A power transaction deviation control method and system

Abstract

The application belongs to the technical field of power distribution network operation control, and relates to a power transaction deviation control method and system, which comprises the following steps: according to the deviation between a day-ahead transaction plan and real-time injection power, a multi-constraint optimal power flow model is solved with power balance and section capacity as constraints to generate a primary power correction instruction; the transient voltage change rate of a converter grid-connected point bus is read, and an active power ramping limit slope is established in combination with the remaining reactive power capacity margin of the converter; when the time derivative of the primary instruction is greater than the slope, rigid truncation is implemented, the power jump component exceeding the slope is filtered out to generate a target active power instruction and adjust the converter output, and the application causes nonlinear linkage between active power deviation suppression action and local reactive power support capability of the grid-connected point and bus voltage stability, effectively avoids the risk of systematic voltage loss, and guarantees the thermal stability safety of the power transmission section.

Description

Technical Field

[0001] This invention belongs to the field of power distribution network operation control technology, and relates to a power trading deviation control method and system. It is used for dynamic power deviation adjustment of power distribution networks facing day-ahead trading plans and real-time load fluctuations, and can improve the transient voltage stability and thermal stability safety of power transmission sections. Background Technology

[0002] Currently, using online simulation and sensitivity analysis techniques to dynamically monitor the stability margin of key transmission sections in power systems has become a core approach to ensuring the safe operation of large power grids. However, existing technologies still have limitations in handling control quantity allocation, automatic physical boundary contraction, and source-load random disturbance compensation under multi-section coupled conditions. First, there is the cross-section over-limit elimination technology based on static sensitivity thresholds. For example, Chinese invention patent CN103779876B discloses a scheduling method for eliminating cross-section over-limits. This method calculates the sensitivity of the generator to the over-limit cross-section and filters the control set. However, the drawback is that this scheme is essentially based on a linearized simplified model. The decision logic lacks a nonlinear representation of the dynamic coupling effect between cross-sections and fails to map the physical compression of non-target cross-sections caused by the adjustment process into the boundary constraints of the subsequent control space. This leads to the fact that under complex network structures, the elimination command for a single cross-section is very likely to induce secondary over-limits of adjacent cross-sections, causing the system to fall into a dilemma of local optima and overall instability.

[0003] Second, there is the stability margin assessment technology based on paired adjustment logic. For example, Chinese invention patent CN105574238A discloses a method for analyzing cross-sectional active power control strategies for online stability margin assessment. It uses DC power flow sensitivity to generate a reverse equal-paired control strategy. Its drawback is that this technology focuses on open-loop prediction under deterministic operating conditions. Although it reduces the reliance on offline experience of experts, it fails to establish a closed-loop correlation between real-time voltage support capacity and remaining resource carrying capacity. When dealing with random power flow fluctuations caused by high proportion of new energy access, this paired logic, which lacks an adaptive correction mechanism, is difficult to correct previous execution deviations in real time, and there is a risk of system instability due to resource overdraft.

[0004] Third, power flow regulation technology based on source-load characteristic decoupling architecture, for example, Chinese invention patent application CN121484883A discloses a power flow limit elimination method based on cross-sectional sensitivity and source-load characteristics. It solves the problem by dividing the cross-section into types such as new energy transmission and load center feed-in. Its drawback is that the existing decoupling solution strategy usually reduces the global security constraint to a local linear decision unit, ignoring the continuous temporal influence of different characteristic regions under strong electromagnetic coupling. When the power grid operation mode fluctuates drastically, if there is no mandatory constraint mechanism of global physical security boundary, the predicted trajectory is prone to cumulative drift, causing the output scheme to lose its practical value due to exceeding the global grid load limit.

[0005] Therefore, how to construct a predictive architecture with global coupling awareness and temporal resource locking capabilities, so that the cross-section control logic converges to the real physical security boundary throughout the entire cycle, is the technical problem to be solved by this invention. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and solve the technical problems of limitations in the prior art in handling control quantity allocation, automatic contraction of physical boundaries and random disturbance compensation of source load under multi-section coupling conditions, and to provide a power trading deviation control method and system.

