New energy power station planning method and system
By combining static and dynamic feature screening in the planning method for new energy power plants, the problem of lack of multi-dimensional evaluation in traditional site selection methods has been solved, realizing the scientificity and reliability of power plant site selection and ensuring the safety and stability of power plants.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-11
- Publication Date
- 2026-04-14
AI Technical Summary
Traditional methods for selecting sites for new energy power plants lack systematic integration and quantitative evaluation of multi-source and multi-dimensional information, leading to problems such as high risk of geological disasters, low power generation efficiency, and high operation and maintenance costs after the construction of power plants under complex geographical and climatic conditions.
A method combining static and dynamic feature screening is adopted. By acquiring the static features (such as earthquake activity frequency, altitude, slope, and groundwater depth) and dynamic features (such as wind speed, solar radiation intensity, ambient temperature, ambient humidity, and precipitation) of candidate points, a dual screening is performed. Combined with cluster analysis and policy similarity evaluation, the final points are selected.
It enables multi-dimensional and systematic evaluation of candidate sites, improves the scientific nature and reliability of site selection decisions, ensures the safety and stability of power plants throughout their entire life cycle, and reduces natural risks and operating costs.
Smart Images
Figure CN121860118A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of new energy power plant planning technology, and more specifically, to a new energy power plant planning method and system. Background Technology
[0002] Site selection for new energy power plants still faces many challenges. Traditional site selection methods often rely on expert experience or single-dimensional data analysis, lacking systematic integration and quantitative evaluation of multi-source and multi-dimensional information. Especially under complex geographical and climatic conditions, the static geological conditions and dynamic climate resources of candidate sites interact with each other. Without scientific screening, this can easily lead to problems such as high risk of geological disasters, low power generation efficiency, and high operation and maintenance costs after the power plant is built. Summary of the Invention
[0003] The purpose of this invention is to provide a method and system for planning new energy power plants to improve the above-mentioned problems.
[0004] To achieve the above objectives, in one aspect, the present invention provides a method for planning new energy power plants, the method comprising:
[0005] Obtain static features of multiple candidate points, wherein the static features include seismic activity frequency, altitude, slope and groundwater depth;
[0006] Calculate the static anomaly score of each candidate point based on its static features, and perform a first round of screening on all candidate points based on the static anomaly scores to obtain the remaining candidate points.
[0007] The dynamic features of each remaining candidate point are obtained, including wind speed, solar radiation intensity, ambient temperature, ambient humidity, and precipitation. Based on the dynamic features of each remaining candidate point and the static anomaly score of each candidate point, a second round of screening is performed on all remaining candidate points to obtain the final location.
[0008] Secondly, the present invention provides a new energy power plant planning system, the system comprising:
[0009] The acquisition module is used to acquire the static features of multiple candidate points, wherein the static features include seismic activity frequency, altitude, slope and groundwater depth;
[0010] The first screening module is used to calculate the static anomaly score of each candidate point based on the static features of the candidate points, and to perform the first round of screening on all candidate points based on the static anomaly scores to obtain the remaining candidate points.
[0011] The second screening module is used to obtain the dynamic characteristics of each remaining candidate point, wherein the dynamic characteristics include wind speed, solar radiation intensity, ambient temperature, ambient humidity and precipitation; based on the dynamic characteristics of each remaining candidate point and the static anomaly score of each candidate point, a second round of screening is performed on all remaining candidate points to obtain the final location.
[0012] Thirdly, the present invention provides a new energy power plant planning device, the device comprising a memory and a processor. The memory is used to store a computer program; the processor is used to execute the computer program to implement the steps of the above-described new energy power plant planning method.
[0013] Fourthly, the present invention provides a readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described new energy power plant planning method.
[0014] The beneficial effects of this invention are as follows:
[0015] (1) This invention achieves a multi-dimensional and systematic evaluation of candidate sites by introducing a dual screening mechanism of static and dynamic anomaly scores. First, based on static characteristics such as seismic activity frequency, altitude, slope, and groundwater depth, candidate sites with unstable geological conditions or unsuitable geographical environments are eliminated, thus avoiding natural risks in engineering construction and long-term operation from the source. Subsequently, by combining historical time-series data of dynamic characteristics such as wind speed, solar radiation intensity, temperature, humidity, and precipitation, sites with large resource fluctuations and unreliable supply are further screened out through moving average prediction and anomaly identification. This method significantly improves the scientificity and reliability of site selection decisions, ensuring the safety and stability of the power station throughout its entire life cycle.
[0016] (2) In the static screening stage, this invention introduces a clustering optimization mechanism. When there are too many candidate points remaining after the initial screening, the most representative points are automatically selected through cluster analysis, which ensures screening efficiency and avoids the omission of high-quality points. In the dynamic screening stage, through standardization and moving average prediction, the dynamic characteristics of different dimensions and benchmarks are unified into comparable parameters, which enhances the accuracy and adaptability of anomaly detection. Finally, among multiple intermediate candidate points, the final point that is technically reliable and economically optimal is selected through policy similarity analysis and comprehensive evaluation of return on investment.
[0017] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings. Attached Figure Description
[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0019] Figure 1 This is a schematic diagram of the new energy power plant planning method described in the embodiments of the present invention;
[0020] Figure 2 This is a schematic diagram of the new energy power plant planning system structure described in this embodiment of the invention;
[0021] Figure 3 This is a schematic diagram of the planning equipment structure for a new energy power plant as described in this embodiment of the invention. Detailed Implementation
[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.
