A method for observable and measurable grid sensing for distributed power sources
By collecting and processing electrical parameter data in low-voltage distribution networks, a voltage-power differential sensitivity matrix is constructed, which solves the interference of static impedance parameter distortion on calculation, realizes accurate perception and reliable scheduling of distribution network status after distributed power source access, and improves power supply security.
Patent Information
- Application Number
- CN202611578068.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-13
- Publication Date
- 2026-11-17
AI Technical Summary
After distributed generation is connected to the low-voltage distribution network, under the existing measurement configuration and asynchronous communication conditions, the distortion of static impedance parameters affects the calculation convergence, making it difficult to accurately obtain the voltage profile and power flow distribution of the entire network, resulting in insufficient reliability of power supply security and active dispatch of the distribution network.
By collecting three-phase reference electrical parameters at the low-voltage side outlet bus of the distribution transformer and non-equal interval measurement data of the distributed power grid-connected inverter, a synchronous time slot measurement sequence is generated using the time scale difference projection operator. A voltage-power differential sensitivity matrix is constructed, a set of physical equilibrium equations is established, and the net active load and net reactive load of the unmeasured nodes are solved. The effective value profile of the phase voltage and the power flow distribution of the feeder are reconstructed by combining the forward power flow model, and a situation warning data package is generated.
It reduces the interference of static impedance parameter distortion on calculation, improves the stability and accuracy of state-aware calculation, ensures the reliability of distributed power source operation supervision and power supply safety, and reduces the power absorption loss and regulation lag caused by reverse power flow.
Smart Images

Figure CN123122003A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of smart grid data processing and power supply operation management technology, and relates to a grid sensing method for observable and measurable distributed power sources. Background Technology
[0002] Low-voltage distribution networks are responsible for transmitting electrical energy to end users. Traditionally, their operation is mainly based on unidirectional power flow and load absorption. Existing distribution monitoring systems typically rely on measurement and control terminals located at the head end of distribution transformers, and are based on the prior knowledge of the entire network topology and line impedance parameters. With the large-scale integration of distributed photovoltaic and energy storage devices in low-voltage distribution areas, the original radial distribution network is gradually transforming into an active distribution network with multiple power sources. The output of distributed power sources is affected by environmental conditions and has significant random fluctuations. When peak power generation and load troughs occur simultaneously, reverse power backfeeding is likely to occur, causing the voltage at the feeder end to rise and exceed limits, and may also cause heavy overload of distribution transformers. Mastering the real-time operating status of all network nodes has become an important foundation for the distribution network management system to implement proactive control.
[0003] Low-voltage distribution lines generally have characteristics such as high resistance-to-inductance ratio, three-phase asymmetry, and conductor impedance variation with temperature. The static line impedance parameters determined offline are prone to continuous deviations from the actual operating state. Improving measurement coverage by densely installing miniature synchronous phasor measurement devices across the entire network is not only costly but also suffers from limited communication capacity of low-voltage carrier waves and narrowband communication channels, making it difficult to handle large amounts of simultaneous measurement data uploads over extended periods. Furthermore, existing state estimation algorithms suffer from computational and convergence defects in data processing and algorithm analysis. For example, Chinese invention patent application CN117477546A discloses a method based on distributed power source pseudo-measurement... The state estimation method based on data weight optimization constructs a pseudo-measurement weight optimization model and combines iterative solutions of node voltage magnitude and phase angle using the Newton-Raphson method. However, this type of method is highly dependent on pre-set offline static admittance parameters and essentially relies on the nonlinear iterative mechanism of inverting the Jacobian matrix. Under complex conditions such as high impedance ratio in low-voltage distribution networks, impedance dynamic drift caused by conductor temperature rise, and asynchronous uploading of distributed power source measurements, it is easy to induce ill-conditioned Jacobian matrix and even lead to divergence in power flow iteration. In addition, it is difficult to effectively decouple the background load without measurement nodes and the fluctuating output of distributed power sources under underdetermined conditions by weight optimization alone, resulting in a lack of rigid physical constraints in reverse power flow deduction.
[0004] Therefore, the technical problem to be solved by this invention is to reduce the impact of static impedance parameter distortion on computational convergence, reduce the interference caused by the superposition of loads at unmeasured nodes and distributed power generation outputs, accurately obtain the voltage profile and feeder power flow distribution of the entire network, and ensure the power supply safety and reliability of active dispatching of the distribution network while maintaining the existing measurement configuration and asynchronous communication conditions. Summary of the Invention
[0005] To address the problems in the background art, the present invention provides the following technical solution: a method for observable and measurable grid sensing for distributed power sources, comprising the following steps: Step S1: Collect the three-phase reference electrical parameter time series at the low-voltage side outlet bus of the distribution transformer, as well as the non-equal interval measurement data packets of the distributed power grid-connected inverter; using the clock of the three-phase reference electrical parameter time series as a reference, resample the non-equal interval measurement data packets using the time scale difference projection operator to generate a synchronous time slot measurement sequence. Step S2: Extract the voltage disturbance of the grid-connected node within adjacent discrete time steps based on the synchronous time slot measurement sequence. Under the constraint of monotonically increasing impedance of the network tree topology, identify the self-sensitivity coefficient and mutual sensitivity coefficient of each branch line online with the low-voltage side outlet bus of the distribution transformer as the reference node, and construct the voltage-power differential sensitivity matrix. Step S3: Establish a set of physical equilibrium equations with the total active power and total reactive power of the low-voltage side outlet bus of the distribution transformer as the boundary, construct an inversion model with the goal of minimizing the power deviation of the unmeasuring nodes, and solve for the net active load and net reactive load of the unmeasuring nodes of the entire network. Step S4: Substitute the solved net active load, net reactive load and synchronous time slot measurement sequence into the forward power flow evolution model, and reconstruct the phase voltage effective value profile of the physical nodes of the entire network and the feeder power flow distribution by combining the voltage-power differential sensitivity matrix. Calculate the reverse power flow impulse ratio and the voltage over-limit margin of the end node, generate a situation warning data packet and send it to the distribution management master station.
[0006] Preferably, step S1 includes the following sub-steps: Step S11, determining that the uploading time stamp of the non-equal interval measurement data packet falls within the adjacent discrete calculation step size interval; Step S12, using two adjacent historical measurement state quantities, performing first-order linear interpolation projection according to the time difference between the current discrete calculation time slot and the uploading time stamp, and determining the effective values of the output active power and terminal voltage after time alignment; when no new measurement data packet of the corresponding node is received within two consecutive discrete calculation steps, maintaining the alignment state quantity of the previous discrete calculation step size, and lowering the measurement covariance weight coefficient of that node.
[0007] Preferably, step S3, which solves for the net active and net reactive loads of all unmeasured nodes in the network, includes the following sub-steps: Step S31, acquire surface meteorological radiance monitoring data, and determine the theoretical power generation upper limit of the distributed power source in the current calculation time slot based on the rated capacity of the distributed power source; minimize the sum of squares of the deviations of the net injected power variation of each unmeasured load node in the network from its previous operating mean as the optimization objective function; Step S32, subject the measured total active power and total reactive power of the low-voltage side outlet bus of the distribution transformer to the physical conservation equations of the algebraic sum of the injected power of all nodes in the network and the power loss of the feeder, respectively, as rigid equation constraints of the Lagrange inversion model; use the Lagrange multiplier method to solve for the analytical extreme values of the optimization objective function under the rigid equation constraints and the theoretical power generation upper limit, and determine the net active and net reactive loads of each unmeasured node in the network.
[0008] Preferably, after determining the net active load and net reactive load of each unmeasuring node in the entire network in step S32, the method further includes: step S33, deriving the current of each branch of the feeder based on the power distribution of the nodes in the entire network and the voltage-power differential sensitivity matrix; calculating the actual Joule heat loss of the entire network based on the derived current of each branch and the line resistance of each section, and recalculating the Lagrange multipliers by filling the actual Joule heat loss back into the physical conservation equation in situ, thereby eliminating the reference power drift caused by the preset loss.
[0009] Preferably, step S4, reconstructing the phase voltage RMS profile and feeder power flow distribution of all physical nodes in the network, includes the following sub-steps: Step S41, accumulating the negative of the net injected power of each node segment by segment from the physical end node of the feeder to the low-voltage side outlet bus of the distribution transformer, and deducing the transmitted active power and transmitted reactive power of each power supply branch line; Step S42, calculating the transmission voltage drop at both ends of each power supply branch line based on the line resistance and line reactance of each power supply branch line, combined with the nominal value of the rated phase voltage; starting from the measured RMS voltage value of the low-voltage side outlet bus of the distribution transformer, subtracting the corresponding transmission voltage drop segment by segment along the feeder power supply path, and reconstructing the phase voltage RMS value of each node of the feeder.
[0010] Preferably, step S4, which calculates the reverse power flow impulse ratio and the voltage over-limit margin of the end node, generates a situation warning data packet, and sends it to the distribution management master station, includes the following sub-steps: Step S43, calculate the reverse power flow impulse ratio based on the ratio of the absolute value of the measured reverse active power transmitted from the low-voltage side outlet bus of the distribution transformer to the rated single-phase capacity of the distribution transformer; Step S44, trigger an over-limit situation alarm when the effective value of the phase voltage of the end node is greater than the preset upper limit threshold of 242.0V, or the reverse power flow impulse ratio is greater than the preset threshold by 10%; Step S45, calculate the active power reduction regulation of the distributed power grid-connected inverter based on the ratio of the voltage difference between the effective value of the phase voltage of the end node and the active power self-sensitivity coefficient of the end node, and generate a situation warning data packet containing the active power reduction regulation and send it to the distribution management master station.
