Distributed photovoltaic power generation prediction method and system based on edge calculation

Abstract

The invention provides a distributed photovoltaic power generation prediction method and system based on edge calculation, and belongs to the technical field of photovoltaic power generation, and the method comprises the steps: collecting the historical power generation data and real-time environment parameter data of each distributed photovoltaic station in a transformer area, and carrying out the preprocessing of the collected data; inputting the preprocessed real-time environment parameter data into the basic prediction model of the station to obtain predicted generating capacity, and forming a generating capacity prediction model by the basic prediction model and a PID control algorithm model; pID parameters of the PID control algorithm model are optimized and adjusted according to comparative analysis of the predicted generating capacity and the real-time generating capacity, and then the generating capacity prediction model is optimized; and based on the optimized generating capacity prediction model, generating a predicted value of the photovoltaic generating capacity of the station, outputting a prediction result, and deducing the theoretical photovoltaic generating capacity of the transformer area according to the prediction result of each distributed photovoltaic station.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power generation, and in particular to a distributed photovoltaic power generation prediction method and system based on edge computing. Background Art

[0002] With the transformation of the global energy structure and the enhancement of environmental protection awareness, the development and utilization of renewable energy has become particularly important. Among various new energy sources, green new energy, led by photovoltaics, has attracted more and more attention due to its advantages of cleanliness, efficiency, and sustainability. However, in the actual operation of photovoltaic power generation systems, the power generation capacity is affected by many factors, including but not limited to light intensity, temperature and humidity, geographical location, equipment hardware aging, line aging, etc. It is precisely because of these factors that the photovoltaic power generation capacity has become significantly uncertain.

[0003] To cope with this uncertainty, researchers and engineers have developed a variety of photovoltaic power generation forecasting methods. Most of these methods rely on centralized computing resources, collect a large amount of historical data and real-time data, and use complex mathematical models and algorithms to make predictions. However, centralized computing methods have some inherent problems, such as poor real-time performance and high latency. These problems are particularly prominent in distributed photovoltaic systems, because distributed systems require fast and high-precision forecasts to optimize the operation of photovoltaic power generation systems to ensure the stability and reliability of the system.

[0004] Prior art document CN119046782A discloses a distributed photovoltaic power generation prediction method and system, the method comprising: obtaining an original power generation data sequence; preprocessing the original power generation data sequence, decomposing the original power generation data sequence by the principle of empirical mode decomposition, and obtaining several subsequences; inputting several subsequences into a recurrent neural network, and performing convolution operations on several subsequences by introducing a causal convolution method, and obtaining the prediction components of each subsequence; aggregating the prediction components, introducing a dot product attention mechanism and using an activation function as a network connection method to control the aggregation degree of different prediction components and obtain the prediction value; regulating the output power based on the prediction value, and setting different output strategies. Its shortcomings are that the collection timeliness of the original power generation data required in the existing invention is not high, the algorithm structure is complex, and it is difficult to implement. Summary of the invention

[0005] In order to solve the problems existing in the prior art, the present invention provides a distributed photovoltaic power generation prediction method and system based on edge computing.

[0006] The present invention adopts the following technical solution.

[0007] A first aspect of the present invention provides a distributed photovoltaic power generation prediction method based on edge computing, comprising the following steps:

[0008] Collect historical power generation data and real-time environmental parameter data of each distributed photovoltaic site in the substation area, and pre-process the collected data;

[0009] The pre-processed real-time environmental parameter data is input into the basic prediction model of the site to obtain the predicted power generation, and then the power generation prediction model is composed of the basic prediction model and the PID control algorithm model;

[0010] According to the comparative analysis of the predicted power generation and the real-time power generation, the PID parameters of the PID control algorithm model are optimized and adjusted, thereby optimizing the power generation prediction model;

[0011] Based on the optimized power generation prediction model, the predicted value of the photovoltaic power generation of the site is generated, and the prediction result is output. According to the prediction result of each distributed photovoltaic site, the theoretical photovoltaic power generation of the substation is derived.

[0012] Optionally, the historical power generation data of the distributed photovoltaic site includes environmental factors and actual power generation.

[0013] Optionally, the real-time environmental parameters of the distributed photovoltaic site include solar radiation intensity, air temperature, humidity, wind speed and cloud changes.