[0023] It should be noted that similar reference numerals or letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.
[0024] Example 1
[0025] like Figure 1 As shown, this embodiment provides a new energy power plant planning method, which includes SSS:
[0026] S1: Obtain the static features of multiple candidate points, wherein the static features include seismic activity frequency, altitude, slope and groundwater depth;
[0027] In this step, static characteristics can be understood as attribute parameters used to describe the inherent, long-term stable geographical, geological, infrastructure, and economic conditions of the candidate site. These parameters remain essentially unchanged or change slowly throughout the power plant's life cycle and are fundamental prerequisites for power plant site selection. The static characteristics include, but are not limited to, seismic activity frequency, altitude, slope, and groundwater depth. The seismic activity frequency can be calculated by statistically analyzing the number of earthquakes that have occurred in the city where the candidate site is located over the past thirty years and then calculating the annual average number of earthquakes.
[0028] S2: Calculate the static anomaly score of each candidate point based on the static features of the candidate points, and perform the first round of screening on all candidate points according to the static anomaly scores to obtain the remaining candidate points;
[0029] In this step, the static anomaly score of each candidate point is calculated based on the static features of the candidate points, and the first round of screening of all candidate points is carried out according to the static anomaly score to obtain the remaining candidate points. The specific implementation steps include S21 and S22.
[0030] S21: Organize the G static features of the collected I candidate points into an I-row, G-column original data matrix A, where each row represents all static features of a candidate point, and each column represents the same feature dimension of all candidate points; for each column in the original data matrix, calculate the mean of the g-th column. and standard deviation ;
[0031] In this step, I represents the total number of candidate points; G represents the number of static features. For example, if the static features are seismic activity frequency, altitude, slope, and groundwater depth, then G is 4. Assuming G is 4, then g is 1, 2, 3, or 4.
[0032] S22: Perform a standardization transformation on each data element in the original data matrix, where for the element in the i-th row and g-th column of the original matrix... Its standardized value The calculation formula is All obtained after standardization Reassemble the data to form a standardized data matrix B with the same dimensions as the original data matrix; calculate the covariance matrix corresponding to the standardized data matrix, and perform eigenvalue decomposition on the covariance matrix to obtain the eigenvector matrix and the diagonal eigenvalue matrix; calculate the static anomaly score of each candidate point based on the eigenvector matrix and the diagonal eigenvalue matrix, and obtain the remaining candidate points based on the static anomaly scores.
[0033] This step can be understood as: calculating the covariance matrix of the standardized matrix B. , Transpose the standardized matrix B to obtain ,according to The formula performs eigenvalue decomposition on the covariance matrix, outputting an eigenvector matrix P: an orthogonal matrix of size G x G, where each column is an eigenvector of the covariance matrix, and these eigenvectors are unit vectors and orthogonal to each other. It also outputs a diagonal eigenvalue matrix D: a diagonal eigenvalue matrix of size G x G, where all off-diagonal elements are zero. (The diagonal elements...) arrive These are the eigenvalues of the covariance matrix. The eigenvalues are arranged in descending order, i.e., satisfying... ≥ ≥…≥ ≥0. It is the transpose of the eigenvector matrix P;
[0034] Simultaneously, in this step, the static anomaly score of each candidate point is calculated based on the eigenvector matrix and the diagonal eigenvalue matrix. The specific implementation steps for obtaining the remaining candidate points based on the static anomaly score include S221:
[0035] S221: Find the smallest j such that ≥0.95, Let be the element in the g-th row and g-th column of the diagonal eigenvalue matrix; select the first j columns of the eigenvector matrix to obtain the mapping matrix; multiply the mapping matrix by the standardized data matrix to obtain the first matrix; multiply the first matrix by the transpose of the mapping matrix to obtain the second matrix; for each candidate point, calculate its static anomaly score, the formula for which is:
[0036] ;
[0037] In the formula, The static anomaly score of the candidate point corresponding to the i-th row in the standardized data matrix; Let be the element in the i-th row and g-th column of the standardized data matrix; Let be the element in the i-th row and g-th column of the second matrix; is the element in the g-th row and g-th column of the diagonal eigenvalue matrix;
[0038] S222: Compare the static anomaly score with the preset static score threshold. If it is greater than the static score threshold, delete the candidate point to obtain the primary remaining candidate points. Count the number of primary remaining candidate points. If it is less than the preset first number, record the primary remaining candidate point as a remaining candidate point. Otherwise, combine all the static features corresponding to each primary remaining candidate point to obtain a combination vector. Based on the combination vectors of all primary remaining candidate points, use a clustering algorithm to cluster them into multiple clusters. For each primary remaining candidate point, calculate the Euclidean distance from its combination vector to the centroid of its own cluster. Sort all primary remaining candidate points in ascending order according to their corresponding Euclidean distance values, and select the first M primary remaining candidate points as remaining candidate points, where M is a positive integer.
[0039] The purpose of this step is to automatically and efficiently select the most representative high-quality points when there are too many points that pass the initial screening. The combination vector can be understood as combining all the static feature values of each primary remaining candidate point into a multi-dimensional vector (such as [seismic activity frequency, altitude, slope, groundwater depth]); or each row in matrix B can be directly used as the combination vector of the primary remaining candidate points.
