Fusion processing system and method for multi-source heterogeneous data of power system under 5g network

Abstract

The application discloses a fusion processing system and method for multi-source heterogeneous data of a power system under a 5G network, relates to the technical field of fusion processing of multi-source heterogeneous data of a power system, and comprises the following steps: collecting adjacent node phase angles, active power, voltage and time deviation on a unified time axis to generate observation paired phase angle difference and relative time deviation estimation; obtaining an electrical expected phase angle difference according to a small-angle power flow approximation, and mapping the relative time deviation to a 5G network projection phase angle according to a power frequency; performing difference on the three and normalizing the difference according to a noise equivalent standard deviation to obtain an inconsistency, complete paired network correction and generate a weight matrix; performing weighted least squares under the participation of the weight to obtain a state vector, and performing consistency review by residual error measurement to realize reliable fusion calculation under time delay fluctuation conditions. The application can separate network errors, and improve fusion precision and robustness.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system multi-source heterogeneous data fusion, and particularly relates to a power system multi-source heterogeneous data fusion processing system and method under a 5G network. BACKGROUND

[0002] With the promotion of power grid digitization and ubiquitous sensing, synchronous phasor measurement devices, dispatching and remote system measurements, distributed power supply and energy storage control, smart meters, video and acoustic inspection, environmental and safety monitoring sensors, and other multi-source data are deployed at substations and distribution sides. The fifth generation mobile communication network is used to carry station area and feeder side data backhaul, and cooperates with edge computing nodes to realize local processing and central analysis. In engineering practice, the phase angle sequence output by the synchronous phasor measurement device, the active power and voltage provided by the dispatching and remote system, and the time synchronization and communication side monitoring time and link state need to be managed on the same time axis; at the same time, the business is usually configured to carry according to different needs, high-definition video usually adopts large bandwidth carrying, protection and control related measurements prefer low latency carrying, and slow sensors go to large connection carrying. Considering that industrial and power communication protocols and time synchronization methods coexist in the field, and the processing link of edge and center cooperation, there are significant differences in sampling rate, observation window and arrival time sequence of different sources, which brings challenges to unified modeling and unified measurement for subsequent state estimation, oscillation identification and operation analysis.

[0003] In multi-source data processing based on the fifth generation mobile communication network, the cross-node phase angle difference at the same time contains the electrical phase difference determined by the power transmission law of the power grid and the phase offset converted from the time error caused by time synchronization and communication link, which are mixed in the measurement sequence. If quantitative separation and correction are not performed before entering the fusion solution, it will cause inconsistent calculation input, thereby causing state estimation bias, false alarm or missing alarm of oscillation identification, and unstable threshold setting. The existing technology relies on periodic time correction, single source pre-alignment and fixed observation window processing flow, and participates in the solution with static threshold or fixed weight; this kind of method often assumes that the link delay is stable and the measurement window is consistent, which is difficult to adapt to the delay jitter caused by dispatching and retransmission under the fifth generation mobile communication network and the window mismatch between different sources, and cannot accurately strip the phase offset introduced by the network on a per-time, per-node pair granularity, so as to ensure the reliability and consistency of the fusion result. SUMMARY

[0004] The purpose of the present application is to solve the problem of mixing electrical changes and network time errors in phase angle difference in the prior art, and to provide a power system multi-source heterogeneous data fusion processing system and method under a 5G network.

[0005] In order to solve the problems in the prior art, the present application adopts the following technical solutions:

[0006] The 5G network power system multi-source heterogeneous data fusion processing system comprises:

[0007] A collection time module is used for collecting the phase angle, active power, voltage and time synchronization deviation measurement of each adjacent node in the 5G network bearing environment, generating the observed paired phase angle difference of the node pair, and calculating the relative time deviation estimate;

[0008] An expected phase angle module is used for obtaining the electrical expected phase angle difference corresponding to the node pair according to the small-angle power flow approximation based on the active power, node voltage and line reactance;

[0009] A network mapping module is used for obtaining the 5G network projection phase angle corresponding to the node pair based on the relative time deviation estimate and the system power frequency;

[0010] An inconsistency module is used for calculating the inconsistency based on the observed paired phase angle difference, the electrical expected phase angle difference and the 5G network projection phase angle, and combining the noise equivalent standard deviation;

[0011] A correction weighting module is used for correcting the observed paired phase angle difference based on the 5G network projection phase angle to obtain the paired corrected phase angle difference, and generating the node pair weight according to the inconsistency and assembling it into a weight matrix;

[0012] A solution back-checking module is used for constructing an observation vector based on the paired corrected phase angle difference, and performing weighted least squares solution combined with the weight matrix to obtain a state vector, calculating the observation residual according to the state vector, and performing consistency back-checking on the observation residual.

[0013] Preferably, the collection of the phase angle, active power, voltage and time synchronization deviation measurement of each adjacent node, the generation of the observed paired phase angle difference of the node pair, and the calculation of the relative time deviation estimate comprise:

[0014] The phase angle of the synchronous phasor measurement device, the active power between the adjacent nodes monitored by the dispatching and remote control system, the voltage and the time synchronization deviation measurement monitored by the communication side are collected, and the collected data are subjected to time axis unification and field standardization to obtain a unified time axis record set;

[0015] In the unified time axis record set, the phase angles of two electrical nodes at the same time are subtracted and subjected to phase continuous unfolding processing to obtain the observed paired phase angle difference;

[0016] The time synchronization deviation measurement of the two electrical nodes at the same time is subjected to inter-node difference to obtain the relative time deviation estimate of the node pair.

