Multi-factor integrated park new energy output prediction method

Abstract

The invention discloses a multi-factor integrated park new energy output prediction method, which comprises a first stage and a second stage, and is characterized in that the first stage is a new equipment installation and operation stage without historical data accumulation; in the second stage, multiple periods of historical data are accumulated for equipment operation; the new energy output prediction in the first stage is based on environmental factors, the new energy output prediction in the second stage is based on environmental factors, operation and maintenance factors, equipment factors and historical factors for output power prediction, and the environmental factors in the second stage are subjected to prediction calculation based on a prediction algorithm in the first stage. And other factors are integrated for result correction. According to the scheme, operation and maintenance factors, equipment factors and historical factors are introduced, the prediction model is comprehensively optimized, and the prediction accuracy and reliability are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of park output prediction, and in particular to a method for predicting park new energy output by integrating multiple factors. Background Art

[0002] Most of the current large-scale smart parks are equipped with new energy photovoltaic power generation systems. Insufficient output data prediction of new energy photovoltaic power generation systems usually leads to a series of problems. For example:

[0003] 1. Unstable power supply The output of photovoltaic power generation is affected by many factors such as weather and season. If the forecast is inaccurate, it may lead to unstable power supply, especially during peak demand periods or when photovoltaic power generation is unstable, which may not be able to meet the power demand of the park, thus affecting the operation of the park.

[0004] 2. Resource management is not optimized If the output of photovoltaic power generation cannot be accurately predicted, the park's energy management may result in resource waste or shortage.

[0005] 3. Increased operating costs Inaccurate forecasts may cause the park to take additional measures in scheduling and operation to compensate for the forecast errors, such as increasing backup generation capacity or conducting frequent electricity market transactions, which will increase operating costs.

[0006] 4. Economic benefits are damaged Inaccurate forecasts may affect the economic benefits of the park. For example, the inability to maximize the use of photovoltaic power generation or missing out on price fluctuations in the electricity market due to inaccurate forecasts will affect the economic benefits of the park.

[0007] 5. Difficulty in optimizing energy structure Smart parks are usually committed to achieving green and low-carbon goals. If the output data of photovoltaic power generation is not predicted enough, it will cause difficulties for the park in formulating and adjusting the energy structure strategy and fail to effectively integrate the use of renewable energy and traditional energy.

[0008] At present, the prediction mechanism of new energy output in smart parks mostly uses a single algorithm for calculation and prediction. For example, vector machines and neural network algorithms are used for data prediction. This mechanism has some shortcomings:

[0009] Algorithm limitations: Each single algorithm has its own unique assumptions and limitations. For example, support vector machines may require complex kernel functions when dealing with nonlinear problems, and although neural networks can handle complex nonlinear relationships, they may require a large amount of training data and computing resources. A single algorithm may not be able to fully capture the complex relationships of all influencing factors.

[0010] Insufficient data features: Different algorithms differ in the way they process data features. An algorithm may not be sensitive enough to certain features and cannot effectively capture important information in the data. If only a single-factor algorithm is used, all available feature information may not be fully utilized.

[0011] Lack of comprehensive perspective: The data sources for analysis are limited. Some consider lighting environment factors and power data, while others use output curves as the entry point for analysis. There is no unified multi-factor approach, and predictions are made in different situations and steps. Predictions can only be made from one perspective, and multiple factors cannot be considered comprehensively.

[0012] Poor adaptability: It can find relationships in specific situations, but it is not comprehensive enough. There are still optimization options at different stages of data sources and processing. For example, different data can be preprocessed differently and different algorithms can be used to find their relevance to the results.

[0013] Lack of interpretability: The prediction results of some algorithms (such as deep learning models) may lack interpretability, making it difficult to understand and analyze the prediction results. Summary of the invention

[0014] In order to solve the above technical problems, the present invention provides a method for predicting the output of new energy in a park by integrating multiple factors.

[0015] The present invention is implemented by adopting the following technical scheme: a multi-factor integrated park new energy output prediction method, including a first stage and a second stage, the first stage is the new installation and operation stage of the equipment, and there is no historical data accumulation; the second stage is the equipment operation accumulation of multiple cycles of historical data; the new energy output prediction of the first stage is based on environmental factors, and the new energy output prediction of the second stage is based on environmental factors, operation and maintenance factors, equipment factors and historical factors to predict the output power, the environmental factors of the second stage are predicted and calculated based on the prediction algorithm of the first stage, and other factors are integrated to correct the results.