[0040] The above screening process employs a two-tiered mechanism of anomaly removal and optimization. First, static screening identifies and removes anomalous candidate points from high-dimensional features, mitigating engineering risks at the source. Subsequently, cluster optimization selects representative candidate points, providing a high-quality foundation for determining the final location.
[0041] The long-term stable operation of a power plant site depends not only on a stable static environment but also on the reliability of its dynamic energy resources. Therefore, based on the static screening, this plan will further conduct dynamic risk assessments on the remaining candidate sites. By analyzing their historical time-series data, it will accurately identify high-risk sites that meet the static conditions but have unstable resource supply, thereby ensuring the efficiency and operational safety of the power plant.
[0042] S3: Obtain the dynamic features of each remaining candidate point, wherein the dynamic features include wind speed, solar radiation intensity, ambient temperature, ambient humidity and precipitation; based on the dynamic features of each remaining candidate point and the static anomaly score of each candidate point, perform a second round of screening on all remaining candidate points to obtain the final location.
[0043] In this step, the final location is the final site selection for the power plant. Dynamic characteristics can be understood as attribute parameters that describe the natural environmental resource conditions of the candidate site and fluctuate frequently over time. These parameters exist in the form of time-series data. Dynamic characteristics include, but are not limited to, wind speed, solar irradiance (solar irradiance), ambient temperature, ambient humidity, and precipitation.
[0044] In this step, a second-round screening is performed on all remaining candidate points based on the dynamic features of each remaining candidate point and the static anomaly score of each candidate point. The specific implementation steps for obtaining the final point position include S31 and S32;
[0045] S31: Obtain the true value of each dynamic feature of each remaining candidate point at each moment within a preset historical period in the past, and form a true value sequence corresponding to each dynamic feature; for each true value sequence, calculate the mean and standard deviation of all true values, subtract the mean from the true value at each moment to obtain a first value, and record the ratio between the first value and the standard deviation as the standardized true value at each moment, so as to obtain a standardized true value sequence corresponding to each dynamic feature; based on the true value sequence corresponding to each dynamic feature, use the simple moving average method to calculate the predicted value of the dynamic feature at each moment within a preset historical period in the past, obtain a predicted value sequence, subtract the mean from the predicted value at each moment to obtain a second value, and record the ratio between the second value and the standard deviation as the standardized predicted value at each moment, so as to obtain a standardized predicted value sequence corresponding to each dynamic feature;
[0046] In this step, the preset historical period in the past can be the past year;
[0047] The true value sequence can be understood as: for each dynamic feature (such as wind speed) of each point position, obtain its data every day within a past period (such as 365 days) to form a sequence X = [x1, x2, ..., xT].
[0048] Based on the true value sequence corresponding to each dynamic feature, using the simple moving average method to calculate the predicted value of the dynamic feature at each moment within a preset historical period in the past, and obtaining a predicted value sequence can be understood as: for the moment t (t ≥ L), L represents the window size in the moving average algorithm, which is a preset positive integer parameter, and its predicted value is equal to the average value of the true values in the past L moments; for the moment t (1 < t < L), its predicted value is equal to the average value of all true values from the start of the sequence to the previous moment (t - 1); for the moment t (t = 1), directly use the true value as the predicted value;
[0049] S32: Calculate the dynamic anomaly score corresponding to each dynamic feature according to the standardized predicted value sequence and the standardized true value sequence; compare the dynamic anomaly score corresponding to each dynamic feature with its corresponding dynamic score threshold. If the dynamic anomaly score is greater than the dynamic score threshold, then determine that this dynamic feature is an abnormal dynamic feature; for each remaining candidate point, count the number of abnormal dynamic features in its corresponding dynamic features. If the number is greater than the preset second number, then determine that this remaining candidate point is abnormal, delete it, and obtain intermediate candidate points; obtain the final point position according to the intermediate candidate points.
[0050] In S31 and S32, firstly, by standardizing and using moving average prediction on historical time-series data, dynamic characteristics of different dimensions and benchmarks are unified into comparable parameters, establishing a reliable foundation for anomaly detection. Subsequently, by calculating dynamic anomaly scores and comparing them with thresholds, anomaly characteristics are effectively identified. Finally, a comprehensive decision-making mechanism based on anomaly characteristic counting filters out high-risk locations that meet static conditions but have unreliable dynamic resource supply.
[0051] In this step, the specific implementation steps for obtaining the final location based on the intermediate candidate points include S321;
[0052] S321: Analyze the intermediate candidate points again. If there is only one intermediate candidate point, it is taken as the final point. If there is more than one intermediate candidate point, the dynamic anomaly scores corresponding to the dynamic features contained in each intermediate candidate point are weighted and summed to obtain the total dynamic anomaly score of each intermediate candidate point. At the same time, the intermediate candidate points are merged and classified, and the final point is selected based on the total dynamic anomaly score of each intermediate candidate point and the merging and classification results.
[0053] In this step, the weighted summation of all dynamic anomaly scores corresponding to the dynamic features contained in each intermediate candidate point can be understood as follows: assuming there are 3 dynamic features, each corresponding to a dynamic anomaly score, then the weighted summation of these three dynamic anomaly scores yields the total dynamic anomaly score. The weights are manually assigned, and the sum of the weights is 1. Simultaneously, in this step, the intermediate candidate points are merged and categorized, and the final point is selected based on the total dynamic anomaly score of each intermediate candidate point and the merging and categorization results. The specific implementation steps include S3211.