[0017] Preferably, the electrical expected phase angle difference corresponding to the node pair is obtained according to a small-angle power flow approximation based on active power, node voltage and line reactance, including:

[0018] A ratio of the product of the line reactance of the electrical node pair at the same time and the effective value of the voltage of each node in the electrical node pair is calculated to construct a voltage-to-reactance ratio factor;

[0019] The active power between the electrical node pair and the voltage-to-reactance ratio factor are multiplied to obtain the electrical expected phase angle difference.

[0020] Preferably, the 5G network projected phase angle corresponding to the node pair is obtained based on the relative time deviation estimate and the system power frequency, including:

[0021] The unit of the relative time deviation estimate is converted into seconds;

[0022] The system power frequency and the circular constant are multiplied to obtain a time-to-phase mapping coefficient;

[0023] The 5G network projected phase angle is obtained according to the product of the relative time deviation estimate and the mapping coefficient.

[0024] Preferably, the inconsistency is calculated based on the observed paired phase angle difference, the electrical expected phase angle difference and the 5G network projected phase angle, and combined with a noise equivalent standard deviation, including:

[0025] The observed paired phase angle difference and the electrical expected phase angle difference are differentially operated to obtain a differential result, and the differential result is subtracted from the 5G network projected phase angle to obtain a phase residual;

[0026] The phase residual is normalized according to the prior noise equivalent standard deviation to obtain the inconsistency.

[0027] Preferably, the observed paired phase angle difference is corrected based on the 5G network projected phase angle to obtain a paired corrected phase angle difference, and the node pair weight is generated according to the inconsistency and assembled into a weight matrix, including:

[0028] The observed paired phase angle difference is subtracted from the 5G network projected phase angle to obtain the paired corrected phase angle difference;

[0029] The inconsistency is squared and then inverted after adding one to obtain the weight of the electrical node pair;

[0030] The weights of all node pairs are diagonalized and arranged in order of the same sampling time to obtain the weight matrix.

[0031] Preferably, an observation vector is constructed based on the paired corrected phase angle difference, and a weighted least squares solution is performed in combination with the weight matrix to obtain a state vector, the observation residual is calculated according to the state vector, and the consistency is reviewed according to the observation residual, including:

[0032] Stacking the observation vectors in the order of the measurement pairs to generate the pair-corrected phase angle difference;

[0033] Under the participation of the weight matrix, the linearization matrix generated based on the network topology relationship and the observation vector are subjected to weighted least square operation to obtain the state vector;

[0034] The linearization matrix and the state vector are subjected to matrix multiplication operation to obtain the predicted measurement;

[0035] The predicted measurement is subtracted from the observation vector to obtain the residual vector;

[0036] The residual vector is subjected to weighted two-norm operation and normalized to obtain the weighted residual measurement;

[0037] If the weighted residual measurement exceeds the statistical threshold, the state vector is marked as over-limit, otherwise the state vector is marked as not over-limit.

[0038] Compared with the prior art, the beneficial effects of the present application are:

[0039] 1、The present application synchronously generates and binds the observation pair phase angle difference, the electrical expected phase angle difference and the 5G network projected phase angle on the same time axis, obtains the phase residual by first subtracting the observation pair phase angle difference from the electrical expected phase angle difference and then deducting the 5G network projected phase angle, and obtains the inconsistency by normalizing the noise equivalent standard deviation, thereby distinguishing and quantifying the phase shift caused by the time error from time service and communication and the electrical phase change determined by the power transmission law from the data source, and avoiding the mixing caused by different sources, different windows and different arrival times.

[0040] 2、The present application corrects the observation pair phase angle difference by the 5G network projected phase angle to obtain the pair-corrected phase angle difference, maps the inconsistency to a numerical weight between zero and one by squaring, adding one and taking the reciprocal, assembles the weight matrix in the form of a diagonal matrix, makes the weighted least square robust in the presence of heteroscedasticity and occasional abnormal measurements, reduces the artificial threshold and the repeated alignment link, ensures that the influence size and the consistency degree of each measurement in solving are matched, thereby reducing the state estimation bias and reducing the false alarm and missed alarm.

[0041] 3、The application generates a linearization matrix based on network topology relationship, stacks the pair correction phase angle difference according to the measurement sequence as an observation vector, performs weighted least squares under the participation of a weight matrix to obtain a state vector, constructs a residual vector with the difference between the predicted measurement and the observation vector, calculates a weighted residual measurement and compares it with a statistical threshold to give an out-of-limit mark, forms a positive and negative mapping closed loop of the measurement space and the state space; ensures the interpretability and traceability of the solution result, provides a machine-readable quality measurement for edge and center collaborative processing, and enables the system to stably output a reliable state vector under the time delay fluctuation and arrival uncertainty brought by the fifth generation mobile communication network. BRIEF DESCRIPTION OF DRAWINGS

[0042] The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and serve to explain the principles of the application. In the drawings:

[0043] Figure 1 The functional module diagram of the fusion processing system of multi-source heterogeneous data of a power system under a 5G network is provided. DETAILED DESCRIPTION

[0044] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application.

