A "two detailed rules" power management improvement system and power management improvement method
By designing the 'two rules' power management improvement system, the problem that the auxiliary service management system of power plants in central China cannot monitor and verify the wind power assessment situation in real time, real-time monitoring of the assessment results of the entire network and optimization of the power prediction model. Improve management efficiency and reliability.
Patent Information
- Application Number
- CN202010291443.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2020-04-14
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2040-04-14
AI Technical Summary
In the prior art, the technical support system cannot monitor the assessment of wind power power in the entire network in real time, verify the correctness of the assessment results, and cannot determine the accuracy of the assessment results. Station unit installation assessment position and ranking.
Design a 'two-standard' power management improvement system, including a data source acquisition module, a 'two-standard' assessment algorithm prediction and actual power deviation calculation module and a visual report generation module, and write an automatic acquisition program through Java to obtain short-term and ultra-short-term , 96 points of power available, power prediction and deviation calculation are performed, and a multi-dimensional BI interface display is generated through Tableau to realize real-time monitoring and algorithm optimization.
Real-time monitoring of the assessment results of the "two details" of the entire network has been realized, which reduces the time and error rate of manual statistics, improves management efficiency and reliability, optimizes the power prediction model, and enhances the industry's benchmarking capabilities.
Smart Images

Figure CN111428939B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a technical field of power development and management, in particular to a "two detailed rules" power management improvement system and a power management improvement method, and belongs to the technical field of lean development of the power market. Background Art
[0002] In order to ensure the safe, high-quality and economical operation of the Central China power grid and safeguard the legitimate rights and interests of power generation enterprises, the Central China Regulatory Bureau has further revised and improved the "Implementation Rules for Auxiliary Management Services of Grid-connected Power Plants in Central China (Trial)" and "Implementation Rules for Grid-connected Operation Management of Grid-connected Power Plants in Central China" (hereinafter referred to as the "Two Rules") in accordance with the requirements of the "Notice on Approval to Print and Issue Implementation Rules for Auxiliary Service Management of Power Plants in Central China, namely Grid-connected Operation Management" issued by the State Electricity Regulatory Commission. The "Two Rules" include 11 assessment items in total, including: safety management assessment, black start assessment, dispatching management assessment, unplanned outage assessment, daily power generation plan assessment, AGC assessment, primary frequency regulation assessment, reactive power regulation assessment, peak load adjustment assessment, and detection management assessment. There are five types of auxiliary service compensation, including: spinning standby compensation, reactive power compensation, black start compensation, paid peak load adjustment compensation, and AGC service compensation.
[0003] Since the launch of the technical support system for the "two detailed rules" of a provincial power grid, each power plant has only been assigned one account to manage the assessment and apportionment results of the "two detailed rules" of a single power plant. In this case, it is impossible to obtain the assessment situation of the entire network, and the apportionment data is also unknown, so it is impossible to verify whether the assessment results are correct, and it is impossible to know and determine the position and ranking of the installed capacity assessment of the station unit. Based on this, there is an urgent need to build a platform to improve management capabilities. Through the big data analysis platform, the wind power assessment situation can be monitored in real time, and the algorithm of the existing wind power model can be optimized. Summary of the invention
[0004] In order to overcome the deficiencies and shortcomings of the prior art, the present invention provides a "two rules" power management improvement system, including:
[0005] The data source acquisition module uses Java to write an automatic acquisition program to acquire data sources, including short-term, ultra-short-term, and 96-point data on available power and the assessment results of the "two detailed rules" published by the dispatching side;
[0006] The "Two Rules" assessment algorithm prediction and actual power deviation calculation module is used to make power predictions and calculate actual power deviations based on 96 points of short-term, ultra-short-term, and available power data of each station, and compare them with the data published by the dispatching side to determine the correctness of the assessment results; and
[0007] The visual report generation module is used to display the collected data sources, 96-point predictions, and actual power calculation assessment results in a multi-dimensional BI interface, including BI interface display by power type, group name, section difference, DC matching, and industry benchmarking.
[0008] Preferably, the automatic acquisition program based on Java is written to obtain the data source, including: simulating user login and page form submission operations, using Java as the development language, using the OkHttp communication framework for client-server communication, first simulating login with the site side account, and then sending data to the server.
[0009] Preferably, the data source acquisition module also includes: modeling based on the collected data, and displaying the data in the Navicat client, wherein the data display includes grouping each electric field in the form of an ID according to the power generation group, the section where it is located, whether it belongs to a DC supporting system, and whether it is self-owned, and different power source type setting fields are distinguished by numbers.
[0010] Preferably, the "two rules" assessment algorithm prediction and actual power deviation calculation module has a built-in assessment algorithm based on two rules, and has an autonomous calculation function. The 96-point prediction and actual power collected are added with the "two rules" assessment algorithm, and the daily assessment points are calculated autonomously, which are compared with the data published by the dispatching side to avoid assessments caused by dispatching system problems. The assessment algorithm based on the two rules includes:
[0011] (1) Short-term power forecast assessment: Wind farms and photovoltaic power stations should submit short-term power forecast curves to power control agencies on time. The maximum error of the daily forecast curve provided by the wind farm shall not exceed 25%, and the maximum error of the daily forecast curve provided by the photovoltaic power station shall not exceed 20%. If the standard is not met, the assessment will be based on the deviation integral electricity of 0.2 points / 10,000 kWh. The calculation formula for the maximum error value of the daily forecast curve is as follows:
[0012] Where: i is the number of points; n is 96 points; is the predicted value of available power at the i-th point; is the actual power at the i-th point.
[0013] when When it is zero, or within 3% of the installed capacity, it will not be assessed; when it exceeds 3% of the installed capacity, the error value will be calculated as 100%.
[0014] when When it is zero, or within 3% of the installed capacity, it will not be assessed; when it exceeds 3% of the installed capacity, the error value will be calculated as 100%.
[0015] The short-term power forecast assessment algorithm rules include:
[0016] ① Excluding the forecast and actual power during the power-limited period, the power-limited period will not be assessed;
[0017] Power restriction period: When wind power output is restricted due to power grid reasons, the power restriction points are excluded. If the number of valid points on the day is ≤24 points, the whole day is exempted from the test; if the number of valid points is greater than 24 points, the assessment is based on the valid points; when photovoltaic power output is restricted due to power grid reasons, if the number of valid points on the day is ≤16 points, the whole day is exempted from the test; if the number of valid points is greater than 16 points, the assessment is based on the valid points;
[0018] ② When the actual power is 0, the predicted power is within 3% of the installed capacity, and the data point is removed. If it exceeds 3% of the installed capacity, it is assessed as 100% deviation;
[0019] ③ When the predicted power is 0, the actual power is within 3% of the installed capacity, and the data point is removed. If it exceeds 3% of the installed capacity, it is assessed as 100% deviation;
[0020] (2) Ultra-short-term assessment: Wind farms and photovoltaic power stations shall submit ultra-short-term power forecast curves to the power control agency on time. The accuracy of the second-hour harmonic mean of the ultra-short-term forecast curves of wind farms and photovoltaic power stations shall not be less than 75%. If it fails to meet the standard, the assessment will be based on the total installed capacity × 0.015 points / 10,000 kilowatts for every 1% reduction;
[0021] Where: i is the number of points; n is 96 points; is the predicted value of available power at the i-th point; is the actual power at the i-th point;
[0022] When the predicted available power and the actual power are both within 3% of the installed capacity, this point is not included in the error calculation;
[0023] The ultra-short-term assessment algorithm rules include:
[0024] ① The forecast and actual power during the power-limited period are excluded, and the power-limited period is not assessed;
[0025] Power restriction period: When wind power output is restricted due to power grid reasons, the power restriction points are excluded. If the number of valid points on the day is ≤24 points, the whole day will be exempted from the test; if the number of valid points is greater than 24 points, the assessment will be conducted based on the valid points. When photovoltaic power output is restricted due to power grid reasons, if the number of valid points on the day is ≤16 points, the whole day will be exempted from the test; if the number of valid points is greater than 16 points, the assessment will be conducted based on the valid points.
[0026] ② When the actual power and predicted power are both less than 3% of the installed capacity, the point is eliminated;
[0027] (3) Available power assessment: The integral amount of available power generation is the available power. The daily accuracy of available power should be no less than 97%. Every 1% decrease will be assessed at the rate of total installed capacity × 0.05 points / 10,000 kW.
