Comprehensive evaluation method and system for sampling data quality of smart substation

Abstract

The present invention provides a method and system for comprehensive evaluation of sampling data quality of smart substations, including: obtaining merging unit data, measurement and control device data, and PMU data in the substation to form multi-source redundant three-phase measurements under a unified time scale in the substation; using multi-source redundant three-phase measurements in the substation and the real-time switching state of the substation model and circuit breaker to perform three-phase linear state estimation of the substation, and using a weighted least squares algorithm to solve the state estimation problem; performing a comprehensive comparison based on the state estimation results, using the Chi-square test to determine whether there is bad data in the measurement quantity, and applying the maximum normalized residual LNR method to identify the most likely bad measurement quantity, and obtaining a comprehensive analysis result of data quality. The present invention is based on multi-interval, multi-type, and obtainable sampling data in the existing integrated monitoring system in the station, to achieve comprehensive evaluation of sampling data quality analysis of smart substations and improve the reliability of substation operation.

Description

Technical Field

[0001] The present invention relates to the technical field of smart substations, and in particular to a method and system for comprehensively evaluating the quality of sampling data of smart substations. Background Art

[0002] The rapid development of smart grids has promoted the scientific and technological progress of power grids and accelerated the application of new theories, technologies and equipment in power systems. Smart substations are one of the important contents of smart grid construction and are also the key "support point" for information interaction in smart grids. Since my country began to study smart grids, it has attached great importance to the research of smart substation technology and equipment and its application in power grids.

[0003] There are many information systems in smart substations, such as data acquisition and monitoring systems, online monitoring systems, fault recording systems, etc. Each system is built independently, and with the increasing number of advanced applications in smart substations, higher requirements are placed on the quality of basic data. The accuracy and reliability of raw data has become a bottleneck issue affecting the development of advanced applications such as distributed state estimation.

[0004] At present, smart substations are built according to the "three layers and two networks" architecture. The "three layers" specifically refer to the station control layer, the bay layer and the process layer. The station control layer equipment includes the smart substation integrated monitoring host, data communication gateway, database server, integrated application server and planned maintenance terminal, etc. The bay layer includes various protection devices, measurement and control devices, etc. The process layer includes merging units, smart terminals, etc. The "two networks" specifically refer to the process layer network (including GOOSE and SV messages) and the station control layer network (including MMS and GOOSE messages).

[0005] The merging unit is used to realize the digitization of analog quantities such as current and voltage; the intelligent terminal is used to realize the intelligence of equipment such as circuit breakers and disconnectors. The dual-configuration protection devices and intelligent terminals are connected to different process layer networks respectively; the single-configuration protection (measurement and control) devices and intelligent terminals are connected to a single process layer network; the main transformer electrical quantity protection is connected to the process layer network on each side at the same time. The protection device and the measurement and control device are connected to the station control layer network using a digital interface that supports the IEC61850 standard. The relevant information is transmitted to the substation station control layer equipment through the network in the form of MMS messages, and the interlocking information between the measurement and control devices is also transmitted through the network in the form of GOOSE messages.

[0006] According to the architecture system of the smart substation, the information of the whole station is collected by the merging unit and intelligent terminal of the process layer network and sent to the integrated monitoring background of the station through the measurement and control device. The relevant data has been time-stamped by the clock synchronization system during the collection stage. The above information covers the whole area, has the characteristics of high redundancy and high information integrity, and provides a good basic platform for the comprehensive utilization of the information of the whole station. At present, there are a lot of errors in the data provided by the background of the station. These errors may come from the collection device or the transmission medium, etc. The reliability of these raw data is very low, and they cannot be directly used for security analysis of the system. In order to improve this situation, on the one hand, from the hardware perspective, the error can be reduced by adding measurement equipment that can ensure accuracy, speed and reliability, but it will undoubtedly increase the construction cost of the smart substation; on the other hand, from the software perspective, the raw data collected by the system can be processed by means of power data analysis and modern state estimation, bad data can be eliminated and random interference can be eliminated, and the accuracy, reliability and integrity of data information can be improved. This method will undoubtedly become an efficient and feasible solution.

[0007] The power system synchronous phasor measurement unit (PMU) system based on the synchronous timing system has quite significant characteristics compared with traditional measurement equipment:

[0008] (1) The measurement accuracy meets the practical requirements of the project; the GPS synchronization accuracy is within 2ps, and the phase angle measurement accuracy can theoretically reach 0.1 degrees, which can meet the data synchronization requirements of power system analysis and control;

[0009] (2) The data has a precise time stamp, and the PMU data at different locations can be strictly synchronized, and the data is transmitted in a packaged manner;

[0010] (3) Direct measurement of power grid state quantities. Measurements in a rectangular coordinate system (including voltage and current phase measurement) are linearly related to the state quantities;

[0011] (4) The data update cycle is short, the transmission speed is fast, and it can track the dynamic changes of the system status. During operation, the PMU transmits 30 phase information per second, that is, the transmission interval can reach 30ms.

[0012] At present, newly built 500kV substations, 220kV and 110kV substations with large-scale renewable energy access are all equipped with synchronous phasor measurement devices. Although PMU has great advantages in technology, due to economic and technical constraints, PMU measurement points cannot achieve observability of the entire power grid or the entire substation. How to make full use of the high data redundancy and good integrity of existing measurement equipment, and effectively connect the technical advantages of PMU to monitor and identify bad power data, and achieve higher performance state estimation is a key issue in the current power system state estimation field. Summary of the invention

[0013] The purpose of the present invention is to provide a method and system for comprehensive evaluation of sampling data quality of an intelligent substation, which comprehensively utilizes multi-interval and multi-type sampling data in the station, and realizes three-phase linear state estimation of the sampling data under a unified time scale according to the clock synchronization system in the station, provides bad data identification for the sampling device in the station, and predicts in advance the probability of occurrence of bad data and the cause of bad data in a certain period of time, automatically pushes early warning signals and establishes bad data record reports, effectively improving the lean level of substation operation and maintenance.