[0054] S3211: Obtain the policy introduction information corresponding to each intermediate candidate point;
[0055] In this step, policy information refers to official policy documents and announcements related to the planning, construction, and operation of new energy power plants, collected, organized, and uploaded to the system by staff. The scope of acquisition is limited to policy information publicly released by local governments, energy bureaus, natural resources bureaus, and other relevant authorities within the municipal administrative region where each intermediate candidate site is located.
[0056] After this step, you can further filter the policy information. The specific steps are as follows:
[0057] The TF-IDF method is used to vectorize each policy introduction information into text vectors. All text vectors are then input into the K-Means clustering algorithm to obtain multiple clusters. For each cluster, the mean of the coordinates of all text vectors in that cluster is calculated to obtain the central feature vector of each cluster. The distance between each text vector and the central feature vector is calculated, and the text vector with the maximum distance is added to the vector set. The remaining text vectors are designated as unprocessed text vectors. The following steps are repeated until the number of text vectors in the vector set reaches the preset target number: Calculate the first distance between each unprocessed text vector and the central feature vector; calculate the second distance between each unprocessed text vector and each text vector in the vector set, and sum all the second distances to obtain the third distance; sum the first and third distances to obtain the total distance; add the unprocessed text vector with the maximum total distance to the vector set.
[0058] The text vectors in the vector set are sequentially combined with the central feature vectors along the row direction of the matrix to obtain a combination matrix, and the corresponding covariance matrix is calculated. Simultaneously, the eigenvectors of the covariance matrix are calculated to obtain the target feature vectors. The target feature vectors corresponding to each cluster are input into a preset analysis model. The analysis model is used to characterize the one-to-one correspondence between the target feature vectors and the clustering scores, obtaining the clustering score for each cluster. Each clustering score is compared with a preset clustering score threshold. The number of clusters with clustering scores less than the threshold is counted. If the number is greater than or equal to the preset target number threshold, the K value in the K-Means clustering algorithm is modified, and clustering is performed again until the number of clusters with clustering scores less than the threshold is less than the preset target number threshold, obtaining the final clusters. Each of the final clusters is designated as the first cluster. For each first cluster, the score of each text vector in the first cluster is calculated, where the score calculation formula is:
[0059] ;
[0060] In the formula, The score is the score of the s-th text vector in the first cluster. The distance between the s-th text vector in the first cluster and the cluster center of the first cluster; Let be the distance between the s-th text vector in the first cluster and the y-th text vector in the same cluster, where z is the total number of text vectors in the first cluster; e is the base of the exponential function; q is the truncation distance. The truncation distance is calculated as follows: calculate the distance between each pair of text vectors in the first cluster to obtain multiple distance values, sort the distance values in ascending order to obtain a distance value sorting list; calculate the product of the preset percentile and the total number of distance values to obtain the index position, where if the index position is not an integer, it is rounded up; find the distance value at that position in the distance value sorting list according to the index position and use it as the truncation distance.
[0061] Text vectors whose scores are greater than the preset text vector score threshold are removed from the first cluster; the policy introduction information and intermediate candidate points corresponding to the remaining text vectors are retained and entered into S3212, and the following steps of S3212 are executed.
[0062] The training of the analysis model can be as follows: obtain the target feature vector for training and the clustering scores of the staff who labeled it; after labeling, use the target feature vector as input and the clustering score label as output to train the convolutional neural network model to obtain the analysis model.
[0063] Convolutional neural network models include:
[0064] Input layer: Receives the target feature vector. One-dimensional convolutional layer: 64 kernels, kernel size 3, stride 1, using ReLU activation function. One-dimensional max-pooling layer: Pooling window size 2. Flattening layer: Converts the multi-dimensional input to one dimension. Fully connected layer: 128 neurons, using ReLU activation function. Output layer: 1 neuron, using linear activation function, outputting the clustering score.
[0065] The specific configuration for model training is as follows: Loss function: Mean squared error. Optimizer: Adam optimizer with an initial learning rate of 0.001. Training parameters: Batch size set to 32, training epochs to 100. During training, training is terminated early if the validation set loss does not decrease for 10 consecutive epochs. Data partitioning: The training dataset is randomly partitioned into training, validation, and test sets in a 7:2:1 ratio.
[0066] In addition to the training methods mentioned above, other conventional training methods can also be used.
[0067] S3212: Perform a merging operation: Calculate the edit distance between each pair of policy introduction information, compare the string lengths of the selected pairs of policy introduction information, and designate the policy introduction information with the larger string length as the target policy introduction information. Calculate the ratio between the edit distance and the string length of the target policy introduction information, and denote the difference between the preset constant and the ratio as the similarity between the pairs of policy introduction information. Merge pairs of policy introduction information with similarity greater than the similarity threshold into one category.
[0068] In this step, the constant is 1;
[0069] S3213: Continue the merging operation until the similarity between each pair of policy information is less than the similarity threshold, and then stop to obtain the final category. In each final category, select the intermediate candidate point with the lowest total anomaly score and set them together to obtain the intermediate candidate point set. The static anomaly score corresponding to the intermediate candidate point is weighted and summed with the total dynamic anomaly score to obtain the total anomaly score of the intermediate candidate point. Select the intermediate candidate point with the highest return on investment from the intermediate candidate point set as the final point.