[0014] Optionally, the preprocessing of the collected data includes data cleaning, missing value filling and format conversion.

[0015] Optionally, the power generation prediction model includes:

[0016] P(t)=p(t)+u(t)

[0017] Where P(t) represents the power generation prediction model, p(t) represents the basic prediction model, which calculates the initial predicted power generation based on the actual environmental parameters; u(t) represents the PID control algorithm model, which calculates the correction value through the PID algorithm and is used to adjust the initial predicted power generation.

[0018] Optionally, the power generation prediction model further includes dynamically adjusting PID parameters of the PID control algorithm model according to the real-time prediction error:

[0019]

[0020] K p (t) = K p ·(1+α·|e(t))

[0021]

[0022] In the formula, α, β, and γ are adjustment coefficients, which are used to adjust the sensitivity of the proportional, integral, and differential terms respectively;

[0023] |e(t)| is the absolute value of the error, reflecting the current deviation of the system;

[0024] is the cumulative value of the error, reflecting the long-term deviation of the system;

[0025] is the error change rate, reflecting the dynamic changes of the system;

[0026] K p is the proportional gain, K i is the integral gain, K d is the differential gain, e(t) is the prediction error, is the predicted value of the power generation of the photovoltaic power generation system, which is predicted by the basic prediction model, y(t) is the actual value of the power generation of the photovoltaic power generation system, and t is the time variable; the final prediction result of the power generation prediction is:

[0027]

[0028] Optionally, the range of the prediction error is ±3%.

[0029] A second aspect of the present invention provides a distributed photovoltaic power generation prediction system based on edge computing. Based on the distributed photovoltaic power generation prediction system based on edge computing described in the first aspect of the present invention, the system includes:

[0030] Data acquisition module, data processing module, PID controller module, prediction model module and user interaction interface;

[0031] The data acquisition module is used to collect the historical power generation data and real-time environmental parameter data of the photovoltaic power generation system. The data processing module is used to pre-process the data collected by the data collection module and transmit it to the prediction model module. The prediction model module is used to calculate the predicted power generation. The PID controller module optimizes and adjusts the PID parameters according to the analysis of the predicted power generation and real-time power generation, and optimizes and adjusts the power generation prediction in the prediction model module. The user interaction interface is used to display the prediction results.

[0032] Optionally, the data acquisition module includes temperature, humidity and light intensity sensors, which are respectively installed at each photovoltaic site to collect environmental parameter data in real time; the data acquisition module also includes an inverter for monitoring electrical parameters of the photovoltaic system.

[0033] Optionally, the user interaction interface supports the user to manually input environmental parameters to generate predicted power generation and display it on the user interaction interface.

[0034] Compared with the prior art, the beneficial effects of the present invention include at least: the algorithm structure is simple and easy to implement, saving costs; edge deployment ensures high efficiency and high accuracy of data collection, and improves the model prediction accuracy; the method is highly adaptable and can predict power generation for different regions and different climatic conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a schematic diagram of the structural model of the system of the present invention. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. The described embodiments are only embodiments of a part of the present invention, not all embodiments. Based on the spirit of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the protection scope of the present invention.

[0037] The present invention provides a distributed photovoltaic power generation prediction method based on edge computing in embodiment 1, comprising the following steps:

[0038] Step 1: Collect historical power generation data and real-time environmental parameter data of each distributed photovoltaic site in the substation area, and pre-process the collected data;

[0039] In a further preferred but non-limiting embodiment, the historical power generation data of the distributed photovoltaic site includes environmental factors and actual power generation, wherein the environmental factors include sunshine intensity, temperature and humidity.

[0040] In a further preferred but non-limiting embodiment, the real-time environmental parameters of the distributed photovoltaic site include solar radiation intensity, air temperature, humidity, wind speed and cloud changes.

[0041] In a further preferred but non-limiting embodiment, the preprocessing of the collected data includes data cleaning, missing value filling and format conversion.

[0042] Step 2, input the preprocessed real-time environmental parameter data into the basic prediction model of the site to obtain the predicted power generation, and then form a power generation prediction model by combining the basic prediction model and the PID control algorithm model;

[0043] In a further preferred but non-limiting embodiment, the PID control algorithm model includes a proportional part P, an integral part I and a differential part D, which are respectively used to adjust the prediction error, the accumulated error and the rate of change.

