Method and system for predicting provincial load of power system

Abstract

The invention discloses a power system provincial level load prediction method, which comprises the steps of obtaining historical data information of each city in a target province, and performing processing and association to construct a data set; constructing a load prediction initial model of each city in the target province based on an XGBoost algorithm, and training to obtain a load prediction model of each city in the target province; and performing load prediction of each city by using the load prediction model of each city, and comprehensively obtaining a load prediction result of the target province to complete the prediction of the provincial load of the power system. The invention also discloses a system for realizing the provincial load prediction method of the power system. According to the method, the provincial load of the power system can be predicted, the reliability is higher, and the accuracy is better.

Description

Technical Field

[0001] The invention belongs to the field of electrical automation, and in particular relates to a provincial load forecasting method and system for an electric power system. Background Art

[0002] With the development of economy and technology and the improvement of people's living standards, electricity has become an indispensable secondary energy source in people's production and life, bringing endless convenience to people's production and life. Therefore, ensuring the stable and reliable supply of electricity has become one of the most important tasks of the power system. The load forecasting of the power system is an important guarantee for the safe and stable operation of the power system. Therefore, accurate and reliable load forecasting results are of great significance to the power system.

[0003] At present, the traditional load forecasting scheme of the power system generally conducts load forecasting from an overall perspective. However, in the process of provincial load forecasting of the power system, there are obvious regional differences among various regions within the province. The existence of such regional differences will make the existing overall load forecasting scheme have the problem of low reliability and poor accuracy when conducting provincial load forecasting. Summary of the invention

[0004] One of the purposes of the present invention is to provide a provincial load forecasting method for a power system with high reliability and good accuracy.

[0005] A second objective of the present invention is to provide a system for implementing the provincial load forecasting method for the power system.

[0006] The provincial load forecasting method for a power system provided by the present invention comprises the following steps:

[0007] S1. Obtain historical data information of each city in the target province;

[0008] S2. Process the historical data information obtained in step S1, and associate the processed load data with the influencing factor data to construct a data set;

[0009] S3. Based on the XGBoost algorithm, an initial load forecasting model for each city in the target province is constructed;

[0010] S4. Using the data set obtained in step S2, the initial load forecasting model constructed in step S3 is trained to obtain the load forecasting model for each city in the target province;

[0011] S5. Using the load forecasting model of each city obtained in step S4, load forecasting of each city is performed;

[0012] S6. Summarize the load forecast results of each prefecture-level city obtained in step S5, and comprehensively obtain the load forecast results of the target province to complete the provincial load forecast of the power system.

[0013] The acquisition of historical data information of each city in the target province described in step S1 specifically includes the following steps:

[0014] Obtain historical data information for each city in the target province;

[0015] The historical data information includes historical load data and historical influencing factor data;

[0016] The influencing factor data include meteorological data, economic data, population data and holiday data;

[0017] The meteorological data includes temperature data, humidity data and precipitation data; the economic data includes actual GDP data and industrial production index data; the population data includes actual total population data and population growth rate data; and the holiday data is data on whether it is a holiday.

[0018] Step S2 processes the historical data information obtained in step S1, associates the processed load data with the influencing factor data, and constructs a data set, which specifically includes the following steps:

[0019] Performing data processing on the historical data information obtained in step S1;

[0020] The data processing includes missing value filling;

[0021] The following formula is used to fill in missing values:

[0022]

[0023] Where φ(e′) is the result after completion; e′ is the time series horizontal coordinate of the missing point; e0 is the time series horizontal coordinate of the non-missing point on the left that is closest to e′; e1 is the time series horizontal coordinate of the non-missing point on the right that is closest to e′, φ(e0) is the function value corresponding to e0; φ(e1) is the function value corresponding to e1.

[0024] The processed load data and influencing factor data are represented by matrices and associated to construct a data set.

[0025] The XGBoost algorithm is used to construct an initial load forecasting model for each city in the target province, which specifically includes the following steps:

[0026] The initial load forecast model for each city in the target province is the same;

[0027] The initial load forecasting model includes a baseline layer XGBoost model and a real-time layer XGBoost model; the baseline layer XGBoost model and the real-time layer XGBoost model are the same;

[0028] The input data of the initial load forecasting model is predicted by the baseline layer XGBoost model to obtain the baseline forecasting result; after calculating the forecast error between the baseline forecasting result and the actual value, the forecast error is input into the real-time layer XGBoost model for forecasting to obtain the final load forecasting result;

[0029] The following formula is used as the processing function of the XGBoost model:

[0030]

[0031] In the formula is the predicted value; h k () is the model of the kth decision tree, and h k ()∈H, H is the function space composed of all decision trees; K is the total number of decision trees; x i is the i-th sample;

[0032] The following formula is used as the loss function of the XGBoost model:

[0033]

[0034] In the formula is the loss function; is the actual value y of the i-th sample i and the predicted value of the i-th sample The error between them; n is the total number of samples; f k is the kth decision tree; K is the total number of decision trees; Ω() is the regularization function, and γ is the first regularization coefficient, λ is the second regularization coefficient, T is the number of leaves, ω is the weight of the leaf node, and |||| is the value.