[0011] Preferably, step S2, when constructing the voltage-power differential sensitivity matrix, further includes: step S21, obtaining the absolute value of the three-phase voltage fluctuation at the low-voltage side outlet bus of the distribution transformer caused by the power fluctuation of the single-phase distributed power source; calculating the disturbance ratio of the absolute value of each phase voltage fluctuation to the sum of the absolute values of the three-phase voltage fluctuations; when the disturbance ratio of any phase is greater than the preset phase determination threshold of 0.65, locking the physical access phase of the single-phase distributed power source as that phase, and updating the three-phase asymmetric network admittance parameters according to the identified physical access phase.
[0012] Preferably, step S2, when constructing the voltage-power differential sensitivity matrix, further includes: step S22, obtaining a conductor electrothermal response model that includes the ambient temperature variation and the root mean square value of the feeder current; separating the continuous impedance drift caused by the conductor resistance temperature effect and the discrete impedance step caused by the feeder connection variation based on the thermal equilibrium time constant; and correcting the resistance parameters in the voltage-power differential sensitivity matrix according to the separated continuous impedance drift to complete the feedforward adaptive compensation.
[0013] Preferably, after step S45 is sent to the power distribution management master station, the following step is also included: Step S46, when the effective value of the phase voltage of the end node is still higher than 242.0V after the distributed power grid-connected inverter reduces the active power, the inductive reactive power absorption compensation of the distributed power grid-connected inverter is calculated based on the ratio of the residual voltage difference of the effective value of the phase voltage of the end node exceeding 242.0V to the reactive power self-sensitivity coefficient of the end node; a coordination control command containing the inductive reactive power absorption compensation is generated and sent to the distributed power grid-connected inverter to control the distributed power grid-connected inverter to absorb inductive reactive power.
[0014] The beneficial effects of this invention are: 1. In the observable and measurable grid sensing of distributed power sources, by aligning the measurement data of the low-voltage side outlet bus of the distribution transformer with the measurement data of the grid-connected inverter of the distributed power source in time, and by using the node voltage disturbance to construct the voltage-power differential sensitivity matrix online, the interference of sudden changes in background power load on parameter identification results can be reduced, and the impedance changes caused by conductor temperature rise and joint contact aging can be reflected in a timely manner. This reduces the dependence on offline static admittance parameters, avoids ill-conditioned calculation matrix or power flow iteration divergence caused by excessive deviation between actual line parameters and static parameters, and improves the stability and accuracy of low-voltage distribution network state sensing calculation.
[0015] 2. Using the measured total active and reactive power of the low-voltage side outlet bus of the distribution transformer as the power balance constraint of the whole network, and combining the upper limit of the theoretical power generation output of distributed generation sources to invert the power of unmeasured nodes, it is possible to distinguish between background power load and distributed generation output when measurement information is insufficient. Compared with the method of generating pseudo-measurements based on historical power consumption curves, it can reduce the estimation bias caused by historical experience data, and make the solution of net active and net reactive loads of unmeasured nodes have clear physical constraints, thereby improving the reliability and calculation convergence of load extrapolation results under underdetermined operating conditions.
[0016] 3. By reconstructing the effective voltage profile of the entire grid phase voltage and the power flow distribution of the feeders based on the net load of the unmeasuring nodes and the measured power of the inverter obtained by inversion, the operating status of the low-voltage distribution network after the distributed generation is connected can be grasped more completely. When reverse power flow or terminal voltage exceeds the limit, the active power reduction regulation required by the grid-connected inverter of the distributed generation can be determined by using the active power self-sensitivity coefficient, so that the regulation can be determined according to the actual grid operating status, reducing the power consumption loss and regulation lag caused by repeated trial and adjustment, and improving the reliability of distributed generation operation supervision, power supply safety early warning and distribution dispatch. Attached Figure Description
[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, wherein: Figure 1 This is a flowchart of the power distribution network situation awareness calculation process of the present invention; Figure 2 This is a topology diagram of the data interaction between the distribution radio stations of this invention. Detailed Implementation
[0018] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.
[0019] A method for observable and measurable grid sensing for distributed generation includes the following steps: Step S1: Collect the three-phase reference electrical parameter time series at the low-voltage side outlet bus of the distribution transformer, as well as the non-equal interval measurement data packets of the distributed power grid-connected inverter; using the clock of the three-phase reference electrical parameter time series as a reference, resample the non-equal interval measurement data packets using the time scale difference projection operator to generate a synchronous time slot measurement sequence. Step S2: Extract the voltage disturbance of the grid-connected node within adjacent discrete time steps based on the synchronous time slot measurement sequence. Under the constraint of monotonically increasing impedance of the network tree topology, identify the self-sensitivity coefficient and mutual sensitivity coefficient of each branch line online with the low-voltage side outlet bus of the distribution transformer as the reference node, and construct the voltage-power differential sensitivity matrix. Step S3: Establish a set of physical equilibrium equations with the total active power and total reactive power of the low-voltage side outlet bus of the distribution transformer as the boundary, construct an inversion model with the goal of minimizing the power deviation of the unmeasuring nodes, and solve for the net active load and net reactive load of the unmeasuring nodes of the entire network. Step S4: Substitute the solved net active load, net reactive load and synchronous time slot measurement sequence into the forward power flow evolution model, and reconstruct the phase voltage effective value profile of the physical nodes of the entire network and the feeder power flow distribution by combining the voltage-power differential sensitivity matrix. Calculate the reverse power flow impulse ratio and the voltage over-limit margin of the end node, generate a situation warning data packet and send it to the distribution management master station.
[0020] Preferably, step S1 includes the following sub-steps: Step S11, determining that the uploading time stamp of the non-equal interval measurement data packet falls within the adjacent discrete calculation step size interval; Step S12, using two adjacent historical measurement state quantities, performing first-order linear interpolation projection according to the time difference between the current discrete calculation time slot and the uploading time stamp, and determining the effective values of the output active power and terminal voltage after time alignment; when no new measurement data packet of the corresponding node is received within two consecutive discrete calculation steps, maintaining the alignment state quantity of the previous discrete calculation step size, and lowering the measurement covariance weight coefficient of that node.
[0021] Preferably, step S3, which solves for the net active and net reactive loads of all unmeasured nodes in the network, includes the following sub-steps: Step S31, acquire surface meteorological radiance monitoring data, and determine the theoretical power generation upper limit of the distributed power source in the current calculation time slot based on the rated capacity of the distributed power source; minimize the sum of squares of the deviations of the net injected power variation of each unmeasured load node in the network from its previous operating mean as the optimization objective function; Step S32, subject the measured total active power and total reactive power of the low-voltage side outlet bus of the distribution transformer to the physical conservation equations of the algebraic sum of the injected power of all nodes in the network and the power loss of the feeder, respectively, as rigid equation constraints of the Lagrange inversion model; use the Lagrange multiplier method to solve for the analytical extreme values of the optimization objective function under the rigid equation constraints and the theoretical power generation upper limit, and determine the net active and net reactive loads of each unmeasured node in the network.
[0022] Preferably, after determining the net active load and net reactive load of each unmeasuring node in the entire network in step S32, the method further includes: step S33, deriving the current of each branch of the feeder based on the power distribution of the nodes in the entire network and the voltage-power differential sensitivity matrix; calculating the actual Joule heat loss of the entire network based on the derived current of each branch and the line resistance of each section, and recalculating the Lagrange multipliers by filling the actual Joule heat loss back into the physical conservation equation in situ, thereby eliminating the reference power drift caused by the preset loss.
[0023] Preferably, step S4, reconstructing the phase voltage RMS profile and feeder power flow distribution of all physical nodes in the network, includes the following sub-steps: Step S41, accumulating the negative of the net injected power of each node segment by segment from the physical end node of the feeder to the low-voltage side outlet bus of the distribution transformer, and deducing the transmitted active power and transmitted reactive power of each power supply branch line; Step S42, calculating the transmission voltage drop at both ends of each power supply branch line based on the line resistance and line reactance of each power supply branch line, combined with the nominal value of the rated phase voltage; starting from the measured RMS voltage value of the low-voltage side outlet bus of the distribution transformer, subtracting the corresponding transmission voltage drop segment by segment along the feeder power supply path, and reconstructing the phase voltage RMS value of each node of the feeder.
[0024] Preferably, step S4, which calculates the reverse power flow impulse ratio and the voltage over-limit margin of the end node, generates a situation warning data packet, and sends it to the distribution management master station, includes the following sub-steps: Step S43, calculate the reverse power flow impulse ratio based on the ratio of the absolute value of the measured reverse active power transmitted from the low-voltage side outlet bus of the distribution transformer to the rated single-phase capacity of the distribution transformer; Step S44, trigger an over-limit situation alarm when the effective value of the phase voltage of the end node is greater than the preset upper limit threshold of 242.0V, or the reverse power flow impulse ratio is greater than the preset threshold by 10%; Step S45, calculate the active power reduction regulation of the distributed power grid-connected inverter based on the ratio of the voltage difference between the effective value of the phase voltage of the end node and the active power self-sensitivity coefficient of the end node, and generate a situation warning data packet containing the active power reduction regulation and send it to the distribution management master station.
[0025] Preferably, step S2, when constructing the voltage-power differential sensitivity matrix, further includes: step S21, obtaining the absolute value of the three-phase voltage fluctuation at the low-voltage side outlet bus of the distribution transformer caused by the power fluctuation of the single-phase distributed power source; calculating the disturbance ratio of the absolute value of each phase voltage fluctuation to the sum of the absolute values of the three-phase voltage fluctuations; when the disturbance ratio of any phase is greater than the preset phase determination threshold of 0.65, locking the physical access phase of the single-phase distributed power source as that phase, and updating the three-phase asymmetric network admittance parameters according to the identified physical access phase.