[0016] Specifically, the new energy output forecasting of the first stage includes the following steps: Step A1: Collect environmental factors and photovoltaic power data according to preset time intervals, integrate them into a data matrix, and initialize the model; Step A2: Use the decision tree to perform n iterations of training, and finally obtain a prediction formula based on residual correction. Calculate the residuals of each round. The round residual calculation formula is: ; in, is the output of the model initialization, ; is the number of iterations, which is equal to the number of model initialization iteration data; Step A3: Use the residuals as training decision tree , the leaf node output of the tree is , the calculation formula is: ; in, It is the influencing factor that determines R; is the jth tree in the feature The output value on ; Step A4: Use the prediction values ​​of j trees to output the prediction values ​​of the j+1th tree to update the model, and repeat the training until the number of iterations stops. The model update formula is: ; Step A5: The final prediction formula is: ; ; in, is the power prediction value at time i, is the initialization model output value, is the result value outputted by the environmental factor C on the jth tree, is the residual influence coefficient.

[0017] Specifically, each row of the data matrix integrated in step A1 is the value of the factor and the result at a certain time t, expressed as: ={ , , , , , , , , }; The integration matrix Z is expressed as: .

[0018] Specifically, the model initialization formula is: ; is the initialization model prediction value, and the model is adjusted after iteration. n is the number of iterative data. It is actual output of strength.

[0019] Specifically, the second phase of new energy output forecasting includes the following steps: Step B1: derive a prediction model based on the first-stage algorithm; Step B2: Query the equipment damage status to calculate the damage rate, calculate the degradation rate based on the operating time, and calculate the factors affecting dust accumulation; Step B3: The model integrates multiple factors and inputs environmental factors to obtain prediction data; Step B4: Query historical data of similar environmental factors and use the weighted method to correct the prediction results.

[0020] Specifically, the step B3 integrates multiple factors, and the specific formula is: ; in, is the power prediction value at time i, is the initialization model output value, is the residual influence coefficient, is the result value outputted by the environmental factor C on the jth tree, represents the annual degradation rate of the equipment, is an important factor controlling the degradation rate. represents the dust accumulation impact rate, It is an important factor controlling the degradation rate.

[0021] Specifically, the step B4 specifically includes: correcting the prediction result by combining the output data of historical factors with similar environmental factors, performing weighted result correction with environmental factors of 5% similarity, and performing weakly weighted result correction with environmental factors of 10% similarity, and the correction formula is: ; in, is the weighting coefficient, is the predicted value, is the historical average value obtained by similar factor query, is the adjusted result.

[0022] The beneficial effects of the present invention are: 1. Improved prediction accuracy: Introducing operation and maintenance factors and equipment factors can make the prediction model more accurately reflect the changes in actual photovoltaic power generation. For example, considering the power loss that may be caused by equipment degradation and failure, the prediction results can be adjusted more finely, thereby improving the accuracy of the prediction. In addition, by weighting and correcting historical data, the long-term impact of changes in environmental factors can be better handled.

[0023] 2. Improved operation and maintenance efficiency: By accurately predicting power generation and timely understanding of equipment status, the operation and maintenance team can optimize maintenance plans and reduce the frequency of sudden failures.

[0024] 3. Enhanced decision support: The optimized model can provide more reliable data support to help management make more informed decisions, including equipment updates, resource allocation, and emergency plans.

[0025] 4. Improve system stability: Through more accurate prediction of photovoltaic power generation, grid resources can be better allocated, supply and demand can be balanced, and the stability and reliability of the entire power system can be improved.

[0026] 5. Improved customer satisfaction: For photovoltaic power generation manufacturers or service providers, providing accurate power generation forecasts and efficient operation and maintenance services can improve customer satisfaction and trust and enhance market competitiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. 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 the structures shown in these drawings without paying creative work.