[0035] Step S4 uses the data set obtained in step S2 to train the initial load forecasting model constructed in step S3 to obtain the load forecasting model of each city in the target province, which specifically includes the following steps:

[0036] Using the data set obtained in step S2, the initial load forecasting model constructed in step S3 is trained to obtain the load forecasting model of each prefecture-level city in the target province;

[0037] During the training process, the data set obtained in step S2 is input into the initial load forecasting model, the model is trained, the parameters of the model are adjusted, and the load forecasting model of each prefecture-level city in the target province after training is obtained;

[0038] After the training is completed, the evaluation index of each load forecasting model is calculated using the following formula for the load forecasting model obtained for each city:

[0039]

[0040] Where L is the first evaluation index; x′ m ′ is the predicted correct load value; h j is the prediction result; W is the second evaluation index; h j ' is the actual result;

[0041] When both the first evaluation index and the second evaluation index meet the set requirements, the load forecasting model is deemed qualified and the subsequent steps are carried out;

[0042] Otherwise, return to step S1 and retrain the load forecasting model for each city until the load forecasting model is deemed qualified.

[0043] Step S5 uses the load forecasting model of each city obtained in step S4 to perform load forecasting for each city, which specifically includes the following steps:

[0044] Obtain the influencing factor data of each city during the forecast period;

[0045] The influencing factor data include meteorological data, economic data, population data and holiday data;

[0046] The meteorological data includes temperature data, humidity data and precipitation data; the economic data includes actual GDP data and industrial production index data; the population data includes actual total population data and population growth rate data; the holiday data is data on whether it is a holiday;

[0047] The acquired influencing factor data are input into the load forecasting model of each prefecture-level city to carry out load forecasting for each prefecture-level city.

[0048] Step S6 summarizes the load forecast results of each city obtained in step S5, and comprehensively obtains the load forecast result of the target province, completing the provincial load forecast of the power system, which specifically includes the following steps:

[0049] The weight of each prefecture-level city is set according to its development situation; the development situation includes historical load ratio, economic development level, population size, population density and climate;

[0050] According to the weighted weights of each city, the load forecast result S of the target province is calculated using the following formula:

[0051]

[0052] Where ω i is the weighted weight of the ith city; s i is the load forecast result of the i-th prefecture-level city; N is the total number of prefecture-level cities in the target province.

[0053] The present invention also provides a system for implementing the provincial load forecasting method of the power system, comprising a data acquisition module, a data processing module, a model construction module, a model training module, a regional forecasting module and a provincial forecasting module; the data acquisition module, the data processing module, the model construction module, the model training module, the regional forecasting module and the provincial forecasting module are connected in series in sequence; the data acquisition module is used to acquire the historical data information of each prefecture-level city in the target province, and upload the data information to the data processing module; the data processing module is used to perform data processing on the acquired historical data information according to the received data information, and associate the processed load data with the influencing factor data, construct a data set, and upload the data information to the model construction module; the model construction module is used to perform data processing on the received data information according to the received data information Based on the XGBoost algorithm, an initial load forecast model for each city in the target province is constructed, and the data information is uploaded to the model training module; the model training module is used to train the constructed initial load forecast model based on the received data information and the obtained data set, and obtain the load forecast model for each city in the target province, and upload the data information to the regional prediction module; the regional prediction module is used to use the obtained load forecast model for each city according to the received data information to perform load forecasting for each city, and upload the data information to the provincial prediction module; the provincial prediction module is used to summarize the load forecast results of each city based on the received data information, and comprehensively obtain the load forecast results of the target province, so as to complete the provincial load forecasting of the power system.

[0054] The provincial load forecasting method and system for the power system provided by the present invention establishes a load forecasting model for each prefecture-level city in the province based on the XGBoost model to perform load forecasting, and finally combines the load forecasting results of each prefecture-level city in combination with their specific conditions to obtain a provincial load forecasting result; therefore, the present invention can not only realize the provincial load forecasting of the power system, but also has higher reliability and better accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 The figure is a schematic diagram of the method flow of the present invention.

