A port clean energy power generation prediction and scheduling method and system

Abstract

The application provides a port clean energy power generation prediction and scheduling method and system, and relates to the technical field of energy scheduling and prediction. First, based on historical operation information of a first port, the correlation between weather data, port operation information and various clean energy power generation information is analyzed, then, based on power demand values and power generation information in a future preset time interval, first clean energy scheduling parameters are determined, finally, based on an integrated target principal component information set, a target correlation data set is determined, and based on the matching degree between the power demand values and the power generation in the target correlation data set, the first clean energy scheduling parameters are adjusted to determine target clean energy scheduling parameters. Through the technical scheme of the application, multi-dimensional information in a port operation scene can be comprehensively utilized to generate clean energy scheduling information, so that the power demand values and the power generation can be highly matched.

Description

Technical Field

[0001] This invention relates to the field of energy dispatch and forecasting technology, and in particular to a method and system for forecasting and dispatching clean energy power generation in ports. Background Technology

[0002] Currently, the clean energy used in ports includes various types such as solar and wind power. The power generation capacity of each type of clean energy is closely related to weather conditions. Port electricity demand is often closely linked to port operations. If the power generation from clean energy sources does not match the port's electricity demand, it can lead to insufficient power supply or energy waste. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for predicting and scheduling clean energy power generation in ports, which can achieve high matching degree scheduling of clean energy in ports.

[0004] This application proposes a method for predicting and scheduling clean energy power generation in ports, the method comprising: S1: Obtain at least one first associated data of the first port within a first time interval, and determine at least one set of first principal component information based on at least one first associated data; S2: Based on the first operation information within the second time interval, predict the first power demand value within the second time interval; S3: Based on the weather data feature vector within the second time interval and the first electricity demand value, the first clean energy dispatch parameters are predicted; S4: Determine the target associated data set based on at least one first principal component information set, and adjust the first clean energy scheduling parameters based on the target associated data set to obtain the target clean energy scheduling parameters.

[0005] Preferably, S1 includes: S11: Perform first associated data collection and processing for the first port, and obtain at least one first associated data set; S12: Perform a matching and filtering process on at least one first associated data set to obtain at least one second associated data set; S13: Perform clustering processing on at least one second associated data set to obtain at least one third associated data set; S14: Perform principal component analysis on each of the third associated data sets to determine at least one set of first principal component information.

[0006] Preferably, S12 includes: S121: Perform a first scheduling problem processing on each of the first associated data sets to obtain at least one first scheduling problem information; S122: Based on at least one of the first scheduling problem information, determine at least one second associated data set from at least one first associated data set.

[0007] Preferably, S13 further includes: The clustering process includes the K-means clustering algorithm.

[0008] Preferably, S2 includes: S21: Based on at least one of the first associated data sets, predict the first job information within the second time interval; S22: Calculate the first power demand value within the second time interval based on the first operation information.

[0009] Preferably, S3 includes: S31: Preprocess the weather data within the second time interval to obtain the weather data feature vector; S32: Input the weather data feature vector into the clean energy power generation prediction model to obtain clean energy power generation information within the second time interval; S33: Based on the first power demand value and the clean energy power generation information, predict the first clean energy dispatch parameters.

[0010] Preferably, the first clean energy dispatch parameters include: solar dispatch parameters and wind dispatch parameters.

[0011] Preferably, S4 includes: S41: Perform a matching process on the weather data feature vector and the clean energy power generation information with at least one set of first principal component information to determine at least one first matching degree; S42: Determine at least one set of second principal component information based on at least one first matching degree, and determine a target principal component information set based on at least one set of second principal component information; S43: Determine a target associated data set from at least one of the second associated data sets based on the target principal component information set, and generate target clean energy scheduling parameters based on the target associated data set and the first clean energy scheduling parameters.

[0012] Preferably, step S41 includes: the matching process is expressed using cosine similarity.

[0013] This application also proposes a port clean energy power generation prediction and scheduling system to implement the aforementioned port clean energy power generation prediction and scheduling method.

[0014] This application proposes a method and system for predicting and scheduling clean energy power generation in ports, relating to the field of energy scheduling and prediction technology. First, based on historical operational information of a first port, the correlation between weather data, port operational information, and various clean energy power generation information is analyzed. Then, based on the electricity demand and power generation information within a preset future time interval, first clean energy scheduling parameters are determined. Finally, based on the integrated target principal component information set, a target associated data set is determined, and based on the matching degree between the electricity demand and power generation in this target associated data set, the first clean energy scheduling parameters are adjusted to determine the target clean energy scheduling parameters. Through the technical solution of this application, multi-dimensional information in port operational scenarios can be comprehensively utilized to generate clean energy scheduling information, enabling a high degree of matching between electricity demand and power generation. Attached Figure Description

[0015] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0016] Figure 1 This is an execution flowchart of a port clean energy power generation prediction and scheduling method in this application.

