Power generation efficiency prediction accuracy evaluation method

Abstract

The invention relates to a power generation efficiency prediction accuracy evaluation method, and belongs to the technical field of resource power prediction. Obtaining a confidence level and probability prediction intervals of a plurality of adjacent time periods, judging whether a power failure event and / or a power limiting event exist in the probability prediction interval of each time period, if yes, judging whether the probability prediction interval of each time period is effective according to the power failure duration, and if yes, selecting a corresponding current effective probability prediction interval, if not, judging the probability prediction interval of the next time period; if the power limiting event exists, selecting an effective probability prediction interval of a historical data time period as a probability prediction interval of a current time period, and comparing the probability prediction interval of the current time period with a corresponding prediction value to obtain a coverage probability of the prediction value; the reliability of the prediction result in the current time period is obtained through the confidence level and the coverage probability, the prediction result is evaluated according to the reliability of the current prediction result, and the reliability obtained through the method is accurate.

Description

Technical Field

[0001] The invention relates to a method for evaluating the accuracy of power generation efficiency prediction, and belongs to the technical field of resource power prediction. Background Art

[0002] At present, the research on renewable energy power generation (wind power generation, photovoltaic power generation, etc.) is constantly deepening, and renewable energy power generation has been widely promoted and used in power grid operation applications. However, the uncertainty of its output power has also brought some operational problems to the power grid operation. In order to solve the scheduling problem caused by the uncertainty of the output power of the energy of power generation, it is necessary to effectively predict the power generation of power generation resources. There are many prediction methods for various types of energy, and the accuracy of the prediction results obtained by various methods is different. Therefore, it is necessary to use multiple indicators to select a full range of accuracy indicators, so as to more accurately and comprehensively evaluate the accuracy of the power generation prediction model and achieve model optimization.

[0003] The current method for evaluating the accuracy of the power generation prediction model mainly normalizes each indicator, determines the weight, obtains the evaluation value matrix based on the weight and the evaluation level membership of each indicator, and uses the matrix to evaluate the accuracy of the power prediction model. The invention patent document with publication number CN116167506A discloses an evaluation method for flexible resource power prediction. This method is based on reliability indicators and sensitivity indicators for comprehensive evaluation. This method can simplify complex calculation formulas and calculation processes and improve the simplicity of the evaluation system. However, this method does not consider the influencing factors of the power generation of new energy stations during the evaluation process, resulting in inaccurate calculation of the reliability of the prediction results. Summary of the invention

[0004] The purpose of the present invention is to provide a method for evaluating the accuracy of power generation efficiency prediction, so as to solve the problem of inaccurate reliability evaluation of prediction results.

[0005] To achieve the above object, the solution of the present invention includes:

[0006] A method for evaluating the accuracy of power generation efficiency prediction of the present invention comprises the following steps:

[0007] Obtaining a confidence level from a power generation prediction model, and reading probability prediction intervals of multiple adjacent time periods in the prediction model;

[0008] Determine whether there is a power outage and / or power restriction event within the probability prediction interval of each time period. If there is a power outage event, determine whether the probability prediction interval of each time period is valid according to the power outage duration. If it is valid, select the corresponding valid probability prediction interval as the probability prediction interval of the current time period. If it is invalid, determine the probability prediction interval of the next time period. If there is a power restriction event, select the valid probability prediction interval of the time period in the historical data as the probability prediction interval of the current time period.

[0009] The predicted values ​​in the power generation prediction model are read in time series, the probability prediction interval of the current time period is compared with the corresponding predicted values, the coverage probability of the predicted values ​​is obtained according to the comparison results, the reliability of the prediction results of the current time period is obtained using the confidence level and the coverage probability, and the prediction results are evaluated according to the reliability of the current prediction results.

[0010] Beneficial effect: The power generation efficiency prediction accuracy evaluation method of the present invention fully considers the impact of factors such as power outages and / or power restrictions on the power generation efficiency of new energy sites, and adds specific influencing factors when evaluating the accuracy of power generation efficiency prediction to eliminate the impact of force majeure or human factors. The reliability calculated by this method is more accurate, and the evaluation of the prediction results obtained by the prediction model is more accurate.

