Photovoltaic power station monitoring system plan curve generation and management method and system

Abstract

The invention relates to the technical field of data processing, and discloses a photovoltaic power station monitoring system plan curve generation and management method and system. The method comprises the following steps: collecting photovoltaic power station environment, equipment and power grid dispatching data, and carrying out filtering, denoising and abnormal value processing to generate a preprocessed data set; generating capacity prediction is carried out through time sequence analysis and meteorological data, a power grid dispatching instruction is received, power distribution is calculated through multi-constraint optimization, and an operation plan curve is generated. Time and space dimension decomposition is carried out on the plan curve, and a control instruction is formulated. And modeling deviation distribution through the capsule mixed density network, optimizing and correcting parameters, collecting operation data to identify performance bottlenecks, and generating a system iteration updating instruction. Optimization adjustment is carried out according to different deviation sources, so that the power generation efficiency and stability of the power station are improved, the power station can better respond to the power grid dispatching instruction, and the power grid income is maximized.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method and system for generating and managing a planning curve for a photovoltaic power station monitoring system. Background Art

[0002] As a clean, renewable energy source, photovoltaic power generation systems are experiencing increasing global adoption and development. With the continuous advancement of photovoltaic technology, the scale and operational efficiency of photovoltaic power plants have significantly increased. However, because photovoltaic power generation is affected by environmental conditions (such as solar irradiance, temperature, and humidity), power generation from power plants can fluctuate significantly, especially during severe weather or sudden changes in weather. Furthermore, the operational status of photovoltaic power plant equipment can be affected by factors such as equipment aging, failures, and maintenance, which can prevent the plant's power generation capacity and efficiency from always being maintained at optimal levels. To optimize power plant operation and improve power generation efficiency and stability, power grid dispatch centers typically issue dispatch instructions to power plants, requiring them to provide a certain level of power generation capacity or adjust power. However, due to factors such as environmental fluctuations, unstable equipment status, and fluctuations in grid load, power plants often struggle to accurately follow dispatch instructions, resulting in deviations from power generation plans and mismatches with grid load, impacting grid stability.

[0003] Although existing technologies have adopted a variety of optimization algorithms and intelligent control methods to improve the operating efficiency of photovoltaic power plants, including data prediction models based on weather forecasts, equipment status monitoring, power allocation algorithms, etc., most of these methods can only be optimized under limited conditions and have certain limitations when processing complex multi-dimensional data. Existing power generation planning optimization systems usually rely on simple mathematical models or traditional optimization algorithms, lacking a full understanding of dynamically changing data and the ability to make real-time adjustments. In addition, existing technologies are often unable to effectively handle the complex deviations caused by the interaction of multiple factors in the power generation process of the power plant, and are unable to provide accurate correction solutions for specific sources of deviations. Summary of the Invention

[0004] The present application provides a photovoltaic power station monitoring system planning curve generation and management method and system, which is used to optimize and adjust according to different sources of deviation, thereby improving the power generation efficiency and stability of the power station, ensuring that the power station can better respond to grid dispatch instructions and maximize grid benefits.

[0005] In the first aspect, the present application provides a method for generating and managing a plan curve for a photovoltaic power station monitoring system, which includes: collecting environmental data, equipment data and grid dispatching data of the photovoltaic power station, filtering and denoising the collected data and performing outlier detection processing to obtain a preprocessed power station operation data set; applying time series analysis to the preprocessed power station operation data set, combining it with meteorological data to perform power generation correlation calculation to generate power generation forecast data for the future period; receiving grid dispatching instructions, matching the power generation forecast data with the dispatching requirements, calculating the power allocation plan through a multi-constraint optimization algorithm, and forming a power station operation plan curve; performing time and space dimension decomposition calculations on the power station operation plan curve, determining the power allocation strategy according to the equipment characteristics, generating specific control instructions and issuing them to each execution unit; using a capsule mixed density network to perform probabilistic modeling of the deviation distribution during the execution of the control instructions, identifying the source of the deviation and calculating the optimal correction parameters; collecting operation process data, analyzing equipment response characteristics and environmental influencing factors, identifying performance bottlenecks and generating parameter optimization plans, and forming system iterative update instructions.

[0006] In a second aspect, the present application provides a photovoltaic power station monitoring system plan curve generation and management system, the photovoltaic power station monitoring system plan curve generation and management system comprising:

[0007] The detection module is used to collect environmental data, equipment data and grid dispatch data of the photovoltaic power station, filter and remove noise and detect outliers on the collected data to obtain a pre-processed power station operation data set;

[0008] An analysis module, configured to apply time series analysis to the preprocessed power plant operation data set, perform power generation correlation calculations in combination with meteorological data, and generate power generation forecast data for future time periods;

[0009] a matching module, configured to receive grid dispatch instructions, match the power plant power generation forecast data with dispatch requirements, calculate a power allocation plan using a multi-constraint optimization algorithm, and form a power plant operation plan curve;

[0010] a decomposition module for performing time and space dimension decomposition calculations on the power plant operation plan curve, determining a power allocation strategy based on equipment characteristics, generating specific control instructions, and issuing them to each execution unit;

[0011] A modeling module is used to use a capsule mixture density network to probabilistically model the deviation distribution during the execution of control instructions, identify the source of the deviation and calculate the optimal correction parameters;

[0012] The generation module is used to collect operation process data, analyze equipment response characteristics and environmental influencing factors, identify performance bottlenecks and generate parameter optimization solutions, and form system iterative update instructions.

[0013] In a third aspect, a photovoltaic power station monitoring system plan curve generation and management device is provided, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the photovoltaic power station monitoring system plan curve generation and management device executes the above-mentioned photovoltaic power station monitoring system plan curve generation and management method.

[0014] In a fourth aspect, a computer-readable storage medium is provided, wherein instructions are stored in the computer-readable storage medium, which, when executed on a computer, enables the computer to execute the above-mentioned photovoltaic power station monitoring system plan curve generation and management method.

[0015] The technical solution provided by this application uses precise data collection and preprocessing to cover environmental data, equipment data, and grid dispatch data, ensuring the accuracy and completeness of the input data. This process, through technical means such as denoising, interpolation and data standardization, ensures that the data input to the model provides a reliable foundation for subsequent prediction and optimization in terms of both temporal resolution and data quality. Through refined data processing, the present invention provides solid data support for subsequent power generation prediction, power allocation, and deviation correction. The present invention utilizes a multi-model fusion power generation prediction system, combining physical models, statistical models, and deep learning models to improve the accuracy of power generation prediction for photovoltaic power plants at different time scales (short-term, medium-term, and long-term). In addition, by combining weather forecast data and real-time power plant operating status, the system can adaptively adjust the weight coefficients of the prediction model to cope with power generation fluctuations under different meteorological conditions, further enhancing the accuracy of the prediction results. By introducing a multi-objective optimization algorithm, the system can find the optimal balance between multiple optimization objectives. In this process, the system not only maximizes power generation revenue but also achieves dual optimization of the economic benefits of power plant operation and equipment life by minimizing deviation power penalties and equipment losses. Especially when facing the diverse needs of grid dispatch, the present invention can accurately calculate the optimal power distribution of each link such as the inverter and energy storage system, ensuring that the power station can flexibly respond to the load fluctuations and dispatching needs of the power grid under strict physical and dispatching constraints, thereby achieving an efficient match between the power generation of the power station and the grid demand. Through a refined deviation correction mechanism, the power generation deviation problem of the power station in the face of environmental changes, equipment fluctuations and grid dispatch instruction errors is effectively solved. By using capsule networks to model the spatial characteristics of deviation data, the system can accurately identify the source of the deviation, and combine the Bayesian optimization algorithm to calculate targeted correction parameters, thereby adjusting the power generation plan in real time to ensure that the power station can maintain efficient and stable operation under any circumstances. In particular, by analyzing the contribution of different deviation sources (such as environmental, equipment and instruction factors) through a dynamic routing algorithm, the system can design the most appropriate correction strategy for each type of deviation, further improving the operational reliability and flexibility of the photovoltaic power station. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] 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 some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0017] Figure 1 A schematic diagram of an embodiment of a method for generating and managing a planning curve for a photovoltaic power station monitoring system in an embodiment of the present application;