[0056] Figure 2 Schematic diagram of the functional modules of the system of the present invention. DETAILED DESCRIPTION

[0057] like Figure 1The method flow chart of the method of the present invention is shown as follows: The provincial load forecasting method of the power system disclosed in the present invention comprises the following steps:

[0058] S1. Obtain historical data information of each city in the target province; specifically includes the following steps:

[0059] Obtain historical data information for each city in the target province;

[0060] The historical data information includes historical load data and historical influencing factor data;

[0061] The influencing factor data include meteorological data, economic data, population data and holiday data;

[0062] The meteorological data includes temperature data, humidity data and precipitation data; the economic data includes actual GDP data and industrial production index data; the population data includes actual total population data and population growth rate data; the holiday data is data on whether it is a holiday, preferably binary data;

[0063] S2. Process the historical data information obtained in step S1, and associate the processed load data with the influencing factor data to construct a data set; specifically, the steps include:

[0064] Performing data processing on the historical data information obtained in step S1;

[0065] The data processing includes missing value filling;

[0066] The following formula is used to fill in missing values:

[0067]

[0068] In the formula, φ(e′) is the result after completion; e′ is the time series horizontal coordinate of the missing point; e0 is the time series horizontal coordinate of the non-missing point on the left closest to e′; e1 is the time series horizontal coordinate of the non-missing point on the right closest to e′, φ(e0) is the function value corresponding to e0; φ(e1) is the function value corresponding to e1;

[0069] The processed load data and influencing factor data are represented by a matrix and associated to construct a data set;

[0070] S3. Based on the XGBoost algorithm, an initial load forecasting model for each city in the target province is constructed; specifically, the following steps are included:

[0071] The initial load forecast model for each city in the target province is the same;

[0072] The initial load forecasting model includes a baseline layer XGBoost model and a real-time layer XGBoost model; the baseline layer XGBoost model and the real-time layer XGBoost model are the same;

[0073] The input data of the initial load forecasting model is predicted by the baseline layer XGBoost model to obtain the baseline forecasting result; after calculating the forecast error between the baseline forecasting result and the actual value, the forecast error is input into the real-time layer XGBoost model for forecasting to obtain the final load forecasting result;

[0074] The following formula is used as the loss function of the XGBoost model:

[0075]

[0076] In the formula is the loss function; is the actual value y of the i-th sample i and the predicted value of the i-th sample The error between them; n is the total number of samples; f k is the kth decision tree; K is the total number of decision trees; Ω() is the regularization function, and γ is the first regularization coefficient, λ is the second regularization coefficient, T is the number of leaves, ω is the weight of the leaf node, and |||| is the value.

[0077] In the double-layer XGBoost model provided by the present invention, the reference layer is used to analyze the load forecast period characteristic quantity and obtain the forecast result of the reference layer; the real-time layer is responsible for comparing the forecast result of the reference layer with the measured value, obtaining the dynamic error, and outputting the forecast result by iteratively calculating the time characteristics and adding the forecast result of the reference layer; in the double-layer collaborative structure, the reference layer will first give the load forecast result, and the real-time layer will correct the forecast result according to the deviation of the reference layer data, and establish the forecast model closest to the actual working condition according to the actual situation, so as to make the forecast result more reliable;

[0078] S4. Using the data set obtained in step S2, the initial load forecasting model constructed in step S3 is trained to obtain the load forecasting model of each city in the target province; specifically, the steps include:

[0079] Using the data set obtained in step S2, the initial load forecasting model constructed in step S3 is trained to obtain the load forecasting model of each city in the target province;

[0080] During the training process, the data set obtained in step S2 is input into the initial load forecasting model, the model is trained, and the parameters of the model are adjusted to improve the prediction accuracy and generalization ability of the model, and the load forecasting model of each prefecture-level city in the target province after training is obtained;

[0081] After the training is completed, the evaluation index of each load forecasting model is calculated using the following formula for the load forecasting model obtained for each city:

[0082]

[0083] Where L is the first evaluation index. The larger the value of L, the better the prediction result. m ′ is the predicted correct load value; h j is the prediction result; W is the second evaluation index, the smaller the value of W is, the better the prediction result is; h j ' is the actual result;

[0084] When both the first evaluation index and the second evaluation index meet the set requirements, the load forecasting model is deemed qualified and the subsequent steps are carried out;

[0085] Otherwise, return to step S1 and retrain the load forecasting model of each city until the load forecasting model is qualified;

[0086] S5. Using the load forecasting model of each city obtained in step S4, load forecasting of each city is performed; specifically comprising the following steps:

[0087] Obtain the influencing factor data of each city during the forecast period;

[0088] The influencing factor data include meteorological data, economic data, population data and holiday data;

[0089] The meteorological data includes temperature data, humidity data and precipitation data; the economic data includes actual GDP data and industrial production index data; the population data includes actual total population data and population growth rate data; the holiday data is data on whether it is a holiday;

[0090] Input the acquired influencing factor data into the load forecasting model of each prefecture-level city to conduct load forecasting for each prefecture-level city;

[0091] S6. Summarize the load forecast results of each city obtained in step S5, and comprehensively obtain the load forecast results of the target province to complete the provincial load forecast of the power system; specifically, the following steps are included:

[0092] The weight of each prefecture-level city is set according to its development situation; the development situation includes historical load ratio, economic development level, population size, population density and climate;

[0093] According to the weighted weights of each city, the load forecast result S of the target province is calculated using the following formula:

[0094]

[0095] Where ω i is the weighted weight of the ith city; s i is the load forecast result of the i-th prefecture-level city; N is the total number of prefecture-level cities in the target province.

[0096] like Figure 2 The functional module schematic diagram of the system of the present invention is shown as follows: the system disclosed in the present invention for realizing the provincial load forecasting method of the power system comprises a data acquisition module, a data processing module, a model construction module, a model training module, a regional forecasting module and a provincial forecasting module; the data acquisition module, the data processing module, the model construction module, the model training module, the regional forecasting module and the provincial forecasting module are connected in series in sequence; the data acquisition module is used to acquire the historical data information of each prefecture-level city in the target province, and upload the data information to the data processing module; the data processing module is used to perform data processing on the acquired historical data information according to the received data information, and associate the processed load data with the influencing factor data, construct a data set, and upload the data information to the model construction module; the model construction module is used to The received data information is used to construct an initial load forecast model for each city in the target province based on the XGBoost algorithm, and the data information is uploaded to the model training module; the model training module is used to train the constructed initial load forecast model based on the received data information and the obtained data set, obtain the load forecast model for each city in the target province, and upload the data information to the regional prediction module; the regional prediction module is used to use the obtained load forecast model for each city based on the received data information to perform load forecast for each city, and upload the data information to the provincial prediction module; the provincial prediction module is used to summarize the load forecast results of each city based on the received data information, and comprehensively obtain the load forecast results of the target province, so as to complete the provincial load forecast of the power system.

Claims

1. A method for provincial load forecasting of a power system, comprising the following steps: S1. Obtain historical data information of each city in the target province; S2. Process the historical data information obtained in step S1, and associate the processed load data with the influencing factor data to construct a data set; S3. Based on the XGBoost algorithm, an initial load forecasting model for each city in the target province is constructed; S4. Using the data set obtained in step S2, the initial load forecasting model constructed in step S3 is trained to obtain the load forecasting model for each city in the target province; S5. Using the load forecasting model of each city obtained in step S4, load forecasting of each city is performed; S6. Summarize the load forecast results of each prefecture-level city obtained in step S5, and comprehensively obtain the load forecast results of the target province to complete the provincial load forecast of the power system.

2. The provincial load forecasting method for power system according to claim 1, characterized in that The acquisition of historical data information of each city in the target province described in step S1 specifically includes the following steps: Obtain historical data information for each city in the target province; The historical data information includes historical load data and historical influencing factor data; The influencing factor data include meteorological data, economic data, population data and holiday data; The meteorological data includes temperature data, humidity data and precipitation data; the economic data includes actual GDP data and industrial production index data; the population data includes actual total population data and population growth rate data; and the holiday data is data on whether it is a holiday.

3. The provincial load forecasting method for power system according to claim 2 is characterized in that Step S2 processes the historical data information obtained in step S1, associates the processed load data with the influencing factor data, and constructs a data set, which specifically includes the following steps: Performing data processing on the historical data information obtained in step S1; The data processing includes missing value filling; The following formula is used to fill in missing values: Where φ(e′) is the result after completion; e′ is the time series horizontal coordinate of the missing point; e0 is the time series horizontal coordinate of the non-missing point on the left that is closest to e′; e1 is the time series horizontal coordinate of the non-missing point on the right that is closest to e′, φ(e0) is the function value corresponding to e0; φ(e1) is the function value corresponding to e1. The processed load data and influencing factor data are represented by matrices and associated to construct a data set.