[0014] In order to achieve the above objectives, the present invention is implemented by the following technical solutions:

[0015] A comprehensive evaluation method for sampling data quality of a smart substation, comprising:

[0016] Acquire the merging unit data, measurement and control device data, and PMU data in the substation to form multi-source redundant three-phase measurement under the unified time scale in the substation;

[0017] The three-phase linear state estimation of the substation is performed by using the redundant three-phase measurement in the substation, the substation model and the real-time switch state of the circuit breaker, and the weighted least squares algorithm is used to solve the state estimation problem.

[0018] A comprehensive comparison is performed based on the state estimation results, and the Chi-square test is used to determine whether there are bad data in the measurement quantity. The maximum normalized residual (LNR) method is applied to identify the most likely bad measurement quantity, and the comprehensive analysis results of the data quality are obtained.

[0019] Further, the obtaining of merging unit data in the substation includes:

[0020] The PT / CT output data of the digital sampling of the merging unit is read, and converted into synchronous phasor data under the same mark through the phasor conversion algorithm to obtain the corresponding voltage and current phasors, wherein the merging unit accepts clock synchronization and outputs sampling pulses that are precisely synchronized with the synchronization pulse to achieve synchronous sampling of the entire station data.

[0021] Furthermore, when performing phase conversion, a filter is used to filter the high-frequency harmonics of the system.

[0022] Furthermore, under system transient conditions, the phasor conversion algorithm adopts a reasonable optimized time window and ensures data processing speed, thereby ensuring that delays do not affect the accuracy of phasor data.

[0023] Furthermore, before performing linear state estimation, the data stream is quality checked in five aspects: communication and message format check, time and timing verification, PMU data state verification, data feature verification, and topology-based error detection;

[0024] If the data is declared as “bad” in any aspect, no further inspection is required, if the data is declared as “good” or “uncertain”, proceed with subsequent data inspection.

[0025] Furthermore, the substation model is in XML format and is defined by the CIM model.

[0026] Furthermore, the substation three-phase linear state estimation includes current state estimation and zero-impedance voltage state estimation, the former using Kirchhoff's current law to verify the circuit breaker current, and the latter being a weighted average of voltage phasor measurements.

[0027] Furthermore, the received circuit breaker switch states are verified before state estimation, and the substation topology is verified by using the current values ​​of all measured data to keep the substation model synchronized with the data being used.

[0028] Furthermore, the linear state estimation uses the synchronized phasor measurement information to check and verify the system topology. If a topology error is identified from the PMU data, this information is used to change the circuit breaker states to update the topology before the next ICCP update.

[0029] A comprehensive evaluation system for sampling data quality of a smart substation, comprising:

[0030] Data acquisition module, used to acquire the merging unit data, measurement and control device data, and PMU data in the substation, forming multi-source redundant three-phase measurement in the substation;

[0031] The state estimation module is used to perform three-phase linear state estimation of the substation using the redundant three-phase measurements in the substation and the substation model and the real-time switch state of the circuit breaker, and use the weighted least squares algorithm to solve the state estimation problem;

[0032] The bad data identification module is used to perform a comprehensive comparison based on the state estimation results, use the Chi-square test to determine whether there is bad data in the measurement quantity, and apply the maximum normalized residual LNR method to identify the most likely bad measurement quantity, and obtain the comprehensive analysis results of data quality.

[0033] Compared with the prior art, the present invention has the following advantages:

[0034] Based on the multi-interval, multi-type, and accessible sampling data in the existing integrated monitoring system in the station, a comprehensive evaluation of the sampling data quality analysis of the smart substation is achieved. It can not only serve as a redundant backup for the sampling device's own bad data identification, but also predict the probability of bad data occurrence and the cause of bad data in a certain period of time in advance, and provide early warning signals and bad data record reports, thereby improving the reliability of substation operation and facilitating the post-analysis of operation and maintenance personnel. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for the description. Obviously, the drawings described below are an embodiment of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work:

[0036] Figure 1 A flow chart of a method for comprehensive assessment of sampling data quality of a smart substation provided by an embodiment of the present invention;

[0037] Figure 2 This is an architecture diagram of a comprehensive evaluation system for sampling data quality of a smart substation provided by an embodiment of the present invention;

[0038] Figure 3 A schematic diagram of system data identification provided by an embodiment of the present invention;

[0039] Figure 4 A flow chart of a bad data discrimination method without a model benchmark provided by an embodiment of the present invention;

[0040] Figure 5 A schematic diagram of a KCL performing a comprehensive and observable PMU inspection in a substation according to an embodiment of the present invention;

[0041] Figure 6 A diagram of PMU measurement data at different sampling frequencies provided by an embodiment of the present invention;

[0042] Figure 7 A schematic diagram of a time scale marking of a multifunctional measurement and control device provided in one embodiment of the present invention;

[0043] Figure 8a A schematic diagram of checking that the sum of the currents injected into the node is 0 according to Kirchhoff's Current Law (KCL) provided in one embodiment of the present invention;

[0044] Figure 8b A schematic diagram of checking the current at both ends of a line using Kirchhoff's Current Law (KCL) provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The solution proposed by the present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0046] The quality of sampling data is an important indicator of substations. If there are errors, jitters or mutations in the sampling data, it may affect the reliability of the operation of the relay protection device, which is related to operational safety and efficiency.