[0044] Step 3, based on the comparative analysis of the predicted power generation and the real-time power generation, the PID parameters of the PID control algorithm model are optimized and adjusted, thereby optimizing the power generation prediction model;

[0045] In a further preferred but non-limiting embodiment, step 3 specifically comprises:

[0046] According to the current PID parameters and environmental parameters, the corresponding theoretical power generation is output. Comparing the theoretical power generation with the actual power generation can generate a new relative coefficient P. By defining the error range, the error range is the comparison between the predicted power generation and the actual power generation, and the error range is ±3%. After repeated iterations and updates, a relatively stable coefficient P can be finally obtained. When the parameter P is stable, a certain period of time is selected to compare the difference between the theoretical power generation and the actual power generation again, and the integral error between the two is further calculated and extracted. The integral coefficient I is used for correction, and the differential coefficient D is used to correct the differential error. Finally, the predicted theoretical power generation is infinitely close to the actual power generation, forming a relatively stable and accurate set of P, I, and D parameters.

[0047] In a further preferred but non-limiting embodiment, the power generation prediction model includes:

[0048] P(t)=p(t)+u(t)

[0049] Where P(t) represents the power generation prediction model, p(t) represents the basic prediction model, which calculates the initial predicted power generation based on the actual environmental parameters; u(t) represents the PID control algorithm model, which calculates the correction value through the PID algorithm and is used to adjust the initial predicted power generation.

[0050] In a further preferred but non-limiting embodiment, the PID parameters of the PID control algorithm model are dynamically adjusted according to the real-time prediction error:

[0051]

[0052] K p (t) = K p ·(1+α·|e(t))

[0053]

[0054] In the formula, α, β, and γ are adjustment coefficients, which are used to adjust the sensitivity of the proportional, integral, and differential terms respectively;

[0055] |e(t)| is the absolute value of the error, reflecting the current deviation of the system;

[0056] is the cumulative value of the error, reflecting the long-term deviation of the system;

[0057] is the error change rate, reflecting the dynamic changes of the system;

[0058] K p is the proportional gain, K i is the integral gain, K d is the differential gain, e(t) is the prediction error, is the predicted value of the power generation of the photovoltaic power generation system, which is predicted by the basic prediction model, y(t) is the actual value of the power generation of the photovoltaic power generation system, and t is the time variable; the final prediction result of the power generation prediction is:

[0059]

[0060] Specifically, the basic prediction model is a model based on physical laws or an empirical model combined with historical data.

[0061] Step 4: Based on the optimized power generation prediction model, generate the predicted value of the photovoltaic power generation of the site, output the prediction result, and derive the theoretical photovoltaic power generation of the area according to the prediction result of each distributed photovoltaic site.

[0062] Specifically, the accuracy of the predicted power generation of each photovoltaic site is first ensured, and then the predicted values ​​are added up to obtain the total predicted power generation of the area or even the cluster.

[0063] The present invention provides a distributed photovoltaic power generation prediction system based on edge computing in embodiment 2. Based on the distributed photovoltaic power generation prediction system based on edge computing described in embodiment 1 of the present invention, the system includes:

[0064] Data acquisition module, data processing module, PID controller module, prediction model module and user interaction interface;

[0065] The data acquisition module is used to collect the historical power generation data and real-time environmental parameter data of the photovoltaic power generation system. The data processing module is used to pre-process the data collected by the data collection module and transmit it to the prediction model module. The prediction model module is used to calculate the predicted power generation. The PID controller module optimizes and adjusts the PID parameters according to the analysis of the predicted power generation and real-time power generation, and optimizes and adjusts the power generation prediction in the prediction model module. The user interaction interface is used to display the prediction results.

[0066] In a further preferred but non-limiting embodiment, the data acquisition module includes temperature, humidity and light intensity sensors, which are installed at each photovoltaic site respectively to collect environmental parameter data in real time; the data acquisition module also includes an inverter for monitoring the electrical parameters of the photovoltaic system.