[0007] To achieve the above-mentioned objective, the present invention provides a method for controlling power trading deviations, comprising the following steps: Step S1: Solve the optimal power flow model to generate the primary correction command: Obtain the active power sequence of the day-ahead transaction plan and the real-time injected active power of each power supply and consumption node from the real-time power dispatch control center of the distribution network. Based on the power deviation between the active power sequence of the day-ahead transaction plan and the real-time injected active power, and with the injected power balance of each node of the distribution network and the cross-sectional capacity as the constraint boundary, solve the multi-constraint optimal power flow model to generate the primary power correction command. Step S2, establish the active power ramping limit slope based on transient voltage: read the transient voltage change rate of the AC bus at the converter grid connection point in real time through the current transformer, and combine it with the remaining physical reactive capacity margin in the current network of the converter, establish the active power ramping limit slope based on the transient voltage drop boundary of the AC bus through the transient voltage safety boundary model. Step S3: Compare the limit slope and implement rigid truncation to generate the target command: Monitor the time derivative of the primary power correction command in real time. When the time derivative is less than or equal to the active power ramping limit slope, issue the primary power correction command to the converter. When the time derivative is greater than the active power ramping limit slope, implement rigid truncation on the primary power correction command at the physical interface of the converter to filter out power jump components that exceed the active power ramping limit slope to generate the target active power command, and use the target active power command to adjust the output active power of the converter.

[0008] Step S2 of the present invention includes: acquiring the transient voltage of the AC bus and calculating the first derivative within a determined sampling time period to obtain the transient voltage change rate; subtracting the current real-time reactive power output from the maximum reactive power capacity to obtain the current remaining physical reactive power capacity margin; inputting the transient voltage change rate and the current remaining physical reactive power capacity margin into a determined transient voltage safety boundary model to output the active power ramping limit slope.

[0009] The multi-constraint optimal power flow model solution in step S1 of this invention also includes congestion control of key transmission sections in the distribution network, including the following sub-steps derived from step S1: Step S11, real-time monitoring of the active power flow margin of key transmission sections in the distribution network; Step S12, when the active power flow margin is lower than a determined safety warning threshold, using the power transfer distribution factor of each power supply node to the key transmission section, dynamically dividing the distribution network into upstream and downstream regions of the key transmission section; Step S13, modifying the constraint logic of the multi-constraint optimal power flow model to prohibit the allocation of power correction commands across key transmission sections, and instead organizing independent local offsetting calculations of power deviations within the upstream and downstream regions respectively.

[0010] The present invention also includes the following sub-steps derived from step S1: Step S14, real-time acquisition of physical capacity boundary parameters of energy storage nodes with frequency regulation capability; Step S15, when the physical capacity boundary parameters approach the determined physical capacity warning boundary, reducing the participation weight of energy storage nodes in the multi-constraint optimal power flow model solution, and smoothly transferring the power correction task to the backup regulation node with sufficient capacity.

[0011] The method for reducing the participation weight of energy storage nodes in the multi-constraint power flow model solution in step S15 of the present invention is as follows: the participation weight decays according to the exponential function law. Specifically, the decayed real-time participation weight is equal to the determined initial participation weight multiplied by the negative exponent of the base of the natural logarithm. The exponent of the negative exponent is the negative value of the product of the determined decay coefficient and the capacity difference characterizing the physical capacity boundary parameter approaching the physical capacity warning boundary.

[0012] The method for obtaining the day-ahead trading plan active power sequence and the real-time injected active power of each power supply and consumption node in step S1 of the present invention includes: reading the day-ahead trading plan power curve through the power trading system interface as the day-ahead trading plan active power sequence; and obtaining the real-time injected power of each power generation device and load node in the distribution network in real time through the data acquisition and monitoring control system and phasor measurement unit as the real-time injected active power of each power supply and consumption node.