[0047] like Figure 1 As shown, the present invention provides a comprehensive evaluation method for the quality of sampling data of a smart substation, comprising the following steps:

[0048] S1, obtain the merging unit data, measurement and control device data, and PMU data in the substation to form multi-source redundant three-phase measurement under the unified time scale in the substation;

[0049] S2, uses the redundant three-phase measurements in the substation, the substation model and the real-time switch status of the circuit breaker to perform three-phase linear state estimation of the substation, and uses the weighted least squares algorithm to solve the state estimation problem;

[0050] S3, based on the state estimation results, a comprehensive comparison is performed, the Chi-square test is used to determine whether there is bad data in the measurement quantity, and the maximum normalized residual (LNR) method is applied to identify the most likely bad measurement quantity, and the comprehensive analysis results of the data quality are obtained.

[0051] Combination Figure 2 As shown in the figure, through the standardized reading interface, the system platform offline imports or collects the merged unit data, SCADA-RTU and PMU measurement data in real time as the data basis for substation data identification. By mining, integrating and comprehensively analyzing relevant information such as data quality labels, data accuracy, time series, etc., and integrating the data itself, circuit breaker switch information, substation model, etc. through real-time analysis of the operating status, data abnormality warning and auxiliary decision-making are realized, and by combing the logical associations such as classification and stratification between various signals, the final alarm and display scheme under the normal fault abnormality of substation data is determined, and the comprehensive analysis results of data quality are obtained, which provide auxiliary decision-making for the operation and maintenance of measurement equipment and power grid operation. Then, through the reverse writing interface of the monitoring system platform, the substation data identification will reverse the calculated results into the real-time database of the platform, display them on the graphical interface, and provide a comprehensive display of alarm events. It also provides intelligent alarm, event information and abnormal analysis decision support for the superior system, and guides and helps the superior dispatcher (or monitoring personnel of the centralized control center) to quickly handle faults. The overall data identification process of the system is as follows Figure 3 shown.

[0052] When the system finds bad data, the key point is to further analyze the rules of the bad data, find the root cause of the problem through pattern recognition, and issue early warnings to the corresponding equipment. In particular, for sudden changes in data, it is necessary to distinguish whether it is caused by changes in system topology or system disturbances, or by faulty transformers, so as to avoid system misjudgment.

[0053] Data quality is a term that is often used but rarely defined. As a result, it is used in different ways and applied in inconsistent ways. For this discussion, the following definition will be used:

[0054] Data quality: “any data that is fit for a specific purpose and meets all requirements for its use”.

[0055] Specifically:

[0056] (1) any aspect of the data, including but not limited to accuracy, corruption or loss of data;

[0057] (2) Associating data with intended uses and ignoring issues that do not affect the intended use.

[0058] This definition is very broad and applies to almost any situation where data used in an application can adversely affect performance. It is also important to note that data may have an impact in one aspect in one application and not in another.

[0059] With the deployment of synchronized phasor data related applications, PMUs have been widely installed in the State Grid Corporation's 330kV and above voltage levels and important 220kV substations. With the popularization of measurement devices, data quality issues have become one of the most concerned issues for local power companies and regional power companies. Synchronous phasor data is estimated by the PMU and sent to the end user through synchronized communication, usually through several levels of real-time processing. Problems may occur in any process including voltage and current transformers, signal conversion equipment, PMU hardware, communication framework, phasor data concentrator (PDC), and the final application. Data errors and losses may occur at any stage.

[0060] PMU data quality issues can generally be divided into four aspects:

[0061] (1) Availability: Data loss from one or more PMUs or signal loss in the PMUs.

[0062] (2) Corruption: Data is damaged by the transmitting or receiving equipment, during communication or processing.

[0063] (3) Timely delivery: Inconsistent data rates and delays. Excessive delays lead to data loss; variable delays can cause losses through buffer overload and application processing errors.

[0064] (4) Validity and accuracy: measurement errors, such as phase errors caused by timing errors, ratio errors and proportional errors, excessive measurement noise, and repeated measurements.

[0065] Poor data quality can impact downstream applications. If application results cannot be trusted, they cannot be used.

[0066] The methods for identifying bad data can be divided into data algorithms with model-free benchmarks and model-based methods. Data algorithms with model-free benchmarks verify the data in the data stream and can create a data quality label for each measurement to indicate problems found in the data. For example, data labels can be divided into "good", "bad" or "uncertain". Good means that no problems were found in the data; bad means that the measurement was found to be wrong, making it unusable. Uncertain means that the data is likely to be wrong, but the test does not have enough conclusions to determine whether it is good or bad. A data frame usually has several measurements, including phasor amplitude and phase angle, frequency, rate of change of frequency, etc. Evaluating and labeling the quality of each measurement can maximize the use of the acquired data and simplify the data processing steps of downstream applications. The bad data identification method of the model-free benchmark contains several consecutive modules, such as Figure 4 As shown. Each module individually checks specific features of the data and records the results in a data quality label that can be included in the data flow. Downstream applications can parse it as a measure of data quality flags. The modules are independent of each other, which facilitates maintenance and troubleshooting. The parameters of all modules in the model-free step are designed to be configurable so that the algorithm can be used in any operational practice. Each module performs a check and passes the data to the next module. If the data is declared "bad" at any stage, no further inspection is required. If the data label detected by the module is "good" or "uncertain", the data inspection continues with other modules.

[0067] (1) Communication and message format module: This module processes any error indications received from the communication interface and checks message characteristics such as message format, length, and destination. If the basic message is wrong at this time, the entire data stream will be untrustworthy, so the data will be judged as "bad" and no further inspection will be performed.