[0028] Figure 1 This is a flowchart of new energy output prediction in the first stage in an embodiment of the present invention; Figure 2 This is the new energy output prediction for the second stage in the embodiment of the present invention. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0030] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0031] The following is combined with Figure 1-2 , some embodiments of the present invention are described in detail. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0032] The present invention proposes a multi-factor integrated park new energy output prediction method, including a first stage and a second stage, the first stage is the new installation and operation stage of the equipment, and there is no historical data accumulation; the second stage is the equipment operation accumulation of multiple cycles of historical data; the new energy output prediction of the first stage is based on environmental factors, and the new energy output prediction of the second stage is based on environmental factors, operation and maintenance factors, equipment factors and historical factors to predict the output power, the environmental factors of the second stage are predicted and calculated based on the prediction algorithm of the first stage, and other factors are integrated to correct the results.

[0033] In a preferred embodiment, the first phase of new energy output prediction includes the following steps: Step A1: Collect environmental factors and photovoltaic power data according to preset time intervals, integrate them into a data matrix, and initialize the model; Step A2: Use the decision tree to perform n iterations of training, and finally obtain a prediction formula based on residual correction. Calculate the residuals of each round. The round residual calculation formula is: ; in, is the output of the model initialization, ; is the number of iterations, which is equal to the number of model initialization iteration data; Step A3: Use the residuals as training decision tree , the leaf node output of the tree is , the calculation formula is: ; in, It is the influencing factor that determines R; is the jth tree in the feature The output value on ; Step A4: Use the prediction values ​​of j trees to output the prediction values ​​of the j+1th tree to update the model, and repeat the training until the number of iterations stops. The model update formula is: ; Step A5: The final prediction formula is: ; ; in, is the power prediction value at time i, is the initialization model output value, is the result value outputted by the environmental factor C on the jth tree, is the residual influence coefficient.

[0034] In this embodiment, each row of the data matrix integrated in step A1 is the value of the factor and the result at a certain time t, expressed as: ={ , , , , , , , , }; The integration matrix Z is expressed as: .

[0035] In this embodiment, the model initialization formula is: ; is the initialization model prediction value, and the model is adjusted after iteration. n is the number of iterative data. It is actual output of strength.

[0036] In this embodiment, the new energy output forecast for the second stage is as follows: Figure 2 As shown, the following steps are included: Step B1: derive a prediction model based on the first-stage algorithm; Step B2: Query the equipment damage status to calculate the damage rate, calculate the degradation rate based on the operating time, and calculate the factors affecting dust accumulation; Step B3: The model integrates multiple factors and inputs environmental factors to obtain prediction data; Step B4: Query historical data of similar environmental factors and use the weighted method to correct the prediction results.

[0037] In this embodiment, step B3 integrates multiple factors, and the specific formula is: ; in, is the power prediction value at time i, is the initialization model output value, is the residual influence coefficient, is the result value outputted by the environmental factor C on the jth tree, represents the annual degradation rate of the equipment, is an important factor controlling the degradation rate. represents the dust accumulation impact rate, It is an important factor controlling the degradation rate.

[0038] In this embodiment, step B4 specifically includes: correcting the prediction result by combining the output data of historical factors with similar environmental factors, performing weighted result correction with environmental factors with 5% similarity, and performing weakly weighted result correction with 10% similarity. The correction formula is: ; in, is the weighting coefficient, is the predicted value, is the historical average value obtained by similar factor query, is the adjusted result.

[0039] In another embodiment, the prediction of new energy output in a smart park refers to the prediction of photovoltaic power generation. There are several factors that affect photovoltaic power generation in a smart park: (1) Environmental factors: time, sunshine, radiation intensity, wind speed, temperature, humidity, air pressure, cloud cover, etc. (2) Equipment factors: equipment degradation, equipment failure rate (3) Operation and maintenance factors: equipment surface cleanliness, regular line inspection and maintenance (4) Historical output data. The integrated multi-factor prediction is based on the above multi-factor data to predict the output of new energy.

[0040] The output forecast of new energy is divided into several stages. The first stage is the new installation and operation stage. At this stage, there is no complete accumulation of historical data. At this time, the output forecast algorithm under environmental factors can be considered. In the second stage, the new energy power generation system has been in operation for many years and has accumulated data for several complete cycles. In this stage, equipment factors, environmental factors, operation and maintenance factors, and historical factors can be considered for forecasting.