[0013] The dynamic division of the distribution network into upstream and downstream regions of the critical transmission section in step S12 of this invention includes the following sub-steps derived from step S12: Step S121, establishing a DC power flow model based on the base state reference power flow and network topology of the distribution network; Step S122, calculating the influence coefficient of the active power injection changes of each power supply node on the active power flow of the critical transmission section through the DC power flow model as the power transfer distribution factor; Step S123, assigning power supply nodes with a power transfer distribution factor greater than zero to the upstream region of the critical transmission section, and assigning power supply nodes with a power transfer distribution factor less than or equal to zero to the downstream region of the critical transmission section.

[0014] The control constraints included in the multi-constraint optimal power flow model solution in step S1 of the present invention include the nonlinear power flow balance constraint that the injected active power and injected reactive power of each node in the distribution network satisfy, the voltage safety constraint that the transient voltage deviation of each node is within the safe operation limit, and the section capacity constraint that the transmission power of each transmission section does not exceed the thermal stability power limit.

[0015] The backup regulation nodes described in this invention include gas turbine generator sets, hydroelectric generator sets, and selected centralized energy storage power stations that are connected to the power distribution network and have the ability to quickly regulate active power ramping. The power trading deviation control method adopts a multi-timescale nested closed-network control mode, wherein the update cycle of the active power sequence of the day-ahead trading plan is 24 hours, the optimization calculation cycle of the primary power correction command is 5 minutes to 15 minutes, and the transient control cycle for real-time monitoring of the time derivative of the primary power correction command and implementation of physical hard truncation of the active power command is 10 ms to 50 ms.

[0016] The present invention also provides a power trading deviation control system, which is used to implement a power trading deviation control method, comprising: The primary power correction instruction generation module is used to obtain the active power sequence of the day-ahead trading plan and the real-time injected active power of each power supply and consumption node from the real-time power dispatch control center of the distribution network. Based on the power deviation between the active power sequence of the day-ahead trading plan and the real-time injected active power, and with the injected power balance and cross-sectional capacity of each node of the distribution network as the constraint boundary, the module solves the multi-constraint optimal power flow model to generate the primary power correction instruction. The active power ramping limit slope determination module is used to read the transient voltage change rate of the AC bus at the converter grid connection point in real time through the current transformer, and combine it with the remaining physical reactive capacity margin in the current network of the converter to establish the active power ramping limit slope based on the transient voltage drop boundary of the AC bus through the transient voltage safety boundary model. The target active power command output module is used to monitor the time derivative of the primary power correction command generated by the primary power correction command generation module in real time. When the time derivative is less than or equal to the active power ramping limit slope established by the active power ramping limit slope establishment module, the primary power correction command is issued to the converter. When the time derivative is greater than the active power ramping limit slope, the primary power correction command is rigidly truncated at the physical interface of the converter to filter out the power jump components that exceed the active power ramping limit slope in order to generate the target active power command, and the output active power of the converter is adjusted using the target active power command.

[0017] Compared with the prior art, the present invention has at least the following beneficial effects: 1. In power trading deviation control, by dynamically comparing the transient voltage change rate with the reactive power support capacity margin, the active power ramping limit slope that does not cause transient instability of AC bus voltage is automatically established. When the time derivative of the primary active power correction exceeds the ramping limit slope, the envelope is rigidly cut off, limiting the large reactive power pumping effect caused by the converter's forced tracking of the active power step command in a short period of time. This establishes a deterministic nonlinear linkage between the active power deviation smoothing action and the local reactive power support capacity of the grid connection point and the bus voltage stability. This reverses the design inertia of traditional control that only focuses on smoothing out the total global supply and demand values ​​while ignoring the transient interaction of micro-level power flow. It ensures that any physical active power adjustment caused by trading plan deviation is limited to the thermal stability and transient voltage safety boundary of the distribution network topology, effectively avoiding the risk of systemic voltage loss induced by sacrificing local grid voltage stability to make up for the power difference.