[0018] Figure 2 This is a schematic diagram of an embodiment of a photovoltaic power station monitoring system plan curve generation and management system in an embodiment of the present application;

[0019] Figure 3 It is a schematic block diagram of the structure of a planning curve generation and management device for a photovoltaic power station monitoring system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0020] The embodiments of the present application provide a method and system for generating and managing a planned curve for a photovoltaic power station monitoring system. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0021] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In one embodiment of the present application, a method for generating and managing a planned curve for a photovoltaic power station monitoring system includes:

[0022] Step S101: Collect environmental data, equipment data, and grid dispatch data of the photovoltaic power station, perform filtering, denoising, and outlier detection on the collected data, and obtain a pre-processed power station operation data set;

[0023] Step S102: Apply time series analysis to the pre-processed power plant operation data set, perform power generation correlation calculation in combination with meteorological data, and generate power generation forecast data for future time periods;

[0024] Step S103: Receive grid dispatch instructions, match power plant power generation forecast data with dispatch requirements, calculate power allocation plans through a multi-constraint optimization algorithm, and form a power plant operation plan curve;

[0025] Step S104: Decompose and calculate the time and space dimensions of the power plant operation plan curve, determine the power allocation strategy based on the equipment characteristics, generate specific control instructions, and issue them to each execution unit;

[0026] Step S105: Use the capsule mixture density network to perform probability modeling on the deviation distribution during the execution of the control instruction, identify the source of the deviation and calculate the optimal correction parameter;

[0027] Step S106: Collect operation process data, analyze equipment response characteristics and environmental influencing factors, identify performance bottlenecks and generate parameter optimization solutions to form system iterative update instructions.

[0028] It is understandable that the execution subject of this application can be a photovoltaic power station monitoring system plan curve generation and management system, or a terminal or server, which is not limited here. The embodiment of this application is described by taking the server as the execution subject as an example.

[0029] Specifically, real-time data is collected from various sensors and devices in the photovoltaic power station. This data includes environmental data (such as irradiance, temperature, and humidity), equipment operating data (such as inverter output power and solar panel temperature), and grid dispatch data (such as grid frequency and voltage). The frequency of data collection varies: environmental data is collected every 10 seconds, equipment data is collected every 5 seconds, and grid dispatch data is received in real time. After acquiring this data, preprocessing is required, primarily involving filtering, denoising, and outlier detection. During the data filtering stage, a combination of wavelet transform and moving average filtering is often used to remove high-frequency noise. By decomposing the signal into different frequency bands, the wavelet transform effectively identifies and removes the inevitable random noise components in the signal. The moving average filter, on the other hand, further smoothes the signal by averaging local regions of the data, reducing the interference of transient fluctuations on subsequent analysis. For outlier detection, the data is first ranged; any data outside the predefined range is considered an outlier. Missing data is supplemented using linear interpolation, which infers the missing value using the linear relationship between known data points, thereby maintaining data continuity and integrity. After these processes, the power plant's real-time operating data becomes a clean dataset suitable for subsequent analysis, with high accuracy and reliability. This preprocessed dataset undergoes further time series analysis and, combined with meteorological data, generates power generation correlations. This step first involves time series analysis of the collected historical power generation data to determine power generation patterns under different meteorological conditions. For example, weather changes (such as irradiance) directly impact the efficiency of photovoltaic panels. By establishing a correlation model between power generation and meteorological data (such as irradiance and temperature), the system can estimate future power generation based on historical data and predicted meteorological conditions. Models used in this calculation include support vector regression (SVR) and deep learning models (such as LSTM). LSTM (Long Short-Term Memory) networks are particularly well-suited for processing time series data, capturing long-term dependencies within the data and enabling more accurate predictions of future power generation trends. This approach enables the system to generate short- to long-term power generation forecasts, which directly influence subsequent power plant operation decisions and grid scheduling.

[0030] Matching grid dispatch instructions with power generation forecast data becomes the core task of this step. Grid dispatch requirements usually include load regulation, peak load regulation and frequency regulation. In order to best match the power generation capacity of the power station with the needs of the power grid, the system uses a multi-objective optimization algorithm, such as the improved particle swarm optimization (PSO) algorithm. This algorithm can simultaneously consider multiple optimization objectives, such as maximizing the power generation revenue of the power station, minimizing the deviation power penalty and optimizing the utilization efficiency of the energy storage system. By setting reasonable objective functions and constraints, the PSO algorithm can find the best power generation plan for the power station under given constraints (such as equipment capacity, grid load, etc.). During the optimization process, by adjusting the output of different subsystems (such as photovoltaic arrays, energy storage systems, etc.), the system generates multiple alternative operation plan curves for the power grid dispatch department to choose.

[0031] The generated power plant operation plan curve is further decomposed according to the time and space dimensions to ensure that each subsystem can execute independently and in coordination. Time decomposition means that the plan curve is divided into smaller time segments (such as minutes and hours), allowing for more flexible adjustments during actual operation. Spatial decomposition allocates the total generated power to different execution units within the power plant, such as multiple photovoltaic arrays, inverters, and energy storage systems. For each subsystem, the system adjusts the load distribution based on its real-time status (such as temperature and power output) to ensure efficient operation. For example, when the irradiance in a certain area is high, the system will prioritize more power generation tasks to that area, thereby maximizing overall power generation efficiency. Deviation processing and correction are key steps. Using mixture density modeling in capsule networks, the system can model and identify deviations in the execution of control instructions. Using a dynamic routing algorithm, the capsule network can identify deviations between different execution units and calculate the optimal correction parameters. Specifically, when the actual power generation of a subsystem deviates significantly from the planned curve, the system automatically adjusts the power output of that subsystem to restore it to the optimal operating state as much as possible. In this way, the system can effectively correct any possible deviations and ensure the matching between the power plant's power generation capacity and the grid dispatching requirements.

