A multi-source intelligent power prediction method and system for new energy power stations

Abstract

The present invention discloses a multi-source intelligent power prediction method and system for a new energy power station, which relates to the technical field of new energy power prediction and optimization, including collecting multi-source data and performing preprocessing, constructing an improved deep echo state network model to predict power demand values; calculating load demand gaps according to power demand values, constructing an objective function and solving the Pareto optimal solution set, executing the Pareto optimal solution set and optimizing the dynamic weight of the objective function. The present invention realizes efficient prediction of power station power demand by constructing an improved deep echo state network model, improves the accuracy and real-time performance of power demand prediction, calculates the weights of total operating costs and load demand gaps by entropy weight method, and then provides multi-objective optimization support for power grid dispatching decisions by dynamic weight adjustment, uses particle swarm optimization algorithm to solve the Pareto optimal solution set, and improves the stability of new energy power station operation and the load management capability of the power grid.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy power prediction and optimization, and in particular to a multi-source intelligent power prediction method and system for a new energy power station. Background Art

[0002] New energy power stations are an important part of the current global energy transformation. Their development and application provide an important way to deal with energy crises and environmental problems. Due to the strong intermittent and volatile nature of new energy resources (such as solar energy and wind energy), the power output of new energy power stations is uncertain and dynamically changing. Accurately predicting the power output of new energy power stations has become a key link in ensuring the stable operation of the power grid, optimizing resource allocation and reducing operating costs. With the rapid development of sensor technology, artificial intelligence and big data technology, the collection, fusion and analysis of multi-source data have gradually become an important technical means to improve the accuracy of power prediction. Due to the large amount of data, diverse types and complex noise of new energy power stations, the existing power prediction methods face great challenges in model construction and optimization.

[0003] Although the existing technology has made certain progress in the field of renewable energy power prediction, it still has the following shortcomings: the existing power prediction models usually use traditional deep learning methods, such as LSTM or simple echo state network, and their activation functions and parameter optimization methods fail to fully consider the dynamic and nonlinear characteristics of renewable energy data, thereby limiting the generalization ability and accuracy of the model. The design of the objective function does not fully consider the comprehensive requirements of the economy and grid balance in the operation of renewable energy power stations. Traditional optimization methods are difficult to balance the trade-offs between multiple objectives, resulting in the prediction results failing to achieve the best effect in practical applications. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a multi-source intelligent power prediction method and system for new energy power stations to solve the problem that existing power prediction models usually use traditional deep learning methods, such as LSTM or simple echo state network, and their activation functions and parameter optimization methods fail to fully consider the dynamic and nonlinear characteristics of new energy data, thereby limiting the generalization ability and accuracy of the model. The design of the objective function does not fully consider the comprehensive requirements of the economy and grid balance in the operation of new energy power stations. Traditional optimization methods are difficult to balance the trade-offs between multiple objectives, resulting in the problem that the prediction results cannot achieve the best effect in practical applications.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a multi-source intelligent power prediction method for a new energy power station, which includes collecting and preprocessing multi-source data, and constructing an improved deep echo state network model to predict power demand values;

[0008] Calculate the load demand gap according to the power demand value, construct the objective function and solve the Pareto optimal solution set, execute the Pareto optimal solution set and optimize the dynamic weight of the objective function;

[0009] Build a visual interface to display the Pareto optimal solution set and store the multi-source data generated by collection and analysis.

[0010] As a preferred solution of the multi-source intelligent power prediction method for a new energy power station described in the present invention, wherein: the collecting of multi-source data and pre-processing thereof refers to using an intelligent sensor to collect multi-source data of a new energy power station and pre-processing thereof;

[0011] The intelligent sensors include voltage, current, resistance, temperature, wind speed and radiation intensity sensors;

[0012] The multi-source data includes voltage, current, resistance, temperature, wind speed and solar radiation data;

[0013] The preprocessing includes using the network time protocol for time synchronization, using wavelet transform for denoising, using the interquartile range method to identify and delete outliers, using the local mean method to fill in missing values, and standardizing multi-source data.

