Distributed photovoltaic and electrical load decoupling bus load prediction method and system

Abstract

The invention discloses a fuzzy evolutionary algorithm-based distributed photovoltaic and electrical load decoupling bus load prediction method and system for a power grid planning theme dynamic iteration method, and relates to the technical field of power dispatching. Comprising the steps of obtaining historical operation data of a to-be-predicted bus region; establishing a photovoltaic power generation and electrical load decoupling model of the access interval; dividing the bus load into time intervals according to the installed capacity proportion; constructing an electrical load prediction reference model of the non-accessed section; performing bus load prediction based on the decoupling model; and performing accuracy evaluation and iterative optimization on the prediction model. According to the method, the reliability and accuracy of prediction are ensured by acquiring the historical operation data and establishing the reference model, the data utilization efficiency is improved, the time intervals are divided according to the installed capacity proportion, the decoupling model of the distributed photovoltaic and electrical load is established, corresponding prediction strategies are adopted in different time intervals, and the adaptability of prediction is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power dispatching, and in particular to a bus load prediction method and system for decoupling distributed photovoltaic and electric loads. Background Art

[0002] Bus load forecasting is an important technical method in the field of power dispatching technology. Its content is to forecast the power load supplied by a specific substation. It is an important basis for the preparation of power generation plans, grid safety verification, and power spot clearing. With the continuous deepening of power market reform, the importance of bus load forecasting in grid operation has increased significantly. Since the bus load forecasting object is the load supplied by a certain substation, its forecasting object has the characteristics of small scale and large fluctuation.

[0003] But at the same time, the large-scale access of distributed photovoltaics has become an important technical challenge for bus load forecasting. Distributed photovoltaics adopt the mode of local consumption and surplus grid access, and its power generation is mixed with the power load. Compared with the case where distributed photovoltaics are not connected, the existing collection systems in most provinces and regions in my country can only monitor the power load and the overall exchange power of distributed photovoltaics, and cannot yet achieve observable and measurable distributed photovoltaic power generation. The dispatching agency can only grasp the installed capacity of distributed photovoltaics, but does not have access to its power generation curve and power generation data. Therefore, at present, a non-decoupling method is often used to comprehensively consider the relevant factors of distributed photovoltaics and power load, and a correlation prediction model between the relevant factors and the overall power load is constructed to realize bus load forecasting.

[0004] Distributed photovoltaic and electricity load are affected by many factors such as temperature, date, and light, and the impact is quite significant. Therefore, it is necessary to use the relevant factor prediction method to realize the prediction of distributed photovoltaic power generation and electricity load. However, there are large differences in the relevant factors of distributed photovoltaic and electricity load. Taking date as an example, this factor has a significant impact on electricity load, but basically has no impact on distributed photovoltaic; at the same time, although some factors have an impact on both, the impact range is quite different. Taking temperature as an example, temperature has a greater impact on electricity load, but has a smaller impact on distributed photovoltaic. This factor may not even be considered in the prediction of distributed photovoltaic related factors.

[0005] Therefore, when the overall bus load is predicted by comprehensively considering distributed photovoltaics and power load-related factors in a non-decoupling manner, the prediction accuracy will be adversely affected due to the selection of related factors.

[0006] On the one hand, if the factors related to distributed photovoltaics and electricity load are aggregated to construct the overall prediction model, it will lead to information interference. For example, the date factor will affect the prediction of the distributed photovoltaic part, resulting in an increase in the overall prediction deviation.

[0007] On the other hand, if we only select some overlapping related factors of distributed photovoltaics and electricity load to build an overall prediction model, important related factors may be missed. Summary of the invention

[0008] In view of the problems existing in the prior art, the present invention is proposed.

[0009] Therefore, the problem to be solved by the present invention is how to solve the problem that the influencing factors of distributed photovoltaics and power loads under the traditional non-decoupling method are quite different, resulting in large overall prediction errors. It has higher prediction accuracy in scenarios where the structure and scale of the power load itself have not changed significantly, and is simple to implement and easy to achieve.