[0068] (2) Time and Timing Module: This module verifies the time stamp and synchronization of the data. For example, the data rate may be inconsistent (i.e., the calculated data rate does not match the one specified in the configuration frame), the delay may be inconsistent or exceed the allowed limit, and the data may be out of order or have abnormal time stamps. Data with time problems can sometimes be corrected and marked, for example, poor time synchronization means that the phase angle measurement is inaccurate, but the phasor amplitude may be good.

[0069] (3) PMU Data Status Module: The four-bit status word that comes with the PMU data provides the status of all measurements contained in the data frame, including: 1) Invalid Data: Set to indicate that the data is bad and cannot be used; 2) PMU Error: Set to indicate internal errors, such as A / D out of calibration, memory errors, and processor errors; 3) PMU Synchronization Error: Indicates that the PMU has detected a loss of external time synchronization, such as satellite tracking loss or timing input connection failure; 4) Sort by Arrival Order: The data for a specific PMU is bad, and the PDC configures its own time stamp for the data (in this case, the measured phase angle is not available). This module applies these status indications to the quality flags created for each measurement. They have different effects depending on the measurement. For example, a synchronization error makes the phase angle unavailable but does not affect the amplitude.

[0070] (4) Data feature module: This module provides additional verification of each measurement value and mainly includes the following submodules:

[0071] 1) Range check, which checks whether a measurement is within a reasonable range. It can be applied to most measurements with reasonable known limits. For example, the frequency measurement value of the system should be within ±5Hz of the normal value, and the phase angle must be within ±π;

[0072] If M is used to represent the measured value of the measured data (electrical quantity amplitude, phase, frequency and other information), and U and L are used to represent the reasonable range determined based on theoretical analysis or historical prior experience, the data quality can be expressed as follows:

[0073]

[0074] 2) Repeated data check, check whether the measured quantity repeats the same value continuously within a certain time window T, which may occur due to problems in data calculation, data mapping or data verification procedures;

[0075] If M represents the measured value of the measured data (electrical quantity amplitude, phase, frequency and other information), D represents the threshold of the allowable measurement deviation, t represents the time when the measured data exceeds the set threshold, and T represents the duration of the sampling time window, then the data quality can be expressed as follows through repeatability inspection:

[0076]

[0077] 3) Noise Check. Random noise is always present in the data, but is expected to be much lower than the signal of interest. High noise caused by a burst higher than normal power system events may indicate equipment failure, communication errors, or other data quality issues. If the noise is passed as data, it may cause control or data analysis errors. The noise may resemble a legitimate oscillation, but can be detected and distinguished by its characteristics of having random variations and having frequencies higher than the normal PMU measurement capability. In this case, it also has a higher amplitude than other power system signals. This detection method is essentially a high-pass filter with a delay.

[0078] Noise detection is available for voltage amplitude, current amplitude, and frequency measurements. The main issue in designing a high pass filter is setting the filter cutoff frequency and the amplitude limit value. If the PMU reporting rate is set to 25 frames per second, the Nyquist frequency limit is 12Hz, and the effective measurement limit is about 7Hz. To detect noise, the cutoff frequency can be configured to, for example, 5-6Hz, which excludes the oscillation frequency range of 0-5Hz where real system oscillations usually occur. The average of the absolute value of the high pass filtered signal over time for the measured quantity is calculated and checked against the user defined limits. The threshold is set high enough so that low amplitude noise of 10Hz will not trigger, but noise of larger amplitude (>10%) will be detected. This threshold and cutoff frequency can be adjusted to more or less noisy conditions. If the limit is exceeded, the value will be considered noise and its data quality flag will be marked.

[0079] (5) Topology-based error detection module: This module considers the topology of the substation to check the rationality of the measured quantity. It calculates a value from the local power system configuration (topology) based on the measurement, and if the calculated value deviates greatly from the expected value, the measurement result is unqualified. In most cases, it is difficult to determine which specific measurement results will cause errors with this method, so all measurement results used for calculation will be marked as "uncertain".

[0080] For example, Figure 5 As shown, current detection based on Kirchhoff's current law (KCL) can be used to verify the measurement results of all currents currently entering a specific bus. In practice, because the measurement may have a small error, a small error ε, such as 10A, is used in the application to replace 0A in the KCL equation. The current measurement that passes this topology check can also be used to verify the circuit breaker status in the model-based substation state estimation step, as described later.

[0081] The wide application of merging units in smart substations provides another set of data sources, which can be used as redundant data to compare with RTU and PMU data to verify data quality problems and find the root cause of bad data.

[0082] The PT / CT output data of the digital sampling of the merging unit needs to be converted into synchronous phasor data under the same mark to obtain the corresponding voltage and current phasors, which can be directly used for comparison with PMU data or as a joint input for the substation state estimation algorithm. When performing phasor conversion, a filter is required to filter the high-frequency harmonics of the system. A small calculation delay will be added during the filtering process. The algorithm needs to use a reasonable optimized time window and ensure the data processing speed to ensure that the delay does not affect the accuracy of the phasor data. When the system is in a disturbed state caused by a fault, especially in a transient state of the system, the accuracy of the phasor conversion algorithm needs to be guaranteed so as not to cause transient errors. It is necessary to ensure that the total phasor error TVE meets the transient operation related standards of the PMU.

[0083] The method of synchronous phasor calculation is as follows:

[0084] Measure vector values ​​with real-time time stamps such as three-phase fundamental voltage, three-phase fundamental current, sequence value, switch status, etc. Send vector values ​​and disturbance record files with real-time time stamps to the data concentrator via the network at a specified rate in accordance with the IEEE std 1344-1995 (R2001) protocol.