[0014] As a preferred solution of the multi-source intelligent power prediction method for a new energy power station described in the present invention, wherein: the construction of an improved deep echo state network model to predict power demand value refers to collecting historical multi-source data and preprocessing it to generate a training set;

[0015] Constructing an improved deep echo state network model including an input layer, an echo layer and an output layer;

[0016] Use the weighted linear combination method to calculate the data weights of multi-source data, and use the ReLU function to construct the multi-source data into a weighted function ;

[0017] The activation function is constructed by multiplying the Tanh and Sigmoid functions. ;

[0018] The input of the improved deep echo state network model is set to be the preprocessed multi-source data;

[0019] Calculate the state vector of multi-source data at time t based on the improved echo layer , the formula is:

[0020] ,

[0021] in is the activation function, is the input connection matrix of the echo layer, is the i-th preprocessed multi-source data at time t, is the state vector of the improved echo layer at time t, is the echo connection matrix of the echo layer, is the multi-source data weight matrix, b is the bias term;

[0022] The state vector at time t , the output layer predicts the power demand value at time t after improvement , the formula is:

[0023] ,

[0024] in is the output layer weight matrix, is the Hadamard product;

[0025] Using the training set to train the improved deep echo state network model, and using the differential evolution algorithm and the weighted loss function to iteratively optimize the parameters of the improved deep echo state network model;

[0026] Input the preprocessed multi-source data into the trained improved deep echo state network model to predict the power demand value at the improved time t .

[0027] As a preferred solution of the multi-source intelligent power prediction method for a new energy power station described in the present invention, wherein: the calculation of the load demand gap according to the power demand value, the construction of the objective function and the solution of the Pareto optimal solution set refer to the collection of the unit electricity price D(t) at time t from the national electricity market trading platform;

[0028] Calculate the charge and discharge power at time t based on the voltage and current in the preprocessed multi-source data ;

[0029] Calculate the charge and discharge energy cost at time t based on the charge and discharge power at time t ;

[0030] The charging and discharging power according to time t and unit electricity price D(t) to calculate the electricity transaction cost ;

[0031] The energy cost of charging and discharging at time t and electricity transaction costs The total running cost of adding up the computation time t ;

[0032] Calculate the transmission loss at time t based on the current and resistance in the preprocessed multi-source data ;

[0033] The charge and discharge power at time t Reduce transmission loss Load demand at time t ;

[0034] The power demand value according to time t Load shedding requirements Calculate the load demand gap at time t ;

[0035] Total operating cost according to time t and load demand gap Calculate the information entropy of the total operating cost and the information entropy of load demand gap , use the entropy weight method to calculate the total operating cost weight and the load demand gap weight respectively, and divide the load demand gap weight by the total operating cost weight to calculate the dynamic weight ;

[0036] The objective function F is constructed using the multi-objective optimization method, and the formula is:

[0037] ,

[0038] Where T is the total time;

[0039] Randomly generate N particles in the population, each particle represents a solution ;

[0040] Perform Pareto sorting on all solutions in the population, identify the dominance relationship, use the non-dominated sorting method to screen, and select the Pareto optimal solution set;

[0041] Calculate the objective function value of the Pareto optimal solution set and identify the maximum value of the objective function value of the jth Pareto optimal solution and minimum value ;

[0042] Use kernel density estimation to calculate the relative probability density of each Pareto optimal solution on the objective function, and use Shannon entropy to calculate the information entropy of the objective function corresponding to the j-th Pareto optimal solution based on the relative probability density. ;

[0043] According to the maximum value of the objective function value of the jth Pareto optimal solution , minimum And the information entropy of the corresponding objective function , calculate the congestion degree of the jth Pareto optimal solution , the formula is:

[0044] ,

[0045] in and are the values ​​of the objective function neighbors corresponding to the Pareto optimal solution of the nth particle in the jth neighbor, is a small constant;

[0046] Select the Pareto optimal solution with the highest congestion to enter the next generation;

[0047] Use particle swarm optimization algorithm to calculate and update particle speed , update the particle solution ;

[0048] The maximum number of iterations is set using the rule of thumb method. When the maximum number of iterations is reached, the iteration is stopped and the Pareto optimal solution set is output, including the optimized total operating cost and load demand gap.