[0010] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0011] In the first aspect, an embodiment of the present invention provides a bus load prediction method for decoupling distributed photovoltaic and power loads, which includes obtaining historical operating data of the bus area to be predicted; establishing a decoupling model of photovoltaic power generation and power load in the connected interval; dividing the bus load into time intervals according to the installed capacity ratio; constructing a benchmark model for power load prediction in the unconnected interval; predicting the bus load based on the decoupling model; and evaluating the accuracy and iteratively optimizing the prediction model.

[0012] As a preferred solution of the bus load forecasting method for decoupling distributed photovoltaic and power loads of the present invention, the historical operation data includes: load forecasting data and related influencing factors; centralized and distributed power generation operation characteristic data; regional installed capacity scale change data;

[0013] Among them, the time interval division is based on the proportion of distributed installed capacity to regional load.

[0014] As a preferred solution of the bus load forecasting method for decoupling distributed photovoltaic and power loads of the present invention, the unconnected interval benchmark model includes: establishing a load forecasting related factor model; determining the influence weight of each factor on the forecast; constructing a load forecasting accuracy evaluation benchmark; and obtaining the forecasting accuracy distribution characteristics of the benchmark model.

[0015] As a preferred solution of the bus load prediction method for decoupling distributed photovoltaic and power loads of the present invention, the decoupling model includes: photovoltaic power generation prediction based on the characteristics of the adjacent area; load prediction based on historical data; introduction of scaling factors for model decoupling; and establishment of a decoupled combined prediction model.

[0016] As a preferred solution of the bus load prediction method for decoupling distributed photovoltaic and power loads of the present invention, the combined prediction model includes: establishing a decoupling relationship between power generation and load; determining the optimal fitting value of the expansion / contraction factor; constructing a complete bus load prediction model; and evaluating the prediction effect of the combined model.

[0017] As a preferred solution of the bus load forecasting method for decoupling distributed photovoltaic and power loads of the present invention, the accuracy evaluation includes daily accuracy calculation and evaluation, and the accuracy is calculated by dividing NT operating periods:

[0018]

[0019] Among them, A d represents the prediction accuracy of the power load forecasting model on the operating day d, NT is the number of time periods divided on the operating day, is the measured value of power load during period t of the operating day;

[0020] The majority method is used to determine the reference value of the accuracy of the non-access time interval; the mode value of the access time interval prediction accuracy is calculated:

[0021]

[0022] Among them, A EA It represents the prediction accuracy evaluation value of the bus load prediction model within the access time interval, and CND represents the access time area date set;

[0023] Serves as a benchmark for evaluating overall forecasting levels;

[0024] Accuracy determination: set the prediction deviation threshold to no more than 2% and evaluate the prediction effect according to the determination formula:

[0025]

[0026] Among them, ε is the given deviation threshold;

[0027] Determine whether to enter the optimization process.

[0028] As a preferred solution of the bus load prediction method for decoupling distributed photovoltaic and power load of the present invention, the prediction model optimization includes: decoupling prediction model construction, establishing a telescopic factor photovoltaic prediction model:

[0029] in, Represents the predicted power generation output of distributed photovoltaic power generation at time t on the operating day, represents the installed capacity of distributed photovoltaic power generation on the operating day, NCP represents the number of centralized photovoltaic power stations in the adjacent area included in the prediction model, is the power generation output per installed capacity of the centralized photovoltaic cp during the operation day d period t, λ cp is the output expansion factor of centralized photovoltaic cp power generation;

[0030] Calculate the predicted output per installed capacity:

[0031]

[0032] in, Represents the predicted power generation output per unit installed capacity of the centralized photovoltaic cp during the operation day d and period t;

[0033] Construct bus load prediction model:

[0034] in, represents the power load forecast value of the operation day d period t, NS is the number of relevant factors included in the relevant factor prediction model, and D s,d,t is the value of the relevant factor s at the operation day d and time period t, α s is the correlation coefficient of the relevant factor s;

[0035] Spatial distance optimization and termination determination:

[0036] Sort the centralized photovoltaics by spatial distance, eliminate the farthest centralized photovoltaics in turn, recalculate the expansion factor and prediction model, repeat the optimization until the accuracy requirements are met, and record the final adjacent photovoltaic range. If the requirements are still not met after all are eliminated, it is determined that the power load characteristics of the area have changed and the method is not applicable.