[0011] Furthermore, if there is a power outage and the duration of the power outage exceeds the set duration, the currently obtained probability prediction interval will be used as an invalid data segment; if there is a power outage and the duration of the power outage does not exceed the set duration, the valid probability prediction interval of the previous time period adjacent to the current time period will be used as the probability prediction interval of the current time period.

[0012] Beneficial effects: When there is a power outage event within the current probability prediction interval, it means that the data within the prediction interval is not accurate. When the power outage duration exceeds the set duration, the currently obtained probability prediction interval is used as an invalid data segment. If the power outage duration does not exceed the set duration, the valid probability prediction interval of the previous time period is used as the probability prediction interval of the current time period, thereby ensuring the accuracy of the reliability calculation.

[0013] Furthermore, the time period in the selected historical data is the same period as the current time period.

[0014] Furthermore, the reliability is:

[0015]

[0016] In the formula, R ACE is the reliability of the prediction results in the prediction model, P ICP is the coverage probability of the predicted value corresponding to the probability prediction interval of the current time period in the prediction model, For confidence.

[0017] Furthermore, the coverage probability of the predicted value in the prediction model is:

[0018]

[0019] Where P ICP is the coverage probability of the predicted value corresponding to the probability prediction interval of the current time period in the prediction model, and x is an indicative function. When the predicted value falls within the prediction interval, the function value is 1, otherwise the function value is 0. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a schematic diagram of the reliability evaluation process of power generation efficiency prediction in the method embodiment of the present invention. DETAILED DESCRIPTION

[0021] The present invention will be further described in detail below in conjunction with the accompanying drawings.

[0022] Example of the method for evaluating the accuracy of power generation efficiency prediction:

[0023] like Figure 1 The method for evaluating the accuracy of power generation efficiency prediction shown in the figure takes full account of the impact of power outages and power restrictions on power generation efficiency when reading the probability prediction interval to obtain an effective probability prediction interval, and uses the comparison between the prediction interval and the prediction value to calculate the reliability of the prediction result. The specific implementation method is as follows:

[0024] First, obtain the confidence level from the power generation prediction model The probability prediction intervals of multiple adjacent time periods in the prediction model are read, wherein the probability prediction intervals are read from the history library in a time series with a set duration, for example, the probability prediction intervals can be read every 15 minutes.

[0025] Secondly, in order to improve the accuracy of calculation reliability, this embodiment fully considers the influencing factors such as power outage events and / or power restriction events, and judges the validity of the probability prediction interval according to whether there are power outage events and / or power restriction events in the interval. Among them, power outage events include planned power outages, temporary power outages, and fault power outages, and power restriction events include planned power restriction and temporary power restriction, etc. The specific judgment method is: judge whether there are power outage events and / or power restriction events in the probability prediction interval of each time period obtained; if there is a power outage event, judge whether the probability prediction interval of each time period is valid according to the length of the power outage. If it is valid, select the corresponding valid probability prediction interval as the probability prediction interval of the current time period. If it is invalid, judge the probability prediction interval of the next time period; if there is a power restriction event, select the valid probability prediction interval of the time period in the historical data as the probability prediction interval of the current time period.

[0026] Among them, if there is a power outage event and the power outage duration exceeds the set duration, the probability prediction interval of the current time period will be used as an invalid data segment. If there is a power outage event and the power outage duration does not exceed the set duration, the valid probability prediction interval of the previous time period adjacent to the current time period will be used as the probability prediction interval of the current time period. The valid probability prediction area usually refers to the probability prediction interval where there is no power outage event. The set duration can be set to one hour or 0.5 hours. The set duration in this embodiment is one hour. As other implementation methods, the specific set duration can be determined according to the actual duration of the probability prediction interval. The invalid data segment (invalid probability prediction interval) in this embodiment no longer determines the prediction accuracy of the time period.

[0027] Among them, when there is a power outage event in the present embodiment, the effective probability prediction interval of the time period selected from the historical data can be an effective probability prediction interval of a time period that is concurrent with the current time period (for example, the same month, day, and time of last year), or it can be an effective probability prediction interval of a time period of the same period in the recent period. The specific time period can be selected according to actual conditions. To improve the accuracy of reliability calculation, the time period adopted in the present embodiment is a time period that is concurrent with the current time period (for example, the same month, day, and time of last year).