[0049] As a preferred solution of the multi-source intelligent power prediction method for a new energy power station described in the present invention, wherein: the execution of the Pareto optimal solution set and the optimization of the dynamic weight of the objective function refers to using a parameter instruction mapping method to convert the Pareto optimal solution set into a scheduling instruction, which is transmitted to the control platform through an API interface, and the control platform receives the scheduling instruction and executes it, and collects the operation data after the execution;

[0050] Calculate the difference between the execution running data and the optimal solution set;

[0051] Use statistical methods to set the judgment threshold, compare the difference with the judgment threshold, mark the difference greater than or equal to the judgment threshold, and use the marked difference to adjust the dynamic weight of the objective function , until the difference is less than the judgment threshold, stop adjusting and continue to monitor the difference results.

[0052] As a preferred solution of the multi-source intelligent power prediction method for a new energy power station described in the present invention, wherein: the construction of a visual interface to display the Pareto optimal solution set refers to using the front-end framework React.js to construct a visual interface, including a main chart area and a top information bar;

[0053] Display the optimization objective function in the main chart area and the Pareto optimal solution set in the top information bar;

[0054] Users who have passed real-name verification are allowed to view the information.

[0055] As a preferred solution of the multi-source intelligent power prediction method for a new energy power station described in the present invention, wherein: the storing of the multi-source data generated by the collection and analysis refers to sorting the collected multi-source data and the Pareto optimal solution set generated in the analysis process according to the timestamp and storing them in a database, the database is sorted in chronological order, and the stored data is marked;

[0056] Synchronize and regularly back up the stored data, regularly perform security and integrity checks on the stored data and backup data, and generate test reports.

[0057] In a second aspect, the present invention provides a multi-source intelligent power prediction system for a new energy power station, comprising: a collection and construction module for collecting and preprocessing multi-source data, and constructing an improved deep echo state network model to predict power demand values;

[0058] A solution execution module is used to calculate the load demand gap according to the power demand value, construct the objective function and solve the Pareto optimal solution set, execute the Pareto optimal solution set and optimize the dynamic weight of the objective function;

[0059] The visualization storage module is used to build a visualization interface to display the Pareto optimal solution set and store the multi-source data generated by collection and analysis.

[0060] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the multi-source intelligent power prediction method for a new energy power station as described in the first aspect of the present invention is implemented.

[0061] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the multi-source intelligent power prediction method for a new energy power station as described in the first aspect of the present invention.

[0062] The beneficial effects of the present invention are as follows: the present invention collects multi-source data and performs preprocessing to construct an improved deep echo state network model to predict power demand values; calculates the load demand gap according to the power demand value, constructs an objective function and solves the Pareto optimal solution set, executes the Pareto optimal solution set and optimizes the dynamic weight of the objective function; improves the accuracy and precision of multi-source intelligent power prediction of new energy power stations, enhances operational efficiency and decision-making support capabilities, and improves the model's ability to fit complex nonlinear relationships in power data of new energy power stations. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0064] Figure 1 This is a flow chart of the multi-source intelligent power prediction method for a new energy power station in Example 1.

[0065] Figure 2 This is a schematic diagram of a multi-source intelligent power prediction system for a new energy power station in Example 1. DETAILED DESCRIPTION

[0066] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0067] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0068] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0069] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and provides a multi-source intelligent power prediction method for a new energy power station, comprising the following steps:

[0070] This embodiment also provides a multi-source intelligent power prediction system for a new energy power station, including:

[0071] S1, collect and pre-process multi-source data, and build an improved deep echo state network model to predict power demand values;

[0072] Specifically, collecting multi-source data and pre-processing it means using smart sensors to collect multi-source data from new energy power stations and pre-processing it;

[0073] The intelligent sensors include voltage, current, resistance, temperature, wind speed and radiation intensity sensors;

[0074] The multi-source data includes voltage, current, resistance, temperature, wind speed and solar radiation data;

[0075] The preprocessing includes using the network time protocol for time synchronization, using wavelet transform for denoising, using the interquartile range method to identify and delete outliers, using the local mean method to fill in missing values, and standardizing multi-source data.

[0076] Through intelligent sensors, the operation status and external environment of the new energy power station are fully monitored, and multi-source data (such as voltage, current, temperature, wind speed, etc.) are collected, which can fully reflect the operation dynamics and external environmental characteristics of the power station. This comprehensive and real-time data collection lays a data foundation for the power prediction of the new energy power station, effectively avoiding the problem of incomplete information caused by a single data source in the traditional method. The collected data is synchronized by the network time protocol (NTP), eliminating the offset or misalignment problem in the time dimension of the data. The wavelet transform can decompose and suppress the complex noise in the multi-source data of the new energy power station, and separate the noise from the effective signal. The interquartile range method is used to identify and delete outliers in the data, which can avoid the deviation caused by outliers, thereby ensuring the authenticity and reliability of the data. The missing values ​​in the multi-source data are filled by the local mean method, which not only maintains the integrity of the data, but also avoids the analysis error caused by data loss. The multi-source data is standardized, which effectively eliminates the scale difference between different eigenvalues, thereby improving the computational efficiency and stability of the prediction model.