[0032] The system continuously collects data during operation, analyzing device response characteristics and external environmental impacts to identify potential performance bottlenecks. For example, if it detects a drop in inverter efficiency or a significant impact on a photovoltaic area's power generation capacity due to weather changes, the system immediately generates optimization instructions and adjusts device parameters, such as adjusting the energy storage system's charging and discharging strategies or optimizing the inverter's power output. Through this process, the system continuously optimizes its control strategies, further improving the plant's operating efficiency and responsiveness.

[0033] In the embodiment of the present application, through precise data collection and preprocessing, environmental data, equipment data and grid dispatch data are covered, ensuring the accuracy and completeness of the input data. This process ensures that the data of the input model can provide a reliable basis for subsequent prediction and optimization in terms of time resolution and data quality through technical means such as denoising, interpolation and data standardization. Through the refined processing of data, the present invention provides solid data support for subsequent power generation prediction, power allocation and deviation correction. The present invention utilizes a multi-model fusion power generation prediction system, which improves the accuracy of power generation prediction of photovoltaic power stations on different time scales (short-term, medium-term and long-term) by combining physical models, statistical models and deep learning models. In addition, combined with weather forecast data and real-time power station operation status, the system can adaptively adjust the weight coefficient of the prediction model to cope with power generation fluctuations under different meteorological conditions, further enhancing the accuracy of the prediction results. By introducing a multi-objective optimization algorithm, the system can find the optimal balance between multiple optimization objectives. In this process, the system not only maximizes power generation revenue, but also achieves dual optimization of the economic benefits of power station operation and equipment service life by minimizing deviation power penalties and equipment losses. Especially when facing the diverse needs of grid dispatch, the present invention can accurately calculate the optimal power distribution of each link such as the inverter and energy storage system, ensuring that the power station can flexibly respond to the load fluctuations and dispatching needs of the power grid under strict physical and dispatching constraints, thereby achieving an efficient match between the power generation of the power station and the grid demand. Through a refined deviation correction mechanism, the power generation deviation problem of the power station in the face of environmental changes, equipment fluctuations and grid dispatch instruction errors is effectively solved. By using capsule networks to model the spatial characteristics of deviation data, the system can accurately identify the source of the deviation, and combine the Bayesian optimization algorithm to calculate targeted correction parameters, thereby adjusting the power generation plan in real time to ensure that the power station can maintain efficient and stable operation under any circumstances. In particular, by analyzing the contribution of different deviation sources (such as environmental, equipment and instruction factors) through a dynamic routing algorithm, the system can design the most appropriate correction strategy for each type of deviation, further improving the operational reliability and flexibility of the photovoltaic power station.

[0034] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0035] Collect environmental data, equipment data, and grid dispatch data at different sampling frequencies, with environmental data collected every 10 seconds, equipment data collected every 5 seconds, and grid dispatch data received in real time;

[0036] Apply wavelet transform to decompose the collected environmental data, equipment data and power grid dispatching data into multiple frequency bands, remove high-frequency noise components, and reconstruct smoothed signals;

[0037] Outlier detection is performed on the smoothed signal based on the improved Mahalanobis distance algorithm, and data points that deviate from the mean by more than 3 standard deviations are marked as outliers;

[0038] The interpolation replacement method is applied to the abnormal points, and the replacement value is calculated based on the data before and after the time series using the cubic spline interpolation method;

[0039] Convert the processed environmental data, equipment data, and grid dispatch data into a standard format and store them in order according to timestamps;

[0040] The quality scores of the data stored in order of timestamps are calculated, including completeness score, accuracy score and timeliness score, to form a preprocessed power plant operation data set.

[0041] Specifically, environmental data from a photovoltaic power plant (such as irradiance, temperature, and humidity) is collected every 10 seconds, while equipment data (such as inverter output power and solar panel temperature) is collected at a higher frequency, every 5 seconds. Grid dispatch data (such as grid frequency and voltage level) is received in real time and continuously updated. The different collection frequencies of these data types are primarily due to their varying contributions to forecast accuracy and response time requirements. Environmental data, which changes more slowly, can be sampled at a lower frequency. Equipment data, on the other hand, requires a higher sampling frequency to ensure that changes in plant equipment status are captured, allowing for more precise adjustments to power generation plans. Grid dispatch data is received in real time to respond to changing grid demands. After collecting these raw data, wavelet transforms are applied to the data for denoising. By decomposing the signal into multiple frequency bands, the wavelet transform effectively removes high-frequency noise components, resulting in a smoother signal. Specifically, the wavelet transform first decomposes the signal, gradually converting it from high to low frequencies. By separating the high and low frequencies, it can remove transient fluctuations or interference from the equipment or environment. This process ensures the smoothness and continuity of the data, eliminates irrelevant fluctuations, thereby improving data quality and providing more precise information for subsequent analysis.

[0042] Outlier detection is achieved using a modified Mahalanobis distance algorithm. By calculating the deviation of a data point from the mean, the Mahalanobis distance algorithm effectively identifies outliers that differ significantly from other data points. In this invention, the system specifies that any data point that deviates from the mean by more than three standard deviations will be considered an outlier. This method can quickly and accurately flag data that does not conform to common patterns, such as erroneous data caused by sensor failure or external interference. In this way, anomalous data that affects analysis results can be effectively removed.

[0043] For data marked as outliers, cubic spline interpolation is used to replace them. Cubic spline interpolation is a widely used interpolation method for time series data correction. It generates a smooth interpolation curve using known preceding and following data points, thereby filling in missing or outlier data. The interpolation process calculates the relationship between known data points and infers reasonable replacement values ​​based on this relationship. This method ensures data smoothness and continuity while minimizing errors caused by data interpolation. After interpolation, all outliers in the dataset are effectively replaced, preventing inaccurate data from affecting subsequent analysis.

[0044] After the data processing is completed, the system converts all environmental data, equipment data, and power grid dispatch data into a standard format to ensure that all data items can be managed and stored according to a unified structure. This step ensures data consistency and facilitates subsequent analysis and processing. All processed data will be sorted and stored according to its timestamp to ensure that the data is arranged in chronological order, which is convenient for subsequent time series-based analysis and prediction. The stored data will be scored for quality to evaluate the completeness, accuracy, and timeliness of the data. The completeness score reflects whether there are missing values ​​in the data. Data with fewer missing values ​​will receive a higher score. The accuracy score evaluates the degree of match between the data and the actual situation. Data with smaller errors will receive a higher score. The timeliness score takes into account the frequency of data updates and time delays. Data updated in a timely manner will receive a higher score.

[0045] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0046] The preprocessed power plant operation data set is divided into a training data set and a validation data set according to the time window, and the time features, meteorological features and equipment operation features are extracted;

[0047] Perform Fourier transform on the time characteristics and decompose them into annual cycle, seasonal cycle, daily cycle and hourly cycle components to obtain periodic feature vectors;

[0048] A recursive feature elimination algorithm is used to sort the importance of all features and the top 80% of features are selected to construct a feature subset;

[0049] Input the feature subset and historical power generation data into the long short-term memory network to train the basic model for power generation time series prediction;

[0050] According to the difference between the output of the basic model and the actual meteorological data, a meteorological correction coefficient matrix is ​​constructed to adjust the predicted value;

[0051] The weather forecast data for the next 72 hours is input into the basic model and combined with the meteorological correction coefficient matrix to calculate the power generation forecast data for the future period with a resolution of 15 minutes.