[0045] Embodiment: The embodiment provides a fusion processing system of multi-source heterogeneous data of a power system under a 5G network, referring to Figure 1 , specifically, comprising:

[0046] The collection time module is used for collecting the phase angle, active power, voltage and time synchronization deviation measurement of each adjacent node under the 5G network bearing environment, generating the observation pair phase angle difference of the node pair, and calculating the relative time deviation estimation;

[0047] In the embodiment of the application, the phase angle, active power, voltage and time synchronization deviation measurement of each adjacent node are collected, the observation pair phase angle difference of the node pair is generated, and the relative time deviation estimation is calculated, comprising:

[0048] The phase angle of the synchronous phasor measurement device, the active power between the adjacent nodes monitored by the dispatching and remote control system, the voltage and the time synchronization and communication side monitored time synchronization deviation measurement are collected, and the collected data are time axis unified and field standardized to obtain a unified time axis record set;

[0049] In the unified time axis record set, the phase angles of two electrical nodes at the same time are subtracted and phase continuous expansion processing is performed to obtain the observation pair phase angle difference;

[0050] The relative time deviation estimation of the node pair is obtained by inter-node difference of the time deviation measurement of the two electrical nodes at the same time;

[0051] Specifically, the synchronous phasor measurement device refers to a measurement device that outputs voltage phasor, current phasor and time stamp under a unified time reference. The local clock is aligned through 5G network time service inside the device, the fundamental positive sequence phasor is estimated at a fixed sampling rate and the phase angle, amplitude, frequency and other measurement values are output, which are used to provide node phase angle data. The phase angle refers to the angle of the fundamental positive sequence phasor of the electrical quantity relative to the selected phase reference, and the unit is degree or radian. The phase angle is given with time and time stamp, which is used to difference the phase angles of two nodes at the same time to form the observed paired phase angle difference. The dispatching and remote control system refers to a monitoring and data acquisition system composed of dispatching master station, front-end and station-side remote control equipment such as remote terminal and intelligent electronic device, which provides active power and voltage measurement values between adjacent nodes. Adjacent nodes refer to two network nodes directly connected by a branch in the power network topology, also known as two bus nodes, and the adjacent relationship is used to determine the calculation object of the paired quantity and the selection of line parameters. Time synchronization and communication side monitoring refers to the process and device used for monitoring the time quality related to time service and communication link, and the monitoring data typically includes 5G network time service deviation, round-trip delay, time stamp quality mark, time service source state and other data, and the output is used to quantify the time consistency level between devices. The time deviation measurement refers to the instantaneous time difference between the local clock of the device and the reference clock, or the relative time difference between two devices.

[0052] Specifically, the phase angle of the synchronous phasor measurement device, the active power and voltage of the dispatching and remote control system, and the time deviation measurement of the time synchronization and communication side monitoring are unified and field standardized on the same time axis, the purpose is to ensure that all quantities can be directly compared at the same time and have consistent dimensions; in the small angle working condition, the active power of the line and the phase angle difference of the two nodes are approximately proportional in a linear manner, and the proportional factor is equal to the ratio of the product of the effective values of the voltages of the two nodes to the line reactance between the two nodes, therefore, the phase angle difference of the two electrical nodes at the same time needs to be differentiated and the phase continuous expansion processing needs to be performed to eliminate the whole cycle jump, so that the observed paired phase angle difference which can participate in the power relationship calculation and subsequent comparison can be obtained; on the other hand, the time error caused by 5G network time service and communication will be converted into phase shift at a fixed ratio, and the ratio is equal to the product of twice the circular constant and the power frequency, so the relative time deviation estimation of the node pair must be obtained by inter-node difference of the time deviation measurement of the two electrical nodes at the same time to obtain the real relative time deviation estimation acting on the node pair, so as to distinguish the phase shift introduced by 5G network time service and link delay from the phase angle change caused by electrical state change at the data level, and provide a directly calculable basis for subsequent correction and weighting.

[0053] Specifically, the phase angle data of each electrical node output by the synchronous phasor measurement device is collected, the active power data between adjacent electrical nodes monitored by the dispatching and remote control system and the voltage data of each electrical node are collected, and the time synchronization and the time deviation measurement data of each electrical node obtained by the communication side monitoring equipment are collected; the time stamp information of each data source data is extracted with reference to the preset high-precision time reference in the power system, the sampling time of different data source data is aligned through the time stamp calibration algorithm, the time deviation caused by the sampling period difference of each device and the transmission time delay difference is eliminated, and all data is corresponded to the same time scale; according to the general specification of power industry data collection and processing, the field format of the collected phase angle, active power, voltage and time deviation measurement data is uniformly regularized, the phase angle is in radian unit, the active power is in megawatt unit, the voltage is in kilovolt unit, and the time deviation is in microsecond unit, so that the unit of the same type of data is unified, the data precision retention bit number is consistent, and the field identification is standardized; all data after the time axis unification and field standardization processing is integrated in time sequence to form a unified time axis record set containing the phase angle of each electrical node, the active power of adjacent nodes, the node voltage and the time deviation measurement data, and each data corresponds to a unique unified time stamp.