[0070] In this step, the return on investment (ROI) is manually input into the system; the weights of the static anomaly score and the total dynamic anomaly score can be 0.6 and 0.4, respectively. This step identifies and merges candidate sites under similar policy environments by analyzing the similarity of policy information, forming "option groups" with equal investment potential. Based on this, the representative site with the best comprehensive technical indicators is first selected within each option group, and then the final site with the highest ROI is determined from all the representatives across groups. This mechanism effectively avoids excessive trade-offs on technical details, ensuring that the site selection conclusion meets the requirements of engineering and technical reliability while also aligning with the optimality of commercial investment, thereby significantly improving the project's feasibility and overall risk resistance.
[0071] Meanwhile, in S32, the specific implementation steps for calculating the dynamic anomaly score corresponding to each dynamic feature based on the standardized predicted value sequence and the standardized true value sequence include S322 and S323.
[0072] S322: Divide the standardized true value sequence into multiple subsequences according to the preset window size; for each subsequence, calculate the absolute value of the difference between the standardized data and the standardized predicted value at each time step to obtain the third value; sum the third values corresponding to each time step in the subsequence to obtain the first factor value corresponding to each subsequence; sum the absolute values of the standardized predicted values corresponding to each time step in the subsequence to obtain the fourth value; take the square root of the fourth value to obtain the second factor value corresponding to each subsequence; and record the ratio between the first factor value and the second factor value as the anomaly score of each subsequence.
[0073] In this step, the preset window size can be 30 time points, and the window size can be customized.
[0074] S323: Sort the anomaly scores of each subsequence corresponding to each dynamic feature in ascending order to obtain a sorted list; calculate the position index H after sorting, with the formula: H = 0.9 (s - 1) + 1, where s is the number of subsequences; if the position index is an integer, find the abnormal score at the Hth position in the sorted list and record it as the dynamic abnormal score corresponding to each dynamic feature; if the position index is not an integer, obtain the integer part Q and the fractional part U of the position index, take out the Qth and Q+1th abnormal scores in the sorted list, multiply the difference between the Q+1th abnormal score and the Qth abnormal score by the fractional part U to obtain the fifth value, add the fifth value to the Qth abnormal score to obtain the dynamic abnormal score corresponding to each dynamic feature.
[0075] S322 and S323 provide reliable dynamic risk quantification basis for the site selection of new energy power plants, improving the reliability of site selection decisions.
[0076] Example 2
[0077] like Figure 2 As shown in the figure, this embodiment provides a new energy power plant planning system, which includes an acquisition module 1, a first screening module 2, and a second screening module 3.
[0078] Acquisition module 1 is used to acquire the static features of multiple candidate points, including seismic activity frequency, altitude, slope and groundwater depth;
[0079] The first screening module 2 is used to calculate the static anomaly score of each candidate point based on the static features of the candidate points, and to perform a first round of screening on all candidate points according to the static anomaly scores to obtain the remaining candidate points.
[0080] The second screening module 3 is used to obtain the dynamic characteristics of each remaining candidate point, including wind speed, solar radiation intensity, ambient temperature, ambient humidity and precipitation; based on the dynamic characteristics of each remaining candidate point and the static anomaly score of each candidate point, a second round of screening is performed on all remaining candidate points to obtain the final location.
[0081] In one specific embodiment of this disclosure, the first screening module 2 further includes a construction unit 21 and a conversion unit 22.
[0082] Construction unit 21 is used to organize the collected static feature data of I candidate points into an I-row G-column original data matrix A, where each row represents all static features of a candidate point, and each column represents the same feature dimension of all candidate points; for each column in the original data matrix, the mean of the g-th column is calculated. and standard deviation ;
[0083] Transformation unit 22 is used to perform standardization transformation on each data element in the original data matrix, wherein for the element in the i-th row and g-th column of the original matrix... Its standardized value The calculation formula is All obtained after standardization Reassemble the data to form a standardized data matrix B with the same dimensions as the original data matrix; calculate the covariance matrix corresponding to the standardized data matrix, and perform eigenvalue decomposition on the covariance matrix to obtain the eigenvector matrix and the diagonal eigenvalue matrix; calculate the static anomaly score of each candidate point based on the eigenvector matrix and the diagonal eigenvalue matrix, and obtain the remaining candidate points based on the static anomaly scores.
[0084] In one specific embodiment of this disclosure, the conversion unit 22 further includes a search unit 221 and a comparison unit 222.
[0085] Search unit 221 is used to find the smallest j such that ≥0.95, Let be the element in the g-th row and g-th column of the diagonal eigenvalue matrix; select the first j columns of the eigenvector matrix to obtain the mapping matrix; multiply the mapping matrix by the standardized data matrix to obtain the first matrix; multiply the first matrix by the transpose of the mapping matrix to obtain the second matrix; for each candidate point, calculate its static anomaly score, the formula for which is:
[0086] ;
[0087] In the formula, The static anomaly score of the candidate point corresponding to the i-th row in the standardized data matrix; Let be the element in the i-th row and g-th column of the standardized data matrix; Let be the element in the i-th row and g-th column of the second matrix; is the element in the g-th row and g-th column of the diagonal eigenvalue matrix;
[0088] The comparison unit 222 is used to compare the static anomaly score with a preset static score threshold. If the score is greater than the static score threshold, the candidate point is deleted to obtain the primary remaining candidate points. The number of primary remaining candidate points is counted. If the number is less than the preset first number, the primary remaining candidate point is recorded as a remaining candidate point. Otherwise, all static features corresponding to each primary remaining candidate point are combined to obtain a combination vector. Based on the combination vectors of all primary remaining candidate points, a clustering algorithm is used to cluster them to form multiple clusters. For each primary remaining candidate point, the Euclidean distance from its combination vector to the centroid of its own cluster is calculated. All primary remaining candidate points are sorted in ascending order according to their corresponding Euclidean distance values. The first M primary remaining candidate points are selected as remaining candidate points, where M is a positive integer.