[0037] In a second aspect, an embodiment of the present invention provides a bus load prediction system for decoupling distributed photovoltaic and power loads, which includes a prediction accuracy evaluation module, a bus load prediction module, and a model optimization module;

[0038] The prediction accuracy evaluation module is used to collect daily accuracy calculation data, interval accuracy statistics, and accuracy determination data to establish an accuracy evaluation system;

[0039] The bus load prediction module is used to construct a decoupled prediction model based on the expansion factor photovoltaic prediction model and the power load prediction model;

[0040] The model optimization module is used to monitor the decoupled prediction process, establish an optimization evaluation system through the spatial distance optimization criterion, and regularly iterate and update the prediction model.

[0041] In a third aspect, an embodiment of the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program instructions are executed by the processor, the steps of the bus load prediction method for decoupling distributed photovoltaic and power loads as in the first aspect of the present invention are implemented.

[0042] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program instructions are executed by a processor, the steps of the bus load prediction method for decoupling distributed photovoltaic and power loads as described in the first aspect of the present invention are implemented.

[0043] The beneficial effects of the present invention are as follows: the bus load prediction method based on the decoupling of distributed photovoltaic and power load provided by the present invention ensures the reliability and accuracy of the prediction by obtaining historical operation data and establishing a benchmark model, thereby improving the data utilization efficiency, reducing the prediction error, and improving the accuracy of the final prediction result; by dividing the time interval according to the installed capacity ratio, a decoupling model of distributed photovoltaic and power load is established, so that the corresponding prediction strategies are adopted in different time intervals, the adaptability of the prediction is improved, the model complexity and calculation overhead are reduced, and the economy and practicality of the load prediction are ensured; an iterative optimization system is established through the accuracy evaluation and optimization mechanism, and the prediction model is regularly updated and improved, which improves the prediction efficiency, reduces the prediction deviation, and ensures the advancement and effectiveness of the prediction method. The present invention achieves better results in terms of prediction accuracy, calculation efficiency and model adaptability, solves the problem that the influencing factors of distributed photovoltaic and power load in the traditional non-decoupling mode are quite different, resulting in a large overall prediction error, and has higher prediction accuracy in scenarios where the structure and scale of the power load itself have not changed significantly, and is simple to implement and easy to achieve. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0045] Figure 1 A flow chart of a bus load prediction method and system for decoupling distributed photovoltaic and power loads; DETAILED DESCRIPTION

[0046] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0047] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0048] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0049] Example 1

[0050] Reference Figure 1 This is the first embodiment of the present invention, which provides a bus load prediction method for decoupling distributed photovoltaic and power loads, comprising:

[0051] S1, obtaining the historical operation data of the bus area to be predicted;

[0052] The purpose of implementing this step is to determine the basic data required for the implementation of the present invention and to clarify the source of the data.

[0053] In one embodiment, the historical operation data includes: load forecast data and its related influencing factors; centralized and distributed power generation operation characteristic data; regional installed capacity scale change data;

[0054] Among them, the time interval division is based on the proportion of distributed installed capacity to regional load.

[0055] Preferably, the data and sources thereof required for the implementation of the present invention include:

[0056] The historical power load of the bus area to be predicted and the prediction model of related factors are obtained by the load prediction management system;

[0057] The centralized photovoltaic prediction model and historical power generation output data in the adjacent area are obtained by the new energy management system;

[0058] The historical data of changes in distributed photovoltaic installed capacity in the bus area to be predicted is obtained by the new energy management system.

[0059] S2, establish a decoupling model of photovoltaic power generation and power load in the access area;

[0060] In one embodiment, the decoupling model includes: photovoltaic power generation prediction based on the characteristics of the adjacent area; load prediction based on historical data; introducing a scaling factor to decouple the model; and establishing a decoupled combined prediction model.