[0054] Specifically, the phase angle data of each electrical node output by the synchronous phasor measurement device is collected, the active power data between adjacent electrical nodes monitored by the dispatching and remote control system and the voltage data of each electrical node are collected, and the time synchronization and the time deviation measurement data of each electrical node obtained by the communication side monitoring equipment are collected; the time stamp information of each data source data is extracted with reference to the preset high-precision time reference in the power system, the sampling time of different data source data is aligned through the time stamp calibration algorithm, the time deviation caused by the sampling period difference of each device and the transmission time delay difference is eliminated, and all data is corresponded to the same time scale; according to the general specification of power industry data collection and processing, the field format of the collected phase angle, active power, voltage and time deviation measurement data is uniformly regularized, the phase angle is in radian unit, the active power is in megawatt unit, the voltage is in kilovolt unit, and the time deviation is in microsecond unit, so that the unit of the same type of data is unified, the data precision retention bit number is consistent, and the field identification is standardized; all data after the time axis unification and field standardization processing is integrated in time sequence to form a unified time axis record set containing the phase angle of each electrical node, the active power of adjacent nodes, the node voltage and the time deviation measurement data, and each data corresponds to a unique unified time stamp.

[0055] The expected phase angle module is used to obtain the electrical expected phase angle difference of the node pair according to the small-angle power flow approximation based on the active power, the node voltage and the line reactance;

[0056] In the embodiment of the application, the electrical expected phase angle difference of the node pair is obtained according to the small-angle power flow approximation based on the active power, the node voltage and the line reactance, which comprises:

[0057] The ratio of the line reactance of the electrical node pair to the product of the voltage effective values of the nodes in the electrical node pair at the same time is calculated to construct the voltage-to-reactance ratio factor;

[0058] The active power between the electrical node pair is multiplied by the voltage-to-reactance ratio factor to obtain the electrical expected phase angle difference;

[0059] Specifically, under the working condition of sinusoidal steady state and small phase angle difference, the classic power flow relationship shows that the active power between two electrical nodes is proportional to the product of the voltage effective values of the two nodes and inversely proportional to the line reactance between the two nodes, and proportional to the phase angle difference between the two nodes; accordingly, the relationship is inversely expressed, that is, the phase angle difference of the two nodes is equal to the product of the line reactance and the product of the voltage effective values of the two nodes multiplied by the active power between the two nodes, so the ratio of the line reactance of the electrical node pair to the product of the voltage effective values of the nodes in the electrical node pair at the same time is calculated to construct the ratio factor for converting the active power into the phase angle difference, and the active power between the electrical node pair is multiplied by the ratio factor to obtain the electrical expected phase angle difference, which reflects the linear relationship between the line transmission power and the phase difference under the condition of sinusoidal steady state and small phase angle difference, and is used as the electrical reference quantity for subsequent difference, correction and weighting.

[0060] Specifically, the line reactance parameter of the electrical node pair at the same time is extracted, and the voltage effective values of the two nodes in the electrical node pair are obtained, the product of the voltage effective values of the two nodes is calculated, and the line reactance is ratioed with the product, that is, the line reactance is the dividend, and the product of the voltage effective values of the two nodes is the divisor, and the result obtained through the operation is the voltage-to-reactance ratio factor; then the active power data between the electrical node pair is extracted, and the active power is multiplied by the voltage-to-reactance ratio factor obtained above, and the operation result is the electrical expected phase angle difference of the electrical node pair at the same time.

[0061] The network mapping module is used to obtain the 5G network projected phase angle corresponding to the node pair based on the relative time deviation estimation and the system power frequency;

[0062] In the embodiment of the application, the 5G network projected phase angle corresponding to the node pair is obtained based on the relative time deviation estimation and the system power frequency, which comprises:

[0063] The unit of the relative time deviation estimation is converted into seconds;

[0064] The system power frequency is multiplied by the circular constant to obtain a time-to-phase mapping coefficient;

[0065] The 5G network projected phase angle is obtained according to the product of the relative time deviation estimation and the mapping coefficient;

[0066] Specifically, the voltage and current of the power system can be regarded as a sinusoidal signal oscillating at the system power frequency, and the angular frequency is equal to the product of two times the constant and the system power frequency. Any time deviation will be converted into an equivalent phase offset according to the product of the angular frequency and the time deviation. Therefore, first, the relative time deviation estimate is converted into seconds to match the unit of the angular frequency and eliminate errors caused by inconsistent dimensions; then, the system power frequency and two times the constant are multiplied to obtain the mapping coefficient from time to phase, which represents the phase increment corresponding to one second; finally, the relative time deviation estimate is multiplied by the mapping coefficient to obtain the 5G network projected phase angle, which is used to represent the deterministic offset of the time error introduced by the communication and timing link in the phase domain, thereby providing a directly callable phase quantity for comparison and subsequent correction of the observed paired phase angle difference and the electrical expected phase angle difference.

[0067] Specifically, first, the original unit of the relative time deviation estimate is determined, and the value of the relative time deviation estimate is converted into a relative time deviation estimate in seconds; the system power frequency of the power system is obtained, and the system power frequency is multiplied by two, and then the obtained result is multiplied by the constant, to obtain the mapping coefficient from time to phase; the relative time deviation estimate converted into seconds is multiplied by the mapping coefficient from time to phase obtained above, and the numerical accuracy is maintained during the multiplication operation to avoid the accumulation of calculation errors. The result of the multiplication operation is the 5G network projected phase angle of the electrical node at the corresponding time.

[0068] The inconsistency quantity module is configured to calculate the inconsistency quantity based on the observed paired phase angle difference, the electrical expected phase angle difference, and the 5G network projected phase angle, and in combination with the noise equivalent standard deviation.