[0077] Furthermore, constructing an improved deep echo state network model to predict power demand values ​​refers to collecting historical multi-source data and preprocessing them to generate a training set;

[0078] Constructing an improved deep echo state network model including an input layer, an echo layer and an output layer;

[0079] Use the weighted linear combination method to calculate the data weights of multi-source data, and use the ReLU function to construct the multi-source data into a weighted function , the formula is;

[0080] ,

[0081] in is the data weight of multi-source data, is the i-th preprocessed multi-source data at time t;

[0082] The activation function is constructed by multiplying the Tanh and Sigmoid functions. ;

[0083] The activation function not only enhances the nonlinear expression ability of the network, but also solves the gradient vanishing problem that may occur in traditional activation functions;

[0084] The input of the improved deep echo state network model is set to be the preprocessed multi-source data;

[0085] Calculate the state vector of multi-source data at time t based on the basic echo layer , the formula is:

[0086] ,

[0087] in is the hyperbolic tangent function, is the input connection matrix of the echo layer, is the state vector of the basic echo layer at time t, is the echo connection matrix of the echo layer, which represents the influence of the hidden state at the previous moment on the hidden state at the current moment, capturing the dynamics in the time series, and b is the bias term;

[0088] Calculate the power prediction value of the base time t based on the base output layer , the formula is:

[0089] ,

[0090] in is the weight matrix of the output layer;

[0091] Calculate the state vector of multi-source data at time t based on the improved echo layer , the formula is:

[0092] ,

[0093] in is the activation function, is the input connection matrix of the echo layer, is the i-th preprocessed multi-source data at time t, is the state vector of the improved echo layer at time t, is the echo connection matrix of the echo layer, is the multi-source data weight matrix, b is the bias term;

[0094] The traditional model only relies on input weights and echo connection weight matrices, ignoring the correlation between external multi-source data. The model's sensitivity to multi-source input is enhanced through external features. The tanh function easily causes gradient vanishing in high-dimensional scenarios, weakening the network's learning ability. The activation function realizes adaptive adjustment of inputs in different ranges, thereby enhancing the ability to extract nonlinear features. The improved echo layer state calculation formula can more accurately capture the complex input and power output relationship in new energy power stations through the fusion of multi-source data, the enhancement of nonlinear expression, and the dynamic weight allocation, significantly improving the power prediction accuracy.

[0095] The state vector at time t , the output layer predicts the power demand value at time t after improvement , the formula is:

[0096] ,

[0097] in is the output layer weight matrix, is the Hadamard product;

[0098] Traditional models directly map input data through input weight matrices, but fail to fully consider the complex interactive relationship between multi-source inputs (such as voltage, current, temperature, etc.). Multi-source data is weighted and fused through weighting functions, and the contribution of each feature to power prediction is dynamically adjusted. The echo layer state is combined with the input feature weights through element-by-element multiplication to achieve dynamic correlation modeling at the feature level. This can more accurately capture the complex dependency between input features and echo layer states. The model can capture the nonlinear relationship between multi-source inputs and power output in a more fine-grained manner, thereby improving prediction accuracy, avoiding the gradient vanishing problem, and ensuring the stability of the model and training efficiency.

[0099] Using the training set to train the improved deep echo state network model, and using the differential evolution algorithm and the weighted loss function to iteratively optimize the parameters of the improved deep echo state network model;

[0100] Input the preprocessed multi-source data into the trained improved deep echo state network model to predict the power prediction value at the improved time t .