[0052] Specifically, the preprocessed power plant operation dataset is divided into a training dataset and a validation dataset based on time windows. These time windows are typically divided based on the length of historical data, for example, hourly, daily, or weekly. The training dataset is used to train the model, while the validation dataset is used to test the model's effectiveness and ensure that it does not lose its generalization ability due to overfitting. After data partitioning, the system extracts key features from each time window. These features are primarily categorized as temporal features, meteorological features, and equipment operation features. Temporal features include information such as date, time, and day of the week; meteorological features include data such as irradiance, temperature, and humidity; and equipment operation features include data such as inverter power output and solar panel temperature. A Fourier transform is applied to the temporal features. This process decomposes the time series into periodic components and then extracts periodic feature vectors. The core concept of the Fourier transform is to decompose complex signals into multiple frequency components. For temporal features, the Fourier transform can decompose them into annual, seasonal, daily, and hourly cycles. These periodic components reflect the natural laws of PV power generation. For example, PV power generation is often affected by seasonal variations and daily changes in solar irradiance. Extracting these periodic features can help forecasting models better capture the cyclical fluctuations in power generation, thereby improving forecast accuracy.

[0053] The system applies a recursive feature elimination (RFE) algorithm to rank the importance of all extracted features. RFE is a feature selection method that gradually removes inefficient features. It optimizes the feature subset by training multiple models and gradually eliminating features that contribute less to model predictions. Each time the least important feature is removed, the RFE algorithm calculates its impact on the model and ultimately selects the most important feature subset. In the present invention, the RFE algorithm aims to select the top 80% of features as the final feature subset, which will be used for subsequent model training. Through feature selection, the system can reduce redundant features, lower the computational complexity of the model, and avoid overfitting, thereby improving prediction accuracy. After the feature subset is determined, the system inputs these features, along with historical power generation data, into a long short-term memory network (LSTM) to train a basic model for power generation time series forecasting. LSTM is a special type of recursive neural network (RNN) with memory capabilities that can capture long-term dependencies in time series. In the present invention, LSTM is used to predict future power generation based on historical data. During the training process, the LSTM model learns how to extract useful information from a subset of input features (such as time, weather, and equipment operation data) to generate power generation forecasts for future time periods.

[0054] After model training is complete, the system constructs a meteorological correction coefficient matrix based on the differences between the base model's output and actual meteorological data. This matrix is ​​used to adjust the model's forecasts to ensure they better adapt to changing weather conditions. The meteorological correction coefficient matrix is ​​constructed by comparing the differences between actual meteorological data and the base model's forecasts, calculating correction coefficients, and applying these coefficients to future forecasts. This process effectively corrects the base model's forecast bias in the face of unexpected weather changes, improving the accuracy of power generation forecasts. The system inputs 72-hour meteorological forecast data into the base model and, combined with the meteorological correction coefficient matrix, calculates power generation forecasts for future periods with a 15-minute resolution. By combining short-term weather forecasts with model outputs, this process makes forecasts more accurate and dynamically reflects changes in power plant generation capacity over time and with weather variations. This approach enables the system to provide the power grid dispatching center with detailed, real-time power generation plans based on weather changes.

[0055] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0056] Receive and parse XML-format dispatch instructions issued by the power grid dispatch center, extract basic load plans, peak and frequency regulation requirements, and voltage support requirements;

[0057] Construct an objective function matrix, including the power generation revenue maximization function, the deviation power penalty minimization function, and the equipment loss minimization function;

[0058] Set a set of constraints, including inverter capacity constraints, energy storage system charge and discharge power constraints, power change rate constraints, and grid dispatch instruction constraints;

[0059] The power plant power generation forecast data, objective function matrix and constraint condition set are input into the improved non-dominated sorting genetic algorithm to generate the initial Pareto optimal solution set;

[0060] The solutions in the Pareto optimal solution set are sorted according to the multi-objective weighted scores, and the solution with the highest comprehensive score is selected as the optimal power allocation solution;

[0061] The optimal power allocation plan is converted into a power plant operation plan curve containing time points, active power values, reactive power values, and energy storage charging and discharging power values.

[0062] Specifically, the system receives and parses XML-formatted dispatch instructions from the grid dispatch center. These XML-formatted instructions contain specific dispatch requirements for the power plant, such as base load planning, peak and frequency regulation requirements, and voltage support requirements. The operation of the PV power plant must be adjusted according to the grid dispatch center's requirements to ensure grid stability and the economic benefits of the power plant. After parsing the XML dispatch instructions, the system extracts this key information and uses it as part of the optimization objective. In this step, the PV power plant needs to calculate the power generation capacity it should provide, load variations, and the operations required for peak and frequency regulation based on the grid dispatch requirements.

[0063] Based on these scheduling requirements, the system constructs an objective function matrix. This objective function matrix is ​​the core of multi-objective optimization and encompasses multiple optimization objectives, including maximizing power generation revenue, minimizing deviation power penalty, and minimizing equipment loss. The power generation revenue maximization function aims to maximize the power plant's power generation revenue by calculating the relationship between the power plant's power generation and the electricity price, ensuring that each kilowatt-hour of power generated by the plant generates the maximum possible revenue. The deviation power penalty minimization function aims to reduce the deviation between actual and expected power generation, lowering grid penalties on the power plant and maintaining the accuracy of the power plant's power generation plan. The equipment loss minimization function considers the efficiency and operating losses of power plant equipment, aiming to optimize power plant operation, reduce losses in the inverter and energy storage system, and improve overall power generation efficiency. After establishing the objective functions, the system must consider a series of constraints to ensure that the power plant operates within its equipment capabilities while meeting grid requirements. Inverter capacity is a key constraint. The power plant's inverter output power cannot exceed its rated power value, so this constraint must be considered when calculating power allocation. The energy storage system's charge and discharge power constraints must also be considered. Energy storage systems have maximum charging and discharging power limits, which must be respected during calculations. Power rate constraints ensure that the power plant's power output does not fluctuate too drastically, preventing rapid power fluctuations from adversely affecting the grid. Furthermore, grid dispatch instructions themselves pose constraints, and power plants must adjust accordingly to ensure their power generation is consistent with grid dispatch requirements.

[0064] The system inputs the power plant's power generation forecast data, the objective function matrix, and the set of constraints into an improved non-dominated sorting genetic algorithm (NSGA-II) for calculation. NSGA-II is a genetic algorithm widely used for multi-objective optimization problems. Through its non-dominated sorting approach, it generates a Pareto-optimal set of solutions, each representing a balance between different objectives. Based on the objective function and constraints, the NSGA-II algorithm calculates multiple possible solutions, effectively finding the optimal trade-off under various constraints. By optimizing multiple objectives, the algorithm ensures optimal economic benefits, power generation accuracy, and equipment efficiency without violating any constraints.