[0061] In one embodiment, the combined prediction model includes: establishing a decoupling relationship between power generation and load; determining the optimal fitting value of the expansion factor; constructing a complete bus load prediction model; and evaluating the prediction effect of the combined model.

[0062] It should be noted that the purpose of constructing a distributed photovoltaic prediction model based on a telescopic factor is to introduce a nearby centralized photovoltaic telescopic factor and construct a distributed photovoltaic prediction model based on the telescopic factor. In the adjacent area, the types of factors related to the power generation output of centralized photovoltaic and distributed photovoltaic are the same. Due to the close location, the correlation coefficients of different related factors with centralized photovoltaic and distributed photovoltaic power generation output are similar. Therefore, the predicted value of centralized photovoltaic power generation output in the adjacent area can be used as the basis for predicting the output of distributed photovoltaic power generation in the same area. However, despite being in the adjacent area, the difference in geographical location between centralized photovoltaic and distributed photovoltaic will still lead to differences in distributed photovoltaic power generation output and centralized photovoltaic power generation output in the adjacent area. To this end, the present invention will introduce a nearby centralized photovoltaic telescopic factor to quantify the difference in geographical location between centralized photovoltaic power generation output and distributed photovoltaic power generation output.

[0063] Preferably, the distributed photovoltaic prediction model with the expansion factor introduced can be expressed as the sum of the product of the centralized photovoltaic power generation output in the adjacent area and the expansion factor, which can be expressed as:

[0064]

[0065] in, Represents the predicted power generation output of distributed photovoltaic power generation at time t on the operating day, represents the installed capacity of distributed photovoltaic power generation on the operating day, NCP represents the number of centralized photovoltaic power stations in the adjacent area included in the prediction model, is the power generation output per installed capacity of the centralized photovoltaic cp during the operation day d period t, λ cp is the output expansion factor of centralized photovoltaic cp power generation;

[0066] Among them, the predicted power generation output per installed capacity of the centralized photovoltaic CP during the operation day of period d is the ratio of the predicted power generation output of the centralized photovoltaic CP during the operation day of period d to its installed capacity, which can be expressed as:

[0067]

[0068] in, Represents the predicted power generation output per unit installed capacity of the centralized photovoltaic (CP) system during period t on the operating day.

[0069] S3, dividing the bus load into time intervals according to the installed capacity ratio;

[0070] It should be noted that the purpose of implementing this step is to construct a power load prediction model based on correlation factors by using the power load characteristics before the distributed photovoltaics are connected, and to calculate the prediction accuracy level before the large-scale access of distributed photovoltaics based on historical data. The present invention is mainly aimed at scenarios where the structure and scale of the power load itself have not changed significantly, so the power load prediction model can be maintained before and after the distributed photovoltaics are connected.

[0071] Preferably, according to the distributed photovoltaic installation access time, the power load is divided into a non-connected time interval and a connected time interval, and the division standard is the proportion of the distributed photovoltaic power load scale in the area. Without loss of generality, the present invention stipulates that the scenario where the distributed photovoltaic installed capacity does not exceed 5% of the average annual power load of the area is defined as the non-connected time area, which can be expressed as:

[0072]

[0073] Where is the average annual electricity load in the region. Due to the gradual development of distributed photovoltaics, the installed capacity of distributed photovoltaics is gradually increasing, and its installed capacity needs to be reported to the power grid company when applying for access. The changes in the installed capacity of distributed photovoltaics can be dynamically grasped based on the reporting situation.

[0074] Preferably, within the non-connected time interval, a power load forecasting model based on relevant factors is constructed. This step includes steps such as relevant factor identification, key factor screening, and relevant factor forecasting model construction. Considering that constructing a power load forecasting model based on relevant factors in the non-connected time interval is a current mature technology, the present invention does not repeat the specific implementation process without loss of generality. The power load forecasting model based on relevant factors can be expressed as:

[0075] in, represents the power load forecast value of the operation day d period t, NS is the number of relevant factors included in the relevant factor prediction model, and D s,d,t is the value of the relevant factor s at the operation day d and time period t, α s is the correlation coefficient of the relevant factor s;

