Distributed photovoltaic group dispatching and group control method

Abstract

The invention relates to a distributed photovoltaic group scheduling and group control method. A D5000 system uses a dedicated channel to transmit a control instruction to an acquisition system through a physical isolation device in a CIME format file; after receiving the control instruction, the collection system immediately calls a specially-arranged photovoltaic control interface program to verify a photovoltaic user, and immediately issues the control instruction to a plurality of corresponding photovoltaic collection terminals to execute after verification is passed, and the collection terminals collect corresponding data; the photovoltaic acquisition terminal transmits acquired result data to the regulated ferry server in real time; the ferry server generates a CIME file from the result data every 20 seconds, and the CIME file passes through the physical isolation device and is fed back to the D5000 system; and the D5000 system generates a corresponding regulation and control instruction according to the result data so as to realize the function of distributed photovoltaic group regulation and group control. According to the invention, minute collection of distributed photovoltaic load data is realized, and the function of group scheduling and group control can be better realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of power control, and in particular to a distributed photovoltaic group modulation and control method. Background Art

[0002] Adapting to the development of active distribution network under the background of new power system construction, ensuring the safe operation and reliable power supply of distribution network, serving the large-scale grid connection of distributed power sources, realizing safe and controllable operation of distribution network, efficient interaction of source, grid, load and storage, and further lean dispatching management. It is necessary to realize the real-time monitoring and observable and controllable functions of distributed energy consumption. Therefore, a distributed photovoltaic user minute-level collection method is urgently needed to obtain relevant data information in time, so as to realize the group dispatching and group control functions more accurately and timely. Summary of the invention

[0003] The purpose of the present invention is to overcome the shortcomings of the prior art and propose a distributed photovoltaic group regulation and control method, which completely relies on the existing metering collection and communication equipment. Without increasing any equipment investment, through the above-mentioned comprehensive technical optimization and improvement from the user master station to the on-site collection equipment, the real-time data of photovoltaic users can be obtained more quickly and accurately in real time, and the minute collection of distributed photovoltaic load data can be successfully realized, thereby better realizing the functions of group regulation and control.

[0004] The present invention solves the technical problem by adopting the following technical solutions:

[0005] A distributed photovoltaic group control method comprises the following steps:

[0006] The S100 and D5000 systems use a dedicated channel to pass control instructions to the acquisition system in the form of CIME format files through the physical isolation device;

[0007] S200, after receiving the control instruction, the collection system immediately calls the dedicated photovoltaic control interface program to verify the photovoltaic user. After the verification is passed, the control instruction is immediately sent to multiple corresponding photovoltaic collection terminals for execution. The collection terminals collect corresponding data, and the photovoltaic collection terminals transmit the collected result data to the control ferry server in real time;

[0008] S300 and the ferry server generate CIME files with the result data every 20 seconds and feed them back to the D5000 system through the physical isolation device;

[0009] The S400 and D5000 systems generate corresponding control instructions based on the result data to realize the function of group control of distributed photovoltaic groups.

[0010] Moreover, the specific implementation method of the acquisition terminal in S200 to collect corresponding data is:

[0011] S210, configuring the corresponding priority for the tasks in the collection terminal, collecting the corresponding data and accurately predicting the output of the photovoltaic power station;

[0012] S220, configuring a data reporting scheme for the collection terminal;

[0013] S230. According to the priority in S210 and the reporting scheme in S220, the photovoltaic data collection terminal transmits the collected result data to the ferry server for control in real time.

[0014] Moreover, the specific implementation method of configuring the corresponding priority in S210 is: according to the relevant technical specifications of the collection terminal and the electric energy meter, the task priority is divided according to the data usage, timeliness and importance.

[0015] Moreover, the specific implementation method of predicting the output of photovoltaic power stations in S210 is: converting the photovoltaic output timing curves of multiple stations and numerical weather forecast data into graph structure input features, establishing a spatiotemporal graph convolutional neural network model that combines the graph convolution method with the gated recurrent unit, and extracting the hidden spatiotemporal dependencies between photovoltaic clusters; using the adjacency matrix adaptive correction method to automatically learn the spatial relationship between different photovoltaic clusters from massive historical data, and introducing the attention mechanism to further improve the prediction model, and finally accurately predicting the output of photovoltaic power stations through the improved prediction model.