[0065] After generating an initial set of Pareto-optimal solutions, the system ranks them based on their overall performance. This ranking is based on a multi-objective weighted scoring method, which assigns different weights to each objective based on its relative importance. This weighted scoring method allows the system to evaluate each solution's performance across all objectives and ultimately select the solution with the highest score as the optimal power allocation solution. This optimal solution comprehensively considers multiple factors, including power generation revenue, power deviation, and equipment losses, providing a balanced and highly feasible operation plan. The system then converts the selected optimal power allocation solution into a power plant's operating plan curve. This plan curve details the active power, reactive power, and energy storage system charge and discharge power values ​​at each point in time. Using this data, the power plant can rationally allocate resources based on grid dispatch requirements and equipment capabilities, ensuring maximum efficiency and meeting grid dispatch requirements. The operating plan curve not only serves as a guide for daily power plant operations but can also be dynamically adjusted to address unexpected situations or demand changes during actual operation. Ultimately, the power plant will generate power according to this plan curve, ensuring stable grid operation and achieving maximum economic benefits.

[0066] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0067] Decompose the power plant operation plan curve into time dimensions according to the control cycle to generate second-level control targets, minute-level control targets, and hour-level control targets;

[0068] Based on the inverter performance parameter library, calculate the current efficiency curve and operating status score of each inverter to obtain the inverter capability index;

[0069] Input the inverter capability index, inverter geographical location and current environmental conditions into the AHP model to calculate the power allocation weight coefficient;

[0070] According to the power allocation weight coefficient, the power value in the minute-level control target is allocated to each inverter to obtain the inverter control instruction set;

[0071] Based on the state of charge and cycle life model of the energy storage system, the optimal charge and discharge strategy is calculated and the energy storage system control instruction set is generated;

[0072] The inverter control instruction set and the energy storage system control instruction set are packaged into a control instruction data packet and sent to each execution unit through an encrypted channel.

[0073] Specifically, the system decomposes the power plant's operating plan curve into time dimensions based on the control cycle, generating second-level, minute-level, and hourly control targets. This time-dimensional decomposition allows for more refined adjustments to the power plant's power generation strategy to adapt to grid demand fluctuations at different timescales. For example, second-level control targets are primarily used to manage rapidly changing loads, minute-level targets are suitable for short- to medium-term load adjustments, and hourly targets are used to plan the power plant's long-term power generation strategy. This decomposition ensures the power plant can respond optimally at each timescale, thereby optimizing its power generation efficiency and stability. The system queries the performance parameter database for each inverter, calculates its current efficiency curve and operating status score, and generates the inverter's capability index. The inverter's capability index reflects its performance level under the current operating environment, including its conversion efficiency, operating status, and compatibility with environmental conditions (such as temperature and irradiance). Based on the inverter's performance parameters, the system can accurately assess each inverter's power generation capacity under current conditions, providing a crucial basis for subsequent power allocation. Through this process, the system can dynamically understand the operating status of each inverter and provide more accurate data support for subsequent control instruction allocation.

[0074] The system inputs the capability index of each inverter, the geographical location of the inverter, and the current environmental conditions into the Analytic Hierarchy Process (AHP) model to calculate the power allocation weight coefficient of each inverter. The Analytic Hierarchy Process is a multi-criteria decision analysis method that performs weighted evaluation on each evaluation criterion by establishing a decision model. In the present invention, the AHP model calculates the power allocation weight coefficient of each inverter by comprehensively considering multiple factors such as the performance, geographical location, and environmental conditions of the inverter. In this way, the system can reasonably allocate power generation tasks to different inverters, ensuring that each inverter within the power station can perform at its maximum efficiency within its respective capabilities. The calculation of this weight coefficient can not only improve the overall efficiency of the power generation of the power station, but also can be flexibly adjusted according to the actual situation of the inverter to meet different operating requirements.

[0075] Based on the calculated power allocation weights, the system distributes the power values ​​within the minute-level control targets to each inverter according to the weights, generating an inverter control instruction set. This instruction set details the power output targets for each inverter over the future time period, ensuring that power is appropriately distributed across the power plant's inverters based on the specified weights and power generation requirements. This approach ensures that each inverter operates within its maximum capacity, preventing overloading or reduced efficiency of certain inverters due to excessive load, while also ensuring optimal power generation capacity across the power plant. Simultaneously, the system calculates the optimal charge and discharge strategy based on the energy storage system's state of charge (SOC) and cycle life model, and generates the corresponding energy storage system control instruction set. The energy storage system's SOC reflects the current charge level of the energy storage device, while the cycle life model considers the number of charge and discharge cycles and efficiency degradation of the energy storage system. Using this data, the system develops an optimal charge and discharge strategy, ensuring appropriate switching between charging and discharging, thereby optimizing the energy storage system's efficiency and extending its service life. This process also ensures that the energy storage system can provide auxiliary services such as peak shaving and frequency regulation to the grid when necessary, while also charging when the power plant generates excess power to prepare for emergencies. The system packages the inverter control instruction set and the energy storage system control instruction set into a control instruction data packet and sends it to each execution unit via an encrypted channel. This control instruction packet includes specific control instructions for each inverter and energy storage system, ensuring that each execution unit in the power plant can accurately execute power generation tasks according to the set plan. Transmission through an encrypted channel ensures data security and confidentiality, preventing data leakage or tampering during transmission. After receiving the control instructions, the execution unit adjusts the inverter power output and the energy storage system's charging and discharging operations accordingly, ultimately achieving stable and efficient operation of the power plant.

[0076] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0077] Construct a capsule network structure with three capsule layers, where each capsule unit contains 16 neurons to capture the spatial features of deviation data;

[0078] The deviation time series data between the control instruction target value and the actual execution value is input into the capsule network to extract the deviation feature vector;

[0079] Based on the deviation eigenvector, eight Gaussian components are generated through the Gaussian mixture model to establish the probability density function of the deviation distribution;

[0080] A dynamic routing algorithm is used to calculate the contribution of different deviation sources and classify the deviations into three categories: environmental factors, equipment factors, and instruction factors.

[0081] Apply the Bayesian optimization algorithm to each type of deviation factor to calculate a targeted correction parameter set, including power adjustment coefficient, response time adjustment coefficient, and reserve margin coefficient;

[0082] According to the current operating status and predicted future state changes, the impact of each correction parameter is evaluated and the optimal correction parameter is selected for deviation correction.

[0083] Specifically, the system constructed a capsule network structure consisting of three capsule layers, with each capsule unit containing 16 neurons, designed to capture the spatial characteristics of deviation data through these neurons. The advantage of capsule networks over traditional neural networks is that they can better understand the spatial hierarchical structure in the data. Each capsule unit learns the spatial transformation characteristics of the data through a specific algorithm during the training process, thereby effectively capturing the complex relationship between power generation control instructions and actual execution values. By building such a network, the system can automatically learn the spatial characteristics of control instruction deviations, thereby providing reliable feature vectors for subsequent correction work.