[0041] The first phase of the forecast of renewable energy output based on environmental factors is as follows: Figure 1 As shown: Collect environmental factors and photovoltaic power data, the time interval can be 15 minutes or 30 minutes. Assume that the collected data is air pressure C1, humidity C2, cloud C3, wind speed C4, wind direction C5, temperature C6, irradiation intensity C7, precipitation C8, power P. Combine the data at different times and integrate the data to get a factor and power matrix Z. Each row of Z is the value of the factor and result at a certain time t. ={ , , , , , , , , }

[0042] There are n moments of data, integrated into a matrix with row elements , , , , , , , , Arrangement method to get Z: Initialize the model: P is the initialization model prediction value, and the model is adjusted after iteration. n is the number of iterative data. It is actual output of strength.

[0043] Use decision trees for training, and finally get a prediction formula based on residual correction. The number of iterations is n, and the residual of each round is calculated. The residual is the difference between the actual value and the predicted value of the current model. For the jth round,

[0044] Use the residuals as training decision trees , the leaf node output of the tree is : in It is the factor that determines R. is the jth tree in the feature The output value on .

[0045] Update the model and use the prediction values ​​of j trees to output the prediction value of the j+1th tree Repeat the training until the number of iterations stops.

[0046] The final prediction formula is: is the power prediction value at time i, is the initialization model output value, is the result value outputted by the environmental factor C on the jth tree, is the residual influence coefficient.

[0047] The second phase of new energy output forecasting mainly adopts the following strategies: Output power forecasting based on environmental factors, operation and maintenance factors, equipment factors, and historical factors.

[0048] Environmental factors are predicted and calculated according to the prediction algorithm in the first stage, and other factors are integrated to correct the results.

[0049] Equipment factors describe equipment degradation factors that have a direct impact on output power.

[0050] In one embodiment, based on the characteristics of photovoltaic panels, the degradation rate is relatively fast in the first five years, generally 1% to 3%, and then tends to be stable, with an annual degradation rate of 0.3% to 0.8%. Here, the range of the degradation factor T is:

[0051] In the first five years, that is, when n<=5, the degradation rate of that year is =a+(n−1) , where a is the initial degradation rate and d is the annual incremental degradation rate.

[0052] When n>5, the annual degradation rate is .

[0053] In addition, consider the equipment damage rate factor percentage K. The damage rate factor is obtained through real-time monitoring of the equipment.

[0054] The primary consideration for operation and maintenance is the cleaning cycle. In this embodiment: a. Assuming the dust cleaning cycle is M years, the peak impact is A, and mod is the remainder operation, then the calculation formula for the dust accumulation impact factor is: J(x)=A(xmodM) / M.

[0055] b. Factors affecting the operation and maintenance cycle. Consider the non-working conditions of the day. The output forecast on non-working days can be analyzed and avoided according to the operation and maintenance schedule to avoid affecting the total output value.

[0056] Model integrating multiple factors: in: is the power prediction value at time i, is the initialization model output value, is the residual influence coefficient, is the result value outputted by the environmental factor C on the jth tree, represents the annual degradation rate of the equipment, α is an important factor controlling the degradation rate, represents the dust accumulation influence rate, and β is an important factor controlling measure of the degradation rate.

[0057] Historical factors are combined with output data of similar environmental factors to correct the prediction results.

[0058] Combined with historical data, the similarity based on important environmental factors can be fully considered. Environmental factors with a similarity of 5% are used for weighted result correction, and those with a similarity of 10% are used for weakly weighted result correction.

[0059] in is the weighting coefficient, is the predicted value, is the historical average value obtained by similar factor query, is the adjusted result.

[0060] In this embodiment, the simulation data, the environmental factor data and the output power data are integrated to obtain an initial data matrix represented as: Its form is shown in Table 1 below: Table 1 Data matrix

[0061] The integrated data is used to train the decision tree to obtain the first-stage model and important influencing factors. The resulting model is as follows:

[0062] is the power prediction value at time i, is the initialization model output value, is the result value outputted by the environmental factor C on the jth tree, is the residual influence coefficient.

[0063] During this stage, important influencing factors will be obtained while the model is being trained. Some of the more prominent factors are listed here: week (time unit is week) 1380.0, relative humidity 1196.0, air pressure 1153.0, precipitation 1145.0, radiation intensity 1105.0, temperature 1091.5, etc.

[0064] Then in the second stage, when drawing on historical factors, it is necessary to focus on comparing the deviations of the above influencing factors in order to find the historical values ​​corresponding to the factors.