[0018] 2. By combining the dynamic status of the active power flow margin of key transmission sections, when the active power flow margin of a transmission section falls below the safety warning threshold, the anti-congestion closed-loop logic is triggered. Utilizing the power transfer distribution factor of each power supply and consumption node to the transmission section, the distribution network is dynamically divided into upstream and downstream regions of the transmission section. By modifying the constraint logic of the algorithm model, the power correction command across the transmission section is prohibited. Instead, independent local offset calculations of power deviations are organized within the upstream and downstream regions. This ensures that the deviation correction action under extreme constraint conditions does not add additional transmission pressure to the transmission section that is already approaching its limit. The spatial congestion risk is limited to the local closed-loop region, thereby ensuring the thermal stability and safety of the transmission section under complex network topology when there are drastic power fluctuations at multiple nodes without relying on large-capacity external energy storage devices.

[0019] 3. By collecting physical capacity boundary parameters of energy storage or power generation nodes with frequency regulation capabilities in real time, when the physical capacity boundary parameters approach the preset physical capacity warning boundary, the weights of the load allocation in the multi-constraint load allocation algorithm model are driven to decay according to the exponential function law. The nonlinear interference of the underlying parameters is used to change the optimization path in advance, and the power correction task is smoothly transferred to the backup regulation node with sufficient capacity. This avoids the regulation failure or reverse overshoot caused by a single power regulation unit reaching the physical output limit, maintains the continuity and stability of the power control chain of the whole network, and alleviates the problem of regulation channel interruption caused by frequent top-off of hardware equipment. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the power trading deviation control method under multi-constraint optimal power flow of the present invention; Figure 2 This is a schematic diagram of the closed-loop transfer of multiple operating states for power trading deviation control according to the present invention. Detailed Implementation

[0021] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments and accompanying drawings.

[0022] Example 1: This embodiment discloses a method for controlling deviations in electricity trading, including the following steps: Step S1: Solve the optimal power flow model to generate the primary correction command: Obtain the active power sequence of the day-ahead transaction plan and the real-time injected active power of each power supply and consumption node from the real-time power dispatch control center of the distribution network. Based on the power deviation between the active power sequence of the day-ahead transaction plan and the real-time injected active power, and with the injected power balance of each node of the distribution network and the cross-sectional capacity as the constraint boundary, solve the multi-constraint optimal power flow model to generate the primary power correction command. Step S2, establish the active power ramping limit slope based on transient voltage: read the transient voltage change rate of the AC bus at the converter grid connection point in real time through the current transformer, and combine it with the remaining physical reactive capacity margin in the current network of the converter, establish the active power ramping limit slope based on the transient voltage drop boundary of the AC bus through the transient voltage safety boundary model. Step S3: Compare the limit slope and implement rigid truncation to generate the target command: Monitor the time derivative of the primary power correction command in real time. When the time derivative is less than or equal to the active power ramping limit slope, issue the primary power correction command to the converter. When the time derivative is greater than the active power ramping limit slope, implement rigid truncation on the primary power correction command at the physical interface of the converter to filter out power jump components that exceed the active power ramping limit slope to generate the target active power command, and use the target active power command to adjust the output active power of the converter.

[0023] Step S2 in this embodiment includes: acquiring the transient voltage of the AC bus and calculating the first derivative within a determined sampling time period to obtain the transient voltage change rate; subtracting the current real-time reactive power output from the maximum reactive power capacity to obtain the current remaining physical reactive power capacity margin; inputting the transient voltage change rate and the current remaining physical reactive power capacity margin into the determined transient voltage safety boundary model to output the active power ramping limit slope.

[0024] The multi-constraint optimal power flow model solution in step S1 of this embodiment also includes congestion control of key transmission sections in the distribution network, including the following sub-steps derived from step S1: Step S11, real-time monitoring of the active power flow margin of key transmission sections in the distribution network; Step S12, when the active power flow margin is lower than a determined safety warning threshold, using the power transfer distribution factor of each power supply node to the key transmission section, dynamically dividing the distribution network into upstream and downstream regions of the key transmission section; Step S13, modifying the constraint logic of the multi-constraint optimal power flow model to prohibit the allocation of power correction commands across key transmission sections, and instead organizing independent local offsetting calculations of power deviations within the upstream and downstream regions respectively.