[0016] Furthermore, the S220 includes the following steps:

[0017] S221. Set the collection and reporting interval of the minute-level data collection task to 1 minute, and the reporting delay to 10 seconds, 15 seconds, or 20 seconds; the reporting execution period is 05:00-20:00 every day;

[0018] S222, setting the minute-level data collection task to report only a single data item, the single data item being an active power item;

[0019] S223, setting the interval for recording the meter data of each household during each meter reading;

[0020] S224, according to the reading interval of each household's electricity meter data, the electricity meters of each household within the same priority level are arranged in reading order, and the electricity meters with the shortest reading interval are preferentially collected in each task cycle;

[0021] S225. After the other electric meters have completed the collection within this task cycle, additional readings are performed. The initial number of additional readings is 3 times.

[0022] S226: If all three attempts have failed, no more reading will be done, and the next meter reading cycle will be started and the reading will be done at the end until the data is successfully collected and merged into the normal meter group, and the reading order will be arranged according to the data reading interval;

[0023] S227: When the communication quality of the collection terminal meets the preset conditions, the historical collection data is supplemented with data; the historical collection data is the minute-level data in a maximum of 5 collection cycles before the current time.

[0024] S228: If the data collection terminal fails to copy data 4 times, obtain the weather and disaster information of the installation area corresponding to the data collection terminal;

[0025] S229: If the weather and disaster information in the installation area meets the preset conditions, the collection terminal sends a collection failure instruction to the preset communication terminal to notify the corresponding staff that the communication in the target installation area may be poor or the equipment may be damaged due to the influence of weather conditions.

[0026] Furthermore, the S400 includes the following steps:

[0027] S401: Sending a control inquiry instruction to the user to be controlled;

[0028] S402: If the control feedback instruction received from the user is to reject the control, then send the control inquiry instruction again after a first preset period of time;

[0029] S403: Generate the current unit price information F of electricity charges for the user to be regulated according to the number of times a that the regulation inquiry instruction is sent to the user to be regulated, and add it to the subsequent regulation inquiry instruction; wherein, the unit price information F of electricity charges = f-min(K*a, Y), f is the basic electricity price of the user, K is the first price reduction coefficient, which can be 0.1 or 0.2, etc., and Y is the maximum price reduction value.

[0030] The advantages and positive effects of the present invention are:

[0031] The present invention uses a dedicated channel through the D5000 system to pass the control instruction to the collection system in the form of a CIME format file through a physical isolation device; after the collection system receives the control instruction, it immediately calls a dedicated photovoltaic control interface program to verify the photovoltaic user. After the verification is passed, the control instruction is immediately issued to multiple corresponding photovoltaic collection terminals for execution, and the collection terminals collect corresponding data. The photovoltaic collection terminals transmit the collected result data to the ferry server for regulation in real time; the ferry server generates a CIME file from the result data every 20 seconds, and feeds it back to the D5000 system through a physical isolation device; the D5000 system generates corresponding regulation instructions based on the result data to realize the function of distributed photovoltaic group regulation and group control. The present invention relies on existing metering collection and communication equipment, and without increasing any equipment investment, through the above all-round technical optimization and improvement from the user collection master station to the on-site collection equipment, the real-time data of photovoltaic users can be obtained more quickly and accurately in real time, and the minute collection of distributed photovoltaic load data is successfully realized, so as to better realize the function of group regulation and group control. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a flow chart of the present invention;

[0033] Figure 2 It is a processing flow chart when multiple supplementary copying fails in the data supplementary copying process of the present invention;

[0034] Figure 3 The present invention is a flow chart of a method for quickly obtaining the consent of the user to be regulated to the regulation instruction. DETAILED DESCRIPTION

[0035] The present invention is further described in detail below with reference to the accompanying drawings.