[0026] Preferably, step S2, when constructing the voltage-power differential sensitivity matrix, further includes: step S22, obtaining a conductor electrothermal response model that includes the ambient temperature variation and the root mean square value of the feeder current; separating the continuous impedance drift caused by the conductor resistance temperature effect and the discrete impedance step caused by the feeder connection variation based on the thermal equilibrium time constant; and correcting the resistance parameters in the voltage-power differential sensitivity matrix according to the separated continuous impedance drift to complete the feedforward adaptive compensation.
[0027] Preferably, after step S45 is sent to the power distribution management master station, the following step is also included: Step S46, when the effective value of the phase voltage of the end node is still higher than 242.0V after the distributed power grid-connected inverter reduces the active power, the inductive reactive power absorption compensation of the distributed power grid-connected inverter is calculated based on the ratio of the residual voltage difference of the effective value of the phase voltage of the end node exceeding 242.0V to the reactive power self-sensitivity coefficient of the end node; a coordination control command containing the inductive reactive power absorption compensation is generated and sent to the distributed power grid-connected inverter to control the distributed power grid-connected inverter to absorb inductive reactive power.
[0028] Example 1: The current rated capacity is 400kVA, the transformation ratio is 10kV / 0.4kV, the system frequency is 50Hz, and the nominal value of the single-phase rated phase voltage is... A 220V low-voltage distribution transformer supplies power to a 380V active distribution substation. A four-core cross-linked polyethylene insulated aluminum core cable is led out from the low-voltage busbar, forming a radial single-chain series power supply path. The resistance of the branch line 0-1 from the low-voltage side output busbar of the distribution transformer to node 1 is... 0.12Ω, inductive reactance The resistance of branch line 1-2 from node 1 to node 2 is 0.04Ω. 0.18Ω, inductive reactance The resistance of the branch line 2-3 from node 2 to node 3 is 0.06Ω. 0.15Ω, inductive reactance The resistance is 0.05Ω. Node 1 along the line is connected to a 15kW three-phase distributed photovoltaic inverter and local residential electricity load. Node 2, located in the middle topology, is a purely residential load node without communication and control meters. Node 3, the physical end of the feeder, is connected to a 20kW three-phase distributed photovoltaic inverter and local commercial electricity load. During the operation phase with sufficient sunlight and light background load, the distributed photovoltaic power generation at the end of the feeder is prone to inducing reverse power flow, causing the end voltage to rise beyond the limit and the distribution transformer to be reverse overloaded. The communication channel bandwidth of the distribution substation is insufficient to support the continuous concurrent uploading of data from the high-frequency synchronous sensing devices of the entire network. In addition, the non-equal interval measurement data packets uploaded by the distributed photovoltaic inverter have a random transmission delay of 10s to 30s and discrete time scale jitter. Furthermore, the low voltage distribution line has a large resistance-to-inductance ratio and the conductor operating temperature rise causes dynamic impedance drift, resulting in ill-conditioned Jacobian matrix and calculation divergence in the state estimation relying on offline static parameters and high-frequency synchronous measurements.
[0029] The monitoring and control terminal deployed on the low-voltage outgoing line side of the distribution transformer continuously collects the three-phase reference electrical parameter time series at the low-voltage side outlet bus of the distribution transformer with a sampling period of 1 second. The main station gateway synchronously receives non-equal interval measurement data packets transmitted back from the inverters at nodes 1 and 3 via the wireless public network. Using the clock of the monitoring and control terminal at the low-voltage side outlet bus of the distribution transformer as the unified reference for the entire network, the time step for the entire network collaborative calculation is set to 1 second. When the uploading time stamp carried by the non-equal interval measurement data packet... Falling into adjacent discrete calculation step size intervals At that time, the inverter's output active power and terminal voltage RMS values are resampled using the time-scale difference projection operator, and its first-order linear interpolation projection relationship is as follows: ,in, To align to discrete computation time slots nodes The inverter outputs active power. To align to discrete computation time slots nodes Effective value of grid connection point voltage and This serves as the reference timestamp for adjacent discrete calculation time slots. During first-order linear interpolation projection, to prevent numerical distortion due to excessively large extrapolation slopes under conditions of drastic changes in illumination such as cloud cover, a power variation slope limiting threshold is pre-set. When the instantaneous change slope obtained by dividing the power difference between two adjacent historical measured state quantities by the time difference exceeds this threshold, the projection slope is truncated to the maximum allowable rate of change. Furthermore, directional consistency is verified by combining the trend of the measured total power change from the bus terminal. If the projection direction is opposite to the direction of the measured power change from the bus, the historical smoothed state quantity from the previous discrete calculation time slot is used. If the corresponding node has not been received within two consecutive discrete calculation steps... When a new measurement data packet is received, the alignment state of the previous discrete calculation step is maintained, and the measurement covariance weighting coefficient of that node is reduced. The reduced measurement covariance weighting coefficient is introduced as a penalty weighting factor into the subsequent inversion model and power flow optimization objective function. For nodes that lose packets, the squared deviation between their expired measurement values and the current inversion state is multiplied by this weighting coefficient to reduce the pulling effect of expired measurement data on the Lagrange solution of the overall network power balance. At the same time, in the forward power flow evolution model, the confidence interval of the node's voltage state is widened proportionally according to the degree of reduction of this weighting coefficient, thereby limiting the spread of expired measurement errors to the state estimation of adjacent branches. s, the effective value of phase A voltage acquired by the busbar monitoring and control terminal The active power of phase A flowing through the outgoing circuit breaker is 225V. The value is -15.8kW, with the negative sign indicating reverse power flow injection into the transformer, representing the total reactive power. The value is 6.2 kvar, and the positive sign indicates that the grid injects inductive reactive power into the feeder; after alignment by the time-scale difference projection operator, the photovoltaic active power output of node 1 is... It has a power output of 12.5kW. The active power consumption of local residents is 0 kvar. It is 3.2kW, with reactive power at load. It is 1.1 kvar, to distinguish it from the aforementioned value representing the inverter's output active power. The net active power injected into the node is denoted as The net reactive power injected into the node is denoted as Thus, the net injected active power of node 1 is obtained. kW, net injected reactive power kvar; Node 3 photovoltaic active power output The inverter generates capacitive reactive power at a power output of 18kW. The active power consumption is -2kvar for local commercial loads. 4kW, reactive power load The value is 1.5 kvar, from which the net injected active power of node 3 is obtained. kW, net injected reactive power kvar.
[0030] Taking node 0 of the low-voltage side outlet busbar of the distribution transformer as the reference node, the power increment caused by distributed photovoltaic regulation within adjacent discrete calculation time slots is extracted. Under the constraint of monotonically increasing impedance of the network tree topology, a voltage-power differential sensitivity matrix considering interphase mutual impedance is constructed: ,in, For nodes The variation in the effective value of the voltage relative to the low-voltage side outlet busbar of the distribution transformer. This refers to the active voltage differential self-sensitivity coefficient or mutual sensitivity coefficient. This refers to the reactive voltage differential self-sensitivity coefficient or mutual sensitivity coefficient. and They are nodes The injected active power increment and reactive power increment; to eliminate the non-steady-state interference of load fluctuations at unmeasuring node 2 on sensitivity identification, the monotonically increasing impedance constraint of the network tree topology is expressed by a rigid algebraic inequality, when node Located at the low-voltage side outlet busbar to node of the distribution transformer When on a common power supply path, the self-sensitivity coefficient is greater than or equal to the mutual sensitivity coefficient and is always greater than zero, that is... When node At node When the downstream branch is connected, the mutual sensitivity coefficient is always equal to the node. The self-sensitivity coefficient, i.e. Under this constraint, a constrained recursive least squares algorithm is used to update the elements of the voltage-power differential sensitivity matrix online, dynamically tracking the impedance drift caused by changes in conductor operating temperature rise and joint contact resistance, reducing matrix ill-conditioning caused by offline static admittance matrix bias. During the discrete calculation time-series advancement process, a time-scale decoupling and alternating advancement mechanism is used to handle the interdependence between sensitivity identification and load inversion: at the initial stage of system startup, the voltage-power differential sensitivity matrix uses the line design nominal parameters or the historical convergence value of the previous discrete calculation time slot as the initial recursive value. Within the current calculation time slot, the prior voltage-power differential sensitivity matrix is first used in conjunction with the bus total power conservation equation to invert the initial net load of each unmeasuring node in a single step. Then, the characteristic power micro-disturbance signal generated by the active voltage regulation action of the grid-connected inverter is extracted as a known excitation source. This signal is then filtered and separated in the frequency domain with the mixed disturbance including the low-frequency drift of the background electrical load. Finally, under the constraint of monotonically increasing impedance of the network tree topology, the self-sensitivity coefficient and the mutual sensitivity coefficient are recursively updated, and the updated voltage-power differential sensitivity matrix is input into the forward power flow evolution model.