[0028] in: E is the actual power generation of the new energy station during the non-power-restriction period of the day; T It is the available power of the new energy station during the non-power-restriction period of the day; when Em is 0, the accuracy of the available power will no longer be assessed;
[0029] The available power assessment algorithm rules include:
[0030] ① Exclude the forecast and actual power during the power-limited period, and the power-limited period will not be assessed
[0031] ② When the actual power consumption is less than 3% of the installed capacity, this point will be eliminated and no longer assessed.
[0032] Preferably, the visualization report generation module is implemented as follows:
[0033] Use BI to quickly create an interface and use the tool Tableau to generate visual reports. Tableau supports direct connection to MySQL data sources, and automatically collects the "two rules" assessment status of the entire network through scheduled tasks every day. New data is stored in the database through the scheduled tasks of the data collection module, and the Tableau dashboard is automatically updated to add new data to the visual report.
[0034] Preferably, the Tableau dashboard is divided into short-term and ultra-short-term power forecast upload reports, monitors the reporting rate of each station, and promptly inquires about the cause of the problem and retransmits data for stations below 100%; 96-point data collection report determines the degree of deviation between the predicted power and the actual power, and promptly contacts the manufacturer to correct the model for large deviations; autonomous calculation and dispatch side statistical data comparison report determines the accuracy of the data published by the dispatch side; the compensation result table of various types of power supply in the whole network automatically collects the assessment and compensation results of various types of power supply in the whole network, determines the assessment trend of this month, and determines the score to be shared by each new energy power field when the compensation is greater than the assessment; the whole network new Energy unit capacity assessment sub-reports can realize real-time assessment result monitoring of short-term, ultra-short-term and available power of new energy unit capacity in Xinjiang, determine the accuracy of power prediction of each electric field, and timely check and fill in the gaps in the monthly assessment result report; any power generation group electric field monthly assessment data report can realize the display of monthly assessment results of new energy electric fields in the whole region, without the need to log in to query each station one by one, and conduct benchmarking of assessment results of stations within the region and under the same section; power generation group unit capacity assessment ranking report can realize the assessment score ranking; power generation hours display report can realize the benchmarking of power generation hours of new energy power stations in different dimensions and understand the power generation situation of the whole network.
[0035] Preferably, all dashboards can directly view reports and export them to Excel for offline file transfer.
[0036] Preferably, it also includes a centralized power prediction system, into which the collected 96-point short-term and ultra-short-term data are imported for training and optimizing the neural network prediction model, thereby improving the accuracy of short-term and ultra-short-term predictions and reducing the assessment scores.
[0037] The present invention also aims to provide a method for improving power management, comprising:
[0038] Step 1: Set up a scheduled task to automatically collect and obtain data sources:
[0039] Based on the daily results published by the technical support system of the "Two Regulations" of the power grid, a program written in Java is used to automatically collect 96 points of data on short-term, ultra-short-term, and available power of each electric field, the upload rate, and the assessment results of the day, and collect them into MySQL. Based on the collected data, the data is displayed on the Navicat client, and each electric field is grouped in the form of ID according to the power generation group, the section where it is located, whether it belongs to the DC supporting system, and whether it is self-owned;
[0040] Step 2: Based on the 96-point prediction and actual power, add an algorithm to calculate the error:
[0041] Based on 96 points of short-term, ultra-short-term and available power data of each station, the "two detailed rules" assessment algorithm is added to make predictions and calculate actual power deviations, and compared with the data published by the dispatching side to determine the correctness of the assessment results;
[0042] Step 3: Display the collected data sources, 96-point predictions, and actual power calculation results in a multi-dimensional BI interface, including displaying the BI interface by power type, group name, section difference, DC matching, and conducting industry benchmarking.
[0043] Preferably, the method further includes step 4, importing the collected 96-point short-term and ultra-short-term data into a centralized power prediction system for training and optimizing a neural network prediction model, improving the accuracy of short-term and ultra-short-term predictions, and reducing assessment scores.
[0044] Beneficial effects of the present invention:
[0045] 1. Save labor costs, improve work efficiency and reduce mistakes:
[0046] The automatic collection platform can monitor the assessment results of the "two rules" of the entire network of a province in real time, including wind and solar power upload rate, short-term, ultra-short-term, and available assessment results. It used to take 10 minutes for each station to manually count the results of a single field. There are 188 wind farms and 336 photovoltaic power stations in Xinjiang, which saves about 87 hours of labor per day, reducing a lot of manual statistics time. The missed statistics and error rates are reduced to 0, and the reliability is significantly improved.
[0047] 2. Add an algorithm to review the assessment results published by the power grid:
[0048] Based on the prediction and actual power uploaded by the station within 15 minutes, the "two detailed rules" assessment algorithm is added to realize the review of the published results and greatly reduce the assessment points caused by errors in the dispatching system.
[0049] 3. Achieve industry benchmarking in multiple dimensions:
[0050] The construction of this platform enables the comparison of the assessment results and online hours of the two rules in multiple dimensions, such as within each group, between different groups, stations under the same section, and DC supporting facilities, to help new energy companies promptly identify detailed problems in the "two rules", draw management ideas, make the industry's power generation situation and the assessment results of the two rules transparent, and facilitate learning from each other's strengths among stations.
[0051] 4. Optimize power prediction model:
[0052] The grid assessment results, 15-minute data line graph, and comparison results of predicted and actual data are transmitted back to the centralized power prediction system for training and optimizing the neural network prediction model to improve prediction accuracy.
[0053] Based on the following detailed description of specific embodiments of the present invention in conjunction with the accompanying drawings, those skilled in the art will become more aware of the above and other objects, advantages and features of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0054] Hereinafter, some specific embodiments of the present invention will be described in detail in an exemplary and non-limiting manner with reference to the accompanying drawings. The same reference numerals in the accompanying drawings indicate the same or similar components or parts. It should be understood by those skilled in the art that these drawings are not necessarily drawn to scale. The objects and features of the present invention will become more apparent in view of the following description in conjunction with the accompanying drawings, in which:
[0055] Figure 1 This is a schematic diagram of an automatic acquisition program interface written in Java according to an embodiment of the present invention;
[0056] Figure 2 A schematic diagram of Navicat client data display according to an embodiment of the present invention;
[0057] Figure 3 A schematic diagram of wind and solar power upload rate prediction according to an embodiment of the present invention;
[0058] Figure 4 This is a schematic diagram of collecting 96 points of short-term data of an arbitrary electric field in a single day according to an embodiment of the present invention;
[0059] Figure 5 This is a schematic diagram of collecting 96 points of ultra-short-term data of an arbitrary electric field in a single day according to an embodiment of the present invention;
[0060] Figure 6 Schematic diagram of 96-point data collection of available power in a single day in any electric field according to an embodiment of the present invention;
[0061] Figure 7 A schematic diagram showing comparison between autonomous computing and data published by the scheduling side according to an embodiment of the present invention;
[0062] Figure 8 A schematic diagram of compensation results of various types of power supplies in the entire network according to an embodiment of the present invention;
[0063] Fig. 9 A schematic diagram of the unit capacity assessment of new energy sources for the entire network according to an embodiment of the present invention;
[0064] Fig.10 This is a schematic diagram of monthly assessment data of an electric field of any power generation group according to an embodiment of the present invention;
[0065] Fig.11 A schematic diagram of ranking of unit capacity assessment scores of power generation groups according to an embodiment of the present invention;
[0066] Fig.12 A diagram showing power generation hours according to an embodiment of the present invention;
[0067] Fig.13 4 is a flow chart of a method for improving power management according to an embodiment of the present invention. DETAILED DESCRIPTION
[0068] In order to more clearly describe the specific implementation mode, the following technical terms appearing in the specific implementation mode are first explained:
[0069] 1. "Two Rules" in Northwest China: In order to ensure the safe, high-quality and economical operation of the Northwest power system, standardize the grid-connected operation management of power plants, safeguard the legitimate rights and interests of power companies, and promote the coordinated development of power grids and power generation companies, according to the "Power Plant Grid-Connected Operation Management Regulations" (Electricity Supervision Market
[2006] No. 42) and relevant national laws and regulations, combined with the actual situation of the Northwest power system, the "Northwest Region Power Plant Grid-Connected Operation Management Implementation Rules" and "Northwest Region Grid-Connected Power Plant Auxiliary Service Management Implementation Rules" are formulated.