[0067] The prediction model module optimizes the photovoltaic power generation prediction model in the edge computing module according to the power generation parameters output by the photovoltaic power generation system after optimization by the PID controller module, and achieves the purpose of accurately predicting the photovoltaic power generation of each site through continuous self-optimization. Specifically, the distributed photovoltaic site model and site PID parameters can form a distributed photovoltaic area power generation prediction model, thereby achieving high-precision prediction of distributed photovoltaic power generation in the area or even the cluster.

[0068] Exemplarily, the prediction model module constructs a 24-hour dynamic model based on meteorological data. By inputting corresponding environmental parameters such as light, temperature, humidity, and the PID control parameters of each photovoltaic component in different time periods, the theoretical power generation of a single photovoltaic power generation system under specific meteorological conditions can be accurately calculated. On this basis, the accurate theoretical photovoltaic power generation of the substation or even the cluster can be derived.

[0069] Optionally, the user interaction interface supports the user to manually input environmental parameters to generate predicted power generation and display it on the user interaction interface.

[0070] Therefore, in order to meet the needs of distributed photovoltaic systems for fast and high-precision prediction, the present invention has developed a new prediction method. This new method should be able to overcome the limitations of centralized computing and have higher real-time performance and lower latency. At the same time, the new method should also have intelligent characteristics, be able to autonomously learn and adapt to various complex environmental changes, thereby improving the overall performance and reliability of photovoltaic power generation systems. In this way, photovoltaic power generation systems will be able to better integrate into modern power systems and make greater contributions to achieving sustainable energy goals.

[0071] In summary, the present invention provides a distributed photovoltaic power generation prediction method and system based on edge computing. This method can efficiently predict photovoltaic power generation with high accuracy and strong reliability by performing high-frequency real-time data collection and computational processing on the edge side. The present invention introduces a standard PID control algorithm on the edge side, and effectively reduces the delay in data transmission and improves the acquisition accuracy of the system's original data by performing high-speed data collection and processing near the data collection point. A relative electric energy reference can be obtained by integrating the collected light intensity over time, and a relative coefficient P can be obtained by comparing the actual electric energy.

[0072]

[0073] Among them, the actual power generation is the actual monitored power generation, and the reference power is the preliminary predicted value based on the influencing factors. By using high-density collection and calculation, the coefficient P is continuously optimized by comparing the reference power with the actual power generation. On this basis, by continuously comparing the curve difference between the reference power and the actual power generation, deep learning is used to calculate relatively accurate I and D coefficients, and a photovoltaic power generation prediction model is established. By inputting light intensity, temperature, humidity and corresponding PID parameters, the corresponding photovoltaic power generation can be output, realizing high-precision prediction of distributed photovoltaic power generation.

[0074] Specifically, in the actual calculation process, environmental parameters are input into the model together to predict power generation. In actual calculation, weighted integration is used to represent the influence of different parameters. For example, the influence of temperature on power generation is greater than that of humidity. Therefore, the coefficient I corresponding to temperature will be larger. In addition, in the preliminary calculation, the integral coefficient I needs to be obtained with the help of historical data or empirical values. Usually, with the help of historical data analysis, through regression analysis, the relationship between factors such as light intensity in historical data and actual power generation is analyzed to obtain a coefficient I.

[0075] For each site in the substation, a corresponding model and PID parameters are established. The site model and the site's PID parameters form a photovoltaic power generation prediction model for the entire substation, thereby achieving high-precision prediction of distributed photovoltaic power generation in the substation or even in the cluster, which is of great significance for improving power generation of power stations, balancing supply and demand, optimizing storage scheduling and performance monitoring.

[0076] Through the above implementation, the present invention is based on edge computing, and high-frequency real-time data collection can be achieved by installing collection equipment on the edge side, which effectively reduces the delay of data transmission, improves the collection accuracy of the original data of the system, and the described PID algorithm structure is simple and easy to implement; based on this significant difference, the technical effects actually achieved by the present invention at least include: realizing high-frequency, efficient and accurate collection of photovoltaic power generation system operation data, effectively supporting the timeliness of data collection required by the PID control algorithm model, being applicable to different types of photovoltaic power generation systems, not being affected by geographical location and environmental factors, and providing reliable prediction results, relying on high-precision and high-timeliness data collection, and this algorithm can ensure the real-time nature of the prediction system by dynamically adjusting the PID parameters. The present invention not only improves the accuracy of photovoltaic power generation prediction, but also provides scientific decision-making support for power station operation, thereby maximizing economic benefits while ensuring energy supply.