[0036] A distributed photovoltaic group control method, such as Figure 1 As shown, the following steps are included:

[0037] The S100 and D5000 systems use dedicated channels to pass control instructions in CIME format files through physical isolation devices to the acquisition system.

[0038] S200, after receiving the control instruction, the collection system immediately calls the dedicated photovoltaic control interface program to verify the photovoltaic user. After the verification is passed, the control instruction is immediately sent to multiple corresponding photovoltaic collection terminals for execution. The collection terminals collect corresponding data, and the photovoltaic collection terminals transmit the collected result data to the control ferry server in real time.

[0039] The specific implementation method of the collection terminal collecting corresponding data in S200 is:

[0040] S210, configuring corresponding priorities for tasks in the collection terminal, collecting corresponding data and accurately predicting the output of the photovoltaic power station.

[0041] The specific implementation method of configuring the corresponding priority is: according to the relevant technical specifications of the collection terminal and the electric energy meter, and according to the data usage, timeliness and importance, divide the task priority, as shown in Table 1.

[0042] Table 1 Terminal execution priority number

[0043]

[0044]

[0045] By giving a higher priority to the 1-minute high-frequency data collection task than the 15-minute high-frequency data collection task, the collection terminal can prioritize the minute-level data collection task for optical priority users at the beginning of each minute. In addition, the full reporting delay is configured to 12 minutes, and the full data collection task can be performed using the remaining time. For example, if the remaining collection time per minute is 50 seconds, there is a total of 12*50=600 seconds of remaining collection time every 12 minutes, which is enough to complete the 15-minute high-frequency data collection task execution for less than 500 households.

[0046] The specific implementation method for accurately predicting the output of photovoltaic power stations is as follows: first, the photovoltaic output timing curves of multiple stations and numerical weather forecast data are converted into graph structure input features, and a spatiotemporal graph convolutional neural network model combining the graph convolution method and the gated recurrent unit is established to extract the hidden spatiotemporal dependencies between photovoltaic clusters.

[0047] Compared with traditional convolutional neural networks, graph convolutional networks generalize convolutional networks to unstructured feature spaces and can process graph structured data of any shape. The spatiotemporal graph convolutional neural network model is represented by G = (V, E, A), where V represents the point set, E represents the edge set, and A is the adjacency matrix of the graph G. The convolution range is node v i Neighborhood B:

[0048] B(v i )={v j |d(v j ,v i )≤D}

[0049] Where: d(v j ,v i ) represents the graph node v i to v j The shortest distance between nodes v. D is the maximum convolution neighborhood. When D = 1, it indicates that the size of the sampling area is the size of node v. i The first-order neighborhood of . At this time, the forward propagation of the graph convolutional layer is as follows:

[0050]

[0051]

[0052] Where: X is the feature matrix of the topological graph G to be predicted; is the adjacency matrix with autocorrelation, that is for The element in row i and column j; is the adjacency matrix after normalization and symmetry. σ(·) is the sigmoid function; W0 is the weight matrix from the upper layer to the graph convolution layer; W1 is the weight matrix from the graph convolution layer to the lower layer, is the node degree matrix. In this paper, the graph convolution method is used to transform the photovoltaic output and numerical weather forecast features into the node parameters of the graph, and the spatial correlation hidden in the output timing curves of different photovoltaic power stations is learned based on the graph convolution layer.

[0053] Secondly, an adaptive correction method of the adjacency matrix is ​​proposed to automatically learn the spatial relationship between different PV clusters from massive historical data, and an attention mechanism is introduced to further improve the prediction model.

[0054] Optimize PV control strategy: Based on the spatiotemporal prediction results between PV clusters, select the dominant nodes that can sensitively reflect the cluster voltage changes, and build an inter-cluster coordination optimization strategy with the goal of ensuring that the node voltage in the entire region does not exceed the limit and the system network loss is minimized.

[0055] Based on the results of the coordinated optimization strategy, the cluster aims to achieve safe operation of node voltage within the cluster, minimize cluster network losses, and maximize local consumption of distributed photovoltaics, and build an autonomous optimization and control strategy within the cluster to improve the safety and economy of the photovoltaic cluster.