[0069] In the embodiments of the present application, the inconsistency quantity is calculated based on the observed paired phase angle difference, the electrical expected phase angle difference, and the 5G network projected phase angle, and in combination with the noise equivalent standard deviation, including:

[0070] The observed paired phase angle difference and the electrical expected phase angle difference are subjected to difference operation to obtain a difference result, and the difference result is subtracted by the 5G network projected phase angle to obtain a phase residual.

[0071] The phase residual is normalized according to the prior noise equivalent standard deviation to obtain the inconsistency quantity.

[0072] Specifically, the observed pair-wise phase angle difference can be seen as superimposed by three parts, i.e. the electrical phase difference determined by the power transmission law of the line, the phase offset converted from the time error of the time service and the communication link and the measurement noise and modeling deviation. According to the small-angle power flow approximation, the electrical expected phase angle difference calculated from the active power and voltage at the same time and the line reactance represents the theoretical value of the electrical component, so the observed pair-wise phase angle difference is first subtracted by the electrical expected phase angle difference to eliminate the main electrical deterministic term; then the 5G network projected phase angle estimated by the relative time deviation using the time-to-phase mapping coefficient is applied to the above difference result, so that the phase offset caused by the link time service can be stripped from it, and the remaining amount is the phase residual, which represents the unmodeled disturbance and random error. Since the noise levels of different measuring points and devices are different, directly comparing the phase residual will lead to distorted evaluation, so the phase residual is normalized by using the prior noise equivalent standard deviation to convert it into a dimensionless inconsistency, so as to realize the same scale comparison between different nodes and different time, and provide stable and interpretable input for subsequent weight generation and weighted solution.

[0073] Specifically, the observed pair-wise phase angle difference and the electrical expected phase angle difference of the electrical node pair at the same time are extracted, the observed pair-wise phase angle difference is subtracted by the electrical expected phase angle difference, the accuracy of the phase angle data is maintained during the operation process to avoid deviation caused by numerical truncation or rounding, and the corresponding difference result is obtained through the difference operation; the 5G network projected phase angle of the electrical node pair at the same time is extracted, and the difference result obtained in the foregoing is subtracted by the 5G network projected phase angle, and the value obtained in this way is the phase residual. The noise equivalent standard deviation is obtained, the noise equivalent standard deviation is determined by the basic value through the preliminary laboratory calibration test, and is dynamically modified by combining the online statistical analysis in the actual operation process of the system, and the modified value is stored in the parameter configuration module of the system in advance; the absolute value of the obtained phase residual is taken to eliminate the influence of the positive and negative signs on the subsequent calculation, and then the phase residual after taking the absolute value is divided by the prior noise equivalent standard deviation, to complete the normalization processing of the phase residual, and the result obtained after the processing is the inconsistency used to represent the contribution ratio of the electrical intrinsic change and the network time deviation.

[0074] The correction weighting module is used for correcting the observed pair-wise phase angle difference based on the 5G network projected phase angle to obtain a pair-wise corrected phase angle difference, and generating a node pair weight according to the inconsistency and assembling the node pair weight into a weight matrix.

[0075] In the embodiments of the present application, the observed pair-wise phase angle difference is corrected based on the 5G network projected phase angle to obtain a pair-wise corrected phase angle difference, and a node pair weight is generated according to the inconsistency and assembled into a weight matrix, which comprises:

[0076] The observed pair-wise phase angle difference is subtracted by the 5G network projected phase angle to obtain a pair-wise corrected phase angle difference.

[0077] The weight of the electrical node pair is obtained by squaring the inconsistent amount and adding one and then taking the reciprocal;

[0078] The weights of all node pairs are arranged in diagonalization according to the order of the same sampling time to obtain the weight matrix;

[0079] Specifically, the observed pair phase angle difference is subtracted from the 5G network projected phase angle based on the deterministic mapping relationship between the time error and the phase offset. This operation can separate the phase offset introduced by the time service and the communication link from the observation, thereby obtaining the pair correction phase angle difference reflecting only the electrical state. Then, the weight is constructed according to the robustness of the weighted least squares. The square operation ensures that the inconsistent degree is non-negative and the increase of the deviation is accelerated. Adding one ensures that the weight is close to one when the inconsistent degree is close to zero without divergence. Taking the reciprocal ensures that the larger the inconsistent degree is, the smaller the weight is to suppress the influence of abnormal observations on the solution. This form is equivalent to the influence function corresponding to the commonly used Cauchy type cost function, which can significantly reduce the influence of large deviations while ensuring high fidelity of small deviations. Finally, the weights of all node pairs are arranged in diagonalization according to the order of the same sampling time to obtain the weight matrix, which is to correspond to the observation vector stacked in the same order and meet the modeling assumption that the errors of each measurement pair are independent of each other. This makes the weighted least squares meet the requirements of Gauss Markov theorem and maintain the sparse structure and calculation stability in numerical implementation, thereby providing input that meets physical constraints and has good numerical properties for subsequent solution.

