x i is the candidate feature set S c No weather features from the selected feature set have been selected.

[0098] Step S15, select so that e n Maximum weather feature x i Add to the current selected feature set, and you get S n =S n-1 +x i .

[0099] Step S16, determine whether n reaches a preset value N. If it does not reach the preset value N, return to step S13 to continue iterating. If it reaches the preset value N, proceed to the next step.

[0100] Step S17, output the final selected feature set S n .

[0101] The Gram-Schmidt Process is used to consider the minimum redundancy between weather features, and the weather features corresponding to the largest orthogonal correlation coefficients are gradually selected to ensure that the features selected each time are the weather features most relevant to the load changes in the current stage, and to eliminate the multicollinearity problem between multiple weather features, thus ensuring the independence of the selected features.

[0102] In an embodiment of the present application, training the prediction model based on the training data includes:

[0103] Inputting the historical weather data into the prediction model and obtaining load prediction data output by the prediction model;

[0104] Obtaining a mean absolute percentage error and a root mean square error of the load forecast data relative to the historical load data;

[0105] When both the mean absolute percentage error and the root mean square error meet preset conditions, training of the prediction model is stopped.

[0106] It should be noted that the sampling time of the historical load data lags behind the sampling time of the historical weather data. The embodiment of the present application collects historical weather data and historical load data every 15 minutes, and considers the electricity consumption habits of office buildings and commercial buildings, and avoids holidays when collecting historical load data of office buildings.

[0107] Considering that different data have different magnitudes, historical weather data and historical load data need to be normalized before model training, that is:

[0108]

[0109] In the formula, is the data after normalization, x is the data before normalization, and x min is the minimum value among the same type of data, x max The maximum value among the data of the same type.

[0110] A good input data feature set is conducive to improving the calculation efficiency of the prediction model and preventing overfitting problems during model operation, which is beneficial to improving the load forecasting accuracy to a certain extent.

[0111] The embodiment of the present application performs model training for two different short-term periods, namely, day-ahead load forecasting and week-ahead load forecasting. For day-ahead load forecasting, the historical weather data of the previous two days is used as training data, and the historical load data of the next day is used as forecast data to verify the model validity, and the input data includes weather data and lagged historical load data. For week-ahead load forecasting, the historical weather data of the previous four weeks is used as training data, and the historical load data of the next week is used as forecast data to verify the model validity, and the input data includes weather data and lagged historical load data.

[0112] During the training process, the mean absolute percentage error and root mean square error are selected to evaluate the model performance. The specific calculation formula is as follows:

[0113]

[0114]

[0115] Where Y i is the predicted load output by the prediction model, is the historical load data, and N is the number of samples of training data.

[0116] By inputting historical weather data into the prediction model and obtaining the load forecast data output by the prediction model, the prediction performance of the model can be monitored and evaluated in real time. By using the two indicators of mean absolute percentage error and root mean square error, the prediction performance of the model can be measured from different angles, thereby improving the training effect of the prediction model.

[0117] In the embodiment of the present application, the prediction model is a bidirectional gated neural network model. The gated neural network (Gated Recurrent Unit, GRU) is an improved structure of the recurrent neural network, which consists of a reset gate and an update gate. The reset gate determines whether to combine the information of the previous moment with the current moment information. The update gate has a certain memory function, which determines the degree of retention of the information of the previous moment, which is conducive to solving the problem of gradient disappearance and gradient explosion in the recurrent neural network. Its function expression is:

[0118] r t =σ(W r ·[h t-1 ,x t ]+b r )

[0119]

[0120] z t =σ(W z ·[h t-1 ,x t ]+b z )

[0121]

[0122] Y t =σ(W o ·h t )

[0123] In the formula, r t is the reset gate; t is the update gate; W is the network weight value, b is the network bias, which can be learned through back propagation; tanh is the activation function to ensure that the candidate hidden state output value The interval is [-1,1]. The output value determines whether the candidate hidden state will be the hidden state h of the previous time step. t- The degree of discarding; the output range of the Sigmoid function σ is 0 to 1, ⊙ is the dot product, x t is the input of the prediction model at time t, Y t is the output of the prediction model at time t.