[0089] In one specific embodiment of this disclosure, the second filtering module 3 further includes an acquisition unit 31 and a statistics unit 32.
[0090] The acquisition unit 31 is used to acquire the true value of each dynamic feature of each remaining candidate point at each moment in the past preset historical period, forming a true value sequence corresponding to each dynamic feature; for each true value sequence, the mean and standard deviation of all true values are calculated, the true value at each moment is subtracted from the mean to obtain a first value, and the ratio between the first value and the standard deviation is recorded as the standardized true value at each moment, thus obtaining a standardized true value sequence corresponding to each dynamic feature; based on the true value sequence corresponding to each dynamic feature, the predicted value of the dynamic feature at each moment in the past preset historical period is calculated using the simple moving average method, thus obtaining a predicted value sequence, the predicted value at each moment is subtracted from the mean to obtain a second value, and the ratio between the second value and the standard deviation is recorded as the standardized predicted value at each moment, thus obtaining a standardized predicted value sequence corresponding to each dynamic feature;
[0091] The statistical unit 32 is used to calculate the dynamic anomaly score corresponding to each dynamic feature based on the standardized predicted value sequence and the standardized true value sequence; compare the dynamic anomaly score corresponding to each dynamic feature with its corresponding dynamic score threshold; if the dynamic anomaly score is greater than the dynamic score threshold, then the dynamic feature is determined to be an abnormal dynamic feature; for each remaining candidate point, count the number of abnormal dynamic features in its corresponding dynamic features; if the number is greater than a preset second number, then the remaining candidate point is determined to be abnormal, deleted, and intermediate candidate points are obtained; the final point is obtained based on the intermediate candidate points.
[0092] In one specific embodiment of this disclosure, the statistical unit 32 further includes an analysis unit 321.
[0093] Analysis unit 321 is used to analyze intermediate candidate points again. If there is only one intermediate candidate point, it is used as the final point. If there is more than one intermediate candidate point, the dynamic anomaly scores corresponding to the dynamic features contained in each intermediate candidate point are weighted and summed to obtain the total dynamic anomaly score of each intermediate candidate point. At the same time, the intermediate candidate points are merged and classified, and the final point is selected based on the total dynamic anomaly score of each intermediate candidate point and the merging and classification results.
[0094] It should be noted that the specific methods by which each module performs operations in the system described in the above embodiments have been described in detail in the embodiments related to the method, and will not be elaborated here.
[0095] Example 3
[0096] Corresponding to the above method embodiments, this disclosure also provides new energy power plant planning equipment. The new energy power plant planning equipment described below and the new energy power plant planning method described above can be referred to each other.
[0097] Figure 3 This is a block diagram illustrating a new energy power plant planning device 300 according to an exemplary embodiment. For example... Figure 3 As shown, the new energy power plant planning equipment 300 may include: a processor 301 and a memory 302. The new energy power plant planning equipment 300 may also include one or more of the following: a multimedia component 303, an I / O interface 304, and a communication component 305.
[0098] The processor 301 controls the overall operation of the new energy power plant planning equipment 300 to complete all or part of the steps in the aforementioned new energy power plant planning method. The memory 302 stores various types of data to support the operation of the new energy power plant planning equipment 300. This data may include, for example, instructions for any application or method operating on the new energy power plant planning equipment 300, and application-related data such as contact data, sent and received messages, pictures, audio, video, etc. The memory 302 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. Multimedia component 303 may include a screen and an audio component. The screen may be, for example, a touchscreen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signals may be further stored in memory 302 or transmitted via communication component 305. The audio component also includes at least one speaker for outputting audio signals. I / O interface 304 provides an interface between processor 301 and other interface modules, such as a keyboard, mouse, buttons, etc. These buttons may be virtual or physical buttons. Communication component 305 is used for wired or wireless communication between the new energy power plant planning equipment 300 and other devices. Wireless communication may include Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, or 4G, or a combination of these. Therefore, the corresponding communication component 305 may include a Wi-Fi module, a Bluetooth module, or an NFC module.
[0099] In an exemplary embodiment, the new energy power plant planning device 300 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to execute the aforementioned new energy power plant planning method.
[0100] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, which, when executed by a processor, implement the steps of the above-described new energy power plant planning method. For example, the computer-readable storage medium may be the memory 302 including program instructions, which may be executed by the processor 301 of the new energy power plant planning device 300 to complete the above-described new energy power plant planning method.
[0101] Example 4
[0102] Corresponding to the above method embodiments, this disclosure also provides a readable storage medium. The readable storage medium described below can be referred to in conjunction with the new energy power plant planning method described above.
[0103] A readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the new energy power plant planning method described in the above method embodiments.
[0104] Specifically, the readable storage medium can be a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, or any other readable storage medium capable of storing program code.