[0080] Specifically, the observed pair phase angle difference and the 5G network projected phase angle of the same electrical node pair at the same sampling time are extracted first to ensure that the two data are completely matched in time dimension and node pair dimension, avoiding correction deviation caused by data mismatch. The observed pair phase angle difference is used as the minuend and the 5G network projected phase angle is used as the subtrahend for subtraction operation. The original precision of the phase angle data is preserved during the operation without unnecessary numerical rounding or truncation. The result obtained by this operation is the pair correction phase angle difference after eliminating the influence of network time deviation. The inconsistent amount of the electrical node pair at the same sampling time is extracted. The inconsistent amount is squared first to amplify the numerical difference of the inconsistent amount and strengthen the weight discrimination. Then, the result of the square operation is added by one to avoid the case that the denominator is zero when taking the reciprocal. Finally, the reciprocal operation is performed on the result after addition to obtain the weight of the electrical node pair at the current sampling time, which is sorted according to the order of the same sampling time. The diagonal matrix is constructed according to the sorted weight, so that each weight corresponds to an element on the diagonal line of the matrix, and the elements on the non-diagonal line of the matrix are all set to zero. The matrix obtained by this diagonalization arrangement is the weight matrix used for subsequent weighted fusion calculation.

[0081] The solving back-checking module is configured to construct an observation vector based on the pair-wise corrected phase angle difference, perform weighted least square solving on the observation vector in combination with a weight matrix to obtain a state vector, calculate an observation residual based on the state vector, and perform consistency back-checking on the observation residual.

[0082] In the embodiment of the application, the observation vector is constructed based on the pair-wise corrected phase angle difference, the state vector is obtained by performing weighted least square solving on the observation vector in combination with the weight matrix, the observation residual is calculated based on the state vector, and the consistency back-checking is performed on the observation residual, including:

[0083] The pair-wise corrected phase angle difference is stacked in the order of the measurement pairs to generate the observation vector;

[0084] The linearization matrix generated based on the network topology relationship and the observation vector are subjected to weighted least square operation under the participation of the weight matrix to obtain the state vector;

[0085] The linearization matrix and the state vector are subjected to matrix multiplication operation to obtain the predicted measurement;

[0086] Specifically, the pair-wise corrected phase angle difference is stacked in the order of the measurement pairs to generate the observation vector, so as to establish a one-to-one correspondence relationship with the diagonal elements of the weight matrix constructed in the same order, so that each measurement has a fixed position and a determined weight in the solving process, thereby avoiding calculation errors caused by sequence mismatch; the linearization matrix generated based on the network topology relationship is used to depict the first-order linear mapping between each measurement pair and the phase angles of the corresponding two nodes, which is derived from the linearization of the relationship between the phase angle difference and the line transmission power under the condition of small angle and the association rule of the power grid nodes and branches; the linearization matrix and the observation vector are subjected to weighted least square operation under the participation of the weight matrix, so that the minimum variance unbiased estimation satisfying the Gauss-Markov theorem can be obtained under the condition that the variances of the measurement noises are not equal, thereby obtaining the state vector, so that the high-confidence measurement obtains greater influence and the disturbance of abnormal quantities on the solution is suppressed; subsequently, the linearization matrix and the state vector are subjected to matrix multiplication to obtain the predicted measurement, which is used to re-map the node phase angles obtained by solving back to the measurement space, so as to verify the consistency between the solution and the observation and provide a unified benchmark for residual calculation and consistency back-checking; the above processing realizes the forward and reverse mapping closed loop between the measurement space and the state space, maintains the consistency of the dimension and key value under the same time and node pair identification, conforms to the power grid topology and the power flow linearization rule, and guarantees the repeatability and numerical stability of the subsequent evaluation and control process.

[0087] Specifically, first, the pair-wise correction phase angle difference of all electrical nodes in the power system at the same sampling time is extracted, and the measurement order is sorted to ensure the accuracy of data correlation; the sorted pair-wise correction phase angle difference is arranged in sequence to form a single column vector structure, and the original numerical precision of each pair-wise correction phase angle difference is preserved without additional numerical truncation or approximation during the arrangement process. The observation vector for subsequent fusion calculation is generated through this stacking method. The linearization matrix generated based on the power system network topology relationship is obtained by analyzing the grid node connection relationship, line reactance parameters and measurement equation, and linearizing near the system operating point, and is updated in real time with the change of the grid topology structure; first, the transpose matrix of the linearization matrix is calculated, and the transpose matrix and the weight matrix are multiplied to obtain the first intermediate matrix; then the first intermediate matrix and the original linearization matrix are multiplied to obtain the second intermediate matrix; the second intermediate matrix is inverted, and if the second intermediate matrix is singular, a small perturbation term is added to ensure the existence of the inverse matrix, and the inverse matrix of the second intermediate matrix is obtained; then the product of the transpose matrix of the linearization matrix and the weight matrix, i.e. the first intermediate matrix, is calculated, and the product and the observation vector are multiplied to obtain the third intermediate matrix; finally, the inverse matrix of the second intermediate matrix and the third intermediate matrix are multiplied, strictly following the dimension matching rules of matrix operation to ensure the legality of each multiplication operation, and the state vector containing the core state information of the grid phase angle, frequency, etc. is obtained through the complete weighted least squares operation. Then, the dimension of the linearization matrix and the state vector is checked to confirm that the number of columns of the linearization matrix and the number of rows of the state vector are completely consistent, avoiding calculation errors caused by dimension mismatch; after meeting the dimension matching condition, the matrix multiplication operation of the linearization matrix and the state vector is performed, and the elements are calculated and accumulated during the operation process, and enough decimal places are preserved to reduce numerical errors. The predicted measurement with the same dimension as the observation vector is obtained through the matrix multiplication operation, which is used for subsequent comparison with the observation vector to verify the consistency of the fusion result.