[0070] 2. Technical support system for the "two detailed rules" of a provincial power grid: The provincial supervision office of the National Energy Administration regularly publishes the "two detailed rules" assessment and apportionment total scores of each power generation enterprise, so that each power generation enterprise can understand the assessment details and check the source. It includes the assessment points caused by inaccurate short-term, ultra-short-term, and available power forecasts, the AGC dead zone pass rate and response time, etc.
[0071] 3. Wind and solar power forecast upload rate: Wind farms and photovoltaic power stations should submit short-term and ultra-short-term power forecast curves and other data files that meet the operation requirements to the power regulation agency on time. The upload rate should be greater than 95%. If it does not meet the standard, it will be assessed at the full-site capacity × 6 points / 10,000 kilowatts for every 1% reduction. If the upload rate does not meet the standard due to the main station, it will not be assessed.
[0072] 4. Accuracy of short-term prediction of wind and solar power: Wind farms and photovoltaic power stations should submit short-term power prediction curves to power regulation agencies on time. The maximum error of the daily prediction curve provided by the wind farm shall not exceed 25%, and the maximum error of the daily prediction curve provided by the photovoltaic power station shall not exceed 20%. If the standard is not met, the deviation integral electricity will be assessed at 0.2 points per 10,000 kWh.
[0073] 5. Accuracy of ultra-short-term power forecast: Wind farms and photovoltaic power stations should submit ultra-short-term power forecast curves to power regulation agencies on time. The accuracy of the second-hour harmonic average of the ultra-short-term forecast curves of wind farms and photovoltaic power stations should be no less than 75%. If it fails to meet the standard, it will be assessed at the rate of total installed capacity × 0.015 points / 10,000 kilowatts for every 1% reduction.
[0074] 6. Accuracy of available power generation: Theoretical power generation of new energy refers to the power that can be generated when all generator sets can operate normally under the current wind and light resource conditions; available power generation refers to the power that can be generated after the equipment failure, defect or maintenance on the site is blocked. The integral power of available power generation is the available power. The daily accuracy of available power should be no less than 97%. Every 1% decrease is assessed at the rate of total installed capacity × 0.05 points / 10,000 kilowatts.
[0075] The "two rules" power management improvement system of this embodiment includes:
[0076] See also Figure 1 , an automatic collection program written in Java realizes 96 points of short-term, ultra-short-term and available power data and the dispatching side publishes the assessment results of the "two rules".
[0077] See also Figure 2, based on the collected data, the model is built and the data is displayed in the Navicat client. This platform creates 6 models, namely 96-point data collection and calculation, daily assessment results, wind and solar power forecast assessment details, various types of power supply assessment summary, assessment compensation summary, public power field assessment compensation results, and each power field is grouped in the form of ID according to the power generation group, the section, whether it belongs to the DC supporting system, and whether it is self-owned. Different power types are set to distinguish fields by numbers.
[0078] 1. Data source acquisition module, based on Java to write automatic collection program to achieve data source acquisition, the data source includes daily assessment upload rate data, short-term wind and solar power 96-point forecast of designated stations, assessment data, ultra-short-term wind and solar power 96-point forecast of designated stations, assessment data, 96-point available power data of designated stations, wind and solar power daily assessment data of designated stations in designated months, wind and solar power daily assessment data of all stations in designated months, wind and solar power daily assessment data of designated stations in multiple months, and wind and solar power daily assessment data of all stations in multiple months;
[0079] The automatic collection program written based on Java to obtain the data source includes:
[0080] The "Two Detailed Rules Technical Support System of a Certain Province" currently does not provide an interface for Xinjiang electric field query services. This system uses simulated user login and page form submission operations. This system uses Java as the development language and uses the OkHttp communication framework for client-server communication. First, use the station side account to simulate login, and then send data to the server. The data collection procedures are as follows:
[0081] (1) A province - two detailed assessment platforms - daily assessment upload rate data collection service
[0082] public class XjkhPredictionDailyEvaluationService {
[0083] @Autowired
[0084] private StationDAO stationDAO;
[0085] @Autowired
[0086] private BaseStationDAO baseStationDAO;
[0087] @Autowired
[0088] private PredictionDailyEvaluationCrawler evaluationCrawler;
[0089] @Autowired
[0090] private PredictionDailyEvaluationDAO evaluationDAO;
[0091] @Autowired
[0092] private PredictionPowerDAO powerDAO;
[0093] (2) Obtain 96-point forecast and assessment data of short-term wind and solar power at designated sites
[0094] * @param xjwcId
[0095] * @param date
[0096] * @return
[0097] * /
[0098] Public Map <LocalDateTime, Map<String, String> >getDayaheadPredictedPower(int xjwcId, LocalDate date) {
[0099] Map <LocalDateTime, Map<String, String> > dataMap = newLinkedHashMap<>();
[0100] StringBuilder dayaheadPowerUrl = new StringBuilder("http: / / 10.218.230.212:8080 / xjkh / fdglyckh_rqfglyc.do?method=getListXml")
[0101] .append("&pdate=").append(DateTimeUtil.convertLocalDateToString(date))
[0102] .append("&pdateend=").append(DateTimeUtil.convertLocalDateToString(date))
[0103] .append("&plantid=").append(xjwcId);
[0104] OkHttpClient client = OkHttpUtil.getClient();
[0105] Response response = OkHttpUtil.doGet(client,dayaheadPowerUrl.toString());
[0106] String html = "";
[0107] try {
[0108] html = response.body().string();
[0109] } catch (Exception e) {
[0110] e.printStackTrace();
[0111] } finally {
[0112] if(response != null) {
[0113] response.close();
[0114] }
[0115] }
[0116] Document document = Jsoup.parse(html);
[0117] Elements rows = document.getElementsByTag("row");
[0118] if(rows.size() != 96) return null;
[0119] for(Element row : rows) {
[0120] Map<String, String> dataItemMap = new LinkedHashMap<>();
[0121] Elements cells = row.getElementsByTag("cell");
[0122] LocalDateTime time = LocalDateTime.parse(cells.get(0).text() + " " + cells.get(1).text() + ":00", DateTimeFormatter.ofPattern("yyyy-MM-dd HH:mm:ss"));
[0123] dataItemMap.put("dayaheadPredictedPower", cells.get(3).text());
[0124] dataItemMap.put("actualPower", cells.get(4).text());
[0125] dataMap.put(time, dataItemMap);
[0126] }
[0127] (3) Obtain 96-point forecast and assessment data of ultra-short-term wind and solar power at designated sites
[0128] * @param xjwcId
[0129] * @param date
[0130] * @return
[0131] * /
[0132] Public Map <LocalDateTime, Map<String, String> >getRealtimePredictedPower(int xjwcId, LocalDate date) {
[0133] Map <LocalDateTime, Map<String, String> > dataMap = newLinkedHashMap<>();
[0134] StringBuilder realtimePowerUrl = new StringBuilder("http: / / 10.218.230.212:8080 / xjkh / fdglyckh_ssfglyc.do?method=getListXml")
[0135] .append("&pdate=").append(DateTimeUtil.convertLocalDateToString(date))
[0136] .append("&pdateend=").append(DateTimeUtil.convertLocalDateToString(date))
[0137] .append("&plantid=").append(xjwcId);
[0138] OkHttpClient client = OkHttpUtil.getClient();
[0139] Response response = OkHttpUtil.doGet(client,realtimePowerUrl.toString());
[0140] String html = "";
[0141] try {
[0142] html = response.body().string();
[0143] } catch (Exception e){
[0144] e.printStackTrace();
[0145] } finally {
[0146] if(response != null) response.close();
[0147] }
[0148] Document document = Jsoup.parse(html);
[0149] Elements rows = document.getElementsByTag("row");
[0150] if(rows == null ||rows.size() != 96) return null;
[0151] for(Element row : rows) {