[0031] For node 2, which does not have independent monitoring and control meters installed, the total active power measured by the monitoring and control terminal on the low-voltage side outlet bus of the distribution transformer is used. Total reactive power A set of physical equilibrium equations is constructed for the boundary, defining the power flowing from the bus to the feeder as negative and the net power injected into the grid from the node as positive. The initial calculation uses the previous time step's setpoint for the total network line loss, and the active power loss... Set to 0.9kW, reactive power loss Set at 0.3 kvar, the active power conservation equation and reactive power conservation equation for the entire feeder are: Substitute time slot For all known values of s, the work conservation relationship is as follows: Combine known terms to obtain The net active power injected into node 2 was calculated. kW; the reactive power conservation relationship is Combine known terms to obtain The net reactive power injected at node 2 is calculated. According to the aforementioned sign convention that net power injected into the grid by a node is positive, a negative value indicates that the node is in a net power consumption state. Therefore, the net active load of node 2 in the current time slot is 6.6kW, and the net reactive load is 1.3kvar. Subsequently, an inversion model is constructed with the goal of minimizing the power deviation of unmeasured nodes. Surface meteorological radiance monitoring data is read in each calculation time slot, and the rated capacity of the distributed power source is calculated according to the ratio of the current surface meteorological radiance to the rated operating condition radiance of the distributed power source. The calculated result is limited to not exceeding the rated capacity, and the obtained value is used as the upper bound of the theoretical power generation output of the current calculation time slot. The optimization objective function is to minimize the sum of squares of the deviations of the net power injection variation of each unmeasured load node in the entire network from its previous operating average. The measured total active power and total reactive power of the distribution transformer low-voltage side outlet bus are respectively controlled by the physical conservation equations of the algebraic sum of the node injections and feeder power losses in the entire network, which are used as rigid equation constraints for the Lagrange inversion model. At that time, the upper bound of theoretical power generation output is processed using effective set decision logic: First, under the assumption that the upper bound is not activated in a relaxed state, the Lagrangian function is constructed according to the physical conservation equation, and the initial algebraic extrema of each unmeasured node are obtained through the first-order partial derivative; then, the boundary validity is checked. If the calculated value does not exceed the upper bound of theoretical power generation output, the analytical solution is adopted. If the calculated value exceeds the upper bound of theoretical power generation output, the output power of the corresponding node is locked at the upper bound, and the Lagrangian function is reconstructed for the remaining unmeasured nodes that have not exceeded the limit. The remaining power conservation residual is allocated, thereby determining the net active load and net reactive load of the unmeasured nodes in the whole network. Then, the current of each branch of the feeder is deduced based on the power distribution of the nodes in the whole network and the voltage-power differential sensitivity matrix. The actual Joule heat loss of the whole network is calculated based on the current of each branch and the line resistance of each section. The actual Joule heat loss is backfilled in situ into the physical equilibrium equation set to recalculate the Lagrange multipliers, complete the loss correction within a single cycle, and eliminate the reference power drift caused by the preset loss.
[0032] Substituting the calculated net active and reactive loads of node 2, along with the synchronous time slot measurement sequences of nodes 1 and 3, into the forward power flow evolution model, and based on the network topology, deducing segment by segment the transmitted active power, transmitted reactive power, and transmitted voltage drop of each power supply branch line from the physical end node 3 of the feeder to the low-voltage side outlet bus of the distribution transformer; the transmitted power flowing through the end branch 2-3 is equal to the negative of the net injected power of the end node 3, corresponding to the branch active power kW, branch reactive power Substituting kvar into the voltage drop calculation model, we obtain the transmission voltage drop across branch 2-3: In this calculation, the active and reactive power of branch transmission are expressed in kW and kvar respectively, the resistance and reactance of branch lines are expressed in Ω, and the rated phase voltage is expressed in V. Therefore, when calculating the transmission voltage drop, the power is multiplied by 1000 for power unit conversion. The voltage is -8.75V. The power transmitted through the intermediate branch 1-2 is equal to the negative of the sum of the net injected power at nodes 2 and 3, corresponding to the active power of the branch. kW, branch reactive power Substituting kvar into the calculation, we obtain the voltage drop across branch 1-2: The transmission power flowing through the first branch 0-1 is equal to the total power flowing out of the bus, corresponding to the active power of the branch. kW, branch reactive power Substituting kvar into the calculation, we obtain the voltage drop across the two ends of branch 0-1: The measured voltage of the low-voltage side outlet busbar of the distribution transformer. Starting from V, the corresponding transmission voltage drop is subtracted segment by segment along the feeder power supply path to reconstruct the effective phase voltage profile of all physical nodes in the network. The effective phase voltage of node 1 is shown below. V, the effective value of phase voltage at node 2 V, the effective value of the three-phase voltage at the terminal node. V.
[0033] The upper limit for safe operation of phase voltage in low-voltage distribution networks is set to +10% of the nominal value, corresponding to a preset upper limit threshold. The rated single-phase capacity of the distribution transformer is 242V. The rated capacity is 133.33 kVA. The reverse power flow impact ratio is calculated based on the ratio of the absolute value of the measured reverse active power supplied to the low-voltage side outlet busbar of the distribution transformer to the rated single-phase capacity of the distribution transformer. The over-limit margin of the terminal node voltage is calculated based on the difference between the effective value of the phase voltage of the terminal node and the preset upper limit threshold. In this time slot, the effective value of the phase voltage of terminal node 3 is 245.99V, therefore the over-limit margin of the terminal node voltage is 3.99V. The system compares the effective value of the phase voltage of the terminal node with 242V and the reverse power flow impulse ratio with 10%. If any condition is met, an over-limit situation alarm is triggered. In this time slot, both the terminal node voltage and the reverse power flow impulse ratio exceed the limit. The active power self-sensitivity coefficient of terminal node 3 is calculated based on the series topology of the entire line. The obtained active power self-sensitivity coefficient is 2.045V / kW. Based on the ratio of the voltage difference between the effective value of the phase voltage at the end node and the preset upper limit threshold to the active power self-sensitivity coefficient at the end node, the active power reduction regulation of the distributed power grid-connected inverter is calculated: The obtained active power reduction adjustment amount is 1.95kW. The system generates a situation warning data packet containing this active power reduction adjustment amount and sends it to the distribution management master station. After reading the active power reduction adjustment amount in the situation warning data packet, the distribution management master station issues an active power reduction command to the distributed photovoltaic inverter at node 3, reducing its active power output by 1.95kW and adjusting the voltage of the end node to the safe operating range of 242V. In the actual control execution process, if the active power reduction adjustment amount calculated according to the active power self-sensitivity coefficient of the end node exceeds the current maximum adjustable output margin of the grid-connected inverter, or touches the minimum allowable grid-connected power output limit preset by the inverter, resulting in an unresolved excess voltage or high resistance after the active power output reduction reaches its limit, In a distributed power grid, a high-order nonlinear transverse voltage drop causes a first-order sensitivity cutoff error, resulting in the effective value of the phase voltage at the end node still exceeding 242V after active power reduction. In this case, the active power regulation channel is blocked, and the remaining voltage difference is used as the input for subsequent inductive reactive power absorption compensation. When the effective value of the phase voltage at the end node still exceeds 242V after the distributed power grid-connected inverter reduces active power, the inductive reactive power absorption compensation is calculated based on the ratio of the remaining voltage difference exceeding 242V to the reactive power self-sensitivity coefficient of the end node. A coordinated control command containing this inductive reactive power absorption compensation is generated and sent to the distributed power grid-connected inverter to control its absorption of inductive reactive power. When a single-phase distributed power grid generates power disturbances, the bus three-phase voltage disturbance ratio... The value is greater than the preset phase identification threshold of 0.65. Therefore, the physical connection phase of the single-phase distributed power source is locked as phase A. The admittance parameters of the three-phase asymmetric network are updated based on the identified physical connection phase. Specifically, the three-phase four-wire network is decoupled into a third-order composite admittance matrix containing the self-admittance of each phase and the inter-phase mutual admittance. Based on the locked physical connection phase, the equivalent grid-connected admittance increment of the single-phase distributed power source is superimposed on the diagonal element self-admittance position corresponding to A in the composite admittance matrix. At the same time, the spatial geometric distance between this phase and the other two phases is considered. The off-diagonal element parameters of the phase-to-phase mutual admittance are corrected, and the phase-to-phase neutral point potential shift caused by the neutral line impedance is recalculated to update the admittance parameters of the three-phase asymmetric network. Combined with the conductor electrothermal response model that includes the ambient temperature variation and the root mean square value of the feeder current, when the conductor temperature rise of 12℃ causes the resistance increment to reach 0.0058Ω, the branch resistance is adaptively updated to 0.1258Ω. The updated resistance parameters are then used for feedforward adaptive compensation of the voltage-power differential sensitivity matrix to reduce the steady-state impact of temperature drift on sensitivity identification.
[0034] This technical approach relies on the conventional measurement and control meters at the distribution transformer outgoing line and the inherent asynchronous low-frequency telemetry data packets of the inverter. Based on the time slot alignment of the time scale difference projection operator, it uses the small power disturbance of the grid-connected inverter as the excitation signal for online impedance sensitivity identification. It also provides a sensitivity mapping matrix for convex optimization decoupling inversion under the power conservation equation constraint, enabling closed-loop interactive calculation between net load inversion without measurement nodes and forward power flow recursion. Under the engineering constraints of maintaining existing metering equipment and low-bandwidth narrowband communication conditions, it avoids the high cost of retrofitting with micro synchronous phasor measurement devices and the contradiction of channel congestion. In numerical calculation, it avoids the ill-conditioned Jacobian matrix and calculation divergence caused by reliance on offline static parameters in high impedance ratio networks. It transforms the load fitting based on empirical curves under underdetermined conditions into an analytical solution based on the bus power conservation boundary and disturbance response, so that the grid situational awareness and anti-limit control decision-making are directly based on the continuous physical response characteristics of the grid, providing a basis for operation status monitoring and active scheduling for distribution networks with distributed power sources.