[0088] The predicted measurement is subtracted from the observation vector to obtain a residual vector;

[0089] The weighted two-norm operation is performed on the residual vector and normalized to obtain a weighted residual measure;

[0090] If the weighted residual measure exceeds the statistical threshold, the state vector is marked as out-of-limit, otherwise the state vector is marked as not out-of-limit;

[0091] Specifically, the predicted measurement is subtracted from the observation vector to obtain a residual vector, because after the linearization matrix maps the node phase angle to the measurement space, the difference between the two is the component not explained by the model, including the remaining phase offset introduced by time service and communication, device measurement noise and unmodeled disturbances. The weighted two-norm of the residual vector is normalized according to the measurement dimension, which is based on the Gaussian noise assumption and the Gaussian Markov theorem to construct a criterion equivalent to the maximum likelihood cost. The weight matrix is derived from the numerical weight based on the inconsistency quantity in the previous sequence, which can approximately correspond to the inverse of the error variance, so that high-confidence measurements contribute more and abnormal measurements are suppressed. The weighted residual measurement obtained is compared with the statistical threshold set according to the chi-square distribution quantile. When the measurement exceeds the threshold, it is determined that the current model and data are incompatible, and the state vector is marked as over-limit. When the measurement does not exceed the threshold, it is determined that the solution result is consistent with the observation, and the state vector is marked as not over-limit. The process integrates the linear relationship between power and phase angle, the mapping of time error to phase offset, and the weighting mechanism based on inconsistency into a closed loop, thereby providing reproducible and statistically reliable result determination in a unified measurement system.

[0092] Specifically, the predicted measurement is subtracted from the observation vector to obtain a residual vector, because after the linearization matrix maps the node phase angle to the measurement space, the difference between the two is the component not explained by the model, including the remaining phase offset introduced by time service and communication, device measurement noise and unmodeled disturbances. The weighted two-norm of the residual vector is normalized according to the measurement dimension, which is based on the Gaussian noise assumption and the Gaussian Markov theorem to construct a criterion equivalent to the maximum likelihood cost. The weight matrix is derived from the numerical weight based on the inconsistency quantity in the previous sequence, which can approximately correspond to the inverse of the error variance, so that high-confidence measurements contribute more and abnormal measurements are suppressed. The weighted residual measurement obtained is compared with the statistical threshold set according to the chi-square distribution quantile. When the measurement exceeds the threshold, it is determined that the current model and data are incompatible, and the state vector is marked as over-limit. When the measurement does not exceed the threshold, it is determined that the solution result is consistent with the observation, and the state vector is marked as not over-limit. The process integrates the linear relationship between power and phase angle, the mapping of time error to phase offset, and the weighting mechanism based on inconsistency into a closed loop, thereby providing reproducible and statistically reliable result determination in a unified measurement system.

[0093] In this embodiment, the fusion processing method of the multi-source heterogeneous data of the power system under the 5G network is as follows:

[0094] S1, collecting the phase angle, active power, voltage and time deviation measurement of each adjacent node in the 5G network bearing environment, generating the observed paired phase angle difference of the node pair, and calculating the relative time deviation estimation;

[0095] S2, based on the active power, node voltage and line reactance, the electrical expected phase angle difference corresponding to the node pair is obtained according to the small angle power flow approximation;

[0096] S3, based on the relative time deviation estimation and the system power frequency, the 5G network projection phase angle corresponding to the node pair is obtained;

[0097] S4, based on the observed paired phase angle difference, the electrical expected phase angle difference and the 5G network projection phase angle, and combined with the noise equivalent standard deviation, the inconsistency is calculated;

[0098] S5, based on the 5G network projection phase angle, the observed paired phase angle difference is corrected to obtain the paired corrected phase angle difference, and the node pair weight is generated according to the inconsistency and assembled into a weight matrix;

[0099] S6, based on the paired corrected phase angle difference, the observation vector is constructed, and the weighted least square solution is executed combined with the weight matrix to obtain the state vector, the observation residual is calculated according to the state vector, and the consistency of the observation residual is checked.

[0100] The above is only the preferred specific embodiment of the present application, but the protection scope of the present application is not limited to this, any skilled person in the art can make equivalent replacement or change according to the technical solution and the inventive concept of the present application within the technical range disclosed by the present application, which should be covered in the protection scope of the present application.

Claims

A fusion processing system for multi-source heterogeneous data in a power system under a 1.5G network, characterized in that: include: The time synchronization module is used to collect the phase angle, active power, voltage and time deviation measurements of adjacent nodes in the 5G network environment, generate the observed pair phase angle difference of node pairs, and calculate the relative time deviation estimate. The desired phase angle module is used to approximate the electrical desired phase angle difference of a node based on active power, node voltage, and line reactance, according to a small-angle power flow. The network mapping module is used to obtain the 5G network projection phase angle of the node pair based on the relative time deviation estimation and the system power frequency; The inconsistency module is used to calculate the inconsistency based on the observed pairwise phase angle difference, the electrical expected phase angle difference, and the 5G network projected phase angle, combined with the noise equivalent standard deviation. Specifically, the observed pairwise phase angle difference and the electrical expected phase angle difference are differentially calculated to obtain the difference result. The 5G network projected phase angle is then subtracted from the difference result to obtain the phase residual. The phase residual is then normalized according to the prior noise equivalent standard deviation to obtain the inconsistency. The correction weighting module is used to correct the observed pairwise phase angle difference based on the 5G network projection phase angle, obtain the pairwise corrected phase angle difference, and generate node pair weights based on the inconsistency and assemble them into a weight matrix. The backtesting module is used to construct observation vectors based on paired corrected phase angle differences, and perform weighted least squares solution to obtain state vectors in combination with weight matrix. The observation residuals are calculated based on the state vectors, and the consistency of the observation residuals is checked.