[0017] Figure 2 This is a method for determining at least one set of first principal component information in this application. Detailed Implementation

[0018] 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 embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. The illustrative embodiments and descriptions are only used to explain the present invention and are not intended to limit the present invention.

[0020] The present invention provides a detailed description of a port clean energy power generation prediction and scheduling method and system.

[0021] This embodiment proposes a method for predicting and scheduling clean energy power generation in ports, the specific process of which is as follows: Figure 1 As shown.

[0022] S1: Obtain at least one first associated data of the first port within a first time interval, and determine at least one first principal component information set based on at least one first associated data.

[0023] Data on various types of clean energy power generation and port operation types are extracted from historical data within the first time interval to obtain the matching degree between the two types of data. The first time interval refers to a historical period, preferably one year in length. For data combinations with values ​​less than a preset value, clustering is performed based on similar issues, and then principal component analysis is used to extract data combinations with strong correlations.

[0024] In this step, it is necessary to extract the power generation of various clean energy sources and the operation information of various port operation types within the first time interval, and combine them with the weather data within the first time interval to obtain the correlation between the above data, so as to prepare for the subsequent clean energy dispatch work.

[0025] like Figure 2 As shown, step S1 includes the following sub-steps: S11: Perform first associated data collection and processing for the first port and obtain at least one first associated data set.

[0026] This step requires the initial collection and preprocessing of clean energy power generation data, port operation information data, and weather data.

[0027] The first associated data collection and processing includes the collection and preprocessing of various types of data in the port.

[0028] To ensure that the relationships between the extracted data conform to the requirements of big data patterns, the first time interval is preferably one year or longer.

[0029] Data collection: Collect multi-dimensional historical data from the port's TOS system, energy management system, and weather station over the past year, with each set of data collected every 60 minutes.

[0030] Clean energy power generation data includes: solar power generation, wind power generation, photovoltaic efficiency coefficient, and wind power generation coefficient.

[0031] Port operation data includes: operation type (specifically may include container / bulk / liquid / special), operation volume, and operation time period.

[0032] Weather data includes average solar irradiance and average wind speed within the data collection period.

[0033] Standardization: Z-score standardization is performed on all numerical data to eliminate the influence of units.

[0034] This step outputs at least one first associated data set, each of which corresponds to a sub-time interval within the first time interval.

[0035] The first associated data set includes clean energy power generation data, port operation data, and weather data within the same time interval.

[0036] Each of the first associated data sets includes: clean energy power generation data, port operation data, and weather data.

[0037] S12: Perform a matching and filtering process on at least one first associated data set to obtain at least one second associated data set.

[0038] In step S11, at least one first associated data set corresponding to the first time interval is obtained. To determine the correlation between the first associated data set and the scheduling problem, this step requires obtaining the matching degree between workload information and power generation data.

[0039] S12 includes the following sub-steps: S121: Perform a first scheduling problem processing on each of the first associated data sets to obtain at least one first scheduling problem information.

[0040] The first scheduling problem processing refers to calculating the first scheduling problem information based on the actual power usage contained in the first associated data set.

[0041] The first scheduling problem information includes: power generation waste rate and power supply insufficiency rate.

[0042] Specifically, the first scheduling problem processing includes calculating the power generation waste rate and power supply insufficiency rate based on the data contained in the first associated data set.

[0043] The power generation waste rate is the ratio of the remaining power generation value to the total power generation, and the power supply shortage rate is the ratio of the deficit power to the total power supply.

[0044] For specific calculation methods, please refer to existing technologies, which will not be elaborated here.

[0045] S122: Based on at least one of the first scheduling problem information, determine at least one second associated data set from at least one first associated data set.

[0046] In this step, a data set with scheduling problems needs to be selected from at least one of the first associated data sets to serve as the second associated data set. Each second associated data set also includes a scheduling problem information tag. The scheduling problem refers to the waste of power generation and insufficient power supply caused by improper scheduling of operations and clean energy power generation.

[0047] Preferably, the selection operation is as follows: based on the power generation waste rate and power supply insufficiency rate contained in the first scheduling problem information, a preset threshold is compared, thereby determining the first associated data set with the power generation waste rate and power supply insufficiency rate higher than the preset threshold as the second associated data set.