[0124] Accordingly, please refer to Figure 3 The embodiment of the present application provides a short-term load forecasting device for a bidirectional gated network based on Copula orthogonal analysis, the device comprising:

[0125] A correlation analysis module 601 is used to analyze the correlation between the weather characteristics of the building to be predicted and the load based on the orthogonal maximum correlation coefficient, and select strongly correlated weather characteristics from the weather characteristics according to the analysis results;

[0126] A model training module 602 is used to use the historical weather data corresponding to the strongly correlated weather features and the historical load data of the building to be predicted as training data, and train the prediction model based on the training data;

[0127] The prediction module 603 is used to input the real-time weather data corresponding to the strongly correlated weather characteristics into the trained prediction model to obtain the short-term load prediction result of the building to be predicted.

[0128] In some optional implementations, the correlation analysis module 601 includes:

[0129] The maximum correlation analysis unit 6011 determines the correlation coefficient between each weather feature of the building to be predicted and the load based on the t-Copula function, and selects the weather feature corresponding to the maximum correlation coefficient to add to the selected feature set;

[0130] An OCC correlation analysis unit 6012 obtains orthogonal vectors of the weather features other than the selected feature set relative to the selected feature set, and calculates orthogonal correlation coefficients of the orthogonal vectors;

[0131] The iteration unit 6013 selects the weather feature corresponding to the largest orthogonal correlation coefficient and adds it to the selected feature set, and repeats the execution steps of the OCC analysis unit 6012 until the number of repetitions reaches a preset value;

[0132] The output unit 6014 uses the weather features in the selected feature set as the strongly correlated weather features.

[0133] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0134] The bidirectional gated network short-term load forecasting device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0135] See also Figure 4 , Figure 4 is a schematic diagram of the structure of a computer device provided in an embodiment of the present application, such as Figure 4 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 4 A processor 10 is taken as an example.

[0136] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0137] The memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0138] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0139] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0140] The computer device further comprises a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0141] The embodiment of the present application also provides a computer-readable storage medium. The above method according to the embodiment of the present application can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0142] The embodiment of the present application provides a computer program product, which includes computer instructions, which are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the method of any embodiment of the present application.

[0143] Although the embodiments of the present application are described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present application, and such modifications and variations are all within the scope defined by the appended claims.

[0144] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0145] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0146] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0147] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0148] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0149] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0150] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0151] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0152] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

[0153] Although the embodiments of the present application have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present application, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A short-term load forecasting method for bidirectional gated networks based on Copula orthogonal analysis, characterized in that: The bidirectional gating network short-term load forecasting method comprises: Analyze the correlation between the weather characteristics of the building to be predicted and the load based on the orthogonal maximum correlation coefficient, and select strongly correlated weather characteristics from the weather characteristics according to the analysis results; Using the historical weather data corresponding to the strongly correlated weather features and the historical load data of the building to be predicted as training data, and training the prediction model based on the training data; The real-time weather data corresponding to the strongly correlated weather features are input into the trained prediction model to obtain a short-term load prediction result of the building to be predicted.

2. The method for short-term load forecasting of bidirectional gated networks based on Copula orthogonal analysis according to claim 1 is characterized in that: The strongly correlated weather characteristics include at least one of temperature, dew point, humidity, air pressure and wind speed.