[0056] The factors that affect photovoltaic output include many complex factors such as season, weather, solar radiation angle, and equipment installation angle. Among them, environmental factors such as weather can be obtained from numerical weather forecasts and used as feature inputs into the prediction model; equipment factors can be learned by the model through massive historical data and reflected in the weights of the graph convolutional neural network.

[0057] Therefore, six features, namely temperature, humidity, air pressure, sky cloudiness, air quality, and weather type, are selected as the input features of the model. The weather type is coded digitally, and all input features are standardized for easy training:

[0058]

[0059] Where: X ref is the standardized input data set, X mean is the mean of each characteristic component of input X; std is the standard deviation of each feature component of input X.

[0060] S220: Configure the data reporting scheme of the collection terminal.

[0061] S221. Set the collection and reporting interval of the minute-level data collection task to 1 minute, and configure the reporting delay to 10 seconds, 15 seconds, or 20 seconds; the reporting execution period is 05:00-20:00 every day.

[0062] To ensure data timeliness, the reporting delay is configured to be 10 seconds, 15 seconds or 20 seconds. The scheme can be selected based on factors such as the size of the substation, on-site communication environment, and the number of photovoltaic households. To adapt to the time characteristics of photovoltaic power generation, the reporting execution period is 05:00-20:00 every day. Different data content schemes can be customized according to business needs.

[0063] S222. Set the minute-level data collection task to report only a single data item, where the single data item is an active power item.

[0064] Taking 20 photovoltaic users as an example, the estimated time for reading one data item in the on-site area is about 16 seconds, the estimated time for reading three data items is about 30 seconds, and the estimated time for reading five data items is about 40 seconds. Therefore, too many data items will compress the meter reading time of the remaining 96 high-frequency data, affecting the collection of 96 high-frequency data during the day. At the same time, in order to ensure the real-time nature of the data, the current configuration of the reporting delay is 10 seconds, that is, the minute-level data is reported at the 10th second of every minute. The more data items are read, the more reported messages are also increased, and it is difficult to ensure that all of them are uploaded to the master station within 10 seconds. At the same time, the increase in data items also increases the pressure of storage and double writing of the master station, and the integrity and real-time nature of the master station data collection are difficult to guarantee. Combined with current needs, the minute-level data of active power can meet the power monitoring of the optimal light users. Therefore, the data item collected at the minute level of photovoltaic users is determined to be a single data item of "active power".

[0065] Furthermore, since there are two main limiting factors that affect the success rate and efficiency of data collection: one is the local communication carrier channel of the terminal, and the other is the remote communication channel between the terminal and the main station. In order to improve the success rate and efficiency of minute-level data collection tasks, the following optimization measures were taken on the collection terminal side.

[0066] S223. Set the interval for recording the meter data of each household during each meter reading.

[0067] S224. Arrange the reading order of each household's electricity meter within the same priority level according to the reading interval of each household's electricity meter data, and give priority to the electricity meter with the shortest reading interval in each task cycle.

[0068] S225. After the data of other electric meters are collected in this task cycle, additional readings are performed. The initial number of additional readings is 3 times.

[0069] S226: If three times of failure, no more meter reading will be performed, and the next meter reading cycle will be entered and the reading will be performed last until the collection is successful and the meter will be merged into the normal meter group and arranged in the order of data reading according to the data reading interval.

[0070] Through the above steps S223-S224, the collection terminal can add a learning function based on the original priority scheduling of the collection task. The electric meter with the shortest collection time within the same priority can be determined and collected first, which can reduce the time consumption of the initial collection and thus improve the collection efficiency of the minute-level data collection task. In addition, through the above steps S225-S226, by adding a supplementary collection mechanism, the electric meter that failed to collect once can be supplemented. The occurrence rate of collection errors can be reduced, and the success rate of the minute-level data collection task can be improved.

[0071] For the remote communication channel between the terminal and the master station, such as Figure 2 As shown, the present invention also includes:

[0072] S227: When the communication quality of the collection terminal meets the preset conditions, the historical collection data is supplemented with data; the historical collection data is the minute-level data in a maximum of 5 collection cycles before the current time.