[0048] For each of the second associated data sets, the scheduling problem information is labeled in the form of [problem type, problem severity].

[0049] For example, in the category of power generation waste: a power generation waste rate > 15% is considered serious waste, 10% < power generation waste rate ≤ 15% is considered moderate waste, and a power generation waste rate ≤ 10% is considered slight waste.

[0050] The label type for insufficient power supply is the same as above.

[0051] S13: Perform clustering on at least one second associated data set to obtain at least one third associated data set.

[0052] In S12, at least one second associated data set has been identified. In this step, it needs to be clustered according to the scheduling problem information label.

[0053] Specifically, the K-means clustering algorithm can be used to cluster the scheduling problem information labels. An example clustering result is as follows: Cluster 1: High power generation waste; Cluster 2: Low power generation waste; Cluster 3: High power supply insufficiency.

[0054] After clustering, at least one third associated data set can be obtained. Each of these third associated data sets contains at least one second associated data set with a label similar to the scheduling problem information.

[0055] S14: Perform principal component analysis on each of the third associated data sets to determine at least one set of first principal component information.

[0056] In this step, principal component analysis is performed on each of the third associated datasets to identify the main influencing factors corresponding to each of the third associated datasets.

[0057] The aforementioned principal component analysis can be performed using PCA (Principal Component Analysis). Each set of the first principal component information corresponds to one of the third associated data sets.

[0058] Exemplary principal component analysis results may include: The first principal component information set 1 consists of four indicators: container operation, daytime operation, photovoltaic efficiency, and wind power generation. The photovoltaic efficiency load factor is 0.9, and the wind power generation load factor is 0.8.

[0059] The first principal component information set 2 consists of three indicators: bulk cargo operations, nighttime operations, and wind power generation, with the wind power generation load factor being 1.0.

[0060] S2: Based on the first operation information within the second time interval, predict the first power demand value within the second time interval.

[0061] In order to determine the scheduling parameters of various clean energy sources based on the workload and power generation of the first port in the second time interval, it is necessary to predict the power demand value in the future second time interval in this step.

[0062] The second time interval refers to the future period to be predicted.

[0063] S2 mainly includes the following sub-steps: S21: Based on at least one of the first associated data sets, predict the first job information within the second time interval.

[0064] Since the electricity demand value in the second time interval is mainly determined by the first operation information of the first port in that period, it is necessary to predict the first operation information in this step.

[0065] In the specific prediction process, it is necessary to use a spatiotemporal graph convolutional network to predict the first job information in the second time interval based on at least one first associated data set determined in S11.

[0066] Preferably, spatial modeling can also be used in the prediction process to capture spatial dependencies through a port operation area map (nodes represent operation areas and edges represent operation associations) so as to make the prediction of the first operation information more accurate.

[0067] The first job information includes at least one job type, the corresponding job quantity, and the job time period.

[0068] S22: Calculate the first power demand value within the second time interval based on the first operation information.

[0069] Since the first job information, which includes job type, job quantity, and job time period, has been determined in S21, the first power demand value required in the second time interval can be calculated in this step according to the correspondence between job type and power consumption.

[0070] The operations can be categorized into container, bulk, liquid, and special types. The power consumption varies for each type of operation.

[0071] Preferably, in the specific calculation process, the first operation information can be calculated using empirical formulas, or the first power demand value can be obtained through a pre-trained machine learning model, without any specific limitations.

[0072] S3: Based on the weather data feature vector within the second time interval and the first electricity demand value, predict the first clean energy dispatch parameters.

[0073] Using port operation information within the second time interval and weather data from the second time interval prior to it, and employing a pre-trained power generation prediction model, the power generation of various clean energy sources and the power dispatch parameters for clean energy sources within the second time interval are predicted.

[0074] In step S2, the amount of electricity required for the normal operation of the first port during the second time interval has been determined. This means that the power generation from various clean energy sources needs to exceed the first electricity demand to maintain the normal operation of the first port. To meet these requirements, the scheduling parameters of various clean energy devices need to be predicted to avoid power shortages and power waste.

[0075] S3 includes the following sub-steps: S31: Preprocess the weather data within the second time interval to obtain the weather data feature vector.

[0076] Since the power generation of clean energy sources such as solar and wind power is mainly affected by weather data, it is necessary to preprocess the weather data in the second time interval in this step.