[0101] By collecting historical multi-source data and preprocessing it, removing noise and filling missing values, the quality of model input is guaranteed and the prediction accuracy is improved. The preprocessed data has strong representativeness and consistency, providing a stable foundation for model training. The combined activation function of Tanh and Sigmoid enhances the network's ability to handle complex nonlinear relationships. The weighted linear combination makes the importance of each data source in the model reasonably reflected, and improves the comprehensive prediction ability of the model. The design of the echo layer can fully capture the time series characteristics of the input data and improve the modeling effect of time series data. Through the Hadamard product, the calculation of the output layer is more efficient and flexible, ensuring the accuracy of power prediction. The differential evolution algorithm can jump out of the local optimum and find the global optimal solution, improving the generalization ability of the model. By continuously optimizing the model parameters, the accurate prediction of power demand is finally achieved to meet the high requirements of new energy power stations for power management. Accurate power prediction can optimize grid scheduling and load distribution, improve energy efficiency, and reduce the uncertainty of power supply through accurate power prediction, enhancing the stability and reliability of new energy power stations.

[0102] S2. Calculate the load demand gap according to the power demand value, construct the objective function and solve the Pareto optimal solution set, execute the Pareto optimal solution set and optimize the dynamic weight of the objective function;

[0103] Specifically, the load demand gap is calculated according to the power demand value, the objective function is constructed and the Pareto optimal solution set is solved. The unit electricity price D(t) at time t is collected from the national electricity market trading platform;

[0104] Calculate the charge and discharge power at time t based on the voltage and current in the preprocessed multi-source data ;

[0105] The charging and discharging power according to time t , use the trapezoidal method to calculate the total energy of charging and discharging, and calculate the charging and discharging efficiency ;

[0106] Calculate the charge and discharge energy cost at time t based on the charge and discharge power at time t , the formula is:

[0107] ,

[0108] in is the charge and discharge efficiency;

[0109] The charging and discharging power according to time t and unit electricity price D(t) to calculate the electricity transaction cost ;

[0110] The energy cost of charging and discharging at time t and electricity transaction costs The total running cost of adding up the computation time t ;

[0111] Calculate the transmission loss at time t based on the current and resistance in the preprocessed multi-source data ;

[0112] The charge and discharge power at time t Reduce transmission loss Get the load demand at time t ;

[0113] The power demand value according to time t Load shedding requirements Calculate the load demand gap at time t ;

[0114] Total operating cost according to time t and load demand gap Calculate the information entropy of the total operating cost and the information entropy of load demand gap , use the entropy weight method to calculate the total operating cost weight and the load demand gap weight respectively, and divide the load demand gap weight by the total operating cost weight to calculate the dynamic weight ;

[0115] The objective function F is constructed using the multi-objective optimization method, and the formula is:

[0116] ,

[0117] Where T is the total time;

[0118] In actual power systems, the balance between supply and demand of load demand is crucial. The load gap indicates that the current output power cannot meet user demand. Through the multi-objective optimization method, the load gap and equipment aging cost are comprehensively considered, and the trade-off parameters are used to flexibly adjust between the two. This method avoids the one-sidedness of single-objective optimization and makes the optimization results more in line with actual needs. In many optimization problems, the load gap in absolute value form may be directly used for optimization, and it is difficult to impose stricter penalties on large deviations. Taking the square of the load gap as part of the objective function can impose stronger constraints on larger load deviations, making the optimization results more balanced, improving the ability of the power station to meet user load demand at any time, and avoiding power shortages or waste. Multi-objective optimization combines the two core indicators of load demand and equipment maintenance, avoiding the local optimal phenomenon that may be caused by a single optimization goal;

[0119] Randomly generate N particles in the population, each particle represents a solution ;

[0120] Perform Pareto sorting on all solutions in the population, identify the dominance relationship, use the non-dominated sorting method to screen, and select the Pareto optimal solution set;

[0121] Calculate the objective function value of the Pareto optimal solution set and identify the maximum value of the objective function value of the jth Pareto optimal solution and minimum value ;

[0122] Under the objective function, the distribution of solutions usually forms an area rather than a single point, and its value forms a range within this area, with the maximum and minimum values ​​appearing on the boundaries of this area;

[0123] Use kernel density estimation to calculate the relative probability density of each Pareto optimal solution on the objective function, and use Shannon entropy to calculate the information entropy of the objective function corresponding to the j-th Pareto optimal solution based on the relative probability density. ;

[0124] According to the maximum value of the objective function value of the jth Pareto optimal solution , minimum And the information entropy of the corresponding objective function , calculate the congestion degree of the jth Pareto optimal solution , the formula is:

[0125] ,

[0126] in and are the objective function values ​​corresponding to the Pareto optimal solution of the nth particle in the jth neighbor, is a small constant;

[0127] The sum in the formula is the calculation of the objective function for each Pareto optimal solution, which is the difference in the objective function value between adjacent particles;

[0128] Select the Pareto optimal solution with the highest congestion to enter the next generation;

[0129] Use particle swarm optimization algorithm to calculate and update particle speed , update the particle solution , the formula is:

[0130] ,

[0131] in is the nth particle at time t;

[0132] The maximum number of iterations is set using the rule of thumb method. When the maximum number of iterations is reached, the iteration is stopped and the Pareto optimal solution set is output, including the optimized total operating cost and load demand gap.

[0133] By collecting electricity prices and equipment operation data in real time, an accurate input data set is constructed, which improves the real-time and reliability of the optimization process. The calculation of energy costs and electricity transaction costs provides accurate operating cost data, which helps the optimization algorithm to achieve reasonable distribution of operating loads while ensuring economy. By accurately modeling the load demand gap, the modeling accuracy of the objective function is significantly improved. The optimization results can better meet the actual operation needs of new energy power stations. The entropy weight method dynamically adjusts the weight distribution, so that the optimization results can flexibly respond to the uncertainty and volatility in the operation of new energy power stations. The multi-objective optimization method ensures the trade-off between total operating costs and load demand gaps. Pareto sorting and congestion calculation improve the distribution quality and global optimality of the solution. The particle swarm optimization algorithm is combined with Pareto sorting to ensure the global optimality of the solution, avoiding the shortcomings of traditional optimization methods that are prone to fall into local optimality. Through the detailed calculation of various costs and demands in the objective function, the interpretability of the optimization results is significantly improved, providing a clear basis for decision support for new energy power stations.

[0134] Furthermore, executing the Pareto optimal solution set and optimizing the dynamic weight of the objective function means converting the Pareto optimal solution set into a scheduling instruction using a parameter instruction mapping method, transmitting the instruction to the control platform through an API interface, and the control platform receiving the scheduling instruction and executing it, and collecting the operation data after the execution;

[0135] Calculate the difference between the running data and the optimal solution set;

[0136] Use statistical methods to set the judgment threshold, compare the difference with the judgment threshold, and use the difference to adjust the dynamic weight of the objective function F when the difference is greater than or equal to the judgment threshold. , until the difference is less than the judgment threshold, stop adjusting and continue to monitor the difference results.

[0137] Through parameter mapping, rapid conversion from theoretical optimization solution to actual dispatch instructions is achieved, which speeds up decision-making. The API interface ensures efficient transmission and execution of dispatch instructions without human intervention, which improves the intelligence level of the system. Through data feedback, a closed-loop mechanism of optimization and execution is formed to ensure the efficient operation of the dispatch system. Difference analysis provides quantitative performance indicators and provides data support for further optimization of the objective function. Through dynamic weight adjustment, the optimization objective function can adapt to changes in the real-time operating environment, which improves the practicality and robustness of the dispatch scheme. The setting of the judgment threshold ensures the scientific nature and stability of the weight adjustment and avoids unnecessary frequent adjustments. The continuous monitoring mechanism ensures that the dispatch system can operate stably for a long time and maintain performance even when environmental conditions change. Through long-term data accumulation and analysis, it provides a reference for future optimization and gradually improves the intelligence level of the overall system.

[0138] S3, build a visual interface to display the optimization objective function, store the multi-source data generated by collection and analysis;

[0139] Specifically, building a visual interface to display the Pareto optimal solution set refers to using the front-end framework React.js to build a visual interface, including the main chart area and the top information bar;

[0140] Display the optimization objective function in the main chart area and the Pareto optimal solution set in the top information bar;

[0141] Users who have passed real-name verification are allowed to view the information.

[0142] The component reusability of React.js simplifies the development process and improves interface development efficiency. Dynamic charts visualize abstract mathematical functions and lower the threshold for users to understand. Users can flexibly analyze the changing trends of optimization goals through interactive operations, improving decision-making efficiency. The multi-objective characteristics of each solution in the Pareto optimal solution set are clearly displayed to help users understand the trade-offs of optimization solutions. Users can quickly find solutions that meet specific needs through the filtering function, improving selection efficiency. The real-name verification mechanism limits unauthorized access and ensures the security of sensitive data.