[0076] Preferably, the distribution characteristics of the prediction accuracy of the power load prediction model based on relevant factors within the non-connected time interval are statistically analyzed. The prediction accuracy of the power load prediction model is statistically analyzed on a daily basis, which can be expressed as:

[0077]

[0078] Among them, A d represents the prediction accuracy of the power load forecasting model on the operating day d, NT is the number of time periods divided on the operating day, is the measured value of power load during period t of the operating day;

[0079] Considering that the power load prediction accuracy may deviate greatly in extreme scenarios, the present invention recommends using the mode as a reference value for the prediction accuracy of the power load prediction model within the non-connected time interval, which can be expressed as:

[0080]

[0081] Among them, A RV It represents the reference value of the prediction accuracy of the power load prediction model within the non-connected time interval, NND represents the date set of the non-connected time area, Represents the mode function.

[0082] S4, constructing a benchmark model for predicting power load in unconnected areas;

[0083] In one embodiment, the unconnected interval benchmark model includes: establishing a load forecasting related factor model; determining the influence weight of each factor on the forecast; constructing a load forecasting accuracy evaluation benchmark; and obtaining the forecasting accuracy distribution characteristics of the benchmark model.

[0084] It should be noted that the purpose of constructing a bus prediction model based on the decoupling of distributed photovoltaics and power load is to summarize the distributed photovoltaic prediction model based on the scaling factor and the power load prediction model based on the correlation factor, and to construct a bus load prediction model for the access time interval.

[0085] Preferably, a bus load prediction model for the access time interval is constructed. The prediction model is the sum of a distributed photovoltaic prediction model based on a scaling factor and a power load prediction model based on a correlation factor, which can be expressed as:

[0086]

[0087] In the formula, The bus load forecast for time period t on the operating day d representing the access time interval is respectively the distributed photovoltaic power generation output forecast value and the power load forecast value in the above formula.

[0088] in:

[0089] (1) Distributed photovoltaic power generation is obtained from the predicted output of centralized photovoltaic power generation in the adjacent area, where the expansion factor is a pending coefficient and is obtained by the least squares method in the subsequent steps;

[0090] (2) Considering that the scale and structure of the power load in the region have not changed significantly, the power load forecasting model for the non-access time interval can be continued.

[0091] Preferably, the least square method is used to fit the power load forecasting model and the undetermined expansion factor in the formula using the access time interval range data. Considering that the least square method is a common method in the statistical field and does not affect the main innovation of the present invention, the specific implementation process of this step is not repeated. The expansion factor coefficient with the smallest deviation can be obtained by using this method.

[0092] S5, bus load forecasting based on decoupling model;

[0093] It should be noted that the purpose of determining the decoupling prediction accuracy is to compare the prediction accuracy of the power load prediction model in the non-connected time interval with the prediction accuracy of the decoupling prediction model in the connected time interval to determine whether the decoupling prediction meets the prediction accuracy requirements.

[0094] Preferably, the bus load prediction accuracy shown by the sum of the power load prediction models in the access time interval is statistically calculated, and the mode of the access time interval prediction accuracy is used as the overall prediction level, which can be expressed as:

[0095]

[0096] Among them, A EA It represents the prediction accuracy evaluation value of the bus load prediction model within the access time interval, and CND represents the access time area date set.

[0097] Preferably, if the prediction accuracy evaluation value of the bus load prediction model in the access time interval is higher than the prediction accuracy reference value of the bus load prediction model in the non-access time interval, or is lower than the prediction accuracy reference value of the bus load prediction model in the non-access time interval by less than a given threshold, then the judgment condition is met, the prediction model is built, and the prediction can be performed accordingly, and the implementation ends; otherwise, it goes to S6 to optimize and adjust the distributed photovoltaic prediction model based on the telescopic factor. The above judgment conditions can be expressed as:

[0098]

[0099] Among them, ε is the deviation threshold given by dispatching and operation professionals, which generally does not exceed 2%.

[0100] S6, evaluate the accuracy and iteratively optimize the prediction model.