[0035] Example 2: This example uses a 380V low-voltage active power distribution network semi-physical simulation test platform for testing. The rated capacity of the distribution transformer is set to 400kVA, the transformation ratio is 10kV / 0.4kV, the system power frequency is 50Hz, and the nominal value of the single-phase rated phase voltage is... The voltage is 220V; a four-core cross-linked polyethylene insulated aluminum core cable feeder is led out from the low-voltage side output busbar of the distribution transformer. The resistance of the branch line 0-1 from the busbar to node 1 is... 0.12Ω, inductive reactance The resistance of branch line 1-2 from node 1 to node 2 is 0.04Ω. 0.18Ω, inductive reactance The resistance of the branch line 2-3 from node 2 to node 3 is 0.06Ω. 0.15Ω, inductive reactance The impedance is 0.05Ω. Node 1 connects a 15kW three-phase distributed photovoltaic inverter simulation source to the local power load. Node 2 connects a programmable resistive-inductive dynamic load simulation unit without any measuring instruments. Node 3, the physical end of the feeder, connects a 20kW three-phase distributed photovoltaic inverter simulation source to the commercial power load. A substation measurement and control terminal simulation unit with a measurement accuracy of 0.2 and a sampling frequency of 10kHz is deployed on the low-voltage outgoing side of the distribution transformer. The three-phase reference electrical parameter time series is continuously acquired in 1s discrete time steps. The non-equal interval measurement data packets between nodes 1 and 3 are transmitted back through a narrowband power line carrier communication channel simulator. The communication delay randomly jumps within the range of 10s to 30s. The channel packet loss rate is set to 3.5%, with a preset upper limit threshold. The nominal value of the single-phase rated phase voltage is taken as +10%, i.e., 242V. The preset threshold for the reverse power flow impact ratio is taken as 10%. In the calibration under light load reverse feeding conditions, when the reverse feeding power reaches 10% of the rated single-phase capacity of the distribution transformer, the top oil temperature and the temperature rise rate of the hottest spot in the winding are recorded to enter the nonlinear acceleration range. Based on this, 10% is used as the warning threshold for the reverse power flow impact ratio, and the single-phase grid-connected phase identification threshold is taken as 0.65. The calibration results show that when the identification ratio is lower than 0.65, it is easy to be affected by inter-phase cross-coupling and produce phase identification misjudgment. When it is higher than 0.75, it is easy to miss the identification under weak light and small disturbance conditions. For the four-core cable trunk feeder, the peak value of the inter-phase cross-coupling voltage fluctuation of the non-connected phase is considered as 0.35 times the absolute value of the voltage change of the connected phase itself. Thus, the theoretical lower limit of the disturbance ratio is 0.588. A 10% noise immunity safety margin is included on this boundary, and the preset phase identification threshold is determined to be 0.65.
[0036] The test setup included five sample groups: the present invention sample group and control sample groups A, B, C, and D. The present invention sample group used a time-scaled difference projection operator for time alignment, identified self-sensitivity and mutual sensitivity coefficients online under the constraint of monotonically increasing impedance in the network tree topology, performed net load inversion of unmeasured nodes using the bus power conservation equation, performed line loss feedback correction, and reconstructed the effective phase voltage profile and feeder power flow distribution using a forward power flow evolution model. Control sample group A removed the time resampling of the time-scaled difference projection operator and directly used asynchronous delayed raw measurement data packets. Control sample group B removed the monotonically increasing impedance constraint in the network tree topology and used a conventional recursive least squares algorithm without algebraic topology constraints to identify the sensitivity coefficients. In control sample group C, the bus power conservation equation and corresponding inversion model were cancelled, and a pseudo-measurement of historical load curves combined with a weighted least squares state estimation algorithm was used instead. In control sample group D, the single-phase grid connection phase identification threshold was adjusted to 0.5, and the online electrothermal temperature correction of conductor impedance was cancelled. The test covered the following conditions: stable and large photovoltaic power generation during sunny daytime, rapid photovoltaic change due to cloud cover at noon, extreme reverse power flow, single-phase asymmetrical access, and conductor temperature rise due to large temperature difference between day and night. The measured electrical quantities of the bus measurement and control terminal and the raw data packets uploaded by each inverter were recorded in each discrete test time slot. The net injected active power, sensitivity identification relative error, effective value of phase 3 voltage at the end node, root mean square error of end voltage, reverse power flow impulse ratio, calculation status, and single-cycle time of node 2 were also recorded.
[0037] Under the condition of high photovoltaic power generation at noon, each sample group adopted the same bus reverse active power of -15.8kW, inverter output of 18kW at node 3, and asynchronous time delay of 18.5s. They executed their respective processing flows under the same input conditions, with a reverse power flow impact ratio of 11.85% for all samples. After time alignment, sensitivity identification, and power conservation inversion, the net injected active power at node 2 was -6.6kW, and the relative error of sensitivity identification was 1.42%. The forward power flow evolution model reconstructed the phase power at the terminal node 3. The effective voltage value was 245.99V, the root mean square error of the terminal voltage was 0.38V, the calculation remained stable and converged, and the single-cycle time was 12.4ms. Control group A directly used asynchronous data without time resampling, the net injected active power at node 2 became -4.1kW, the relative error of sensitivity identification increased to 14.85%, the reconstructed effective value of the three-phase voltage at the terminal node was 241.15V, the root mean square error of the terminal voltage was 4.84V, the calculation state was out of sync, and the single-cycle time was 11.8ms; Control group B took... After eliminating the monotonically increasing impedance constraint of the tree-like network topology, the net injected active power at node 2 is -8.9kW, the relative error of sensitivity identification is 22.3%, the effective value of the three-phase voltage at the end node is 249.2V, the root mean square error of the end voltage is 6.75V, the calculation shows matrix ill-conditioning, and the single-cycle time is 48.6ms. In control group C, after using pseudo-measurement weighted least squares state estimation, the net injected active power at node 2 is -3.2kW, the relative error of sensitivity identification is 31.6%, and the effective value of the three-phase voltage at the end node is... The value is 237.8V, the root mean square error of the terminal voltage is 8.19V, and the calculation status is iterative divergence / non-convergence. As can be seen from the calculation results under the same input conditions, when the asynchronous measurement is not aligned with the time-scaled difference projection operator, the actual over-limit state of 245.99V for node 3 is calculated as 241.15V. After removing the monotonically increasing constraint of the network tree topology impedance, the sensitivity identification error and the terminal voltage reconstruction error increase simultaneously. After removing the bus power conservation equation and the inversion model, the calculation does not form a stable convergence result.
[0038] Under the sudden change of cloud cover, the bus reverse active power is -4.2kW, the inverter output at node 3 is 6.5kW, the asynchronous delay is 24.2s, and the corresponding reverse power flow impact ratio is 3.15%. After resampling by the time-scaled difference projection operator, the sample group of this invention continues to perform sensitivity identification and inversion of the unmeasured node. The net injected active power at node 2 is -6.4kW, the relative error of sensitivity identification is 1.85%, the reconstructed effective value of the phase voltage at the end node 3 is 236.42V, the root mean square error of the end voltage is 0.45V, the calculation remains stable and converges, and the single-cycle time is 13.1ms. Control sample group A directly uses asynchronous measurement data under the same original input, and the net injected active power at node 2 is -1.8kW, the sensitivity identification... The relative error of the identification was 28.4%, the effective value of the reconstructed three-phase voltage at the end node was 244.1V, and the root mean square error of the end voltage was 7.68V. The calculation result resulted in a false over-limit, and the single-cycle time was 12.2ms. The net injected active power at node 2 of the control sample group C was -5kW, the relative error of the sensitivity identification was 38.2%, the effective value of the three-phase voltage at the end node was 239.5V, and the root mean square error of the end voltage was 9.35V. The calculation state was iterative oscillation, and the single-cycle time was 85.4ms. Under the condition of 24.2s asynchronous delay, the voltage calculation deviation caused by the lack of time resampling in the control sample group A directly changed the over-limit judgment result, while the sample group of the present invention maintained a root mean square error of 0.45V at the end voltage under the same operating conditions.
[0039] In extreme reverse power flow conditions, the bus reverse active power increases to -30.64kW, the inverter output at node 3 is 20kW, the asynchronous delay is 15s, and the corresponding reverse power flow impulse ratio is 22.98%. After completing sensitivity identification under the constraint of monotonically increasing network tree topology impedance, the net injected active power at node 2 is -1.32kW, the relative error of sensitivity identification is 1.15%, and the reconstructed effective value of the phase voltage at the end node 3 is 247.85V, with a root mean square error of 0.32V. The calculation remains stable and converges. The single-cycle time was 12.8ms. After the constraint was removed, the net injected active power of node 2 in control group B became 2.45kW, the relative error of sensitivity identification reached 45.1%, the effective value of the phase voltage of the terminal node 3 was 254.3V, the root mean square error of the terminal voltage was 12.45V, and a negative impedance singularity appeared in the calculation, with a single-cycle time of 62.3ms. Under the same reverse power flow impulse ratio of 22.98%, whether or not the monotonically increasing impedance constraint of the network tree topology was used directly corresponds to the difference between the two sets of sensitivity identification results and subsequent phase voltage reconstruction results.
[0040] Under conductor heating and asymmetrical operating conditions, the bus reverse active power is -12.1kW, the inverter output at node 3 is 16kW, the asynchronous delay is 21s, and the corresponding reverse power flow impulse ratio is 9.08%. After retaining single-phase physical connection phase identification and feedforward adaptive compensation of the conductor electrothermal response model in the sample group of this invention, the net injected active power at node 2 is -5.8kW, the relative error of sensitivity identification is 1.62%, and the reconstructed effective value of the phase voltage at the end node 3 is 243.65V. The root mean square error of the terminal voltage was 0.41V, the calculation remained stable and converged, and the single-cycle time was 14.5ms. After the control sample group D adopted a single-phase grid-connected phase identification threshold of 0.5 and canceled the online electrothermal temperature correction of conductor impedance, the net injected active power of node 2 was -8.1kW, the relative error of sensitivity identification was 19.5%, the effective value of the phase voltage of the terminal node 3 was 238.9V, the root mean square error of the terminal voltage was 5.82V, the phase identification misjudgment occurred, and the single-cycle time was 14.1ms.