[0028] Finally, read the predicted value in the power generation prediction model according to the time series The probability prediction interval of the current time period is compared with the corresponding prediction value, and the comparison result is recorded as x. The coverage probability of the prediction value is obtained according to the comparison result. The reliability of the prediction result of the current time period is obtained using the confidence level and the coverage probability. The prediction result is evaluated according to the reliability of the current prediction result.

[0029] Specifically, the coverage probability of the predicted value in the prediction model is:

[0030]

[0031] Where P ICP is the coverage probability of the predicted value in the prediction model, x is an indicative function, when the predicted value falls within the prediction interval, the function value x is 1, otherwise the function value is 0.

[0032] The reliability of this forecast result is:

[0033]

[0034] In the formula, R ACE For the reliability of the prediction results, P ICP is the coverage probability of the predicted value in the prediction model, For confidence.

[0035] The above-mentioned power generation efficiency prediction accuracy evaluation method is mainly aimed at the reliability of the prediction results in the prediction model, and fully considers the impact of factors such as power outages and / or power restrictions on the power generation efficiency of new energy sites. When evaluating the accuracy of power generation efficiency prediction, specific influencing factors are added to eliminate the impact of force majeure or human factors. The reliability calculated by this method is more accurate, and the evaluation of the prediction results obtained by the prediction model is more accurate.

[0036] Specific implementation methods are given above, but the present invention is not limited to the described implementation methods. The basic idea of ​​the present invention lies in the above basic scheme. For ordinary technicians in this field, it does not take creative work to design various deformed models, formulas, and parameters according to the teachings of the present invention. Changes, modifications, substitutions, and variations of the implementation methods without departing from the principles and spirit of the present invention still fall within the scope of protection of the present invention.

Claims

1. A method for evaluating the accuracy of power generation efficiency prediction, characterized in that: The following steps are involved: Obtaining a confidence level from a power generation prediction model, and reading probability prediction intervals of multiple adjacent time periods in the prediction model; Determine whether there is a power outage and / or power restriction event within the probability prediction interval of each time period. If there is a power outage event, determine whether the probability prediction interval of each time period is valid according to the power outage duration. If it is valid, select the corresponding valid probability prediction interval as the probability prediction interval of the current time period. If it is invalid, determine the probability prediction interval of the next time period. If there is a power restriction event, select the valid probability prediction interval of the time period in the historical data as the probability prediction interval of the current time period. The predicted values ​​in the power generation prediction model are read in time series, the probability prediction interval of the current time period is compared with the corresponding predicted values, the coverage probability of the predicted values ​​is obtained according to the comparison results, the reliability of the prediction results of the current time period is obtained using the confidence level and the coverage probability, and the prediction results are evaluated according to the reliability of the current prediction results.

2. The method for evaluating the accuracy of power generation efficiency prediction according to claim 1, characterized in that: If there is a power outage and the duration of the power outage exceeds the set duration, the currently obtained probability prediction interval will be used as an invalid data segment. If there is a power outage and the duration of the power outage does not exceed the set duration, the valid probability prediction interval of the previous time period adjacent to the current time period will be used as the probability prediction interval of the current time period.

3. The method for evaluating the accuracy of power generation efficiency prediction according to claim 1, characterized in that: The time period of the selected historical data is the same period as the current time period.

4. The method for evaluating the accuracy of power generation efficiency prediction according to claim 1, characterized in that: Reliability is: In the formula, R ACE is the reliability of the prediction results in the prediction model, P ICp is the coverage probability of the predicted value corresponding to the probability prediction interval of the current time period in the prediction model, For confidence.

5. The method for evaluating the accuracy of power generation efficiency prediction according to claim 4, characterized in that: The coverage probability of the predicted value in this prediction model is: Where P ICP is the coverage probability of the predicted value corresponding to the probability prediction interval of the current time period in the prediction model, and x is an indicative function. When the predicted value falls within the prediction interval, the function value is 1, otherwise the function value is 0.

Images