[0073] S228: If the data collection terminal fails to copy the data for 4 times, the weather and disaster information of the installation area corresponding to the data collection terminal is obtained.

[0074] The weather and disaster information may be the weather and disaster information of the installation area within 3-5 days before the date of the time period of failed collection. The weather and disaster information may include precipitation, temperature drop, thunderstorm, wind force, earthquake and other information.

[0075] These meteorological disasters may have serious impacts on the power system, such as line short circuit, tripping, equipment damage, etc., thus affecting the stability and reliability of the power system.

[0076] Thunderstorms: Thunderstorms may cause short circuits, tripping, or even equipment damage in transmission lines, thus affecting the stability of the power system.

[0077] Storms: Strong winds may knock down power poles, damage transmission lines and equipment, and cause power outages and power failures. At the same time, storms may also trigger natural disasters such as floods and mudslides, further exacerbating the damage to the power system.

[0078] Earthquakes: Seismic waves can severely damage power facilities such as substations and transmission lines, leading to power outages and supply shortages.

[0079] S229: If the weather and disaster information in the installation area meets the preset conditions, the collection terminal sends a collection failure instruction to the preset communication terminal to notify the corresponding staff that the communication in the target installation area may be poor or the equipment may be damaged due to the influence of weather conditions.

[0080] The acquisition failure instruction may include the terminal name of the corresponding acquisition terminal, the location information of the acquisition terminal, and the time of acquisition failure, so that the above information can be summarized into a text message notification information. Then the information is sent to the mobile phone (i.e., communication terminal) of the operation and maintenance personnel responsible for operating and maintaining this acquisition terminal, so that the corresponding operation and maintenance personnel can know the damage information of the equipment more timely, so as to carry out manual inspection and maintenance more timely.

[0081] The above step S227 can add a minute task re-reporting mechanism for the loss of points caused by abnormal data reporting caused by the data communication of the master station. If there is a large network delay or the communication module is disconnected, the terminal will re-report the data after restoring normal communication, thereby reducing the possibility of data loss and improving the success rate of minute-level data collection tasks.

[0082] After optimizing the collection terminal program, the minute collection power increased from about 95% in the initial test to an average of 99.5%. The specific experimental data is shown in Table 2:

[0083] Table 2: xxx Company's minute collection success rate on January 13

[0084]

[0085] S230. According to the priority in S210 and the reporting scheme in S220, the photovoltaic data collection terminal transmits the collected result data to the ferry server for control in real time.

[0086] S300 and the ferry server generate CIME files with the result data every 20 seconds and feed them back to the D5000 system through the physical isolation device.

[0087] The S400 and D5000 systems generate corresponding control instructions based on the result data to realize the function of group control of distributed photovoltaic groups.

[0088] like Figure 3 As shown, S400 includes the following steps:

[0089] S401: Sending a control inquiry instruction to the user to be controlled;

[0090] S402: If the control feedback instruction received from the user is to reject the control, then send the control inquiry instruction again after a first preset period of time;

[0091] S403: Generate the current unit price information F of electricity charges for the user to be regulated according to the number of times a that the regulation inquiry instruction is sent to the user to be regulated, and add it to the subsequent regulation inquiry instruction; wherein, the unit price information F of electricity charges = f-min(K*a, Y), f is the basic electricity price of the user, K is the first price reduction coefficient, which can be 0.1 or 0.2, etc., and Y is the maximum price reduction value.

[0092] It should be emphasized that the embodiments described in the present invention are illustrative rather than restrictive. Therefore, the present invention includes but is not limited to the embodiments described in the specific implementation manner. Any other implementation manners derived by those skilled in the art based on the technical solution of the present invention also fall within the scope of protection of the present invention.