[0143] Furthermore, storing the multi-source data generated by the collection and analysis means sorting the collected multi-source data and the Pareto optimal solution set generated during the analysis process according to timestamps and storing them in a database, the database is sorted in chronological order, and the stored data is marked;

[0144] Synchronize and regularly back up the stored data, regularly perform security and integrity checks on the stored data and backup data, and generate test reports.

[0145] Timestamp sorting ensures the time consistency of data storage and provides a reliable foundation for subsequent dynamic analysis. Through marking and sorting, users can quickly locate target data, improving data query efficiency. The backup mechanism provides a recovery path for data loss or damage, enhancing the system's disaster recovery capabilities. Even in unexpected situations (such as hardware failure, network interruption), the system can still recover quickly to ensure the availability of key data. Security and integrity testing provides technical guarantees for the authenticity and reliability of data. The generated test report provides system administrators with a comprehensive data health status reference, facilitating timely troubleshooting and repair of problems. Dynamic management reduces the storage space occupied by invalid data and improves the storage efficiency of the system.

[0146] This embodiment also provides a computer device, which is suitable for use in a multi-source intelligent power prediction method for a new energy power station, and includes: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the multi-source intelligent power prediction method for a new energy power station as proposed in the above embodiment.

[0147] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0148] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the multi-source intelligent power prediction method for a new energy power station proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (Static Random Access Memory, referred to as SRAM), an electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), an erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), a programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), a read-only memory (Read-Only Memory, referred to as ROM), a magnetic memory, a flash memory, a magnetic disk or an optical disk.

[0149] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A multi-source intelligent power prediction method for a new energy power station, characterized in that: include, Collect and pre-process multi-source data, and build an improved deep echo state network model to predict power demand values; Calculate the load demand gap according to the power demand value, construct the objective function and solve the Pareto optimal solution set, execute the Pareto optimal solution set and optimize the dynamic weight of the objective function; The calculation of the load demand gap according to the power demand value, the construction of the objective function and the solution of the Pareto optimal solution set refer to the unit electricity price D(t) collected at time t from the national electricity market trading platform; Calculate the charge and discharge power at time t based on the voltage and current in the preprocessed multi-source data ; Calculate the charge and discharge energy cost at time t based on the charge and discharge power at time t ; The charging and discharging power according to time t and unit electricity price D(t) to calculate the electricity transaction cost ; The charge and discharge energy cost of time t and electricity transaction costs The total running cost of adding up the computation time t ; Calculate the transmission loss at time t based on the current and resistance in the preprocessed multi-source data ; The charge and discharge power at time t Reduce transmission loss Load demand at time t ; The power demand value according to time t Load shedding requirements Calculate the load demand gap at time t ; Total operating cost according to time t and load demand gap Calculate the information entropy of the total operating cost and the information entropy of load demand gap , use the entropy weight method to calculate the total operating cost weight and the load demand gap weight respectively, and divide the load demand gap weight by the total operating cost weight to calculate the dynamic weight ; The objective function F is constructed using the multi-objective optimization method, and the formula is: ， Where T is the total time; Randomly generate N particles in the population, each particle represents a solution , Perform Pareto sorting on all solutions in the population, identify the dominance relationship, use the non-dominated sorting method to screen, and select the Pareto optimal solution set; Calculate the objective function value of the Pareto optimal solution set and identify the maximum value of the objective function value of the jth Pareto optimal solution and minimum value ; Use kernel density estimation to calculate the relative probability density of each Pareto optimal solution on the objective function, and use Shannon entropy to calculate the information entropy of the objective function corresponding to the j-th Pareto optimal solution based on the relative probability density. ; According to the maximum value of the objective function value of the jth Pareto optimal solution , minimum And the information entropy of the corresponding objective function , calculate the congestion degree of the jth Pareto optimal solution , the formula is: ， in and are the values ​​of the objective function neighbors corresponding to the Pareto optimal solution of the nth particle in the jth neighbor, is a small constant; Select the Pareto optimal solution with the highest congestion to enter the next generation; Use particle swarm optimization algorithm to calculate and update particle speed , update the particle solution ; Use the rule of thumb method to set the maximum number of iterations, stop the iteration when the maximum number of iterations is reached, and output the Pareto optimal solution set including the optimized total operating cost and load demand gap; Build a visual interface to display the Pareto optimal solution set and store the multi-source data generated by collection and analysis.