[0101] It should be noted that the optimization of the decoupled prediction model is implemented to optimize the distributed photovoltaic prediction model based on the telescopic factor to meet the prediction accuracy requirements. When the decoupled prediction accuracy does not meet the requirements, the main reason is that the distributed photovoltaic prediction model based on the telescopic factor affects the overall prediction effect. The main reason is that there are differences in the characteristics of centralized photovoltaics and distributed photovoltaics in the adjacent area. The reason for the difference is the difference in their spatial location distance, which leads to different related factors and correlation coefficients. To this end, the prediction accuracy of distributed photovoltaics is improved by eliminating centralized photovoltaics that do not meet the requirements.

[0102] Preferably, the centralized photovoltaics in S2 are sorted from far to near according to their spatial distances to the distributed photovoltaics.

[0103] Preferably, the centralized photovoltaic with the farthest distance and the highest ranking is eliminated, and the process is repeated back to S2 until a prediction model that meets the judgment requirements is obtained; if all centralized photovoltaics still cannot meet the above judgment conditions, the characteristics of the power load in the area have changed, and the method proposed in the present invention is not applicable.

[0104] Furthermore, this embodiment also provides a bus load prediction system for decoupling distributed photovoltaic and power loads, including a prediction accuracy evaluation module, a bus load prediction module, and a model optimization module;

[0105] The prediction accuracy evaluation module is used to collect daily accuracy calculation data, interval accuracy statistics, and accuracy determination data to establish an accuracy evaluation system;

[0106] The bus load prediction module is used to construct a decoupled prediction model based on the expansion factor photovoltaic prediction model and the power load prediction model;

[0107] The model optimization module is used to monitor the decoupled prediction process, establish an optimization evaluation system through the spatial distance optimization criterion, and regularly iterate and update the prediction model.

[0108] This embodiment also provides a computer device, which is suitable for the bus load prediction method of decoupling distributed photovoltaic and power load, including a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the bus load prediction method of decoupling distributed photovoltaic and power load as proposed in the above embodiment.

[0109] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0110] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the bus load prediction method for decoupling distributed photovoltaic and power loads as proposed in the above embodiment is implemented.

[0111] In summary, the bus load prediction method based on decoupling of distributed photovoltaic and power load provided by the present invention ensures the reliability and accuracy of the prediction by obtaining historical operation data and establishing a benchmark model, improves data utilization efficiency, reduces prediction errors, and improves the accuracy of the final prediction results; by dividing the time interval according to the installed capacity ratio, a decoupling model of distributed photovoltaic and power load is established, so that different time intervals adopt corresponding prediction strategies, improves the adaptability of the prediction, reduces the model complexity and calculation overhead, and ensures the economy and practicality of load prediction; through the accuracy evaluation and optimization mechanism, an iterative optimization system is established, and the prediction model is regularly updated and improved, which improves the prediction efficiency, reduces the prediction deviation, and ensures the advancement and effectiveness of the prediction method. The present invention achieves better results in prediction accuracy, calculation efficiency and model adaptability, solves the problem that the influencing factors of distributed photovoltaic and power load in the traditional non-decoupling method are quite different, resulting in a large overall prediction error, and has higher prediction accuracy in scenarios where the structure and scale of the power load itself have not changed significantly, and is simple to implement and easy to achieve.

[0112] Example 2

[0113] This is the second embodiment of the present invention, which provides a bus load prediction method for decoupling distributed photovoltaic and power loads. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0114] In order to verify the effectiveness of the method of the present invention, a 220kV substation under a provincial power grid dispatching center was selected as the test object. In 2023, multiple distributed photovoltaic projects were connected to the power supply area of ​​the substation, with a total installed capacity of 15MW, accounting for about 12% of the average load in the region. The test selected operating data from January 2022 to March 2024, of which 2022 was the period when distributed photovoltaics were not connected, and January 2023 was the period when photovoltaics were gradually connected.

[0115] The following information was collected during the data collection phase:

[0116] Load forecast related data: including historical load data, temperature data (maximum temperature, minimum temperature, average temperature), date type (weekdays, holidays), weather conditions, etc.