[0077] The meteorological data to be processed includes meteorological data within the second time interval and real-time data from the port meteorological station.

[0078] During the processing, it is necessary to extract the main features of various weather characteristics for subsequent power generation prediction.

[0079] Specifically, wavelet packet decomposition can be used to extract weather time-frequency features: Wavelet packet decomposition is performed on light intensity and wind speed to construct weather feature vectors.

[0080] Preferably, feature discrimination can be enhanced through contrastive learning to optimize features.

[0081] After this step, the weather data feature vector for the second time interval can be obtained.

[0082] S32: Input the weather data feature vector into the clean energy power generation prediction model to obtain clean energy power generation information within the second time interval.

[0083] The clean energy power generation information refers to the power generation information of various clean energy power generation equipment under normal operating conditions, including the absolute value information and proportion information of power generation of various clean energy sources. For example, for the same total power generation, the proportion of solar power generation and wind power generation will be different under sunny and cloudy weather conditions.

[0084] The clean energy power generation prediction model in this step is obtained by training a convolutional neural network and sample data is obtained from at least one of the first associated data sets. The weather data contained in each of the first associated data sets is processed into feature vectors and used as input to the model. The power generation values ​​and power generation ratio information of various clean energy sources are used as output.

[0085] S33: Based on the first power demand value and the clean energy power generation information, predict the first clean energy dispatch parameters.

[0086] In order to ensure that the first power demand value output in S22 and the clean energy power generation information output in S32 match the specified requirements, the first clean energy dispatch parameters need to be predicted based on the above data in this step.

[0087] The first clean energy dispatch parameters include solar energy dispatch parameters and wind energy dispatch parameters. The solar energy dispatch parameters include the photovoltaic power output factor and photovoltaic availability rate. The wind energy dispatch parameters include the wind power output factor and wind power availability rate. By reasonably setting the above dispatch parameters, the service life of clean energy power generation equipment can be extended while meeting power generation requirements.

[0088] The first clean energy dispatching parameter can be obtained through a clean energy dispatching parameter determination model. The clean energy dispatching parameter determination model is obtained by training a convolutional neural network model. The training data also comes from at least one of the first associated data sets, that is, from the historical data of the first port within the first time interval. The predicted power generation value and power demand value within the specified time interval are used as inputs, and the actual dispatching parameters are used as output data to complete the training.

[0089] S4: Determine the target associated data set based on at least one first principal component information set, and adjust the first clean energy scheduling parameters based on the target associated data set to obtain the target clean energy scheduling parameters.

[0090] In step S3, the first clean energy dispatch parameters for the future second time interval have been predicted based on at least one set of the first associated data for the first port within the first time interval. However, the actual dispatch data under the clean energy power dispatch parameters may produce mismatch problems. Therefore, in this step, the first clean energy dispatch parameters need to be corrected based on the dispatch parameter prediction model in step S15 to obtain the target clean energy dispatch parameters.

[0091] S4 includes the following sub-steps: S41: Perform a matching process on the weather data feature vector and the clean energy power generation information with at least one set of first principal component information to determine at least one first matching degree.

[0092] At least one of the first principal component information sets corresponds to different scheduling problem types. Therefore, by performing matching processing on the weather data feature vector and the clean energy power generation information in the second time interval with at least one of the first principal component information sets, multiple first matching degrees between multiple first principal component information sets and the second time interval can be obtained. Thus, based on the multiple first matching degrees, the corresponding first principal component information set can be selected in subsequent steps to adjust the clean energy scheduling parameters.

[0093] Preferably, the matching process is expressed using cosine similarity.

[0094] S42: Determine at least one set of second principal component information based on at least one first matching degree, and determine a target principal component information set based on at least one set of second principal component information.

[0095] In step S41, the clean energy scheduling information within the second time interval is matched with at least one set of the first principal component information to determine at least one first matching degree. In this step, at least one set of the first principal component information that meets the first matching degree requirement needs to be selected and integrated to form a target principal component information set.

[0096] Preferably, the first principal component information set with the highest matching degree can be used as the second principal component information set, and also as the target principal component information set.

[0097] Preferably, the first principal component information set whose first matching degree is in the top N positions can be used as at least one second principal component information set, and the at least one second principal component information set can be subjected to a union operation to obtain the target principal component information set.

[0098] The identified target principal component information set contains multiple pieces of information related to the dispatch of various types of clean energy.

[0099] S43: Determine a target associated data set from at least one of the second associated data sets based on the target principal component information set, and generate target clean energy scheduling parameters based on the target associated data set and the first clean energy scheduling parameters.