[0065] Use the model to predict the results. By inputting the environmental influencing factor set C, the predicted result value is obtained. Multiple Cs will result in multiple prediction results. The form is shown in Table 2 below:

[0066] Table 2 Prediction results

[0067] After the predicted value is obtained, the actual predicted value is obtained through coefficient conversion. P is the time division of each day, and each day is divided into 96 time periods.

[0068] If it is the second stage of prediction, it is necessary to integrate historical record factors, operation and maintenance factors, and equipment factors. Then the prediction result Z needs to be put into the following formula:

[0069] The coefficients are adjusted based on the test set data of the year to achieve the best prediction effect. represents the annual degradation rate of the equipment, α is an important factor controlling the degradation rate, represents the dust accumulation influence rate, β is an important factor control measure of the degradation rate, and K is the equipment damage rate.

[0070] The last step, after obtaining the above results, is to find similar historical values ​​of the important influencing factors in step (1), such as the average of 10 groups of power values, and use the historical similar factor result values ​​to adjust the current prediction results. Use the following formula to calibrate the results.

[0071] in is the weighting coefficient, is the predicted value, The historical averages corresponding to similar factors, is the adjusted result.

[0072] final This is the more accurate prediction result we want.

[0073] In this embodiment, the preparation work required is: connecting the photovoltaic operation monitoring system and the meter reading system to obtain data, obtain specific operation data and meteorological environment data, and accumulate at least one year of monitoring data.

[0074] In this embodiment, the algorithm call design and development are as follows: 1Implement the algorithm based on Python and publish the calling interface through http.

[0075] 2. Automatically generate forecast tasks based on the task scheduling framework. Before generating tasks, refer to the operation and maintenance plan for the forecast period to see if there are any operations such as shutdown and maintenance that affect large-scale output.

[0076] 3 The prediction task begins. Each prediction task corresponds to photovoltaic equipment and output points in different regions.

[0077] Query the weather forecast data for the forecast date and divide the time period into 15-minute or 30-minute intervals.

[0078] Use data from different time periods to input into the forecasting model for forecasting.

[0079] Through important environmental factors (similarity 5% or 10%), query and obtain the last 10 historical data, calculate the average value, and use the weighted average method to calibrate the prediction results.

[0080] Display the prediction results to the platform.

[0081] 4. Maintenance personnel can continuously compare the errors between the data in the model and the actual values, and constantly adjust the iteration parameters and weight coefficients in the algorithm to improve the accuracy of the prediction results.

[0082] System Integration and Testing: Data integration, integrating algorithm modules into photovoltaic monitoring or energy management systems.

[0083] Functional testing tests the various functions of the system to ensure the correct calling of data processing and data prediction.

[0084] Invite actual users to use the system, collect feedback on actual values, and continuously adjust the algorithm.

[0085] Application effect evaluation: Using the only experimental data, the root mean square error is used to test the prediction result error. By increasing iterations and data cross-validation, the training data can be within 5% and the prediction error can be within 14%. In actual use, as the factor data and result data continue to accumulate, the error will further decrease.

[0086] Based on the prediction based on environmental factors in the first phase, the solution comprehensively optimizes the prediction model by introducing operation and maintenance factors, equipment factors and historical factors. This multi-factor comprehensive modeling method can more comprehensively consider various factors that affect photovoltaic power generation, rather than a single-dimensional prediction. This method can effectively improve the accuracy and reliability of the prediction.

[0087] Integration of dynamic operation and maintenance factors: Introducing operation and maintenance factors (such as dust cleaning and shutdown maintenance) into the prediction model. This approach can incorporate the impact of actual operation and maintenance into the prediction, so that the model can reflect the power generation capacity of the equipment under different maintenance conditions. This not only enhances the practicality of the model, but also improves the efficiency of operation and maintenance management.

[0088] Equipment degradation and failure rate modeling: By considering equipment degradation and failure factors, it is possible to better simulate the change of equipment performance over time. This dynamic modeling can help predict the long-term performance of equipment and provide a basis for maintenance decisions, which has great application value in practice.

[0089] Weighted correction of historical data: This is a new data fusion method that uses historical data of similar environmental factors to perform weighted correction on current test results. By combining historical data with current data, the model can more accurately handle the long-term trends and variations of environmental factors, thereby improving the robustness and accuracy of predictions.

[0090] Optimization application of decision tree: Based on decision tree and residual correction, the first-stage environmental factor prediction model is generated. On this basis, other factor algorithms are enhanced, which reflects the adaptability and scalability of the algorithm in complex and dynamic systems. By improving the original model, it can better adapt to the actual situation of photovoltaic power generation system and improve the overall performance of the model.