[0084] Next, the system inputs the deviation time series data of the control instruction target value and the actual execution value into the constructed capsule network. Through the multi-layer capsule structure, the capsule network can extract the spatial feature vectors of the deviation data. These deviation feature vectors include various deviation information generated by factors such as environmental changes, device state fluctuations, and instruction differences during the execution process. In this way, the capsule network can effectively capture complex deviation patterns from the time series data and provide accurate input for subsequent deviation correction and optimization. Based on the extracted deviation feature vectors, the system further generates 8 Gaussian components through the Gaussian mixture model (GMM), and then establishes the probability density function of the deviation distribution. The Gaussian mixture model is a common probabilistic model for representing complex data distribution. It can effectively model multidimensional data and divide the data into several Gaussian components. In the present invention, the GMM can identify different types of deviation distributions by analyzing the deviation feature vectors and convert them into multiple Gaussian components, thereby forming a complete deviation distribution probability density function. Through this method, the system can describe the distribution of deviation data in more detail and provide a mathematical basis for subsequent correction parameter calculations.

[0085] The system then uses a dynamic routing algorithm to calculate the contribution of different deviation sources and categorizes the deviation data into three categories: environmental factors, equipment factors, and instruction factors. The core idea of ​​the dynamic routing algorithm is to dynamically adjust the connections between different capsule units in the capsule network in a hierarchical manner, allowing the network to flexibly learn the contributions of different deviation sources. In the present invention, by calculating the contribution of the deviation sources, the system can accurately identify the main causes of the deviation and classify them. For example, environmental factors may include meteorological changes (such as irradiance changes), equipment factors may include inverter efficiency fluctuations, and instruction factors are related to changes in grid dispatch instructions. Through this classification, the system can accurately correct for different types of deviations. The system applies a Bayesian optimization algorithm to each type of deviation factor and calculates a targeted set of correction parameters. Bayesian optimization is an optimization method based on a probabilistic model that can effectively find the optimal solution with limited experimental data. In the present invention, the Bayesian optimization algorithm is applied to calculate a set of correction parameters for different deviation factors. These correction parameters include: power adjustment coefficient, response time adjustment coefficient, and reserve margin coefficient. The power adjustment factor is used to appropriately adjust the inverter's power output, the response time adjustment factor optimizes the response delay of grid dispatch, and the reserve margin factor ensures that the power plant has sufficient margin for adjustment in extreme situations. The system evaluates the calculated correction parameters based on the current power plant's operating status and predicted future state changes, selecting the optimal correction parameters for actual deviation correction. During the evaluation process, the system conducts a detailed analysis of the impact of each correction parameter and determines the most appropriate correction parameters based on the power plant's real-time operating data and expected future operating status. This process not only ensures the effectiveness of deviation correction but also ensures that the power plant can flexibly respond to changes in grid dispatch requirements under various operating conditions.

[0086] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0087] Collect operational process data at three time granularities: seconds, minutes, and hours, including comparison data before and after deviation correction and device response time data;

[0088] The operation process data is divided into a set of typical operation scenarios, including stable power generation scenarios on sunny days, sudden changes from cloudy to sunny days, and nighttime energy storage operation scenarios;

[0089] Apply principal component analysis to each typical scenario to extract key influencing factors and calculate the contribution of each factor to system performance;

[0090] Based on key influencing factors and contribution rates, a decision tree model is constructed to identify performance bottlenecks and generate a set of bottleneck solutions.

[0091] Input the bottleneck solution set into the Monte Carlo simulator, conduct multiple rounds of simulation tests, and calculate the effectiveness score of each solution;

[0092] The optimal solution is selected based on the effect score and converted into parameter adjustment instructions, algorithm update instructions and strategy optimization instructions to form system iterative update instructions.

[0093] Specifically, the system collects operational data at three time granularities: seconds, minutes, and hours. This data includes comparisons before and after deviation correction, as well as device response time data. Second-level data is used to capture rapidly changing deviation corrections and device responses, minute-level data is used for short- and medium-term operational status monitoring, and hourly data is used for overall performance evaluation. By collecting data at these different time granularities, the system comprehensively reflects the operational status and performance changes of the power plant at various time scales. The system divides the collected operational data into several typical operating scenarios: stable power generation on sunny days, sudden weather changes, and nighttime energy storage operation. Each scenario represents a typical operating state of the power plant under different environmental and load conditions. For example, in the stable power generation scenario on sunny days, the power plant generates stable and efficient power. In the sudden weather changes, the power plant may face rapidly changing irradiance, requiring rapid response to adjust power generation. The nighttime energy storage operation scenario focuses on the charging and discharging operations of the energy storage system, ensuring stable operation in the absence of sunlight. By dividing these typical scenarios, the system can better understand the performance of the power station under different operating conditions, and provide targeted scenario data for subsequent analysis and optimization. In each typical scenario, the system applies principal component analysis (PCA) to extract key influencing factors. Principal component analysis can extract the most representative features from multidimensional data by reducing dimensionality, thereby reducing the complexity of the data. In the present invention, PCA is used to identify the main factors affecting the performance of power stations in different scenarios. These influencing factors may include weather changes, equipment failures, energy storage system efficiency, etc. Through PCA, the system can calculate the contribution rate of each factor to the performance of the power station, that is, the degree of influence of each factor on the overall performance. This process enables the system to clearly identify which factors have a greater impact on the operation of the power station, thereby providing data support for subsequent performance optimization.

[0094] Based on the extracted key influencing factors and their contribution rates, the system constructs a decision tree model to identify bottlenecks in power plant performance. By analyzing key influencing factors and their contribution rates, the decision tree model can pinpoint the most critical bottlenecks in power plant operation. For example, certain factors may cause equipment efficiency to decline, or environmental changes may lead to significant fluctuations in power generation. By analyzing the branches of the decision tree, the system can accurately identify performance bottlenecks in power plants under different operating scenarios. Based on these bottlenecks, the system generates a set of bottleneck solutions, providing a series of targeted solutions for the power plant. These solutions may include equipment adjustments, policy optimization, or other related measures.

[0095] After generating a set of bottleneck solutions, the system inputs them into a Monte Carlo simulator for multiple rounds of simulation testing. Using random sampling and statistical methods, the Monte Carlo simulator simulates the performance of different solutions under various uncertainties. Through these rounds of simulation testing, the system calculates a performance score for each solution. These scores reflect the potential effectiveness and performance improvement of each solution in real-world applications. Through Monte Carlo simulation, the system assesses the reliability and effectiveness of each solution in addressing diverse operating conditions and uncertainties. Based on the performance scores from the simulation results, the system selects the optimal solution and converts it into actual parameter adjustment instructions, algorithm update instructions, and strategy optimization instructions. These instructions are then issued as system iterative update instructions to each execution unit, guiding the power plant to make corresponding adjustments and optimizations. For example, adjustments to inverter output parameters or optimization of the energy storage system's charging and discharging strategies may be necessary to ensure optimal performance under the new operating conditions. Through this series of optimizations and adjustments, the power plant can continuously improve its operating efficiency, reduce performance bottlenecks, and ensure long-term stable operation.