[0105] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A method for planning new energy power plants, characterized in that, include: Obtain static features of multiple candidate points, wherein the static features include seismic activity frequency, altitude, slope and groundwater depth; Calculate the static anomaly score of each candidate point based on its static features, and perform a first round of screening on all candidate points based on the static anomaly scores to obtain the remaining candidate points. The dynamic features of each remaining candidate point are obtained, including wind speed, solar radiation intensity, ambient temperature, ambient humidity, and precipitation. Based on the dynamic features of each remaining candidate point and the static anomaly score of each candidate point, a second round of screening is performed on all remaining candidate points to obtain the final location.
2. The new energy power plant planning method according to claim 1, characterized in that, Based on the static features of the candidate points, a static anomaly score is calculated for each candidate point. The remaining candidate points are then filtered in the first round based on these static anomaly scores, resulting in: The collected static feature data of the I candidate points are organized into an I-row, G-column original data matrix A, where each row represents all static features of a candidate point, and each column represents the same feature dimension of all candidate points. For each column in the original data matrix, the mean of the g-th column is calculated. and standard deviation ; Perform a standardization transformation on each data element in the original data matrix, where the transformation is applied to the element in the i-th row and g-th column of the original matrix. Its standardized value The calculation formula is All obtained after standardization Reassemble the data to form a standardized data matrix B with the same dimensions as the original data matrix; calculate the covariance matrix corresponding to the standardized data matrix, and perform eigenvalue decomposition on the covariance matrix to obtain the eigenvector matrix and the diagonal eigenvalue matrix; calculate the static anomaly score of each candidate point based on the eigenvector matrix and the diagonal eigenvalue matrix, and obtain the remaining candidate points based on the static anomaly scores.
3. The new energy power plant planning method according to claim 2, characterized in that, The static anomaly score of each candidate point is calculated based on the eigenvector matrix and the diagonal eigenvalue matrix. The remaining candidate points are then obtained based on these static anomaly scores, including: Find the smallest j such that ≥0.95, Let be the element in the g-th row and g-th column of the diagonal eigenvalue matrix; select the first j columns of the eigenvector matrix to obtain the mapping matrix; multiply the mapping matrix by the standardized data matrix to obtain the first matrix; multiply the first matrix by the transpose of the mapping matrix to obtain the second matrix; for each candidate point, calculate its static anomaly score, the formula for which is: ； In the formula, The static anomaly score of the candidate point corresponding to the i-th row in the standardized data matrix; Let be the element in the i-th row and g-th column of the standardized data matrix; Let be the element in the i-th row and g-th column of the second matrix; is the element in the g-th row and g-th column of the diagonal eigenvalue matrix; The static anomaly score is compared with a preset static score threshold. If it is greater than the static score threshold, the candidate point is deleted, resulting in primary remaining candidate points. The number of primary remaining candidate points is counted. If it is less than a preset first number, the primary remaining candidate point is recorded as a remaining candidate point. Otherwise, all static features corresponding to each primary remaining candidate point are combined to obtain a combination vector. Based on the combination vectors of all primary remaining candidate points, a clustering algorithm is used to cluster them into multiple clusters. For each primary remaining candidate point, the Euclidean distance from its combination vector to the centroid of its own cluster is calculated. All primary remaining candidate points are sorted in ascending order according to their corresponding Euclidean distance values, and the first M primary remaining candidate points are selected as remaining candidate points, where M is a positive integer.
4. The new energy power plant planning method according to claim 1, characterized in that, A second round of screening is performed on all remaining candidate points based on the dynamic characteristics and static anomaly scores of each candidate point to obtain the final locations, including: For each remaining candidate point, obtain the true value of each dynamic feature at each moment within a preset historical period, forming a true value sequence for each dynamic feature. For each true value sequence, calculate the mean and standard deviation of all true values. Subtract the mean from the true value at each moment to obtain a first value. Record the ratio between the first value and the standard deviation as the standardized true value at each moment, thus obtaining a standardized true value sequence for each dynamic feature. Based on the true value sequence for each dynamic feature, use the simple moving average method to calculate the predicted value of the dynamic feature at each moment within a preset historical period, obtaining a predicted value sequence. Subtract the mean from the predicted value at each moment to obtain a second value. Record the ratio between the second value and the standard deviation as the standardized predicted value at each moment, thus obtaining a standardized predicted value sequence for each dynamic feature. Calculate the dynamic anomaly score corresponding to each dynamic feature based on the standardized predicted value sequence and the standardized true value sequence; compare the dynamic anomaly score corresponding to each dynamic feature with its corresponding dynamic score threshold. If the dynamic anomaly score is greater than the dynamic score threshold, the dynamic feature is determined to be an abnormal dynamic feature; for each remaining candidate point, count the number of abnormal dynamic features in its corresponding dynamic features. If the number is greater than a preset second number, the remaining candidate point is determined to be abnormal and deleted to obtain intermediate candidate points; the final point is obtained based on the intermediate candidate points.
5. The new energy power plant planning method according to claim 4, characterized in that, The final location is derived from the intermediate candidate points, including: The intermediate candidate points are analyzed again. If there is only one intermediate candidate point, it is used as the final point. If there is more than one intermediate candidate point, the dynamic anomaly scores corresponding to the dynamic features contained in each intermediate candidate point are weighted and summed to obtain the total dynamic anomaly score of each intermediate candidate point. At the same time, the intermediate candidate points are merged and classified, and the final point is selected based on the total dynamic anomaly score of each intermediate candidate point and the merged classification result.