[0085] The device synchronizes the analog and digital signal sampling pulses according to the 1PPS output of the time synchronization module, and the synchronization error of the sampling pulse is no more than 1μs. During the synchronous sampling process, the data sampling pulse is locked with the second pulse of the time synchronization module to ensure that the synchronous sampling pulses are evenly distributed within each second, and the corresponding time scales of the phasors are also evenly distributed within each second.

[0086] For the multifunctional measurement and control device of the intelligent station, the synchronization of its sampling pulse is guaranteed by the merging unit device. The receiving side of the device calculates the time stamp of the synchronous phasor according to the second division value and local time in the IEC61850-9-2 message, and directly marks its time stamp when the device receives it to ensure the accuracy of the time stamp.

[0087] The multifunctional measurement and control device supports a synchronous sampling frequency of 4kHz, and can synchronously measure the three-phase fundamental voltage, three-phase fundamental current, fundamental positive-sequence phasor of voltage and current, frequency and switching signal of the installation point, and can transmit all the primary values ​​of voltage fundamental positive-sequence phasor, primary values ​​of current fundamental positive-sequence phasor and frequency to the dynamic data concentrator in the station in real time.

[0088] IEEE C37.118 places high demands on PMU performance under dynamic conditions, and multifunctional measurement and control devices must also meet this requirement. Synchronous phasor calculation requires synchronous sampling, discrete Fourier transform calculation, pre-filtering, phasor calculation, and up-transmission filtering. For intelligent station simulation sampling and the previous links are completed by the merging unit.

[0089] After anti-aliasing filtering, the analog quantity avoids spectral aliasing caused by components other than the Nyquist frequency during A / D conversion. The merging unit introduces a time stamp during analog-to-digital conversion to ensure the accuracy of the phasor time stamp. The DFT operation result is passed through a pre-filter to attenuate the spectrum leakage component and improve the calculation accuracy of the phasor. The calculation results of the phasor, frequency and frequency change rate are then sent to the filter to limit the impact of out-of-band interference and improve accuracy.

[0090] Analysis of sampling frequency requirements

[0091] The sampling rate of the sampled data output by the merging unit in the current smart substation is 80 points / cycle. Steady-state telemetry requires that the measurement error of AC voltage and current is no more than 0.2%, and the 2nd to 13th harmonics need to be calculated. According to the sampling theorem, taking into account factors such as the attenuation of the filter network, the sampling rate is generally selected to be 48 or 64 points / cycle. Synchronous phasor measurement requires the measurement of the fundamental phasor and frequency of voltage and current, with the phasor amplitude error less than 0.2%, the phase error less than 0.5°, and the frequency measurement error no more than 0.002Hz. PMU mainly monitors the fundamental component of the system.

[0092] At present, conventional PMU measurement devices in China mainly use sampling rates such as 96 points / cycle and 200 points / cycle. The sampling rate of the digital PMU is determined by the merging unit. The PMU measurement of different amplitudes at 80 points / cycle and 200 points / cycle sampling rates is simulated. The PMU data at the two sampling rates are as follows: Figure 6 As shown in the figure, the triangular data points are at 80 sampling rates, and the circular data points are at 200 sampling rates. The simulation results show that the measurement accuracy of the fundamental phasor is basically the same at the two sampling rates. The 80-point / cycle sampling rate can already meet the PMU measurement accuracy requirements. The sampling rate requirements for electric energy calculation are basically the same as those for steady-state telemetry. Therefore, the 80-point / cycle sampling rate currently provided by the merging unit can meet the accuracy requirements of the three measurement data.

[0093] High-precision synchronous sampling is as follows:

[0094] The AC sampling of the measurement and control device in the smart substation adopts a digital and networked approach. The merging unit centrally collects the AC output of the electromagnetic or electronic transformer, and then frames it according to the IEC61850-9-2 sampling value transmission standard and sends it to the relevant devices through the network.

[0095] The merging unit receives clock synchronization and outputs sampling pulses that are precisely synchronized with the timing pulses to achieve synchronous sampling of the entire station data. To ensure the synchronization of data collected by the entire station, its sampling pulses and timing pulses are always synchronized. When the system frequency fluctuates, the interval layer device cannot achieve frequency tracking sampling of the system by adjusting the sampling frequency. In order to meet the full-cycle sampling and reduce the errors caused by spectrum leakage and fence effect, it is necessary to process the sampling values ​​sent by the merging unit. In addition, the current sampling frequency of the merging unit is 80 points / cycle. The interval layer device mostly uses Fourier algorithm for calculation, and the sampling rate is generally 32 points, 64 points, etc., which are not equal. In order not to change the mature algorithm of the original device, the received merging unit sampling values ​​must be resampled.

[0096] The time scales of steady-state and dynamic data are as follows:

[0097] The multifunctional measurement and control device also puts forward the requirement of marking the time stamp of steady-state telemetry data, so that the master station can use the telemetry data of the same time section for advanced applications of state estimation. The requirements of each functional module in the multifunctional integrated measurement and control device for data time stamp are different. The functions that require accurate data time stamp are steady-state measurement and control function and dynamic PMU function, but the telemetry data time stamp is a steady-state data time stamp, while the PMU function data time stamp is a dynamic data time stamp. The two functions have different requirements for the interval of data time stamping. If the two functions are time stamped separately, it will not only be inefficient, but also take up more cache space. In order to optimize the collected data, the unified processing of the collected data time stamp must be carried out to share the data time stamp.