[0152] Elements cells = row.getElementsByTag("cell");
[0153] LocalDateTime time = LocalDateTime.of(LocalDate.parse(cells.get(0).text()),
[0154] LocalTime.parse(cells.get(1).text()));
[0155] Map<String, String> dataItemMap = new LinkedHashMap<>();
[0156] dataItemMap.put("realtimePredictedPower", cells.get(3).text());
[0157] dataItemMap.put("actualPower", cells.get(4).text());
[0158] dataMap.put(time, dataItemMap);
[0159] }
[0160] StringBuilder realtimeEvaluationUrl = new StringBuilder("http: / / 10.218.230.212:8080 / xjkh / fdglyckh_ssfglyckhjg.do?method=getListXmlRkhjg")
[0161] .append("&pdate=").append(DateTimeUtil.convertLocalDateToString(date))
[0162] .append("&pdateend=").append(DateTimeUtil.convertLocalDateToString(date))
[0163] .append("&plantid=").append(xjwcId);
[0164] Response response2 = OkHttpUtil.doGet(client,realtimeEvaluationUrl.toString());
[0165] String html2 = "";
[0166] try {
[0167] html2 = response2.body().string();
[0168] } catch (Exception e) {
[0169] e.printStackTrace();
[0170] } finally {
[0171] if(response2 != null) response2. close();
[0172] }
[0173] Document document2 = Jsoup.parse(html2);
[0174] Elements rows2 = document2.getElementsByTag("row");
[0175] for(Element row : rows2) {
[0176] Elements cells = row.getElementsByTag("cell");
[0177] LocalDateTime time = LocalDateTime.of(LocalDate.parse(cells.get(1).text()), LocalTime.parse(cells.get(2).text()));
[0178] Map<String, String> dataItemMap = dataMap.get(time);
[0179] dataItemMap.put("isLimited", "0");
[0180] dataItemMap.put("realtimeAbsoluteError", cells.get(6).text().replace("%", ""));
[0181] }
[0182] for(Map.Entry<LocalDateTime, Map<String, String>> entry :dataMap.entrySet()) {
[0183] Map<String, String> dataItemMap = entry.getValue();
[0184] if(! dataItemMap.containsKey("isLimited")) {
[0185] dataItemMap.put("isLimited", "1");
[0186] dataItemMap.put("realtimeAbsoluteError", "0");
[0187] }
[0188] }
[0189] return dataMap;
[0190] (4) Obtain the 96-point available power data of the specified station
[0191] * @param xjwcId
[0192] * @param date
[0193] * @return
[0194] * /
[0195] public Map<LocalDateTime, Map<String, String>> getAvailablePower(int xjwcId, LocalDate date) {
[0196] Map<LocalDateTime, Map<String, String>> dataMap = newLinkedHashMap<>();
[0197] StringBuilder url = new StringBuilder("http: / / 10.218.230.212:8080 / xjkh / fdglyckh_kyglcx.do?method=getListXmlRkhjg")
[0198] .append("&pdate=").append(DateTimeUtil.convertLocalDateToString(date))
[0199] .append("&pdateend=").append(DateTimeUtil.convertLocalDateToString(date))
[0200] .append("&plantid=").append(xjwcId);
[0201] OkHttpClient client = OkHttpUtil.getClient();
[0202] Response response = OkHttpUtil.doGet(client, url.toString());
[0203] String html = "";
[0204] try {
[0205] html = response.body().string();
[0206] } catch (Exception e) {
[0207] e.printStackTrace();
[0208] } finally {
[0209] if(response != null) response.close();
[0210] }
[0211] Document document = Jsoup.parse(html);
[0212] Elements rows = document.getElementsByTag("row");
[0213] if(rows == null || rows.size() == 0) return null;
[0214] for(Element row : rows) {
[0215] Elements cells = row.getElementsByTag("cell");
[0216] LocalDateTime time = LocalDateTime.of(LocalDate.parse(cells.get(1).text()), LocalTime.parse(cells.get(2).text()));
[0217] Map<String, String> dataItemMap = new LinkedHashMap<>();
[0218] dataItemMap.put("availablePower", cells.get(4).text());
[0219] dataItemMap.put("actualPower", cells.get(5).text());
[0220] dataItemMap.put("isLimited", "0");
[0221] dataMap.put(time, dataItemMap);
[0222] }
[0223] return dataMap;
[0224] }
[0225] (5) Obtain the daily wind and solar power assessment data for a specified station and a specified month
[0226] * @param station
[0227] * @param month
[0228] * /
[0229] public void getDailyEvaluation(final Station station, finalLocalDate month) {
[0230] LocalDate startDate = LocalDate.of(month.getYear(),month.getMonthValue(), 1);
[0231] LocalDate endDate = startDate.plusMonths(1).minusDays(1);
[0232] / / Crawler collects data
[0233] Map <LocalDate, Map<String, String> > uploadRateDataMap =evaluationCrawler.getUploadRate(station.getXjwcId(), month);
[0234] Map <LocalDate, Map<String, String> >predictionEvaluationDataMap = evaluationCrawler.getDailyPredictionEvaluation(station.getXjwcId(), month);
[0235] List <predictiondailyevaluationdo>evaluations = new ArrayList<>();
[0236] / / Merge data and store it in the database
[0237] while(! startDate.isAfter(endDate)) {
[0238] / / If the data for the current day is not collected, continue to process the data for the next day
[0239] if((uploadRateDataMap == null || !uploadRateDataMap.containsKey(startDate)) ||
[0240] predictionEvaluationDataMap == null && ! predictionEvaluationDataMap.containsKey(startDate)) {
[0241] startDate = startDate.plusDays(1);
[0242] continue;
[0243] }
[0244] / / Prevent duplicate storage. If the data has already been stored, skip it.
[0245] PredictionDailyEvaluationDO oldEvaluation = evaluationDAO.findByStationIdAndDate(station.getStationId(), startDate);
[0246] if(oldEvaluation != null) {
[0247] startDate = startDate.plusDays(1);
[0248] continue;
[0249] }
[0250] PredictionDailyEvaluationDO evaluationDO = newPredictionDailyEvaluationDO();
[0251] evaluationDO.setStationId(station.getStationId());
[0252] evaluationDO.setDate(startDate);
[0253] evaluationDO.setCreatedTime(LocalDateTime.now());
[0254] evaluationDO.setIsCalculated(0);
[0255] if(uploadRateDataMap != null &&uploadRateDataMap.containsKey(startDate)) {
[0256] evaluationDO.setDayaheadUploadRate(Double.parseDouble(uploadRateDataMap.get(startDate).get("dayaheadUploadRate")));
[0257] evaluationDO.setRealtimeUploadRate(Double.parseDouble(uploadRateDataMap.get(startDate).get("realtimeUploadRate")));
[0258] }
[0259] if(predictionEvaluationDataMap != null && predictionEvaluationDataMap.containsKey(startDate)) {
[0260] Map<String, String> dataItem =predictionEvaluationDataMap.get(startDate);
[0261] if(dataItem.containsKey("realtimeAccuracyRate")) {
[0262] try {
[0263] evaluationDO.setRealtimeAccuracyRate(NumberUtil.round(Double.parseDouble(dataItem.get("realtimeAccuracyRate")) / 100, 4));
[0264] evaluationDO.setRealtimePunishedScore(Double.parseDouble(dataItem.get("realtimePunishedScore")));
[0265] evaluationDO.setAvailableAccuracyRate(NumberUtil.round(Double.parseDouble(dataItem.get("availableAccuracyRate")) / 100, 4));
[0266] evaluationDO.setAvailablePunishedScore(Double.parseDouble(dataItem.get("availablePunishedScore")));
[0267] } catch (Exception e) {
[0268] e.printStackTrace();
[0269] }
[0270] }
[0271] }
[0272] evaluations.add(evaluationDO);
[0273] startDate = startDate.plusDays(1);
[0274] }
[0275] evaluationDAO.addAll(evaluations);
[0276] System.out.println("Has been collected #" + station.getStationId() + " "+ DateTimeFormatter.ofPattern("yyyy-MM").format(month)
[0277] + "Daily assessment data of wind and solar power forecast");
[0278] }
[0279] (6) Obtain the daily wind and solar power assessment data for all stations in a specified month
[0280] * @param month
[0281] * /
[0282] public void getDailyEvaluation(final LocalDate month) {
[0283] List <station>stations = stationDAO.findNewEnergy();
[0284] for(Station station : stations) {
[0285] this.getDailyEvaluation(station, month);
[0286] }
[0287] }
[0288] (7) Obtain daily wind and solar power assessment data for multiple months at a designated site
[0289] * @param station
[0290] * @param startMonth
[0291] * @param endMonth
[0292] * /
[0293] public void getDailyEvaluation(final Station station, finalLocalDate startMonth, final LocalDate endMonth) {
[0294] LocalDate month = startMonth;
[0295] while(! month.isAfter(endMonth)) {
[0296] this.getDailyEvaluation(station, startMonth, month);
[0297] month = month.plusMonths(1);
[0298] }
[0299] }
[0300] (8) Obtain daily wind and solar power assessment data for all stations in multiple months
[0301] * @param startMonth
[0302] * @param endMonth
[0303] * /
[0304] public void getDailyEvaluation(final LocalDate startMonth, finalLocalDate endMonth) {
[0305] LocalDate month = startMonth;
[0306] while(! month.isAfter(endMonth)) {
[0307] this.getDailyEvaluation(month);
[0308] month = month.plusMonths(1);
[0309] }
[0310] }
[0311] 2. The "two detailed rules" assessment algorithm prediction and actual power deviation calculation module is used to make power predictions and calculate actual power deviations based on 96 points of short-term, ultra-short-term and available power data of each station, and compare them with the data published by the dispatching side to determine the correctness of the assessment results.