[0096] The above describes the method for generating and managing the planned curve of the photovoltaic power station monitoring system in the embodiment of the present application. The following describes the method for generating and managing the planned curve of the photovoltaic power station monitoring system in the embodiment of the present application. Figure 2 In one embodiment of the present application, a photovoltaic power station monitoring system plan curve generation and management system includes:

[0097] The detection module 201 is used to collect environmental data, equipment data and grid dispatch data of the photovoltaic power station, and perform filtering, denoising and outlier detection on the collected data to obtain a pre-processed power station operation data set;

[0098] Analysis module 202, for applying time series analysis to the pre-processed power plant operation data set, performing power generation correlation calculations in combination with meteorological data, and generating power generation forecast data for future time periods;

[0099] The matching module 203 is used to receive the grid dispatch instruction, match the power plant power generation forecast data with the dispatch demand, calculate the power allocation plan through a multi-constraint optimization algorithm, and form a power plant operation plan curve;

[0100] A decomposition module 204 is configured to perform time and space decomposition calculations on the power plant operation plan curve, determine a power allocation strategy based on equipment characteristics, generate specific control instructions, and issue them to each execution unit;

[0101] Modeling module 205, for using capsule mixture density network to perform probabilistic modeling on the deviation distribution during the execution of control instructions, identify the source of the deviation and calculate the optimal correction parameters;

[0102] The generation module 206 is used to collect operation process data, analyze equipment response characteristics and environmental influencing factors, identify performance bottlenecks and generate parameter optimization solutions, and form system iterative update instructions.

[0103] Through the collaborative efforts of the aforementioned components and precise data collection and preprocessing, environmental data, equipment data, and grid dispatch data are included, ensuring the accuracy and integrity of the input data. This process, through technical means such as denoising, interpolation, and data standardization, ensures that the data input to the model, in terms of both temporal resolution and data quality, provides a reliable foundation for subsequent prediction and optimization. Through refined data processing, the present invention provides solid data support for subsequent power generation forecasting, power allocation, and bias correction. This invention utilizes a multi-model fusion power generation forecasting system, combining physical, statistical, and deep learning models to improve the accuracy of power generation forecasts for photovoltaic power plants across different timescales (short-term, medium-term, and long-term). Furthermore, by integrating weather forecast data and real-time power plant operating status, the system can adaptively adjust the weight coefficients of the prediction model to account for power generation fluctuations under different meteorological conditions, further enhancing the accuracy of the prediction results. By incorporating a multi-objective optimization algorithm, the system is able to find the optimal balance between multiple optimization objectives. This process not only maximizes power generation revenue but also achieves dual optimization of power plant economic benefits and equipment life by minimizing bias power penalties and equipment losses. Especially when facing the diverse needs of grid dispatch, the present invention can accurately calculate the optimal power distribution of each link such as the inverter and energy storage system, ensuring that the power station can flexibly respond to the load fluctuations and dispatching needs of the power grid under strict physical and dispatching constraints, thereby achieving an efficient match between the power generation of the power station and the grid demand. Through a refined deviation correction mechanism, the power generation deviation problem of the power station in the face of environmental changes, equipment fluctuations and grid dispatch instruction errors is effectively solved. By using capsule networks to model the spatial characteristics of deviation data, the system can accurately identify the source of the deviation, and combine the Bayesian optimization algorithm to calculate targeted correction parameters, thereby adjusting the power generation plan in real time to ensure that the power station can maintain efficient and stable operation under any circumstances. In particular, by analyzing the contribution of different deviation sources (such as environmental, equipment and instruction factors) through a dynamic routing algorithm, the system can design the most appropriate correction strategy for each type of deviation, further improving the operational reliability and flexibility of the photovoltaic power station.

[0104] above Figure 2 The photovoltaic power station monitoring system plan curve generation and management system in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The photovoltaic power station monitoring system plan curve generation and management device in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0105] Figure 3FIG3 is a schematic diagram of the structure of a photovoltaic power plant monitoring system plan curve generation and management device provided by an embodiment of the present invention. The photovoltaic power plant monitoring system plan curve generation and management device 300 may vary significantly due to different configurations or performance. The device may include one or more central processing units (CPUs) 310 (e.g., one or more processors), a memory 320, and one or more storage media 330 (e.g., one or more mass storage devices) storing application programs 333 or data 332. The memory 320 and storage medium 330 may be either transient or persistent storage. The program stored in the storage medium 330 may include one or more modules (not shown), each of which may include a series of instruction operations within the photovoltaic power plant monitoring system plan curve generation and management device 300. Furthermore, the processor 310 may be configured to communicate with the storage medium 330, and the photovoltaic power plant monitoring system plan curve generation and management device 300 may execute the series of instruction operations stored in the storage medium 330 to implement the steps of the photovoltaic power plant monitoring system plan curve generation and management method described above.

[0106] The photovoltaic power plant monitoring system plan curve generation and management device 300 may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input and output interfaces 360, and / or one or more operating systems 331, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. It will be understood by those skilled in the art that Figure 3 The structure of the photovoltaic power station monitoring system plan curve generation and management device shown does not constitute a limitation on the photovoltaic power station monitoring system plan curve generation and management device provided by the present invention, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0107] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions. When the instructions are executed on a computer, the computer executes the steps of the photovoltaic power station monitoring system plan curve generation and management method.

[0108] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0109] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a photovoltaic power station monitoring system plan curve generation and management device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc., various media that can store program code.

[0110] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for generating and managing a planning curve for a photovoltaic power station monitoring system, characterized in that: The method comprises: Collect environmental data, equipment data, and grid dispatch data from the photovoltaic power station, perform filtering, denoising, and outlier detection on the collected data, and obtain a pre-processed power station operation data set; Applying time series analysis to the preprocessed power plant operation data set, performing power generation correlation calculation in combination with meteorological data, and generating power generation forecast data for future time periods; Receive grid dispatch instructions, match the power plant power generation forecast data with dispatch requirements, calculate the power allocation plan through a multi-constraint optimization algorithm, and form a power plant operation plan curve; Decomposing and calculating the time and space dimensions of the power plant operation plan curve, determining the power allocation strategy based on the equipment characteristics, generating specific control instructions and issuing them to each execution unit; The capsule mixture density network is used to probabilistically model the deviation distribution during the execution of control instructions, identify the source of the deviation and calculate the optimal correction parameters; Collect operation process data, analyze equipment response characteristics and environmental influencing factors, identify performance bottlenecks and generate parameter optimization solutions, and form system iterative update instructions.

2. The photovoltaic power station monitoring system planning curve generation and management method according to claim 1, characterized in that: The method collects environmental data, equipment data and grid dispatch data of the photovoltaic power station, performs filtering, denoising and outlier detection on the collected data, and obtains a pre-processed power station operation data set, including: Collect environmental data, equipment data, and grid dispatch data at different sampling frequencies, with environmental data collected every 10 seconds, equipment data collected every 5 seconds, and grid dispatch data received in real time; Apply wavelet transform to decompose the collected environmental data, equipment data and power grid dispatching data into multiple frequency bands, remove high-frequency noise components, and reconstruct smoothed signals; Performing outlier detection on the smoothed signal based on an improved Mahalanobis distance algorithm, marking data points that deviate from the mean by more than 3 standard deviations as outliers; Applying interpolation replacement method to the abnormal point to calculate the replacement value based on the previous and next time series data using cubic spline interpolation method; Convert the processed environmental data, equipment data, and grid dispatch data into a standard format and store them in order according to timestamps; The quality scores of the data stored in order of timestamps are calculated, including completeness score, accuracy score and timeliness score, to form a preprocessed power plant operation data set.