[0117] The operating data of three centralized photovoltaic power stations in the adjacent area are located 15km southeast, 12km southwest and 20km northeast of the target substation.

[0118] Record of changes in distributed photovoltaic installed capacity, monthly statistics from January 2023.

[0119] The implementation process is as follows:

[0120] First, based on the proportion of distributed photovoltaic installed capacity to the average regional load, the time interval is divided into the unconnected interval (the whole year of 2022) and the connected interval (January 2023 to March 2024). In the unconnected interval, a load forecasting benchmark model including factors such as temperature, date type, and seasonal characteristics was constructed, and the weight of each factor was determined through multivariate regression analysis.

[0121] Subsequently, the power generation characteristics of three centralized photovoltaic power stations in the adjacent area were analyzed, and a photovoltaic prediction model with a telescopic factor was established. Considering the impact of spatial distance on lighting conditions, a distance-weighted telescopic factor was introduced to correct the output characteristics differences of photovoltaic power stations in different locations.

[0122] On this basis, a decoupled prediction model is constructed. The model predicts the power load and distributed photovoltaic power generation respectively, and then obtains the final bus load prediction value by combining the prediction models. To verify the effectiveness of the method, comparative tests are carried out using the traditional non-decoupled prediction method and the decoupled prediction method of the present invention.

[0123]

[0124] By comparing the test data, the following analysis conclusions can be drawn:

[0125] Improved prediction accuracy: The decoupled prediction method of the present invention significantly reduces the average prediction error. When using data from three nearby photovoltaic sites, the average prediction error is 2.1%, which is 56.25% lower than the 4.8% of the traditional non-decoupled method. The maximum prediction error is also reduced from 12.5% ​​to 6.8%, indicating that the stability of the prediction results has been significantly improved.

[0126] Optimization of model generalization ability: The model generalization ability coefficient of the decoupled prediction method reached 0.92 (full score 1.0), which is much higher than the 0.75 of the traditional method, indicating that this method has stronger generalization ability and can better cope with new data and unknown scenarios. Especially in the early stage of distributed photovoltaic access and the stage of rapid growth of installed capacity, the prediction performance remains stable.

[0127] Computational efficiency trade-off: Although the decoupled prediction method slightly increases the computational time (3.2s / time vs 2.5s / time), this increase in computational cost is acceptable considering the significant improvement in prediction accuracy. Moreover, by optimizing the selection of neighboring sites, the amount of computation can be reduced while maintaining high prediction accuracy.

[0128] Influence of the number of sites: As the number of reference sites decreases, the prediction accuracy shows a decreasing trend, but even if only one nearest photovoltaic site is used, the prediction accuracy still reaches 94.2%, which is better than the traditional method. This shows that the method of the present invention has good practicality and flexibility.

[0129] Method stability: During the 15-month trial period, the decoupled forecasting method maintained a high forecast accuracy, indicating that the method has good time adaptability. Even when the seasons change and the load characteristics change greatly, the forecast performance remains stable.

[0130] These data fully demonstrate the superiority of the method of the present invention in practical applications, which not only significantly improves the prediction accuracy, but also has good adaptability and practicality. Especially in the context of large-scale access to distributed photovoltaics, this method provides more reliable load prediction support for power grid dispatching.

[0131] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A bus load forecasting method for decoupling distributed photovoltaic and power load, characterized by: include, Obtain historical operation data of the bus area to be predicted; Establish a decoupling model of photovoltaic power generation and power load in the access area; Divide the bus load into time intervals according to the proportion of installed capacity; Construct a benchmark model for electricity load prediction in unconnected areas; Bus load forecasting based on decoupling model; The prediction model is evaluated for accuracy and iteratively optimized.

2. The bus load forecasting method for decoupling distributed photovoltaic and power loads according to claim 1, characterized in that: The historical operation data include: load forecast data and related influencing factors; centralized and distributed power generation operation characteristic data; regional installed capacity scale change data; Among them, the time interval division is based on the proportion of distributed installed capacity to regional load.