[0098] The data time-scale resolution of PMU is 10ms. Steady-state telemetry is time-stamped at intervals of 200ms when the telemetry is stable. When the telemetry value changes suddenly, the time-stamp of the change should be recorded. The time-stamp accuracy requirement is 10ms. Therefore, the time-stamp of the measurement data in the multifunctional measurement and control device is unified according to the requirements of PMU and marked at intervals of 10ms. Steady-state telemetry selects two data of the PMU calculation interval to determine whether the data has changed. When the telemetry does not change, it is marked at intervals of 200ms. When the telemetry value changes, the time-stamp selects the PMU time-stamp of the change moment. This not only meets the functional requirements of PMU, but also meets the requirements of steady-state measurement function, and realizes the time-stamp sharing of collected data. Figure 7 shown.

[0099] The multifunctional measurement and control device adopts a frequency measurement algorithm that compensates for the proportional phase shift caused by spectrum leakage terms and frequency deviations. It can simultaneously meet the accuracy of steady-state measurement and the fast characteristics of dynamic measurement. It then completes the integration of steady-state measurement, synchronous phasor and electric energy metering functions through FFT, low-pass filtering and frequency domain integral calculation.

[0100] The phasor conversion algorithm generally uses the discrete Fourier transform (DFT) algorithm to calculate the phasor. The traditional DFT algorithm is widely used because it can eliminate integer harmonics under static conditions and accurately calculate the phasor of the rated frequency. However, under dynamic conditions such as system failure, power oscillation, and low-frequency oscillation, the amplitude, frequency, and phase angle of the voltage and current in the system change with time, resulting in the traditional DFT algorithm being unable to accurately calculate the dynamic phasor.

[0101] Non-steady-state conditions, such as transients, may occur in power systems. Transients can be broadly classified as electromagnetic transients or electromechanical transients. The former are caused by faults and other switching operations, while the dynamic motion of the power system due to rotor motion of generators and motors causes the latter. Phasor measurements under transient conditions are directly related to many phasor measurement applications.

[0102] Phasor Conversion Algorithms for Electromagnetic Transients: These electromagnetic transients are usually very brief and do not affect the speed of the generator rotor. When the phasor measurement data window includes the portion before and after the disturbing signal, the measured phasor is ambiguous. Consider an ideal electromagnetic transient where the input signal has a step change due to some switching operation. When the data window is completely in the early or late measurement phase, the synchronized phasors reflect the two states of the system. However, when the data window spans the transient, the phasor measurement is a combination of states that are not well defined. The phasors calculated during electromagnetic transients often show a step change in phase angle. This abrupt change in phase angle causes the frequency estimate to exhibit a sharp peak when estimating the rate of change of the system frequency. This is indeed the frequency observed from the waveform, but may not reflect any change in the machine rotor speed. This phenomenon needs to be noted when the frequency estimate is used.

[0103] Study on phasor conversion under electromechanical transients: When considering synchronized phasor measurements under electromechanical transients, we can better understand the signal model during system transients: (t) =X (t) *cos(θ (t) ), where the cosine function acts as a modulator of the input. (t) is the “amplitude” signal, and θ (t) is the "phase" signal, both embedded in the input. A particular solution of the formula under steady-state conditions is when θ (t) =2*π*f0*t+θ0x (t) =X0. During electromechanical transients, the amplitude and phase angle change slowly relative to the system rated frequency. So the phasor magnitude and speed are almost constant within a short observation window. Since the rotor may deviate by 0.1-5 Hz in modern power systems, the phase angle behavior of the synchronous speed in the phasor estimation window is roughly linear.

[0104] The synchronized phasor measurement system can directly measure the amplitude and phase of voltage and current, while the phase information in the traditional state estimation algorithm needs to be obtained through nonlinear state estimation using information such as node voltage, node injection power, and branch power. If only the measurement information of the synchronized phasor measurement system is used for state estimation, the relationship between the estimated system state and the input measurement quantity becomes linear, and the corresponding state estimator is called a "linear state estimator".

[0105] Linear state estimation has significant advantages over traditional SCADA-based state estimation. Time-synchronized fast sampling rate data streams (typically 10-50 samples per second) allow for faster state estimation and observability of system dynamics. Traditional state estimation is not able to quickly resolve the true dynamics of the power system. In addition, the linear nature of the PMU-only state estimation almost always ensures that the system has a solution. Traditional state estimation may sometimes diverge due to its nonlinear nature.

[0106] Quadratic Models for Prediction, Filtering and Smoothing of Synchrophasor Data

[0107] The voltage phasor at any node in the power system follows a quadratic trajectory in the complex plane for changing loads at a constant power factor. Determining the inverse of a certain Vandermonde matrix of a second order polynomial yields a set of coefficients that can be interpolated (or predicted) for the next data point from three previous data points. This prediction polynomial is quadratic in nature and can be used to predict the future complex voltages from previous complex voltages in the power system. The quadratic prediction model can be used in conjunction with filtering and smoothing algorithms to improve the data quality of the raw synchrophasor data. These algorithms are able to fill in missing data, reduce gross errors in measurements, and provide optimal values ​​for data points. The quadratic model algorithm can be used for pre-processing of the synchrophasor data prior to receipt by a linear state estimator.

[0108] The filtered and smoothed data output from the quadratic module is sent to the linear state estimation module for actual state estimation.The linear state estimation problem can be thought of as solving for the system state, considering each new set of measurements as a new problem relative to the previous set of measurements.

[0109] The three-phase linear state estimation of the substation includes current state estimation and zero-impedance voltage state estimation. The former uses Kirchhoff's current law (KCL) to verify the circuit breaker current, while the latter is actually a weighted average of voltage phasor measurements.