[0091] Closed loop of comprehensive prediction and actual operation: Closed loop optimization of prediction results and actual operation (such as operation and maintenance and equipment management) makes the prediction not only a theoretical output, but can be directly applied to actual operation. This closed loop mechanism can adjust the prediction model and operation and maintenance strategy in real time to achieve more efficient energy management and resource utilization

[0092] For the aforementioned embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the order of the actions described, because according to the present application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are preferred embodiments, and the actions involved are not necessarily required by the present application.

[0093] The above embodiments describe the basic principles and main features of the present invention and the advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the changes and modifications made by those skilled in the art shall be within the scope of protection of the appended claims of the present invention without departing from the spirit and scope of the present invention.

Claims

1. A multi-factor integrated park new energy output prediction method, characterized in that: It includes the first stage and the second stage. The first stage is the new installation and operation stage of the equipment, and there is no historical data accumulation; the second stage is the accumulation of historical data of multiple cycles of equipment operation; the new energy output prediction of the first stage is based on environmental factors, and the new energy output prediction of the second stage is based on environmental factors, operation and maintenance factors, equipment factors and historical factors to predict the output power. The environmental factors in the second stage are predicted and calculated based on the prediction algorithm of the first stage, and other factors are integrated to correct the results.

2. A multi-factor integrated park new energy output prediction method as claimed in claim 1, characterized in that: The first phase of new energy output forecasting includes the following steps: Step A1: Collect environmental factors and photovoltaic power data according to preset time intervals, integrate them into a data matrix, and initialize the model; Step A2: Use the decision tree to perform n iterations of training, and finally obtain a prediction formula based on residual correction. Calculate the residuals of each round. The round residual calculation formula is: ； in, is the output of the model initialization, ; is the number of iterations, which is equal to the number of model initialization iteration data; Step A3: Use the residuals as training decision tree , the leaf node output of the tree is , the calculation formula is: ； in, It is the influencing factor that determines R; is the jth tree in the feature The output value on ; Step A4: Use the prediction values ​​of j trees to output the prediction values ​​of the j+1th tree to update the model, and repeat the training until the number of iterations stops. The model update formula is: ； Step A5: The final prediction formula is: ； ； in, is the power prediction value at time i, is the initialization model output value, is the result value outputted by the environmental factor C on the jth tree, is the residual influence coefficient.

3. A multi-factor integrated park new energy output prediction method as claimed in claim 2, characterized in that: Each row of the data matrix integrated in step A1 is the value of the factor and result at a certain time t, expressed as: ; The integration matrix Z is expressed as: 。 4. A multi-factor integrated park new energy output prediction method as claimed in claim 3, characterized in that: The model initialization formula is: ； is the initialization model prediction value, and the model is adjusted after iteration. n is the number of iterative data. It is actual output of strength.

5. A multi-factor integrated park new energy output prediction method as claimed in claim 4, characterized in that: The second phase of new energy output forecasting includes the following steps: Step B1: derive a prediction model based on the first-stage algorithm; Step B2: Query the equipment damage status to calculate the damage rate, calculate the degradation rate based on the operating time, and calculate the factors affecting dust accumulation; Step B3: The model integrates multiple factors and inputs environmental factors to obtain prediction data; Step B4: Query historical data of similar environmental factors and use the weighted method to correct the prediction results.

6. A multi-factor integrated park new energy output prediction method as claimed in claim 5, characterized in that: The step B3 integrates multiple factors, and the specific formula is: ； in, is the power prediction value at time i, is the initialization model output value, is the residual influence coefficient, is the result value outputted by the environmental factor C on the jth tree, represents the annual degradation rate of the equipment, is an important factor controlling the degradation rate. represents the dust accumulation impact rate, It is an important factor controlling the degradation rate.

7. A multi-factor integrated park new energy output prediction method as claimed in claim 6, characterized in that: The step B4 specifically includes: correcting the prediction result by combining the output data of historical factors with similar environmental factors, performing weighted result correction with environmental factors of 5% similarity, and performing weakly weighted result correction with environmental factors of 10% similarity, and the correction formula is: ； in, is the weighting coefficient, is the predicted value, is the historical average value obtained by similar factor query, is the adjusted result.

Images