[0025] This embodiment also includes the following sub-steps derived from step S1: Step S14, real-time acquisition of physical capacity boundary parameters of energy storage nodes with frequency regulation capability; Step S15, when the physical capacity boundary parameters approach the determined physical capacity warning boundary, reducing the participation weight of energy storage nodes in the multi-constraint optimal power flow model solution, and smoothly transferring the power correction task to the backup regulation node with sufficient capacity.

[0026] In step S15 of this embodiment, the method for reducing the participation weight of energy storage nodes in the multi-constraint power flow model solution is as follows: the participation weight decays according to an exponential function. Specifically, the decayed real-time participation weight is equal to the determined initial participation weight multiplied by the negative exponent of the base of the natural logarithm. The exponent of the negative exponent is the negative value of the product of the determined decay coefficient and the capacity difference characterizing the physical capacity boundary parameter approaching the physical capacity warning boundary.

[0027] The method for obtaining the day-ahead trading plan active power sequence and the real-time injected active power of each power supply and consumption node in step S1 of this embodiment includes: reading the day-ahead trading plan power curve through the power trading system interface as the day-ahead trading plan active power sequence; and obtaining the real-time injected power of each power generation device and load node in the distribution network in real time through the data acquisition and monitoring control system and phasor measurement unit as the real-time injected active power of each power supply and consumption node.

[0028] In this embodiment, step S12, which dynamically divides the distribution network into upstream and downstream regions of the critical transmission section, includes the following sub-steps derived from step S12: Step S121, establishing a DC power flow model based on the base state reference power flow and network topology of the distribution network; Step S122, calculating the influence coefficient of the active power injection changes of each power supply and consumption node on the active power flow of the critical transmission section using the DC power flow model as the power transfer distribution factor; Step S123, assigning power supply and consumption nodes with a power transfer distribution factor greater than zero to the upstream region of the critical transmission section, and assigning power supply and consumption nodes with a power transfer distribution factor less than or equal to zero to the downstream region of the critical transmission section.

[0029] The control constraints included in the multi-constraint optimal power flow model solution in step S1 of this embodiment include the nonlinear power flow balance constraint that the injected active power and injected reactive power of each node in the distribution network satisfy, the voltage safety constraint that the transient voltage deviation of each node is within the safe operation limit, and the section capacity constraint that the transmission power of each transmission section does not exceed the thermal stability power limit.

[0030] The backup regulation nodes described in this embodiment include gas turbine generator sets, hydroelectric generator sets, and selected centralized energy storage power stations that are connected to the power distribution network and have the ability to quickly regulate active power ramping. The power trading deviation control method adopts a closed-network control mode with multiple time scales nested. The update cycle of the active power sequence of the day-ahead trading plan is 24 hours, the optimization calculation cycle of the primary power correction command is 5 minutes to 15 minutes, and the transient control cycle of real-time monitoring of the time derivative of the primary power correction command and implementation of physical hard truncation of the active power command is 10 ms to 50 ms.

[0031] This embodiment also provides a power trading deviation control system, including: The primary power correction instruction generation module is used to obtain the active power sequence of the day-ahead trading plan and the real-time injected active power of each power supply and consumption node from the real-time power dispatch control center of the distribution network. Based on the power deviation between the active power sequence of the day-ahead trading plan and the real-time injected active power, and with the injected power balance and cross-sectional capacity of each node of the distribution network as the constraint boundary, the module solves the multi-constraint optimal power flow model to generate the primary power correction instruction. The active power ramping limit slope determination module is used to read the transient voltage change rate of the AC bus at the converter grid connection point in real time through the current transformer, and combine it with the remaining physical reactive capacity margin in the current network of the converter to establish the active power ramping limit slope based on the transient voltage drop boundary of the AC bus through the transient voltage safety boundary model. The target active power command output module is used to monitor the time derivative of the primary power correction command generated by the primary power correction command generation module in real time. When the time derivative is less than or equal to the active power ramping limit slope established by the active power ramping limit slope establishment module, the primary power correction command is issued to the converter. When the time derivative is greater than the active power ramping limit slope, the primary power correction command is rigidly truncated at the physical interface of the converter to filter out the power jump components that exceed the active power ramping limit slope in order to generate the target active power command, and the output active power of the converter is adjusted using the target active power command.