[0100] In S1, at least one of the second associated data sets is identified, which contains matching information between clean energy power generation and power demand. Therefore, the second associated data set with the highest matching degree to the target principal component information set can be used as the target associated data set most relevant to the target principal component information set. Furthermore, the target clean energy dispatch parameters can be generated based on the target associated data set.

[0101] For example, the target associated data set includes the label [Power generation waste rate, power generation waste rate 16%].

[0102] The first clean energy dispatch parameters can be adjusted based on the label to reduce parameters such as the photovoltaic efficiency coefficient and wind power generation coefficient of the clean energy power generation equipment, so that the power demand value and the clean energy power generation value can be highly matched.

[0103] This application also proposes a port clean energy power generation prediction and scheduling system for implementing the aforementioned port clean energy power generation prediction and scheduling method.

[0104] This application proposes a method and system for predicting and scheduling clean energy power generation in ports, relating to the field of energy scheduling and prediction technology. First, based on historical operational information of a first port, the correlation between weather data, port operational information, and various clean energy power generation information is analyzed. Then, based on the electricity demand and power generation information within a preset future time interval, first clean energy scheduling parameters are determined. Finally, based on the integrated target principal component information set, a target associated data set is determined, and based on the matching degree between the electricity demand and power generation in this target associated data set, the first clean energy scheduling parameters are adjusted to determine the target clean energy scheduling parameters. Through the technical solution of this application, multi-dimensional information in port operational scenarios can be comprehensively utilized to generate clean energy scheduling information, enabling a high degree of matching between electricity demand and power generation.

[0105] The above description is only a preferred embodiment of the present invention. Therefore, all equivalent changes or modifications made to the structure, features and principles described in the claims of this patent application are included in the scope of this patent application.

Claims

1. A method for predicting and scheduling clean energy power generation in ports, characterized in that, The method includes: S1: Obtain at least one first associated data of the first port within a first time interval, and determine at least one set of first principal component information based on at least one first associated data; S2: Based on the first operation information within the second time interval, predict the first power demand value within the second time interval; S3: Based on the weather data feature vector within the second time interval and the first electricity demand value, the first clean energy dispatch parameters are predicted; S4: Determine a target associated data set based on at least one first principal component information set, and adjust the first clean energy scheduling parameters based on the target associated data set to obtain the target clean energy scheduling parameters; S1 includes: S11: Perform first associated data collection and processing for the first port, and obtain at least one first associated data set; S12: Perform a matching and filtering process on at least one first associated data set to obtain at least one second associated data set; S13: Perform clustering processing on at least one second associated data set to obtain at least one third associated data set; S14: Perform principal component analysis on each of the third associated data sets to determine at least one set of first principal component information; S12 includes: S121: Perform a first scheduling problem processing on each of the first associated data sets to obtain at least one first scheduling problem information; S122: Based on at least one of the first scheduling problem information, determine at least one second associated data set from at least one first associated data set; S2 includes: S21: Based on at least one of the first associated data sets, predict the first job information within the second time interval; S22: Calculate the first power demand value within the second time interval based on the first operation information; S3 includes: S31: Preprocess the weather data within the second time interval to obtain the weather data feature vector; S32: Input the weather data feature vector into the clean energy power generation prediction model to obtain clean energy power generation information within the second time interval; S33: Based on the first power demand value and the clean energy power generation information, the first clean energy dispatch parameters are predicted; S4 includes: S41: Perform a matching process on the weather data feature vector and the clean energy power generation information with at least one set of first principal component information to determine at least one first matching degree; S42: Determine at least one set of second principal component information based on at least one first matching degree, and determine a target principal component information set based on at least one set of second principal component information; S43: Determine a target associated data set from at least one of the second associated data sets based on the target principal component information set, and generate target clean energy scheduling parameters based on the target associated data set and the first clean energy scheduling parameters.

2. The method for predicting and scheduling clean energy power generation in ports according to claim 1, characterized in that, S13 further includes: The clustering process includes the K-means clustering algorithm.

3. The method for predicting and scheduling clean energy power generation in ports according to claim 1, characterized in that, The first clean energy dispatch parameters include: solar dispatch parameters and wind dispatch parameters.

4. The method for predicting and scheduling clean energy power generation in ports according to claim 3, characterized in that, S41 includes: the matching process is expressed using cosine similarity.

5. A port clean energy power generation prediction and scheduling system, used to implement the port clean energy power generation prediction and scheduling method as described in any one of claims 1-4.

Images