3. The photovoltaic power station monitoring system planning curve generation and management method according to claim 1, characterized in that: The applying time series analysis to the pre-processed power plant operation data set and performing power generation correlation calculation in combination with meteorological data to generate power generation forecast data for future time periods of the power plant includes: Dividing the preprocessed power plant operation data set into a training data set and a validation data set according to a time window, and extracting time features, meteorological features, and equipment operation features; Performing Fourier transform on the time feature to decompose it into annual cycle, seasonal cycle, daily cycle and hourly cycle components to obtain a periodic feature vector; A recursive feature elimination algorithm is used to sort the importance of all features and the top 80% of features are selected to construct a feature subset; Inputting the feature subset and historical power generation data into a long short-term memory network to train a basic model for power generation time series prediction; Constructing a meteorological correction coefficient matrix based on the difference between the output of the basic model and the actual meteorological data to adjust the predicted value; The weather forecast data for the next 72 hours is input into the basic model and combined with the weather correction coefficient matrix to calculate the power generation forecast data of the power station in the future period with a resolution of 15 minutes.

4. The photovoltaic power station monitoring system planning curve generation and management method according to claim 1, characterized in that: The receiving of the grid dispatch instruction, matching the power plant power generation forecast data with the dispatch demand, calculating the power distribution plan through a multi-constraint optimization algorithm, and forming a power plant operation plan curve, includes: Receive and parse XML-format dispatch instructions issued by the power grid dispatch center, extract basic load plans, peak and frequency regulation requirements, and voltage support requirements; Construct an objective function matrix, including the power generation revenue maximization function, the deviation power penalty minimization function, and the equipment loss minimization function; Set a set of constraints, including inverter capacity constraints, energy storage system charge and discharge power constraints, power change rate constraints, and grid dispatch instruction constraints; Inputting the power plant power generation forecast data, the objective function matrix and the constraint condition set into an improved non-dominated sorting genetic algorithm to generate an initial Pareto optimal solution set; Sort the solutions in the Pareto optimal solution set according to the multi-objective weighted scores, and select the solution with the highest comprehensive score as the optimal power allocation solution; The optimal power allocation scheme is converted into a power station operation plan curve including time points, active power values, reactive power values ​​and energy storage charging and discharging power values.

5. The photovoltaic power station monitoring system planning curve generation and management method according to claim 1, characterized in that: The decomposition calculation of the time and space dimensions of the power plant operation plan curve is performed, the power allocation strategy is determined according to the equipment characteristics, and specific control instructions are generated and issued to each execution unit, including: Decomposing the power plant operation plan curve in time dimension according to the control period to generate second-level control targets, minute-level control targets and hour-level control targets; Based on the inverter performance parameter library, calculate the current efficiency curve and operating status score of each inverter to obtain the inverter capability index; Inputting the inverter capability index, the inverter geographical location and the current environmental conditions into an analytic hierarchy process model to calculate a power allocation weight coefficient; Allocating the power value in the minute-level control target to each inverter according to the power allocation weight coefficient to obtain an inverter control instruction set; Based on the state of charge and cycle life model of the energy storage system, the optimal charge and discharge strategy is calculated and the energy storage system control instruction set is generated; The inverter control instruction set and the energy storage system control instruction set are packaged into a control instruction data packet and sent to each execution unit through an encrypted channel.

6. The photovoltaic power station monitoring system planning curve generation and management method according to claim 1, characterized in that: The capsule mixture density network is used to perform probabilistic modeling of the deviation distribution during the execution of control instructions, identify the source of the deviation and calculate the optimal correction parameter, including: Construct a capsule network structure with three capsule layers, where each capsule unit contains 16 neurons to capture the spatial features of deviation data; Inputting the deviation time series data between the control instruction target value and the actual execution value into the capsule network to extract the deviation feature vector; Based on the deviation feature vector, eight Gaussian components are generated through a Gaussian mixture model to establish a probability density function of the deviation distribution; A dynamic routing algorithm is used to calculate the contribution of different deviation sources and classify the deviations into three categories: environmental factors, equipment factors, and instruction factors. Apply the Bayesian optimization algorithm to each type of deviation factor to calculate a targeted correction parameter set, including power adjustment coefficient, response time adjustment coefficient, and reserve margin coefficient; According to the current operating status and predicted future state changes, the impact of each correction parameter is evaluated and the optimal correction parameter is selected for deviation correction.

7. The photovoltaic power station monitoring system planning curve generation and management method according to claim 1, characterized in that: The collection of operation process data, analysis of equipment response characteristics and environmental factors, identification of performance bottlenecks and generation of parameter optimization solutions, and formation of system iterative update instructions include: Collect operational process data at three time granularities: seconds, minutes, and hours, including comparison data before and after deviation correction and device response time data; Dividing the operation process data into a set of typical operation scenarios, including a stable power generation scenario on a sunny day, a sudden change from cloudy to sunny scenario, and a nighttime energy storage operation scenario; Apply principal component analysis to each typical scenario to extract key influencing factors and calculate the contribution of each factor to system performance; Based on the key influencing factors and contribution rates, a decision tree model is constructed to identify performance bottlenecks and generate a set of bottleneck solutions; Input the bottleneck solution set into the Monte Carlo simulator, conduct multiple rounds of simulation tests, and calculate the effectiveness score of each solution; The optimal solution is selected based on the effect score, converted into parameter adjustment instructions, algorithm update instructions and strategy optimization instructions, and formed into system iterative update instructions.

8. A photovoltaic power station monitoring system planning curve generation and management system, characterized in that: The method for generating and managing a planned curve for a photovoltaic power station monitoring system according to any one of claims 1 to 7 is configured to implement the method, wherein the planned curve generating and managing system for a photovoltaic power station monitoring system comprises: The detection module is used to collect environmental data, equipment data and grid dispatch data of the photovoltaic power station, filter and remove noise and detect outliers on the collected data to obtain a pre-processed power station operation data set; An analysis module, configured to apply time series analysis to the preprocessed power plant operation data set, perform power generation correlation calculations in combination with meteorological data, and generate power generation forecast data for future time periods; a matching module, configured to receive grid dispatch instructions, match the power plant power generation forecast data with dispatch requirements, calculate a power allocation plan using a multi-constraint optimization algorithm, and form a power plant operation plan curve; a decomposition module for performing time and space dimension decomposition calculations on the power plant operation plan curve, determining a power allocation strategy based on equipment characteristics, generating specific control instructions, and issuing them to each execution unit; A modeling module is used to use a capsule mixture density network to probabilistically model the deviation distribution during the execution of control instructions, identify the source of the deviation and calculate the optimal correction parameters; The generation module is used to collect operation process data, analyze equipment response characteristics and environmental influencing factors, identify performance bottlenecks and generate parameter optimization solutions, and form system iterative update instructions.

9. A photovoltaic power station monitoring system planning curve generation and management device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, the method for generating and managing a planned curve of a photovoltaic power station monitoring system according to any one of claims 1 to 7 is 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 processor is enabled to execute the method for generating and managing a planning curve for a photovoltaic power station monitoring system according to any one of claims 1 to 7.

Images