[0312] The "two detailed rules" assessment algorithm prediction and actual power deviation calculation module is built-in based on the two detailed rules assessment algorithm and has independent calculation function.
[0313] The 96-point prediction and actual power collected are added with the "two detailed rules" assessment algorithm, and the daily assessment points are calculated independently. They are compared with the data published by the dispatching side to avoid assessments caused by dispatching system problems.
[0314] (1) Short-term power forecast assessment clause: Wind farms and photovoltaic power stations shall submit short-term power forecast curves to the power control agency on time. The maximum error of the daily forecast curve provided by the wind farm shall not exceed 25%, and the maximum error of the daily forecast curve provided by the photovoltaic power station shall not exceed 20%. If the standard is not met, the assessment shall be based on the deviation integral electricity of 0.2 points / 10,000 kWh. The calculation formula for the maximum error value of the daily forecast curve is as follows:
[0315] Where: i is the number of points; n is 96 points; is the predicted value of available power at the i-th point; is the actual power at the i-th point.
[0316] when When it is zero, or within 3% of the installed capacity, it will not be assessed; when it exceeds 3% of the installed capacity, the error value will be calculated as 100%.
[0317] when When it is zero, or within 3% of the installed capacity, it will not be assessed; when it exceeds 3% of the installed capacity, the error value will be calculated as 100%.
[0318] Program construction principles:
[0319] ①. Exclude the forecast and actual power during the power-limited period (power-limited period is not assessed)
[0320] Power restriction period: When wind power output is restricted due to power grid reasons, the power restriction points are excluded. If the number of valid points is ≤24 points on the day, the whole day will be exempted from the test; if the number of valid points is greater than 24 points, the assessment will be conducted based on the valid points. When photovoltaic power output is restricted due to power grid reasons, if the number of valid points is ≤16 points on the day, the whole day will be exempted from the test; if the number of valid points is greater than 16 points, the assessment will be conducted based on the valid points.
[0321] ② When the actual power is 0, the predicted power is within 3% of the installed capacity and the data point is eliminated. If it exceeds 3% of the installed capacity, it is assessed at 100% deviation.
[0322] ③. When the predicted power is 0, the actual power is within 3% of the installed capacity and the data point is eliminated. If it exceeds 3% of the installed capacity, it is assessed at 100% deviation.
[0323] Calculate the daily assessment score and daily average accuracy rate based on the 96-point short-term power curve of the power grid (self-calculation)
[0324] * @param powers
[0325] * @param station
[0326] * @return
[0327] * /
[0328] Private Map<String, Double> calculateDayaheadEvaluationBySelf(final List <predictionpowerdo>powers, final BaseStation station) {
[0329] Map<String, Double> map = new HashMap<>();
[0330] int limitedPointCount = 0;
[0331] / / Eliminate power restriction points
[0332] Iterator <predictionpowerdo>powerDOIterator = powers.iterator();
[0333] while(powerDOIterator.hasNext()) {
[0334] PredictionPowerDO power = powerDOIterator.next();
[0335] if(power.getIsLimited() == 1) {
[0336] limitedPointCount++;
[0337] powerDOIterator.remove();
[0338] continue;
[0339] }
[0340] if(power.getActualPower() < 0 ) power.setActualPower(0);
[0341] }
[0342] / / Determine whether to exempt from assessment throughout the day
[0343] if(station.getType() == 3 && limitedPointCount > 72) {
[0344] map.put("dayaheadAccuracyRateBySelf", 1.0);
[0345] map.put("dayaheadPunishedScoreBySelf", 0.0);
[0346] return map;
[0347] } else if(station.getType() == 4 && limitedPointCount > 80) {
[0348] map.put("dayaheadAccuracyRateBySelf", 1.0);
[0349] map.put("dayaheadPunishedScoreBySelf", 0.0);
[0350] return map;
[0351] }
[0352] / / Calculate the assessment score
[0353] double dayaheadPunishedScore = 0.0;
[0354] for(PredictionPowerDO power : powers) {
[0355] double absoluteError;
[0356] if(power.getDayaheadPredictedPower() <=0 &&power.getActualPower() >= 0.03 * station.getCapacity()) {
[0357] absoluteError = 1;
[0358] } else if(power.getActualPower() <=0 &&power.getDayaheadPredictedPower() >= 0.03 * station.getCapacity()){
[0359] absoluteError = 1;
[0360] } else {
[0361] absoluteError = Math.abs((power.getDayaheadPredictedPower() - power.getActualPower()) / power.getDayaheadPredictedPower());
[0362] }
[0363] if(station.getType() == 3 && absoluteError <= 0.25) { / / Wind power exemption assessment threshold
[0364] continue;
[0365] } else if (station.getType() == 4 && absoluteError <= 0.2) { / / PV non - assessment threshold
[0366] continue;
[0367] } else if (power.getDayaheadPredictedPower() <= 0 && power.getActualPower() <= 0.03 * station.getCapacity()) {
[0368] continue;
[0369] } else if (power.getActualPower() <= 0 && power.getDayaheadPredictedPower() <= 0.03 * station.getCapacity()) {
[0370] continue;
[0371] } else {
[0372] double coefficient = station.getType() == 3? 0.25 : 0.2;
[0373] dayaheadPunishedScore += NumberUtil.round(((Math.abs(power.getDayaheadPredictedPower() - power.getActualPower()) - power.getDayaheadPredictedPower() * coefficient) * 0.25 / 10000 * 0.2), 4);
[0374] dayaheadPunishedScore = NumberUtil.round(dayaheadPunishedScore, 4);
[0375] }
[0376] }
[0377] map.put("dayaheadPunishedScoreBySelf",dayaheadPunishedScore);
[0378] map.put("dayaheadAccuracyRateBySelf", 0.0);
[0379] return map;
[0380] }
[0381] (2) Ultra-short-term assessment clause: Wind farms and photovoltaic power stations shall submit ultra-short-term power forecast curves to the power control agency on time. The accuracy of the second-hour harmonic average of the ultra-short-term forecast curves of wind farms and photovoltaic power stations shall not be less than 75%. If the standard is not met, the assessment will be based on the total installed capacity × 0.015 points / 10,000 kilowatts for every 1% reduction.
[0382] Where: i is the number of points; n is 96 points; is the predicted value of available power at the i-th point; is the actual power at the i-th point.
[0383] When the predicted available power and the actual power are both within 3% of the installed capacity, this point is not included in the error calculation.
[0384] Program construction principles:
[0385] ①. Exclude the forecast and actual power during the power-limited period (power-limited period is not assessed)
[0386] Power restriction period: When wind power output is restricted due to power grid reasons, the power restriction points are excluded. If the number of valid points is ≤24 points on the day, the whole day will be exempted from the test; if the number of valid points is greater than 24 points, the assessment will be conducted based on the valid points. When photovoltaic power output is restricted due to power grid reasons, if the number of valid points is ≤16 points on the day, the whole day will be exempted from the test; if the number of valid points is greater than 16 points, the assessment will be conducted based on the valid points.
[0387] ② When the actual power and predicted power are both less than 3% of the installed capacity, the point is eliminated.