2. The multi-source intelligent power prediction method for a new energy power station according to claim 1, characterized in that: The collecting of multi-source data and pre-processing thereof refers to using smart sensors to collect multi-source data of new energy power stations and pre-processing thereof; The intelligent sensors include voltage, current, resistance, temperature, wind speed and radiation intensity sensors; The multi-source data includes voltage, current, resistance, temperature, wind speed and solar radiation data; The preprocessing includes using the network time protocol for time synchronization, using wavelet transform for denoising, using the interquartile range method to identify and delete outliers, using the local mean method to fill in missing values, and standardizing multi-source data.

3. The multi-source intelligent power prediction method for a new energy power station according to claim 2, characterized in that: The said constructing an improved deep echo state network model to predict the power demand value refers to collecting historical multi-source data and preprocessing it to generate a training set; Constructing an improved deep echo state network model including an input layer, an echo layer and an output layer; Use the weighted linear combination method to calculate the data weight of multi-source data, and use the ReLU function to construct the multi-source data as a weighted function of time t ; The activation function is constructed by multiplying the Tanh and Sigmoid functions. ; The input of the improved deep echo state network model is set to be the preprocessed multi-source data; Calculate the state vector of multi-source data at time t based on the improved echo layer , the formula is: ， in is the activation function, is the input connection matrix of the echo layer, is the i-th preprocessed multi-source data at time t, is the state vector of the improved echo layer at time t, is the echo connection matrix of the echo layer, is the multi-source data weight matrix, b is the bias term; The state vector at time t , the output layer predicts the power demand value at time t after improvement , the formula is: ， in is the output layer weight matrix, is the Hadamard product; Using the training set to train the improved deep echo state network model, and using the differential evolution algorithm and the weighted loss function to iteratively optimize the parameters of the improved deep echo state network model; Input the preprocessed multi-source data into the trained improved deep echo state network model to predict the power demand value at the improved time t .

4. The multi-source intelligent power prediction method for a new energy power station according to claim 3, characterized in that: The executing of the Pareto optimal solution set and optimizing the dynamic weight of the objective function refers to converting the Pareto optimal solution set into a scheduling instruction using a parameter instruction mapping method, transmitting the instruction to the control platform through an API interface, and the control platform receiving the scheduling instruction and executing it, and collecting the operation data after the execution; Calculate the difference between the execution running data and the optimal solution set; Use statistical methods to set the judgment threshold, compare the difference with the judgment threshold, mark the difference greater than or equal to the judgment threshold, and use the marked difference to adjust the dynamic weight of the objective function , until the difference is less than the judgment threshold, stop adjusting and continue to monitor the difference results.

5. The multi-source intelligent power prediction method for a new energy power station according to claim 4, characterized in that: The said constructing a visualization interface to display the Pareto optimal solution set refers to using the front-end framework React.js to construct a visualization interface, including a main chart area and a top information bar; Display the optimization objective function in the main chart area and the Pareto optimal solution set in the top information bar; Users who have passed real-name verification are allowed to view the information.

6. The multi-source intelligent power prediction method for a new energy power station according to claim 5, characterized in that: The storing of the multi-source data generated by the collection and analysis refers to sorting the collected multi-source data and the Pareto optimal solution set generated during the analysis process according to timestamps and storing them in a database, the database is sorted in chronological order, and the stored data is marked; Synchronize and regularly back up the stored data, regularly perform security and integrity checks on the stored data and backup data, and generate test reports.

7. A multi-source intelligent power prediction system for a new energy power station, based on the multi-source intelligent power prediction method for a new energy power station according to any one of claims 1 to 6, characterized in that: include, A collection and construction module is used to collect and preprocess multi-source data and construct an improved deep echo state network model to predict power demand values; A solution execution module is used to calculate the load demand gap according to the power demand value, construct the objective function and solve the Pareto optimal solution set, execute the Pareto optimal solution set and optimize the dynamic weight of the objective function; The visualization storage module is used to build a visualization interface to display the Pareto optimal solution set and store the multi-source data generated by collection and analysis.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the multi-source intelligent power prediction method for a new energy power station described in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the multi-source intelligent power prediction method for a new energy power station described in any one of claims 1 to 6 are implemented.

Images