4. The provincial load forecasting method for power system according to claim 3 is characterized in that The XGBoost algorithm is used to construct an initial load forecasting model for each city in the target province, which specifically includes the following steps: The initial load forecast model for each city in the target province is the same; The initial load forecasting model includes a baseline layer XGBoost model and a real-time layer XGBoost model; the baseline layer XGBoost model and the real-time layer XGBoost model are the same; The input data of the initial load forecasting model is predicted by the baseline layer XGBoost model to obtain the baseline forecasting result; after calculating the forecast error between the baseline forecasting result and the actual value, the forecast error is input into the real-time layer XGBoost model for forecasting to obtain the final load forecasting result; The following formula is used as the processing function of the XGBoost model: In the formula is the predicted value; h k () is the model of the kth decision tree, and h k ()∈H, H is the function space composed of all decision trees; K is the total number of decision trees; x i is the i-th sample; The following formula is used as the loss function of the XGBoost model: In the formula is the loss function; is the actual value y of the i-th sample i and the predicted value of the i-th sample The error between them; n is the total number of samples; f k is the kth decision tree; K is the total number of decision trees; Ω() is the regularization function, and γ is the first regularization coefficient, λ is the second regularization coefficient, T is the number of leaves, ω is the weight of the leaf node, and || || is the value.

5. The provincial load forecasting method for power system according to claim 4 is characterized in that Step S4 uses the data set obtained in step S2 to train the initial load forecasting model constructed in step S3 to obtain the load forecasting model of each city in the target province, which specifically includes the following steps: Using the data set obtained in step S2, the initial load forecasting model constructed in step S3 is trained to obtain the load forecasting model of each city in the target province; During the training process, the data set obtained in step S2 is input into the initial load forecasting model, the model is trained, the parameters of the model are adjusted, and the load forecasting model of each prefecture-level city in the target province after training is obtained; After the training is completed, the evaluation index of each load forecasting model is calculated using the following formula for the load forecasting model obtained for each city: Where L is the first evaluation index; x′ m ′ is the predicted correct load value; h j is the prediction result; W is the second evaluation index; h j ' is the actual result; When both the first evaluation index and the second evaluation index meet the set requirements, the load forecasting model is deemed qualified and the subsequent steps are carried out; Otherwise, return to step S1 and retrain the load forecasting model for each city until the load forecasting model is deemed qualified.

6. The provincial load forecasting method for power system according to claim 5, characterized in that Step S5 uses the load forecasting model of each city obtained in step S4 to perform load forecasting for each city, which specifically includes the following steps: Obtain the influencing factor data of each city during the forecast period; The influencing factor data include meteorological data, economic data, population data and holiday data; The meteorological data includes temperature data, humidity data and precipitation data; the economic data includes actual GDP data and industrial production index data; the population data includes actual total population data and population growth rate data; the holiday data is data on whether it is a holiday; The acquired influencing factor data are input into the load forecasting model of each prefecture-level city to carry out load forecasting for each prefecture-level city.

7. The provincial load forecasting method for power system according to claim 6, characterized in that Step S6 summarizes the load forecast results of each city obtained in step S5, and comprehensively obtains the load forecast result of the target province, completing the provincial load forecast of the power system, which specifically includes the following steps: The weight of each prefecture-level city is set according to its development situation; the development situation includes historical load ratio, economic development level, population size, population density and climate; According to the weighted weights of each city, the load forecast result S of the target province is calculated using the following formula: Where ω i is the weighted weight of the ith city; s i is the load forecast result of the i-th prefecture-level city; N is the total number of prefecture-level cities in the target province.

8. A system for implementing the provincial load forecasting method for a power system according to any one of claims 1 to 7, characterized in that It includes a data acquisition module, a data processing module, a model building module, a model training module, a regional prediction module and a provincial prediction module; the data acquisition module, the data processing module, the model building module, the model training module, the regional prediction module and the provincial prediction module are connected in series in sequence; the data acquisition module is used to obtain the historical data information of each city in the target province, and upload the data information to the data processing module; The data processing module is used to process the acquired historical data information according to the received data information, associate the processed load data with the influencing factor data, construct a data set, and upload the data information to the model construction module; The model building module is used to build the initial load forecast model for each city in the target province based on the received data information and the XGBoost algorithm, and upload the data information to the model training module; The model training module is used to train the constructed initial load forecast model based on the received data information and the obtained data set, obtain the load forecast model of each prefecture-level city in the target province, and upload the data information to the regional forecast module; the regional forecast module is used to use the obtained load forecast model of each prefecture-level city based on the received data information to perform load forecast for each prefecture-level city, and upload the data information to the provincial forecast module; The provincial forecasting module is used to summarize the load forecast results of each city based on the received data information, and comprehensively obtain the load forecast results of the target province to complete the provincial load forecast of the power system.

Images