[0388] Calculate daily assessment points based on the 96-point curve of the power grid's ultra-short-term power forecast (self-calculation)
[0389] * @param powers
[0390] * @return
[0391] * /
[0392] private Map<String, Double> calculateRealtimeEvaluationBySelf(final List <predictionpowerdo>powers, final BaseStation station) {
[0393] Map<String, Double> realtimeResultMap = new HashMap<>();
[0394] int limitedPointCount = 0;
[0395] / / Eliminate power restriction points
[0396] Iterator <predictionpowerdo>powerDOIterator = powers.iterator();
[0397] while(powerDOIterator.hasNext()) {
[0398] PredictionPowerDO power = powerDOIterator.next();
[0399] if(power.getIsLimited() == 1) {
[0400] limitedPointCount++;
[0401] powerDOIterator.remove();
[0402] continue;
[0403] }
[0404] if(power.getActualPower() < 0.03 * station.getCapacity()&& power.getRealtimePredictedPower() < 0.03 * station.getCapacity()) {
[0405] powerDOIterator.remove();
[0406] continue;
[0407] }
[0408] }
[0409] / / Determine whether it is exempt from assessment throughout the day. The rules for wind power and photovoltaic are different
[0410] if(station.getType() == 3 && limitedPointCount > 72) {
[0411] realtimeResultMap.put("realtimeAccuracyRateBySelf", 0.0);
[0412] realtimeResultMap.put("realtimePunishedScoreBySelf",0.0);
[0413] return realtimeResultMap;
[0414] } else if(station.getType() == 4 && limitedPointCount > 80){
[0415] realtimeResultMap.put("realtimeAccuracyRateBySelf", 0.0);
[0416] realtimeResultMap.put("realtimePunishedScoreBySelf",0.0);
[0417] return realtimeResultMap;
[0418] }
[0419] / / Calculate the ultra - short - term prediction accuracy rate (harmonic mean algorithm)
[0420] double cumulatedError = 0.0;
[0421] for(PredictionPowerDO power : powers) {
[0422] cumulatedError += Math.abs(power.getActualPower() -power.getRealtimePredictedPower());
[0423] }
[0424] double accuracyRate = 1.0;
[0425] for(PredictionPowerDO power : powers) {
[0426] double factor1 = Math.abs(power.getActualPower() / (power.getActualPower() + power.getRealtimePredictedPower()) - 0.5);
[0427] double factor2 = Math.abs(power.getActualPower() -power.getRealtimePredictedPower()) / cumulatedError;
[0428] accuracyRate -= 2 * factor1 * factor2;
[0429] }
[0430] accuracyRate = NumberUtil.round(accuracyRate, 4);
[0431] if(accuracyRate < 0) accuracyRate = 0; / / The lower limit of accuracy is 0
[0432] / / Calculate the assessment score
[0433] double realtimePunishedScore = 0.0;
[0434] if(accuracyRate >= 0.75) { / / Short-term power prediction assessment-free threshold
[0435] realtimePunishedScore = 0.0;
[0436] } else {
[0437] realtimePunishedScore = NumberUtil.round((0.75 -accuracyRate) * 100, 0) * 0.015 * station.getCapacity() / 10000;
[0438] realtimePunishedScore = NumberUtil.round(realtimePunishedScore,4);
[0439] }
[0440] realtimeResultMap.put("realtimeAccuracyRateBySelf",accuracyRate);
[0441] realtimeResultMap.put("realtimePunishedScoreBySelf",realtimePunishedScore);
[0442] return realtimeResultMap;
[0443] }
[0444] (3) Available power assessment clause: Theoretical power generation of new energy refers to the power that can be generated when all generator sets can operate normally under the current wind and light resource conditions; available power generation refers to the power that can be generated after the equipment failure, defect or maintenance on the site is blocked. The integral amount of available power generation is the available power. The daily accuracy of available power should be no less than 97%. Every 1% decrease is assessed at the rate of total installed capacity × 0.05 points / 10,000 kilowatts.
[0445] in: E is the actual power generation of the new energy station during the non-power-restriction period of the day; T It is the available power of the new energy station during the non-power-restriction period of the day. When Em is 0, the accuracy of the available power will no longer be assessed.
[0446] Program construction principles:
[0447] ① Eliminate the forecast and actual power during the power-limited period (power-limited period is not assessed)
[0448] ② When the actual power consumption is less than 3% of the installed capacity, this point will be eliminated and no longer assessed.
[0449] Calculate daily assessment points based on the 96-point curve of available power of the power grid (self-calculation)
[0450] * @param powers
[0451] * @return
[0452] * /
[0453] Private Map<String, Double> calculateAvailableEvaluationBySelf(final List <predictionpowerdo>powers, final BaseStation station) {
[0454] / / Eliminate power-limited points and points where actual output is less than 3%
[0455] Iterator <predictionpowerdo>powerDOIterator = powers.iterator();
[0456] while(powerDOIterator.hasNext()) {
[0457] PredictionPowerDO power = powerDOIterator.next();
[0458] if(power.getIsLimited() == 1) {
[0459] powerDOIterator.remove();
[0460] continue;
[0461] }
[0462] if(power.getActualPower() <= 0.03 * station.getCapacity()) {
[0463] powerDOIterator.remove();
[0464] continue;
[0465] }
[0466] }
[0467] / / Calculate the integrated power
[0468] double actualQuantity = 0.0; / / Actual power generation
[0469] double availableQuantity = 0.0; / / Available power generation
[0470] for(PredictionPowerDO power : powers) {
[0471] actualQuantity += power.getActualPower() * 0.25;
[0472] availableQuantity += power.getAvailablePower() * 0.25;
[0473] }
[0474] / / Calculate the accuracy of available power
[0475] double accuracyRate;
[0476] if(actualQuantity <= 0) {
[0477] accuracyRate = 1; / / When the actual power generation is not greater than 0, the accuracy is 100% and no assessment is performed
[0478] } else {
[0479] accuracyRate = 1 - (Math.abs(actualQuantity -availableQuantity) / actualQuantity);
[0480] if(accuracyRate < 0) accuracyRate = 0; / / Avoid negative values
[0481] }
[0482] accuracyRate = NumberUtil.round(accuracyRate, 4);
[0483] / / Calculate available power assessment score
[0484] double punishedScore = 0.0;
[0485] if(accuracyRate >= 0.97) { / / Available power exemption threshold
[0486] punishedScore = 0.0;
[0487] } else {
[0488] punishedScore = NumberUtil.round((0.97 - accuracyRate) *100, 0) * 0.05 * station.getCapacity() / 10000;
[0489] punishedScore = NumberUtil.round(punishedScore, 4);
[0490] }
[0491] 3. Visual report generation module, which is technically implemented as follows in this embodiment:
[0492] Use BI to quickly create interfaces, save development costs, and develop faster, and respond quickly to changing needs. This solution uses the popular BI tool Tableau to generate visual reports. Tableau can support direct connection to MySQL data sources. The "two rules" assessment situation of the entire network is automatically collected through scheduled tasks every day. New data is stored in the database through the scheduled tasks of the data collection module. The Tableau dashboard can be automatically updated and new data is added to the visual report.
[0493] The Tableau dashboard is divided into short-term and ultra-short-term power forecast upload reports, which monitor the reporting rate of each station. For stations with a reporting rate below 100%, the cause of the problem is promptly queried and the data is supplemented; the 96-point data collection report determines the degree of deviation between the predicted power and the actual power. For problems with large deviations, the manufacturer is contacted in time to correct the model; the self-calculated and dispatch-side statistical data comparison report determines the accuracy of the data published by the dispatch side; the evaluation and compensation result table of various types of power sources in the entire network automatically collects the evaluation and compensation results of various types of power sources in a province, determines the evaluation trend of this month, and determines the score to be shared by each new energy power field when the compensation is greater than the evaluation; the evaluation report of new energy unit capacity in the entire network realizes the short-term evaluation of new energy unit capacity in Xinjiang Real-time assessment result monitoring of long-term, ultra-short-term and available power, judging the accuracy of power prediction of each electric field, and timely checking and filling in the gaps in the monthly assessment result report; Monthly assessment data report of any power generation group's electric field, realizing the display of monthly assessment results of new energy electric fields in Xinjiang, no longer needing to log in to query each station one by one, and benchmarking the assessment results of the stations within the group and under the same section; Unit capacity assessment ranking report of power generation group, realizing the assessment score ranking of major power generation groups in Xinjiang, which is conducive to mutual communication and learning among major groups; Power generation hours display report, realizing the benchmarking of power generation hours of new energy power stations in different dimensions in Xinjiang, understanding the power generation situation of the whole network, facilitating benchmarking of major groups in the same industry, and improving the accuracy of power prediction;
[0494] All dashboards can directly view reports and export them to Excel, which is convenient for offline file transfer and conducive to work. Figure 3-12 As shown:
[0495] Figure 3 In the process, wind power and photovoltaic power stations upload a short-term file to the provincial dispatching department once a day, and upload an ultra-short-term file every 15 minutes. If the upload rate does not reach 95%, it will be assessed by the dispatching department. Tableau is used to realize daily upload monitoring of power plants across Xinjiang. If the reporting rate is less than 100%, it can be uploaded in time.