[0077] Finally, 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 above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A distributed photovoltaic power generation prediction method based on edge computing, characterized in that: The steps include: Collect historical power generation data and real-time environmental parameter data of each distributed photovoltaic site in the substation area, and pre-process the collected data; The pre-processed real-time environmental parameter data is input into the basic prediction model of the site to obtain the predicted power generation, and then the power generation prediction model is composed of the basic prediction model and the PID control algorithm model; According to the comparative analysis of the predicted power generation and the real-time power generation, the PID parameters of the PID control algorithm model are optimized and adjusted, thereby optimizing the power generation prediction model; Based on the optimized power generation prediction model, the predicted value of the photovoltaic power generation of the site is generated, and the prediction result is output. According to the prediction result of each distributed photovoltaic site, the theoretical photovoltaic power generation of the substation is derived.

2. A distributed photovoltaic power generation prediction method based on edge computing according to claim 1, characterized in that: The historical power generation data of the distributed photovoltaic site includes environmental factors and actual power generation.

3. The distributed photovoltaic power generation prediction method based on edge computing according to claim 1 is characterized in that: The real-time environmental parameters of the distributed photovoltaic site include solar radiation intensity, temperature, humidity, wind speed and cloud changes.

4. The distributed photovoltaic power generation prediction method based on edge computing according to claim 1 is characterized in that: The preprocessing of the collected data includes data cleaning, missing value filling and format conversion.

5. A distributed photovoltaic power generation prediction method based on edge computing according to claim 1 or 4, characterized in that: The power generation prediction model includes: P(t)=p(t)+u(t) Where P(t) represents the power generation prediction model, p(t) represents the basic prediction model, which calculates the initial predicted power generation based on the actual environmental parameters; u(t) represents the PID control algorithm model, which calculates the correction value through the PID algorithm and is used to adjust the initial predicted power generation.

6. A distributed photovoltaic power generation prediction method based on edge computing according to claim 5, characterized in that: The power generation prediction model also includes that the PID parameters of the PID control algorithm model are dynamically adjusted according to the real-time prediction error: K p (t)=K p ·(1+α·|e(t)) In the formula, α, β, and γ are adjustment coefficients, which are used to adjust the sensitivity of the proportional, integral, and differential terms respectively; e(t)| is the absolute value of the error, reflecting the current deviation of the system; is the cumulative value of the error, reflecting the long-term deviation of the system; is the error change rate, reflecting the dynamic changes of the system; K p is the proportional gain, K i is the integral gain, K d is the differential gain, e(t) is the prediction error, is the predicted value of the power generation of the photovoltaic power generation system, which is predicted by the basic prediction model, y(t) is the actual value of the power generation of the photovoltaic power generation system, and t is the time variable; the final prediction result of the power generation prediction is:

7. A distributed photovoltaic power generation prediction method based on edge computing according to claim 6, characterized in that: The prediction error range is ±3%.

8. A distributed photovoltaic power generation prediction system based on edge computing, based on a distributed photovoltaic power generation prediction system based on edge computing according to any one of claims 1 to 7, characterized in that: The system includes: Data acquisition module, data processing module, PID controller module, prediction model module and user interaction interface; The data acquisition module is used to collect the historical power generation data and real-time environmental parameter data of the photovoltaic power generation system. The data processing module is used to pre-process the data collected by the data collection module and transmit it to the prediction model module. The prediction model module is used to calculate the predicted power generation. The PID controller module optimizes and adjusts the PID parameters according to the analysis of the predicted power generation and real-time power generation, and optimizes and adjusts the power generation prediction in the prediction model module. The user interaction interface is used to display the prediction results.

9. A distributed photovoltaic power generation prediction system based on edge computing according to claim 8, characterized in that: The data acquisition module includes temperature, humidity and light intensity sensors, which are installed at each photovoltaic site respectively to collect environmental parameter data in real time; the data acquisition module also includes an inverter for monitoring the electrical parameters of the photovoltaic system.

10. A distributed photovoltaic power generation prediction system based on edge computing according to claim 8, characterized in that: The user interaction interface supports the user to manually input environmental parameters to generate predicted power generation and display it on the user interaction interface.

Images