[0041] In all test conditions, the absolute deviation of the load extrapolation at node 2 of the present invention sample group without measurement was no greater than 0.2kW, and stable convergence results were obtained for each condition. After canceling the timescale difference projection operator for asynchronous measurement data, a random delay of 10s to 30s caused the inverter measurement timescale to be misaligned with the bus reference time slot, and the root mean square error of the terminal voltage in the cloud-shaded condition reached 7.68V. After canceling the monotonically increasing impedance constraint of the network tree topology, a negative impedance singularity appeared in the extreme reverse power flow condition, and the root mean square error of the terminal voltage reached 12.45V. When using pseudo-measurements of historical load curves combined with weighted least squares state estimation, iterative divergence and iterative oscillation occurred respectively under the condition of rapid changes in photovoltaic output. When comparing the reverse power flow boundary, when When the feeder end voltage is within the 242V safety threshold, after the photovoltaic backfeed power exceeds 13.33kW per phase and the reverse power flow impact ratio exceeds 10%, the end node voltage enters the over-limit range of 245.99V. When comparing the phase identification of single-phase physical access, a judgment threshold of 0.5 produces a phase identification misjudgment probability of 18.6%, while a judgment threshold of 0.65 results in a phase identification accuracy of 100%. These test data correspond to the output changes of each processing stage, including time alignment, monotonically increasing network tree topology impedance constraint, inversion without measurement nodes, single-phase physical access phase identification, and conductor electrothermal compensation.
[0042] Example 3: The current distribution transformer has a rated capacity of 400kVA, a transformation ratio of 10kV / 0.4kV, a system power frequency of 50Hz, and a single-phase rated phase voltage nominal value. The voltage is 220V. A monitoring and control terminal for the low-voltage side of the distribution transformer is installed on the outgoing side of the busbar. The main feeder is sequentially configured with branches 0-1, 1-2, and 2-3 from its beginning to its end. The line resistance of branch 0-1 is... 0.12Ω, inductive reactance The resistance of branch 1-2 is 0.04Ω. 0.18Ω, inductive reactance The resistance of branch lines 2-3 is 0.06Ω. 0.15Ω, inductive reactance The resistance is 0.05Ω. Node 1 and terminal node 3 are connected to three-phase distributed photovoltaic inverters with rated power generation capacities of 15kW and 20kW, respectively. Node 2 is a pure power load node without communication measurement and control meters. Node 3 is connected to commercial power loads in parallel. The non-equal interval measurement data packets uploaded by each distributed photovoltaic inverter are transmitted to the measurement and control terminal of the low-voltage side outlet bus of the distribution transformer via narrowband power line carrier. The random transmission delay is 10s to 30s. The change of ambient temperature during the day and night causes the line resistance to drift continuously. The measurement and control terminal of the low-voltage side outlet bus of the distribution transformer collects the time series of three-phase reference electrical parameters with a reference time step of 1s. After receiving the non-equal interval measurement data packets of each distributed photovoltaic inverter, it resamples them to a unified discrete time slot through the time scale difference projection operator to form a synchronous time slot measurement sequence.
[0043] Using node 0 of the low-voltage side outlet bus of the distribution transformer as the reference node, the disturbance excitation generated by the inverter power variation within adjacent discrete calculation time slots is extracted. Unconstrained intermediate estimates of the self-sensitivity coefficient and mutual sensitivity coefficient of each branch are obtained through recursive least squares operations. Then, algebraic truncation is performed according to the monotonically increasing impedance constraint of the network tree topology for the nodes in the feeder. and nodes When node Located at the low-voltage side outlet busbar to node of the distribution transformer When on a common power supply path, the active power sensitivity coefficient is limited to a lower limit of 0.0001V / kW and a lower limit of [missing value]. Within the upper limit range; where 0.0001V / kW is determined according to the physical magnitude of the minimum inherent contact resistance at the connection between the low-voltage distribution network outgoing circuit breaker and the busbar. When the sensitivity estimate obtained by the unconstrained recursive least squares operation exceeds the range limited by the monotonically increasing impedance constraint of the network tree topology, the criterion is that the minimum Euclidean distance from the unconstrained estimate to the constraint boundary is used, and the nearest boundary value is taken as the effective sensitivity parameter after truncation; when the node Located at node When supplying power to downstream branches, ensure that the active power mutual sensitivity coefficient meets the requirements. Therefore, an algebraic boundary consistent with the tree-like power supply path is applied to the recursive results of the self-sensitivity coefficient and the mutual sensitivity coefficient, so as to avoid the occurrence of zero or negative values in the unconstrained identification results.
[0044] For unmeasuring nodes without independent monitoring and control meters, the total active and reactive power measured by the monitoring and control terminal of the low-voltage side outlet bus of the distribution transformer are used as the boundaries of the physical balance equations. The upper limit of the theoretical power generation output of the current calculation time slot is converted to the rated capacity of the distributed power source according to the ratio of surface meteorological radiance to rated operating radiance, and the conversion result is limited to not exceeding the corresponding rated capacity of the distributed power source. The optimization objective is to minimize the sum of squares of the deviations of the net injected power variation of each unmeasuring load node from its previous operating mean. The active power conservation relationship and the reactive power conservation relationship are used as rigid equality constraints to construct a Lagrange inversion model. When the same penalty weight is used for each unmeasuring load node, the distribution of the power conservation residual among the unmeasuring nodes is determined by the first-order Lagrange extreme value condition. The current operating section only contains one unmeasuring node, node 2, and the initial active power loss is... The reactive power loss is 0.9kW. The measured total active power of the distribution transformer's low-voltage side outlet busbar is -15.8kW, and the total reactive power is 6.2kvar, with a net active power injection at node 1. The net active power injected at node 1 is 0.3kvar. 9.3kW, net injected reactive power The net active power injected into node 3 is -1.1 kvar. 14kW, net injected reactive power The value is -3.5 kvar. Substituting the above data into the physical equilibrium equations, we obtain the net active power injected at node 2. -6.6kW, net injected reactive power The value is -1.3 kvar; according to the sign convention that the net power injected into the grid by a node is positive, the net active load of node 2 is 6.6 kW and the net reactive load is 1.3 kvar.
[0045] After obtaining the initial net power of the unmeasured node, a single-step feedback correction is performed on the line loss. Based on the initial net power of node 2 and the synchronous time slot measurement sequence of each distributed photovoltaic inverter, the negative of the net injected power of each node is accumulated segment by segment from the physical end of the feeder to the low-voltage side outlet bus of the distribution transformer to obtain the transmitted active power and transmitted reactive power of each power supply branch line. Then, the current of each power supply branch is deduced by combining the effective value of the phase voltage of each physical node. Based on the current of each branch and the corresponding line resistance, the actual Joule heat loss of the entire network is calculated. The absolute value of the algebraic difference between the actual Joule heat loss and the preset line loss exceeds 0.02kJ. When W, the actual Joule heat loss is backfilled into the physical equilibrium equations in situ and the Lagrange multipliers are recalculated. The net power of the unmeasured node is calibrated a second time to bring the residual loss calculation to within 0.005kW. The convergence threshold of 0.02kW is determined based on the 0.2-level power metering accuracy of the low-voltage outgoing line side measurement and control terminal of the distribution transformer. In a 400kVA distribution transformer area, this value corresponds to 0.005% of the full-scale measurement range. The power difference below this value is within the quantization noise range of the transformer. When this convergence threshold is used, the root mean square error of the phase voltage reconstruction at the end of the entire network is controlled within 0.5V.
[0046] To address the continuous drift in line resistance caused by changes in ambient temperature and load heating, a conductor electrothermal response model incorporating ambient temperature variation and the root mean square (RMS) value of the feeder current is dynamically updated. The Joule heat generation per unit time of the conductor is determined by the RMS value of the feeder current and the current conductor resistance. The heat dissipated from the conductor to the surrounding environment is determined by the comprehensive heat dissipation coefficient of the conductor surface, the effective heat dissipation area, and the difference between the conductor temperature and the ambient temperature. This difference acts on the heat capacity per unit length of the conductor, yielding a first-order electrothermal equilibrium process of conductor temperature change over time. In each collaborative calculation time slot, the ambient temperature variation is used as the boundary input, and the derived RMS value of the feeder current is substituted into this process to recursively obtain the dynamic temperature rise of the conductor relative to the reference temperature. This, combined with the conductor resistance temperature coefficient and the initial nominal resistance, determines the continuous impedance drift and the thermal equilibrium time constant of the four-core cross-linked polyethylene insulated aluminum core cable. The calibration time is 900s; when the absolute value of the impedance change rate is less than or equal to 0.05mΩ / s, this change is taken as the continuous impedance drift caused by the change in the conductor's operating temperature, and is calculated according to the temperature coefficient of resistance of aluminum conductors. The resistance temperature rise increment is calculated to 0.004℃, and the resistance parameters in the voltage-power differential sensitivity matrix are corrected to complete the feedforward adaptive compensation. When the absolute value of the identified impedance change rate is greater than 0.05mΩ / s and less than 0.5mΩ / s, the voltage-power differential sensitivity matrix parameters and the three-phase asymmetric network admittance parameters of the previous discrete calculation time slot are maintained, and the identified impedance of 3 discrete calculation time steps is continuously recorded within the sliding time window. If the average change rate of the 3 discrete calculation time steps gradually decreases and falls back to within 0.05mΩ / s, then feedforward adaptive compensation is performed according to the continuous impedance drift. If the change rate continues to be maintained or continues to increase, then the discrete impedance step judgment is entered. When the absolute value of the identified impedance change rate is greater than or equal to 0.5mΩ / s, the change is judged as a discrete impedance step caused by a physical relocation of the feeder or a sudden change in the contact state of the switch contact. The electrothermal temperature correction is stopped and the three-phase asymmetric network admittance parameters are updated.