[0496] Figure 4 This is the short-term upload file details, with 1 forecast and actual power data every 15 minutes. If the single point deviation is greater than 25%, the deviation part will be assessed. Using Tableau to achieve real-time deviation monitoring of short-term forecast upload data is conducive to timely adjustment of the short-term forecast model for deviation correction.
[0497] Figure 5 This is the ultra-short-term upload file details, with 1 forecast and actual power data every 15 minutes. The accuracy of the whole-day harmonized average is less than 75%, and the deviation part is assessed. Using Tableau to achieve real-time deviation monitoring of ultra-short-term forecast upload data is conducive to timely adjustment of the ultra-short-term forecast model for deviation correction.
[0498] Figure 6 This is the available power file details. Through AGC, 1 prediction and actual power data is generated every minute. The dispatch side takes 96 points throughout the day. The overall available power accuracy shall not be less than 97%. The deviation part shall be assessed. Use tableau to realize real-time deviation monitoring of available power upload data and timely rectification.
[0499] Figure 7 In the process, the collected dispatching side assessment scores are compared with the assessment scores independently calculated based on the 96-point data deviation. This function is to verify the accuracy of the assessment results announced by the dispatching side.
[0500] Figure 8 In the process, the assessment and compensation results of various types of power sources in the entire province are automatically collected to judge the assessment trend of this month. When the compensation is greater than the assessment, the score that each new energy power field needs to share is determined.
[0501] Fig. 9 This function realizes the real-time monitoring of the assessment results of short-term, ultra-short-term and available power of new energy unit capacity in Xinjiang, determines the accuracy of power prediction of each electric field, and promptly detects and fills gaps.
[0502] Fig.10 This function can realize the display of monthly assessment results of new energy power plants in Xinjiang. There is no need to log in to query each station one by one, and the assessment results of stations within the group and under the same section can be compared.
[0503] Fig.11 This function realizes the assessment score ranking of major power generation groups across Xinjiang, which is conducive to mutual communication and learning among major groups and improves the accuracy of power prediction.
[0504] Fig.12 This function is implemented to benchmark the power generation hours of new energy power stations in different dimensions across Xinjiang, understand the power generation situation of the entire network, and facilitate benchmarking among major groups in the same industry.
[0505] 4. It also includes a centralized power prediction system, which imports the collected short-term and ultra-short-term data of 96 points into the centralized power prediction system for training and optimizing the neural network prediction model, improving the accuracy of short-term and ultra-short-term predictions and reducing the assessment scores.
[0506] See also Fig.13 The present invention also aims to provide a method for improving power management, comprising:
[0507] Step 1: Set up a scheduled task to automatically collect and obtain data sources;
[0508] In this embodiment, based on the daily results published by the technical support system of the "two detailed rules" of a provincial power grid, a program written in Java is used to automatically collect 96 points of short-term, ultra-short-term, and available power data of each electric field in the province, the upload rate, and the assessment results of the day, and collect them into MySQL. Based on the collected data, the data is displayed on the Navicat client, and each electric field is grouped in the form of id according to the power generation group to which it belongs, the section where it is located, whether it belongs to the DC supporting system, and whether it is self-owned.
[0509] Step 2, based on the 96-point prediction and actual power, add an algorithm to calculate the error;
[0510] In this embodiment, based on 96 points of data on short-term, ultra-short-term and available power of each station, the "two detailed rules" assessment algorithm is added to perform prediction and actual power deviation calculation, and compared with the data published by the dispatching side to determine the correctness of the assessment results.
[0511] Step 3: Display the data source, 96-point prediction and actual power calculation results on a multi-dimensional BI interface
[0512] In this embodiment, the data source and 96-point prediction and actual power calculation assessment results are collected and obtained, and the BI interface is displayed according to multiple dimensions such as power source type, group name, section difference, DC matching, etc., and industry benchmarking is carried out.
[0513] In this embodiment, step 4 is added to import the collected 96 points of short-term and ultra-short-term data into the centralized power prediction system for training and optimizing the neural network prediction model, thereby improving the accuracy of short-term and ultra-short-term predictions and reducing the assessment scores.
[0514] See also Fig.13 The method implementation steps of this embodiment specifically include:
[0515] (1) Automatically collect short-term, ultra-short-term, and 96 available data points and dispatch-side assessment results;
[0516] (2) Add an algorithm to independently calculate the assessment results based on the 96-point data;
[0517] (3) Determine whether the assessment results are consistent; if they are consistent, execute steps (4) and (5); if they are inconsistent, check the reasons for the data differences with the dispatching side and return to step (1);
[0518] (4) Use Tableau to prepare visual reports;
[0519] (5) Send the data back to the centralized power prediction system optimization model.
[0520] Beneficial effects of the present invention:
[0521] 1. Save labor costs, improve work efficiency and reduce mistakes:
[0522] The automatic collection platform can monitor the assessment results of the "two rules" of the entire network of a province in real time, including wind and solar power upload rate, short-term, ultra-short-term, and available assessment results. It used to take 10 minutes for each station to manually count the results of a single field. There are 188 wind farms and 336 photovoltaic power stations in Xinjiang, which saves about 87 hours of labor per day, reducing a large amount of manual statistics time. The missed statistics and error rates are reduced to 0, and the reliability is significantly improved.
[0523] 2. Add an algorithm to review the assessment results published by the power grid:
[0524] Based on the prediction and actual power uploaded by the station within 15 minutes, the "two detailed rules" assessment algorithm is added to realize the review of the published results and greatly reduce the assessment points caused by errors in the dispatching system.
[0525] 3. Achieve industry benchmarking in multiple dimensions:
[0526] The construction of this platform enables the comparison of the assessment results and online hours of the two rules in multiple dimensions, such as within each group, between different groups, stations under the same section, and DC supporting facilities, to help new energy companies promptly identify detailed problems in the "two rules" and gain management ideas, making the industry's power generation situation and the assessment results of the two rules transparent, facilitating learning from each other's strengths.
[0527] 4. Optimize power prediction model:
[0528] The grid assessment results, 15-minute data line graph, and comparison results of predicted and actual data are transmitted back to the centralized power prediction system for training and optimizing the neural network prediction model to improve prediction accuracy.