[0110] Current State Estimation: In the current linear state estimator, we redefine the state as the circuit breaker current. In this model, all nodes and circuit breakers of the same voltage level in the substation form a zero-impedance power system. Under the assumption that the phasor measurement of each circuit breaker and branch current is available, the measurement equation is established by applying KCL. For each branch current, if it is a half-breaker mode, it is a function of the two circuit breaker currents. For each circuit breaker current, it is a function relative to itself. The final estimated circuit breaker current can be directly used to verify the opening and closing state of the corresponding circuit breaker to correct possible topological errors. This will be an iterative process. After correcting the bad circuit breaker state, a new association matrix is ​​determined in the new topology. In addition, the bad data detection and identification module will collect any abnormal situation as bad data and output and process its corresponding device ID. Another major benefit is that the current state estimation is used to determine the circuit breaker state, so the topological error of the substation can be avoided, which is conducive to the accurate calculation of the system-level state estimation.

[0111] Zero impedance voltage state estimation: After obtaining the estimated open and closed state of each circuit breaker in the substation, it becomes simple and easy to construct the topology for each voltage level. The output will contain the number of buses of this voltage level and all the connected elements and branches. We can then estimate the bus voltage based on the voltage measurements of all the nodes that constitute this bus. If all the nodes are connected through zero impedance devices such as circuit breakers / switches, then all the nodes have the same voltage. This is essentially a weighted average, which is represented here as a zero impedance voltage state estimation. The states are the voltages of each bus, and the measurements are all the voltage phasor measurements at the nodes belonging to the bus. The measurement function equation is as follows: Z = Hx + r, where H is a unit vector and the solution is the weighted average of all the voltage measurements. The three-phase voltage state estimation solution is solved similarly. The solution of the weighted least squares state estimation can be easily obtained. Similarly, the bad data detection and identification module will capture any abnormal situation as bad data and output the result with the corresponding device ID to the post-processing module.

[0112] The verification method of the circuit breaker switch status is as follows:

[0113] Circuit breaker switch states and measurements usually follow different data streams. Since the circuit breaker state determines the topology of the system, it is necessary to update the model to match the system in order to make good estimates. Keeping the model synchronized with the data being used becomes a problem, especially for real-time estimation.

[0114] The real-time system topology may cause system disturbances due to planned outages, switching events, dispatch responses to load changes, and protection and relay operations. If the circuit breaker status is not reported through phasor data, it can be obtained through SCADA using communication networks and protocols such as ICCP. However, since the interval of ICCP updates may be once every 4 seconds, the system topology may have changed before a new ICCP update is received. This situation makes it possible that the linear state estimation will use an outdated and incorrect model to verify and adjust the PMU data, which will lead to differences in the calculated results.

[0115] In addition, in real-time operation, the switch status of the circuit breaker may be misreported. This requires that the current state estimation algorithm cannot blindly use the switch status of the circuit breaker, and needs to verify the received switch status of the circuit breaker before state estimation. This work can be completed by using the current values ​​of all measured data to verify the substation topology.

[0116] To address this challenge, the system design implements a solution to address real-time topology change identification caused by time deviations in circuit breaker state updates. The linear state estimation will use the synchronized phasor measurement information to check and verify the system topology. If a topology error is identified from the PMU data, the linear state estimation will use this information to change the circuit breaker state directly inside the state estimation engine to update the topology before the next ICCP update.

[0117] In this case, the PMU data used to update the topology in real time needs to be very reliable. Therefore, in the model-free data quality check module, a topology check based on current measurement can be defined. For example, according to Kirchhoff's current law (KCL), the sum of the current injected into the node (here is the bus) should be 0, such as Figure 8a Similarly, the current at both ends of the same line should be similar, as shown in Figure 8b If the PMU data passes these checks, the corresponding data will be marked as good and reliable so that the downstream linear state estimation can update the topology with real-time information.

[0118] The internal mapping of circuit breakers and related lines is in the properties of the substation linear state estimation model. In each PMU updated data frame, the circuit breaker open or closed state will be confirmed by the related line current (where available). The system will set a tolerance close to zero for each transmission line current magnitude (default is 10A). As long as the line current with good data quality meets the specified standards, the linear state estimation can internally process the corresponding line topology analysis to update the corresponding circuit breaker state to an open or closed state. The circuit breaker state change thus updated will be retained for subsequent PMU data frames until the next circuit breaker state update is received through the ICCP protocol, and then the circuit breaker state verification is performed again, and its open and closed state is updated or overwritten accordingly.

[0119] The bad data detection algorithm for substation state estimation is as follows:

[0120] A very important function of state estimation is to detect and eliminate bad measurements. In this substation linear state estimation, the commonly used chi-square test method is used to detect bad measurements. Once bad data is detected, the maximum normalized residual (LNR) method is applied to identify the most likely bad measurements. If continuous bad measurements are detected from a specific device, it may indicate a health problem or failure of the device. Through post-processing, an alarm can be issued and maintenance personnel can inspect the equipment. In other cases, the characteristics of abnormal data may be subtle and exhibit currently unknown sizes and patterns, so there is an uncertain threshold. Therefore, signal processing and anomaly detection should be carefully performed to capture currently unknown abnormal situations.

[0121] The substation model is saved in XML format and defined by the CIM model. The CIM model is widely used in the power industry for the configuration of energy management system network models and model exchange between power bureaus. For the linear state estimation of the substation, the required model information from the CIM network model XML file will be extracted and stored in the local database for easy retrieval and linking. The system needs to access and transform the substation model, and transform the model part with considerable data to establish a model for three-phase state estimation. In addition, the three-phase state estimation of the substation needs to unify each phase in a linear state estimation problem and calculate the unified array, which requires a large amount of calculation. When the substation topology changes, it is necessary to re-compose the new state estimation array and calculate it. Especially when the substation has a split busbar situation, the algorithm needs to support parallel computing and calculate the state estimation of each independent subsystem at the same time.