3. The method for short-term load forecasting of bidirectional gated networks based on Copula orthogonal analysis according to claim 1 or 2, characterized in that: The correlation between the weather characteristics of the building to be predicted and the load is analyzed based on the orthogonal maximum correlation coefficient, and the strongly correlated weather characteristics are screened out from the weather characteristics according to the analysis results, including: Step 1: determining the correlation coefficient between each weather feature of the building to be predicted and the load based on the t-Copula function, and selecting the weather feature corresponding to the largest correlation coefficient to add to the selected feature set; Step 2, obtaining orthogonal vectors of the weather features other than the selected feature set relative to the selected feature set, and calculating orthogonal correlation coefficients of the orthogonal vectors; Step 3, selecting the weather feature corresponding to the largest orthogonal correlation coefficient and adding it to the selected feature set, and repeating the above step 2 until the number of repetitions reaches a preset value; Step 4: Using the weather features in the selected feature set as the strongly correlated weather features.

4. The method for short-term load forecasting of bidirectional gated networks based on Copula orthogonal analysis according to claim 3 is characterized in that: The determining the correlation coefficient between each weather characteristic of the building to be predicted and the load based on the t-Copula function respectively includes: The correlation coefficient between each weather feature and the load is calculated based on formula (1): Wherein, u is the marginal distribution function of the weather characteristic, v is the marginal distribution function of the load, τ(u, v) is the Kendall correlation coefficient between the weather characteristic and the load, and C(u, v) is the t-Copula function with u and v as independent variables.

5. The method for short-term load forecasting of bidirectional gated networks based on Copula orthogonal analysis according to claim 3 is characterized in that: The obtaining of orthogonal vectors of the weather features other than the selected feature set relative to the selected feature set, and calculating orthogonal correlation coefficients of the orthogonal vectors, comprises: Based on formula (2), the orthogonal vector β of the weather feature outside the selected feature set and the selected feature set is obtained. n : In the formula, α n is the historical data vector corresponding to the nth weather feature outside the selected feature set, β1, ..., β n-1 is α n Relative to the orthogonal vectors of each weather feature in the selected feature set, e n β n The normalized orthogonal vector of , <·,·> represents the vector inner product operation; Based on formula (3), the orthogonal vector β is obtained: n The orthogonal correlation coefficient is: OCC=τ(e n ,y) (3) Where, OCC is the orthogonal correlation coefficient, y is the load of the building to be predicted, τ(e n , y) is e n Kendall correlation coefficient between y and y.

6. The method for short-term load forecasting of bidirectional gated networks based on Copula orthogonal analysis according to claim 1 is characterized in that: The sampling time of the historical load data lags behind the sampling time of the historical weather data.

7. The method for short-term load forecasting of bidirectional gated networks based on Copula orthogonal analysis according to claim 1 is characterized in that: The training of the prediction model based on the training data includes: Inputting the historical weather data into the prediction model and obtaining load prediction data output by the prediction model; Obtaining a mean absolute percentage error and a root mean square error of the load forecast data relative to the historical load data; When both the mean absolute percentage error and the root mean square error meet preset conditions, training of the prediction model is stopped.

8. The method for short-term load forecasting of bidirectional gated networks based on Copula orthogonal analysis according to claim 1 or 7, characterized in that: The prediction model is a bidirectional gated neural network model.

9. A short-term load forecasting device for a bidirectional gated network based on Copula orthogonal analysis, characterized in that: The bidirectional gated network short-term load forecasting device comprises: A correlation analysis module, used for analyzing the correlation between the weather characteristics of the building to be predicted and the load based on the orthogonal maximum correlation coefficient, and screening out strongly correlated weather characteristics from the weather characteristics according to the analysis results; A model training module, used to use the historical weather data corresponding to the strongly correlated weather features and the historical load data of the building to be predicted as training data, and train the prediction model based on the training data; The prediction module is used to input the real-time weather data corresponding to the strongly correlated weather characteristics into the trained prediction model to obtain the short-term load prediction result of the building to be predicted.

10. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the short-term load forecasting method for a bidirectional gated network based on Copula orthogonal analysis as described in any one of claims 1 to 8 by executing the computer instructions.

Images