[0047] The net active and reactive loads of node 2 after loss correction, along with the synchronous time-slot measurement sequences of nodes 1 and 3, are substituted into the forward power flow evolution model. Combined with the updated voltage-power differential sensitivity matrix, the phase voltage RMS profiles of all physical nodes and the feeder power flow distribution are reconstructed. The system compares the RMS phase voltage of terminal node 3 with the preset upper limit threshold of 242V and the reverse power flow impulse ratio with the preset threshold of 10%. When any of these conditions is met, an over-limit alarm is triggered, and the active power reduction adjustment is calculated based on the ratio of the voltage difference between the RMS phase voltage of terminal node 3 and 242V to the active power self-sensitivity coefficient of the terminal node. Subsequently, a situational early warning data packet containing the active power reduction adjustment amount is generated and sent to the distribution management master station. The distribution management master station sends an active power adjustment command to the corresponding distributed power grid-connected inverter based on the active power reduction adjustment amount. After the active power adjustment, if the effective value of the three-phase voltage of the end node is still higher than 242V, the inductive reactive power absorption compensation amount is calculated based on the ratio of the remaining voltage difference to the reactive power self-sensitivity coefficient of the end node. A coordinated control command containing the inductive reactive power absorption compensation amount is generated and sent to the distributed power grid-connected inverter to control its absorption of inductive reactive power. When the grid connection of a single-phase distributed power source causes the three-phase voltage of the bus to fluctuate, and the disturbance ratio of phase A is greater than the preset phase identification threshold of 0.65, its physical access phase is locked as phase A, and the three-phase asymmetrical network admittance parameters are updated based on the physical access phase.
[0048] Example 4: This example combines Figures 1 to 2 This paper describes a method for observable and measurable grid sensing for distributed power sources, such as... Figure 1As shown, step S1 collects the three-phase reference electrical parameter time series at the low-voltage side outlet bus of the distribution transformer, as well as the non-equal interval measurement data packets of the distributed power grid-connected inverter. Using the clock of the three-phase reference electrical parameter time series as a reference, the non-equal interval measurement data packets are resampled using the time scale difference projection operator to generate a synchronous time slot measurement sequence. Step S2 is executed to extract the grid-connected node voltage disturbance in adjacent discrete time steps based on the synchronous time slot measurement sequence. Under the constraint of monotonically increasing impedance of the network tree topology, the self-sensitivity coefficient and mutual sensitivity coefficient of each branch line are identified online with the low-voltage side outlet bus of the distribution transformer as the reference node, and a voltage-power differential sensitivity coefficient is constructed. Sensitivity matrix; proceed to step S3, establish a set of physical balance equations with the total active power and total reactive power of the low-voltage side outlet bus of the distribution transformer as the boundary, construct an inversion model with the goal of minimizing the power deviation of unmeasured nodes, and solve for the net active load and net reactive load of the unmeasured nodes in the whole network; proceed to step S4, substitute the solved net active load, net reactive load and synchronous time slot measurement sequence into the forward power flow evolution model, combine the voltage-power differential sensitivity matrix to reconstruct the phase voltage effective value profile and feeder power flow distribution of the physical nodes in the whole network, calculate the reverse power flow impulse ratio and the voltage over-limit margin of the end node, generate a situation warning data packet and send it to the distribution management master station.
[0049] like Figure 2 As shown, the low-voltage side outlet busbar of the distribution transformer includes a monitoring and control terminal. The low-voltage side outlet busbar of the distribution transformer is physically and electrically connected to node 1 via branch 0-1. Node 1 is electrically connected to a three-phase distributed photovoltaic inverter and local residential loads. Node 1 is physically and electrically connected to node 2 (without measurement) via branch 1-2. Node 2 (without measurement) is electrically connected to a purely residential load node. Node 2 (without measurement) is physically and electrically connected to feeder physical end node 3 via branch 2-3. Feeder physical end node 3 is electrically connected to a three-phase distributed photovoltaic inverter. The distribution transformer connects to local commercial power loads; in the signal transmission and control interaction network, the measurement and control terminal included in the low-voltage side outlet bus of the distribution transformer transmits the three-phase reference electrical parameter time series to the distribution management master station; the three-phase distributed photovoltaic inverter connected at node 1 sends non-equal interval measurement data packets to the distribution management master station; the three-phase distributed photovoltaic inverter connected at feeder physical end node 3 sends non-equal interval measurement data packets to the distribution management master station; and the distribution management master station issues situation warning data packets and coordinated control commands to the three-phase distributed photovoltaic inverter connected at feeder physical end node 3.
[0050] Example 5: During the on-site deployment and initialization phase of the low-voltage distribution transformer area, the basic parameters for grid sensing are established under the condition of photovoltaic shutdown at night. The bus reference voltage and branch no-load current are collected by the monitoring and control terminal of the low-voltage side outlet bus of the distribution transformer, and the resistance of the branch 0-1 line is set in combination with the nominal parameters of cable impedance. 0.12Ω, inductive reactance The resistance of branch 1-2 is 0.04Ω. 0.18Ω, inductive reactance The resistance of branch lines 2-3 is 0.06Ω. 0.15Ω, inductive reactance The initial value is 0.05Ω, which provides a recursive initial value for the voltage-power differential sensitivity matrix. At the same time, historical load records are retrieved, and the average active power of the unmeasured node 2 in the corresponding time slot is 6.6kW and the average inductive reactive power is 1.3kvar, which are used as the load prior reference values to limit the prior reference in the inversion model to the physical power consumption range.
[0051] To address the transmission delay inherent in power line carrier and wireless communication channels, the monitoring and control terminal on the low-voltage side of the distribution transformer's outlet bus sends time-stamped probe data packets to the measurement units of each distributed power grid-connected inverter during the on-site commissioning phase. The measured round-trip transmission delay fluctuates between 10 and 30 seconds. Based on this, the time alignment window of the time-stamped difference projection operator is set to 30 seconds, ensuring that the uploading time-stamped probe data packets of non-equidistant intervals are synchronized. The data falls within the discrete resampling time window. During actual deployment and operation, after receiving non-uniformly spaced measurement data packets with random communication transmission delays of 10s to 30s, the distribution management master station stores them in a sliding historical time-stamped buffer queue with a preset time depth. Using the actual measurement timestamp carried by each data packet as the alignment basis, it locates the historical discrete calculation time slot to which it belongs. Then, it uses the data packet and its adjacent historical measurement state quantities to perform interpolation projection resampling at the corresponding historical time slot boundary points, generating a synchronous time slot measurement sequence on the corresponding historical time section. The distribution management master station uses this historical alignment section as a reference to perform power grid situational awareness state reconstruction, and executes [further details needed] in the current collaborative calculation time slot based on the latest arriving measurement increment. The forward state update of the sliding time window is as follows: when a distributed power grid-connected inverter fails to upload new non-equal interval measurement data packets for two consecutive discrete calculation steps, the alignment state quantity of the previous discrete calculation step remains unchanged, and the measurement covariance weight coefficient corresponding to the node is reduced from the nominal reference value of 1 to 0.5. If no new non-equal interval measurement data packets are received for three consecutive discrete calculation steps, the measurement covariance weight coefficient of the node is further reduced to 0.2 until a new non-equal interval measurement data packet is received and then restored to the nominal reference value of 1. This reduces the impact of expired measurement data on subsequent state calculations and maintains the numerical convergence stability of the effective value profile of the phase voltage of all nodes and the power flow distribution of the feeders under the condition of measurement communication disturbance.
[0052] Example 6: This example extracts the measured reverse active power at the low-voltage side outlet busbar of the distribution transformer under light load conditions during midday sunlight. -30.64kW, reactive power Using 1.6kvar as the power flow boundary for the entire network, the active power output of the distributed power grid-connected inverter at node 1 is... 15kW, local consumption active power The active power output of the distributed power grid-connected inverter at the end node 3 is 0.64kW. 20kW, local active power consumption The net injected active power at node 2 without measurement is 0.8kW. The active power loss of the feeder is -1.32kW. With an operating cross-section of 1.6kW, the net injected active power at node 1 and node 3 are respectively kW and kW.
[0053] Based on the rated single-phase capacity of the distribution transformer Calculate the reverse power flow surge ratio for a 133.33 kVA system: The reverse power flow impact ratio was 22.98%, exceeding the preset threshold of 10%. Substituting the node power of the current operating section into the forward power flow evolution model, the effective values of the three-phase voltage at the terminal node were obtained. The voltage is 247.85V, which exceeds the preset upper limit threshold of 242V by 5.85V. The system compares the effective value of the three-phase voltage at the end node with 242V and the reverse power flow impulse ratio with 10%. If any condition is met, an over-limit situation alarm is triggered. In this operating section, both conditions are met simultaneously.
[0054] Based on the resistance of branch line 0-1 from the bus to node 1 The resistance of branch 1-2 is 0.12Ω. The resistance is 0.18Ω and the resistance of branch lines 2-3. Given an impedance of 0.15Ω, the active power self-sensitivity coefficient of terminal node 3 is calculated as follows: The active power mutual sensitivity coefficient between node 1 and terminal node 3 is: By combining the rated capacity weights and sensitivity coefficient ratio matrix of the two distributed power grid-connected inverters, the collaborative reduction amount of the two nodes is calculated, and the active power reduction amount of the distributed power grid-connected inverter at the end node 3 is determined. The active power reduction of the distributed power grid-connected inverter at node 1 is 2.15kW. The voltage is 2.66kW. Substitute this value into the voltage drop response model for verification: Based on this, the system generates a situation warning data packet containing the active power reduction adjustment amount and sends it to the distribution management master station. After reading the active power reduction adjustment amount in the situation warning data packet, the distribution management master station synchronously sends the corresponding active power reduction command to the distributed power grid-connected inverters of node 1 and node 3, so that the two inverters perform active power reduction at 2.66kW and 2.15kW respectively.