3. The bus load forecasting method for decoupling distributed photovoltaic and power loads as claimed in claim 2, characterized in that: The unconnected interval benchmark model includes: establishing a load forecasting related factor model; determining the influence weight of each factor on the forecast; constructing a load forecasting accuracy evaluation benchmark; and obtaining the forecasting accuracy distribution characteristics of the benchmark model.

4. The bus load forecasting method for decoupling distributed photovoltaic and power loads as claimed in claim 3 is characterized by: The decoupling model includes: photovoltaic power generation prediction based on the characteristics of adjacent areas; load prediction based on historical data; introducing a scaling factor to decouple the model; and establishing a decoupled combined prediction model.

5. The bus load forecasting method for decoupling distributed photovoltaic and power loads as claimed in claim 4, characterized in that: The combined prediction model includes: establishing a decoupling relationship between power generation and load; determining the optimal fitting value of the expansion factor; constructing a complete bus load prediction model; and evaluating the prediction effect of the combined model.

6. The bus load forecasting method for decoupling distributed photovoltaic and power loads as claimed in claim 5, characterized in that: The accuracy evaluation includes daily accuracy calculation and evaluation, and the accuracy is calculated by dividing NT operating periods: Among them, A d represents the prediction accuracy of the power load forecasting model on the operating day d, NT is the number of time periods divided on the operating day, is the measured value of power load during period t of the operating day; The majority method is used to determine the reference value of the accuracy of the non-access time interval; the mode value of the access time interval prediction accuracy is calculated: Among them, A EA It represents the prediction accuracy evaluation value of the bus load prediction model within the access time interval, and CND represents the access time area date set; Serves as a benchmark for evaluating overall forecasting levels; Accuracy determination: set the prediction deviation threshold to no more than 2% and evaluate the prediction effect according to the determination formula: Among them, ε is the given deviation threshold; Determine whether to enter the optimization process.

7. The bus load forecasting method for decoupling distributed photovoltaic and power loads according to claim 6, characterized in that: The prediction model optimization includes decoupling the prediction model construction and establishing the telescopic factor photovoltaic prediction model: in, Represents the predicted power generation output of distributed photovoltaic power generation at time t on the operating day, represents the installed capacity of distributed photovoltaic power generation on the operating day, NCP represents the number of centralized photovoltaic power stations in the adjacent area included in the prediction model, is the power generation output per installed capacity of the centralized photovoltaic cp during the operation day d period t, λ cp is the output expansion factor of centralized photovoltaic cp power generation; Calculate the predicted output per installed capacity: in, Represents the predicted power generation output per unit installed capacity of the centralized photovoltaic cp during the operation day d and period t; Construct bus load prediction model: in, represents the power load forecast value of the operation day d period t, NS is the number of relevant factors included in the relevant factor prediction model, and D s,d,t is the value of the relevant factor s at the operation day d and time period t, α s is the correlation coefficient of the relevant factor s; Spatial distance optimization and termination determination: Sort the centralized photovoltaics by spatial distance, eliminate the farthest centralized photovoltaics in turn, recalculate the expansion factor and prediction model, repeat the optimization until the accuracy requirements are met, and record the final adjacent photovoltaic range. If the requirements are still not met after all are eliminated, it is determined that the power load characteristics of the area have changed and the method is not applicable.

8. A bus load prediction system for decoupling distributed photovoltaic and power load, based on the bus load prediction method for decoupling distributed photovoltaic and power load according to any one of claims 1 to 7, characterized in that: It also includes a prediction accuracy evaluation module, a bus load prediction module, and a model optimization module; The prediction accuracy evaluation module is used to collect daily accuracy calculation data, interval accuracy statistics data, and accuracy determination data to establish an accuracy evaluation system; The bus load prediction module is used to construct a decoupled prediction model based on the telescopic factor photovoltaic prediction model and the power load prediction model; The model optimization module is used to monitor the decoupling prediction process, establish an optimization evaluation system through the spatial distance optimization criterion, and regularly iterate and update the prediction model.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the bus load prediction method for decoupling distributed photovoltaic and power loads as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the bus load prediction method for decoupling distributed photovoltaic and power loads as described in any one of claims 1 to 7 are implemented.

Images