[0529] Although the present invention has been described with reference to specific illustrative embodiments, it is not limited to these embodiments but only to the appended claims. It should be understood by those skilled in the art that the embodiments of the present invention can be changed and modified without departing from the scope and spirit of the present invention.< / predictionpowerdo> < / predictionpowerdo> < / predictionpowerdo> < / predictionpowerdo> < / predictionpowerdo> < / predictionpowerdo> < / station> < / predictiondailyevaluationdo>
Claims
1. A "two rules" power management improvement system, characterized by include: The data source acquisition module uses Java to write an automatic acquisition program to acquire data sources, including short-term, ultra-short-term, and 96-point data on available power and the assessment results of the "two detailed rules" published by the dispatching side; The "two detailed rules" assessment algorithm prediction and actual power deviation calculation module is used to make power predictions and calculate actual power deviations based on 96 points of short-term, ultra-short-term and available power data of each station, and compare them with the data published by the dispatching side to determine the correctness of the assessment results; as well as Visual report generation module, used to display the data source, 96-point forecast and actual power calculation results in a multi-dimensional BI interface, including BI interface display by power type, group name, section difference, DC matching and other dimensions, and conduct industry benchmarking; The method of writing an automatic collection program based on Java to obtain the data source includes: simulating user login and page form submission operations, using Java as the development language, and using the OkHttp communication framework to communicate between the client and the server. First, the station side account is used to simulate login, and then data is sent to the server; The data source acquisition module also includes: modeling based on the collected data, and displaying the data in the Navicat client, wherein the data display includes grouping each electric field in the form of an ID, including the power generation group to which it belongs, the section where it is located, whether it belongs to the DC supporting system, and whether it is self-owned, and different power source type setting fields are distinguished by numbers; The "two rules" assessment algorithm prediction and actual power deviation calculation module has a built-in assessment algorithm based on two rules, which has an autonomous calculation function. The 96-point prediction and actual power collected are added with the "two rules" assessment algorithm to autonomously calculate the daily assessment points and compare them with the data published by the dispatching side to avoid assessments caused by dispatching system problems. The assessment algorithm based on the two rules includes: (1) Short-term power forecast assessment: Wind farms and photovoltaic power stations shall submit short-term power forecast curves to the power control agency on time. The maximum error of the daily forecast curve provided by the wind farm shall not exceed 25%, and the maximum error of the daily forecast curve provided by the photovoltaic power station shall not exceed 20%. If the standard is not met, the assessment shall be based on the deviation integral power of 0.2 points / 10,000 kWh; the maximum error value of the daily forecast curve is calculated as follows: Where: i is the number of points; n is 96 points; is the predicted value of available power at the i-th point; is the actual power at the i-th point; when When it is zero, and is within 3% of the installed capacity, no assessment will be made; when it exceeds 3% of the installed capacity, the error value will be calculated as 100%; when When it is zero, and is within 3% of the installed capacity, no assessment will be made; when it exceeds 3% of the installed capacity, the error value will be calculated as 100%; The short-term power prediction assessment algorithm rules include: ① Excluding the forecast and actual power during the power-limited period, the power-limited period will not be assessed; Power restriction period: When wind power output is restricted due to power grid reasons, the power restriction points are excluded. If the number of valid points on the day is ≤24 points, the whole day is exempted from the test; if the number of valid points is greater than 24 points, the assessment is based on the valid points; when photovoltaic power output is restricted due to power grid reasons, if the number of valid points on the day is ≤16 points, the whole day is exempted from the test; if the number of valid points is greater than 16 points, the assessment is based on the valid points; ② When the actual power is 0, the predicted power is within 3% of the installed capacity, and the data point is removed. If it exceeds 3% of the installed capacity, it is assessed as 100% deviation; ③ When the predicted power is 0, the actual power is within 3% of the installed capacity, and the data point is removed. If it exceeds 3% of the installed capacity, it is assessed as 100% deviation; (2) Ultra-short-term assessment: Wind farms and photovoltaic power stations shall submit ultra-short-term power forecast curves to the power control agency on time. The accuracy of the second-hour harmonic mean of the ultra-short-term forecast curves of wind farms and photovoltaic power stations shall not be less than 75%. If it fails to meet the standard, the assessment will be based on the total installed capacity × 0.015 points / 10,000 kilowatts for every 1% reduction; Where: i is the number of points; n is 96 points; is the predicted value of available power at the i-th point; is the actual power at the i-th point; When the predicted available power and the actual power are both within 3% of the installed capacity, this point is not included in the error calculation; The principles for building the ultra-short-term assessment algorithm include: ① The forecast and actual power during the power-limited period are excluded, and the power-limited period is not assessed; Power restriction period: When wind power output is restricted due to power grid reasons, the power restriction points are excluded. If the number of valid points on the day is ≤24 points, the whole day is exempted from the test; if the number of valid points is greater than 24 points, the assessment is based on the valid points; when photovoltaic power output is restricted due to power grid reasons, if the number of valid points on the day is ≤16 points, the whole day is exempted from the test; if the number of valid points is greater than 16 points, the assessment is based on the valid points; ② When the actual power and predicted power are both less than 3% of the installed capacity, the point is eliminated; (3) Available power assessment: The integral amount of available power generation is the available power. The daily accuracy of available power should be no less than 97%. Every 1% decrease will be assessed at the rate of total installed capacity × 0.05 points / 10,000 kW. in: E is the actual power generation of the new energy station during the non-power-restriction period of the day; T It is the available power of the new energy station during the non-power-restriction period of the day; when Em is 0, the accuracy of the available power will no longer be assessed; The available power assessment algorithm rules include: ① Excluding the forecast and actual power during the power-limited period, the power-limited period will not be assessed; ② When the actual power consumption is less than 3% of the installed capacity, this point will be eliminated and no longer assessed.
2. According to claim 1, a "two rules" power management improvement system is characterized in that The visualization report generation module is implemented as follows: Use BI to quickly create an interface, and use the tool Tableau to generate visual reports. Tableau supports direct connection to MySQL data sources, and automatically collects the "two rules" assessment results of the entire network through scheduled tasks every day. New data is stored in the database through the scheduled tasks of the data collection module, and the Tableau dashboard is automatically updated to add new data to the visual report.
3. According to claim 2, a "two rules" power management improvement system is characterized in that The Tableau dashboard is divided into short-term and ultra-short-term power forecast upload reports, which monitor the reporting rate of each station. For stations with a reporting rate lower than 100%, the cause of the problem is promptly checked and data is re-transmitted; 96-point data collection report to determine the degree of deviation between predicted power and actual power. If the deviation is large, contact the manufacturer in time to correct the model; self-calculation and dispatch side statistical data comparison report to determine the accuracy of the data published by the dispatch side; the assessment and compensation result table of various types of power sources in the whole network, automatically collects the assessment and compensation results of various types of power sources in the whole network, determines the assessment trend of this month, and determines the score to be shared by each new energy power field when the compensation is greater than the assessment; The whole network's new energy unit capacity assessment report can realize the real-time assessment result monitoring of the short-term, ultra-short-term and available power of the new energy unit capacity in Xinjiang, judge the accuracy of the power prediction of each electric field, and timely check for omissions and fill in the monthly assessment result report; Monthly assessment data report of power plants of any power generation group can realize the display of monthly assessment results of new energy power plants in the entire region, without the need to log in to query each station one by one, and benchmark the assessment results of stations within the region and under the same section; the unit capacity assessment ranking report of the power generation group can realize the assessment score ranking; the power generation hours display report can realize the benchmarking of power generation hours of new energy power stations in different dimensions and understand the power generation situation of the entire network.
4. According to claim 3, a "two rules" power management improvement system is characterized in that All dashboards can directly view reports and export them to Excel for offline file transfer.
5. A "two rules" power management improvement system according to any one of claims 1-4, characterized in that It also includes a centralized power prediction system, which imports the collected 96-point short-term and ultra-short-term data into the centralized power prediction system for training and optimizing the neural network prediction model, improving the accuracy of short-term and ultra-short-term predictions and reducing assessment scores.
6. A "two rules" power management improvement method, using the "two rules" power management improvement system as described in any one of claims 1-5, characterized in that include: Step 1: Set up a scheduled task to automatically collect and obtain data sources: Based on the daily results published by the technical support system of the "Two Regulations" of the power grid, a program written in Java is used to automatically collect 96 points of data on short-term, ultra-short-term, and available power of each electric field, the upload rate, and the assessment results of the day, and collect them into MySQL. Based on the collected data, the data is displayed on the Navicat client, and each electric field is grouped in the form of ID according to the power generation group, the section where it is located, whether it belongs to the DC supporting system, and whether it is self-owned; Step 2: Based on the 96-point prediction and actual power, add an algorithm to calculate the error: Based on 96 points of short-term, ultra-short-term and available power data of each station, the "two detailed rules" assessment algorithm is added to make predictions and calculate actual power deviations, and compared with the data published by the dispatching side to determine the correctness of the assessment results; Step 3: Display the collected data sources, 96-point predictions, and actual power calculation results in a multi-dimensional BI interface, including displaying the BI interface by power type, group name, section difference, DC matching, and conducting industry benchmarking.
7. The method according to claim 6, characterized in that: It also includes step 4, importing the collected 96 points of short-term and ultra-short-term data into the centralized power prediction system for training and optimizing the neural network prediction model, improving the accuracy of short-term and ultra-short-term predictions, and reducing the assessment scores.