[0055] After the distributed generation grid-connected inverter completes the active power reduction, the effective value of the three-phase voltage at the end node is still 243.5V, which is higher than 242V. The remaining voltage difference is 243.5-242=1.5V, based on the branch inductive reactance. 0.04Ω 0.06Ω and Calculate the reactive power self-sensitivity coefficient of the terminal node 3 with a resistance of 0.05Ω: The inductive reactive power absorption compensation of the grid-connected inverter of the distributed power source at the end node 3 is calculated based on the ratio of the remaining voltage difference to the reactive power self-sensitivity coefficient. Based on this, the power distribution management master station generates a coordinated control command that includes compensation for inductive reactive power absorption and sends it to the grid-connected inverter of the distributed power source at node 3, controlling it to absorb 2.2 kvar of inductive reactive power, so that the effective value of the phase voltage at the end node 3 drops to within the safe upper limit threshold of 242V.
[0056] During the above control process, when contact oxidation occurs at the connector of branch 2-3, causing the branch resistance to jump from the nominal value of 0.15Ω to 0.23Ω, the differential rate of change of impedance is: The differential rate of change of impedance exceeds the step change criterion threshold of 0.5 mΩ / s, while the ambient temperature rise rate measured by the temperature sensor at the same time is less than 0.1℃ / min. Therefore, this impedance change is determined to be a discrete impedance step caused by the contact resistance of the junction, rather than a continuous impedance drift caused by the temperature effect of the conductor. The correction resistance of branch 2-3 is updated to 0.23Ω, and the active power self-sensitivity coefficient of the terminal node 3 is re-determined. Subsequently, the voltage-power differential sensitivity matrix is corrected based on the updated branch resistance parameters, and the admittance parameters of the three-phase asymmetrical network are updated; at the same time, a maintenance work order containing the parameters of abnormal impedance rise at the contact is generated and sent to the power distribution management master station.
[0057] The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the protection scope of the present invention.
Claims
1. A method for observable and measurable grid sensing for distributed power sources, characterized in that, Includes the following steps: Step S1: Collect the three-phase reference electrical parameter time series at the low-voltage side outlet bus of the distribution transformer, as well as the non-equal interval measurement data packets of the distributed power grid-connected inverter; using the clock of the three-phase reference electrical parameter time series as a reference, resample the non-equal interval measurement data packets using the time scale difference projection operator to generate a synchronous time slot measurement sequence. Step S2: Extract the voltage disturbance of the grid-connected node within adjacent discrete time steps based on the synchronous time slot measurement sequence. Under the constraint of monotonically increasing impedance of the network tree topology, identify the self-sensitivity coefficient and mutual sensitivity coefficient of each branch line online with the low-voltage side outlet bus of the distribution transformer as the reference node, and construct the voltage-power differential sensitivity matrix. Step S3: Establish a set of physical equilibrium equations with the total active power and total reactive power of the low-voltage side outlet bus of the distribution transformer as the boundary, construct an inversion model with the goal of minimizing the power deviation of the unmeasuring nodes, and solve for the net active load and net reactive load of the unmeasuring nodes of the entire network. Step S4: Substitute the solved net active load, net reactive load and synchronous time slot measurement sequence into the forward power flow evolution model, and reconstruct the phase voltage effective value profile of the physical nodes of the entire network and the feeder power flow distribution by combining the voltage-power differential sensitivity matrix. Calculate the reverse power flow impulse ratio and the voltage over-limit margin of the end node, generate a situation warning data packet and send it to the distribution management master station.
2. The grid sensing method for observable and measurable distributed power generation according to claim 1, characterized in that, Step S1 includes the following sub-steps: Step S11, determine that the uploading time stamp of the non-equal interval measurement data packet falls within the adjacent discrete calculation step size interval; Step S12, use two adjacent historical measurement status quantities to complete the first-order linear interpolation projection according to the time difference between the current discrete calculation time slot and the uploading time stamp, and determine the effective value of the output active power and terminal voltage after time alignment. If no new measurement data packet is received from the corresponding node within two consecutive discrete calculation steps, the alignment state of the previous discrete calculation step is maintained, and the point measurement covariance weight coefficient is adjusted downward.
3. The grid sensing method for observable and measurable distributed power generation according to claim 1, characterized in that, Step S3, which solves for the net active load and net reactive load of the entire network without measurement nodes, includes the following sub-steps: Step S31, obtain surface meteorological radiance monitoring data, and determine the upper limit of the theoretical power generation output of the distributed power source in the current calculation time slot based on the rated capacity of the distributed power source. The optimization objective function is to minimize the sum of squares of the deviations of the net injected power variation of each unmeasured load node in the entire network from its preceding operating mean. In step S32, the measured total active power and total reactive power of the low-voltage side outlet bus of the distribution transformer are respectively controlled by the physical conservation equations of the algebraic sum of the injected power of the entire network nodes and the power loss of the feeder, which are used as rigid equation constraints for the Lagrange inversion model. The analytical extrema of the optimization objective function under the rigid equation constraints and the upper bound of the theoretical power generation output are solved using the Lagrange multiplier method to determine the net active load and net reactive load of each unmeasured load node in the entire network.
4. The grid sensing method for observable and measurable distributed power sources according to claim 3, characterized in that, After determining the net active load and net reactive load of each unmeasuring node in the entire network in step S32, the following steps are also included: Step S33, deduce the current of each branch of the feeder based on the power distribution of the nodes in the entire network and the voltage-power differential sensitivity matrix; calculate the actual Joule heat loss of the entire network based on the deduced current of each branch and the line resistance of each section, and recalculate the Lagrange multipliers by filling the actual Joule heat loss back into the physical conservation equation in situ, thereby eliminating the reference power drift caused by the preset loss.
5. The grid sensing method for observable and measurable distributed power generation according to claim 1, characterized in that, Step S4 reconstructs the effective phase voltage profile and feeder power flow distribution of all physical nodes in the network, including the following sub-steps: Step S41, from the physical end node of the feeder to the low-voltage side outlet bus of the distribution transformer, the negative number of the net injected power of each node is accumulated segment by segment to deduce the active power and reactive power transmitted by each power supply branch line; Step S42, based on the line resistance and line reactance of each power supply branch line, combined with the nominal value of the rated phase voltage, the voltage drop across the two ends of each power supply branch line is calculated. Starting from the measured effective voltage value of the low-voltage side outlet bus of the distribution transformer, the corresponding transmission voltage drop is subtracted segment by segment along the feeder power supply path to reconstruct the effective phase voltage value of each node of the feeder.
6. The grid sensing method for observable and measurable distributed power generation according to claim 1, characterized in that, Step S4 calculates the reverse power flow impulse ratio and the voltage over-limit margin of the end node, generates a situation warning data packet and sends it to the distribution management master station, including the following sub-steps: Step S43, calculate the reverse power flow impulse ratio based on the ratio of the absolute value of the measured reverse active power of the low-voltage side outlet bus of the distribution transformer to the rated single-phase capacity of the distribution transformer; Step S44, trigger an over-limit situation alarm when the effective value of the phase voltage of the end node is greater than the preset upper limit threshold of 242.0V, or the reverse power flow impulse ratio is greater than the preset threshold of 10%. Step S45: Based on the ratio of the voltage difference between the effective value of the phase voltage of the end node and the preset upper limit threshold, to the active power self-sensitivity coefficient of the end node, calculate the active power reduction regulation of the distributed power grid-connected inverter, and generate a situation warning data packet containing the active power reduction regulation and send it to the distribution management master station.
7. The grid sensing method for observable and measurable distributed power generation according to claim 1, characterized in that, Step S2, when constructing the voltage-power differential sensitivity matrix, also includes: Step S21, obtaining the absolute value of the three-phase voltage fluctuation at the low-voltage side outlet bus of the distribution transformer caused by the power fluctuation of the single-phase distributed power source; calculating the disturbance ratio of the absolute value of each phase voltage fluctuation to the sum of the absolute values of the three-phase voltage fluctuations; when the disturbance ratio of any phase is greater than the preset phase determination threshold of 0.65, locking the physical access phase of the single-phase distributed power source as that phase, and updating the three-phase asymmetric network admittance parameters according to the identified physical access phase.
8. The grid sensing method for observable and measurable distributed power generation according to claim 1, characterized in that, Step S2, when constructing the voltage-power differential sensitivity matrix, also includes: Step S22, obtaining the conductor electrothermal response model containing the ambient temperature variation and the root mean square value of the feeder current; separating the continuous impedance drift caused by the conductor resistance temperature effect and the discrete impedance step caused by the feeder connection variation based on the thermal equilibrium time constant; and correcting the resistance parameters in the voltage-power differential sensitivity matrix according to the separated continuous impedance drift to complete the feedforward adaptive compensation.
9. A grid sensing method for observable and measurable distributed power sources according to claim 6, characterized in that, After step S45 is sent to the power distribution management master station, the following step is also included: Step S46, when the effective value of the phase voltage of the terminal node is still higher than 242.0V after the distributed power grid-connected inverter reduces the active power, the inductive reactive power absorption compensation of the distributed power grid-connected inverter is calculated based on the ratio of the residual voltage difference of the effective value of the phase voltage of the terminal node exceeding 242.0V to the reactive power self-sensitivity coefficient of the terminal node; a coordinated control command containing the inductive reactive power absorption compensation is generated and sent to the distributed power grid-connected inverter to control the distributed power grid-connected inverter to absorb inductive reactive power.
Citation Information
Patent Citations
State estimation method based on distributed power supply pseudo measurement data weight optimization
CN117477546A