Claims

1. A distributed photovoltaic group control method, characterized by: The following steps are involved: The S100 and D5000 systems use a dedicated channel to pass control instructions to the acquisition system in the form of CIME format files through the physical isolation device; S200, after receiving the control instruction, the collection system immediately calls the dedicated photovoltaic control interface program to verify the photovoltaic user. After the verification is passed, the control instruction is immediately sent to multiple corresponding photovoltaic collection terminals for execution. The collection terminals collect corresponding data, and the photovoltaic collection terminals transmit the collected result data to the control ferry server in real time; S300 and the ferry server generate CIME files with the result data every 20 seconds and feed them back to the D5000 system through the physical isolation device; The S400 and D5000 systems generate corresponding control instructions based on the result data to realize the function of group control of distributed photovoltaic groups.

2. A distributed photovoltaic group control method according to claim 1, characterized in that: The specific implementation method of the acquisition terminal in S200 to collect corresponding data is: S210, configuring the corresponding priority for the tasks in the collection terminal, collecting the corresponding data and accurately predicting the output of the photovoltaic power station; S220, configuring a data reporting scheme for the collection terminal; S230. According to the priority in S210 and the reporting scheme in S220, the photovoltaic data collection terminal transmits the collected result data to the ferry server for control in real time.

3. A distributed photovoltaic group control method according to claim 2, characterized in that: The specific implementation method of configuring the corresponding priority in S210 is: dividing the task priority according to the relevant technical specifications of the collection terminal and the electric energy meter, and according to the data usage, timeliness and importance.

4. A distributed photovoltaic group control method according to claim 2, characterized in that: The specific implementation method of predicting the output of photovoltaic power stations in S210 is: converting the photovoltaic output timing curves of multiple stations and numerical weather forecast data into graph structure input features, establishing a spatiotemporal graph convolutional neural network model that combines the graph convolution method with the gated recurrent unit, and extracting the hidden spatiotemporal dependencies between photovoltaic clusters; using the adjacency matrix adaptive correction method to automatically learn the spatial relationship between different photovoltaic clusters from massive historical data, and introducing the attention mechanism to further improve the prediction model, and finally accurately predicting the output of photovoltaic power stations through the improved prediction model.

5. A distributed photovoltaic group control method according to claim 2, characterized in that: The S220 includes the following steps: S221. Set the collection and reporting interval of the minute-level data collection task to 1 minute, and the reporting delay to 10 seconds, 15 seconds, or 20 seconds; the reporting execution period is 05:00-20:00 every day; S222, setting the minute-level data collection task to report only a single data item, the single data item being an active power item; S223, setting the interval for recording the meter data of each household during each meter reading; S224, according to the reading interval of each household's electricity meter data, the electricity meters of each household within the same priority level are arranged in reading order, and the electricity meters with the shortest reading interval are preferentially collected in each task cycle; S225. After the other electric meters have completed the collection within this task cycle, additional readings are performed. The initial number of additional readings is 3 times. S226: If all three attempts have failed, no more reading will be done, and the next meter reading cycle will be started and the reading will be done at the end until the data is successfully collected and merged into the normal meter group, and the reading order will be arranged according to the data reading interval; S227: When the communication quality of the acquisition terminal meets the preset conditions, the historical acquisition data is supplemented with data; the historical acquisition data is the minute-level data in a maximum of 5 acquisition cycles before the current time; S228: If the data collection terminal fails to copy data 4 times, obtain the weather and disaster information of the installation area corresponding to the data collection terminal; S229: If the weather and disaster information in the installation area meets the preset conditions, the collection terminal sends a collection failure instruction to the preset communication terminal to notify the corresponding staff that the communication in the target installation area may be poor or the equipment may be damaged due to the influence of weather conditions.

6. A distributed photovoltaic group control method according to claim 1, characterized in that: The S400 includes the following steps: S401: Sending a control inquiry instruction to the user to be controlled; S402: If the control feedback instruction received from the user is to reject the control, then send the control inquiry instruction again after a first preset period of time; S403: Generate the current unit price information F of electricity charges for the user to be regulated according to the number of times a that the regulation inquiry instruction is sent to the user to be regulated, and add it to the subsequent regulation inquiry instruction; wherein, the unit price information F of electricity charges = f-min(K*a, Y), f is the basic electricity price of the user, K is the first price reduction coefficient, which can be 0.1 or 0.2, etc., and Y is the maximum price reduction value.

Images