2. The fusion processing system for multi-source heterogeneous data in a power system under a 5G network according to claim 1, characterized in that, The phase angle, active power, voltage, and time deviation measurements of adjacent nodes are collected to generate the observed pairwise phase angle difference between node pairs and to calculate the relative time deviation estimate, including: The phase angle of the synchronous phasor measurement device, the active power and voltage between adjacent nodes monitored by the scheduling and telemetry system, and the time deviation measurement monitored by the time synchronization and communication side are collected. The collected data are then unified in time axis and standardized in fields to obtain a unified time axis record set. In a unified timeline record set, the phase angles of two electrical nodes at the same moment are subtracted and the phase is continuously expanded to obtain the observed pairwise phase angle differences; By performing inter-node differential measurements on the time deviation measurements of two electrical nodes at the same time, the relative time deviation estimate of the node pair is obtained.

3. The fusion processing system for multi-source heterogeneous data in a power system under a 5G network according to claim 2, characterized in that, Based on active power, node voltage, and line reactance, the electrical expected phase angle difference corresponding to the node is obtained according to the small-angle power flow approximation, including: Calculate the ratio of the line reactance of an electrical node pair at the same moment to the product of the effective values ​​of the voltages of each node in the electrical node pair, in order to construct the voltage-reactance scaling factor; The desired electrical phase angle difference is obtained by multiplying the active power and voltage between electrical node pairs by the proportionality factor of the reactance.

4. The fusion processing system for multi-source heterogeneous data in a power system under a 5G network according to claim 2, characterized in that, Based on relative time deviation estimation and system power frequency, the 5G network projection phase angles corresponding to the node pairs are obtained, including: Convert the relative time deviation estimate to seconds; The time-to-phase mapping coefficient is obtained by multiplying the system power frequency and the circumferential constant. The projected phase angle of the 5G network is obtained by multiplying the relative time deviation estimate with the mapping coefficient.

5. The fusion processing system for multi-source heterogeneous data in a power system under a 5G network according to claim 1, characterized in that, Based on the 5G network projection phase angle, the observed pairwise phase angle difference is corrected to obtain the pairwise corrected phase angle difference. Then, node pair weights are generated according to the inconsistency and assembled into a weight matrix, including: Subtract the 5G network projected phase angle from the observed pairwise phase angle difference to obtain the pairwise corrected phase angle difference; The weight of the electrical node pair is obtained by squaring the inconsistency, adding one, and taking the reciprocal. Diagonalize the weights of all node pairs according to the order of the same sampling time to obtain the weight matrix.

6. The fusion processing system for multi-source heterogeneous data in a power system under a 5G network according to claim 5, characterized in that, An observation vector is constructed based on paired-correction phase differences, and a weighted least squares solution is performed using the weight matrix to obtain the state vector. The observation residuals are then calculated based on the state vectors, and a consistency check is performed on the observation residuals, including: The paired phase angle differences are stacked in the order of measurement pairs to generate observation vectors; With the participation of the weight matrix, a weighted least squares operation is performed on the linearized matrix generated based on the network topology relationship and the observation vector to obtain the state vector; Matrix multiplication is performed on the linearized matrix and the state vector to obtain the predicted measurement; Subtract the observation vector from the predicted measurement to obtain the residual vector; The weighted residual measure is obtained by performing a weighted L2 norm operation on the residual vector and then normalizing it. If the weighted residual metric exceeds the statistical threshold, the state vector is marked as out of bounds; otherwise, the state vector is marked as not out of bounds. A method for fusing and processing multi-source heterogeneous data of a power system under a 7.5G network, applied to the fusion and processing system for multi-source heterogeneous data of a power system under a 5G network as described in any one of claims 1-6, the method comprising: S1. Under the 5G network environment, collect the phase angle, active power, voltage and time deviation measurements of adjacent nodes, generate the observed pair phase angle difference of node pairs, and calculate the relative time deviation estimate. S2. Based on active power, node voltage, and line reactance, the electrical expected phase angle difference corresponding to the node is obtained by approximating the small-angle power flow. S3. Based on the relative time deviation estimation and the system power frequency, the 5G network projection phase angle corresponding to the node pair is obtained; S4. Based on the observed pairwise phase angle difference, the electrical expected phase angle difference, and the 5G network projected phase angle, and combined with the noise equivalent standard deviation, calculate the inconsistency quantity; S5. Based on the 5G network projection phase angle, the observed pair phase angle difference is corrected to obtain the pair corrected phase angle difference, and node pair weights are generated according to the inconsistency and assembled into a weight matrix. S6. Construct observation vectors based on paired correction phase angle differences, and perform weighted least squares solution to obtain state vectors in combination with weight matrix. Calculate observation residuals based on state vectors, and perform consistency back-check on observation residuals.

Images