6. A new energy power plant planning system, characterized in that, include: The acquisition module is used to acquire the static features of multiple candidate points, wherein the static features include seismic activity frequency, altitude, slope and groundwater depth; The first screening module is used to calculate the static anomaly score of each candidate point based on the static features of the candidate points, and to perform a first round of screening on all candidate points according to the static anomaly scores to obtain the remaining candidate points. The second screening module is used to obtain the dynamic characteristics of each remaining candidate point, wherein the dynamic characteristics include wind speed, solar radiation intensity, ambient temperature, ambient humidity and precipitation; based on the dynamic characteristics of each remaining candidate point and the static anomaly score of each candidate point, a second round of screening is performed on all remaining candidate points to obtain the final location.
7. The new energy power plant planning system according to claim 6, characterized in that, The first filtering module includes: The construction unit is used to organize the collected static feature data of I candidate points into an I-row, G-column original data matrix A, where each row represents all static features of a candidate point, and each column represents the same feature dimension of all candidate points; for each column in the original data matrix, the mean of the g-th column is calculated. and standard deviation ; The transformation unit is used to perform a standardization transformation on each data element in the original data matrix, where the transformation is performed on the element in the i-th row and g-th column of the original matrix. Its standardized value The calculation formula is All obtained after standardization Reassemble the data to form a standardized data matrix B with the same dimensions as the original data matrix; calculate the covariance matrix corresponding to the standardized data matrix, and perform eigenvalue decomposition on the covariance matrix to obtain the eigenvector matrix and the diagonal eigenvalue matrix; calculate the static anomaly score of each candidate point based on the eigenvector matrix and the diagonal eigenvalue matrix, and obtain the remaining candidate points based on the static anomaly scores.
8. The new energy power plant planning system according to claim 7, characterized in that, The conversion unit includes: The search unit is used to find the smallest j such that ≥0.95, Let be the element in the g-th row and g-th column of the diagonal eigenvalue matrix; select the first j columns of the eigenvector matrix to obtain the mapping matrix; multiply the mapping matrix by the standardized data matrix to obtain the first matrix; multiply the first matrix by the transpose of the mapping matrix to obtain the second matrix; for each candidate point, calculate its static anomaly score, the formula for which is: ； In the formula, The static anomaly score of the candidate point corresponding to the i-th row in the standardized data matrix; Let be the element in the i-th row and g-th column of the standardized data matrix; Let be the element in the i-th row and g-th column of the second matrix; is the element in the g-th row and g-th column of the diagonal eigenvalue matrix; The comparison unit compares the static anomaly score with a preset static score threshold. If the score is greater than the threshold, the candidate point is deleted, resulting in primary remaining candidate points. The number of primary remaining candidate points is counted. If the number is less than a preset first number, the primary remaining candidate point is recorded as a remaining candidate point. Otherwise, all static features corresponding to each primary remaining candidate point are combined to obtain a combination vector. Based on the combination vectors of all primary remaining candidate points, a clustering algorithm is used to cluster them into multiple clusters. For each primary remaining candidate point, the Euclidean distance from its combination vector to the centroid of its own cluster is calculated. All primary remaining candidate points are sorted in ascending order according to their corresponding Euclidean distance values, and the first M primary remaining candidate points are selected as remaining candidate points, where M is a positive integer.
9. The new energy power plant planning system according to claim 6, characterized in that, The second filtering module includes: The acquisition unit is used to acquire the true value of each dynamic feature of each remaining candidate point at each moment in the past preset historical period, forming a true value sequence corresponding to each dynamic feature; for each true value sequence, the mean and standard deviation of all true values are calculated, the true value at each moment is subtracted from the mean to obtain a first value, and the ratio between the first value and the standard deviation is recorded as the standardized true value at each moment, thus obtaining a standardized true value sequence corresponding to each dynamic feature; based on the true value sequence corresponding to each dynamic feature, the predicted value of the dynamic feature at each moment in the past preset historical period is calculated using the simple moving average method, thus obtaining a predicted value sequence, the predicted value at each moment is subtracted from the mean to obtain a second value, and the ratio between the second value and the standard deviation is recorded as the standardized predicted value at each moment, thus obtaining a standardized predicted value sequence corresponding to each dynamic feature; The statistical unit is used to calculate the dynamic anomaly score corresponding to each dynamic feature based on the standardized predicted value sequence and the standardized true value sequence; compare the dynamic anomaly score corresponding to each dynamic feature with its corresponding dynamic score threshold; if the dynamic anomaly score is greater than the dynamic score threshold, the dynamic feature is determined to be an abnormal dynamic feature; for each remaining candidate point, count the number of abnormal dynamic features in its corresponding dynamic features; if the number is greater than a preset second number, the remaining candidate point is determined to be abnormal, deleted, and intermediate candidate points are obtained; the final point is obtained based on the intermediate candidate points.
10. The new energy power plant planning system according to claim 9, characterized in that, The statistical unit includes: The analysis unit is used to analyze the intermediate candidate points again. If there is only one intermediate candidate point, it is used as the final point. If there is more than one intermediate candidate point, the dynamic anomaly scores corresponding to the dynamic features contained in each intermediate candidate point are weighted and summed to obtain the total dynamic anomaly score of each intermediate candidate point. At the same time, the intermediate candidate points are merged and classified, and the final point is selected based on the total dynamic anomaly score of each intermediate candidate point and the merging and classification results.