[0122] Based on the same inventive concept, the present invention also provides a smart substation sampling data quality comprehensive evaluation system, comprising:

[0123] Data acquisition module, used to acquire the merging unit data, measurement and control device data, and PMU data in the substation, forming multi-source redundant three-phase measurement in the substation;

[0124] The state estimation module is used to perform three-phase linear state estimation of the substation using the redundant three-phase measurements in the substation and the substation model and the real-time switch state of the circuit breaker, and use the weighted least squares algorithm to solve the state estimation problem;

[0125] The bad data identification module is used to perform a comprehensive comparison based on the state estimation results, use the Chi-square test to determine whether there is bad data in the measurement quantity, and apply the maximum normalized residual LNR method to identify the most likely bad measurement quantity, and obtain the comprehensive analysis results of data quality.

[0126] Although the content of the present invention has been described in detail through the above preferred embodiments, it should be appreciated that the above description should not be considered as a limitation of the present invention. After reading the above content, it will be apparent to those skilled in the art that various modifications and substitutions of the present invention will occur. Therefore, the protection scope of the present invention should be limited by the appended claims.

Claims

1. A comprehensive evaluation method for sampling data quality of a smart substation, characterized in that: include: Acquire the merging unit data, measurement and control device data, and PMU data in the substation to form multi-source redundant three-phase measurement under the unified time scale in the substation; The three-phase linear state estimation of the substation is performed by using the redundant three-phase measurement in the substation, the substation model and the real-time switch state of the circuit breaker, and the weighted least squares algorithm is used to solve the state estimation problem. A comprehensive comparison is performed based on the state estimation results. The Chi-square test is used to determine whether there is bad data in the measurement quantity. The maximum normalized residual (LNR) method is applied to identify the most likely bad measurement quantity, and the comprehensive analysis results of the data quality are obtained. Before performing linear state estimation, the data stream is quality checked in the following five aspects: communication and message format check, time and timing verification, PMU data state verification, data feature verification, and topology-based error detection; If the data is declared as "bad" in any aspect, no further inspection is required. If the data is declared as "good" or "uncertain", proceed with subsequent data inspection; The three-phase linear state estimation of the substation includes current state estimation and zero-impedance voltage state estimation. The former uses Kirchhoff's current law to verify the circuit breaker current, and the latter is a weighted average of voltage phasor measurements. Verify the received circuit breaker switch states before state estimation, verify the substation topology by using the current values ​​of all measured data to keep the substation model synchronized with the data being used; Linear state estimation uses the synchrophasor measurement information to check and verify the system topology. If a topology error is identified from the PMU data, this information is used to change the circuit breaker states to update the topology before the next ICCP update.

2. The method for comprehensive evaluation of sampling data quality of smart substation according to claim 1, characterized in that: The obtaining of merging unit data in the substation includes: The PT / CT output data of the digital sampling of the merging unit is read, and converted into synchronous phasor data under the same mark through the phasor conversion algorithm to obtain the corresponding voltage and current phasors, wherein the merging unit accepts clock synchronization and outputs sampling pulses that are precisely synchronized with the synchronization pulse to achieve synchronous sampling of the entire station data.

3. The method for comprehensive assessment of sampling data quality of smart substations according to claim 2, characterized in that: When performing phase conversion, filters are used to filter out the system's high-frequency harmonics.

4. The method for comprehensive assessment of sampling data quality of smart substations according to claim 3, characterized in that: Under system transient conditions, the phasor conversion algorithm adopts a reasonable optimized time window and guarantees the data processing speed, thereby ensuring that the delay does not affect the accuracy of the phasor data.

5. The method for comprehensive assessment of sampling data quality of smart substation according to claim 1, characterized in that: The substation model is in XML format and is defined by the CIM model.

6. A comprehensive evaluation system for sampling data quality of smart substations, characterized in that: include: The data acquisition module is used to acquire the merging unit data, measurement and control device data, and PMU data in the substation to form multi-source redundant three-phase measurement in the substation; The state estimation module is used to perform three-phase linear state estimation of the substation using the redundant three-phase measurements in the substation and the substation model and the real-time switch state of the circuit breaker, and use the weighted least squares algorithm to solve the state estimation problem; The bad data identification module is used to perform a comprehensive comparison based on the state estimation results, use the Chi-square test to determine whether there is bad data in the measurement quantity, and apply the maximum normalized residual LNR method to identify the most likely bad measurement quantity, and obtain the comprehensive analysis results of data quality; Before performing linear state estimation, the data stream is quality checked in the following five aspects: communication and message format check, time and timing verification, PMU data state verification, data feature verification, and topology-based error detection; If the data is declared as "bad" in any aspect, no further inspection is required. If the data is declared as "good" or "uncertain", proceed with subsequent data inspection; The three-phase linear state estimation of the substation includes current state estimation and zero-impedance voltage state estimation. The former uses Kirchhoff's current law to verify the circuit breaker current, and the latter is a weighted average of voltage phasor measurements. Verify the received circuit breaker switch states before state estimation, verify the substation topology by using the current values ​​of all measured data to keep the substation model synchronized with the data being used; Linear state estimation uses the synchrophasor measurement information to check and verify the system topology. If a topology error is identified from the PMU data, this information is used to change the circuit breaker states to update the topology before the next ICCP update.

Images