[0032] Example 2: In this embodiment, in a regional distribution network operation scenario with a high proportion of distributed power sources, the system consists of a real-time power dispatch control center, phasor measurement units installed at each power supply and consumption node, and underlying converter execution units. When the distribution network faces a dynamic power deviation between the day-ahead trading plan and real-time load fluctuations, the real-time power dispatch control center obtains the active power sequence of the day-ahead trading plan and synchronously collects the real-time injected power of each power generation device and load node through the monitoring and data acquisition system. It calculates the dynamic power deviation value of each power supply and consumption node. The dispatch control center generates a primary power correction command by solving the multi-constraint optimal power flow model, with the overall network injected power balance and the thermal stability transmission capacity of key transmission sections as constraint boundaries.

[0033] To address the transient reactive power pumping effect in the converter caused by the issuance of the primary power correction command, the control system collects the AC bus voltage sequence in real time through the AC bus voltage transformers at each power supply and consumption node grid connection point, calculates the transient voltage change rate within a 50ms sampling period, and simultaneously reads the remaining reactive power support capacity margin of the corresponding node's underlying physical execution unit. Based on the transient voltage change rate and reactive power support capacity margin, the system calculates the current active power physical ramp-up limit slope of the bus without transient instability using the transient voltage safety boundary model. The transient voltage safety boundary model uses the core processor in the grid control equipment to read the lookup table stored in the non-volatile memory sector to calculate the active power ramp-up limit slope. The core processor substitutes the physical reactive power capacity margin and transient voltage change rate obtained in the current sampling period into the following relationship: ,in, The maximum slope for active climbing. This is the voltage stability sensitivity coefficient. For physical reactive power capacity margin, The voltage change rate adjustment constant is For transient voltage change rate, the core processor retrieves the matching value from the lookup table based on the current node load rate. and The system numerically calculates and outputs the active power ramp-up limit slope. It monitors the time derivative of the primary power correction command in real time, compares it with the physical ramp-up limit slope of active power, and rigidly truncates the primary power correction command at the physical interface when the time derivative exceeds the physical ramp-up limit slope. This filters out power jump components exceeding the limit slope to generate the final target active power command. Rigid truncation of the primary power correction command is also implemented at the converter's physical interface. The control chip performs discretized limiting control on the active power reference value in the real-time interrupt service routine. The control chip calculates the difference between the primary power correction command in the current sampling period and the target active power command sent in the previous sampling period, and divides it by the transient control period. The time derivative is obtained. When the time derivative is greater than the active power ramping limit slope, the control chip adds the target active power command of the previous sampling period to the product of the active power ramping limit slope and the transient control period to generate the target active power command of the current sampling period. When the time derivative is less than or equal to the active power ramping limit slope and greater than or equal to the negative value of the active power ramping limit slope, the current primary power correction command is determined as the target active power command of the current sampling period, and the target active power command is converted into the active current setpoint of the current inner loop regulator to complete the converter output power regulation, thereby driving the converter to execute output. The time derivative of the active power command is used to limit and block the destructive transmission of high-frequency trading correction commands to the transient voltage stability of the distribution network.

[0034] When the active power flow margin of a critical transmission section in the distribution network falls below a preset safety warning threshold, the system triggers anti-congestion self-closing logic. Based on the power transfer distribution factors of each power supply node to the critical transmission section, the system calculates the specific values ​​of each factor, dynamically dividing the distribution network into upstream and downstream regions of the congestion surface. The system then modifies the constraint logic of the multi-constraint optimal power flow model to prohibit power allocation correction commands across critical transmission sections. Specifically, the dispatch control center splits the admittance matrix of the global multi-constraint optimal power flow model into independent upstream and downstream sub-matrices according to the spatial topological boundary of the critical transmission section. It removes the cross-section power interaction term from the model's objective function and introduces a zero-value constraint for cross-section power increment. This ensures that the power deviation offsetting calculation results of each power supply node in the upstream region satisfy the upstream active power balance equation, and that the power deviation offsetting calculation results of each power supply node in the downstream region satisfy the downstream active power balance equation. The balance equation, at the matrix operation level, blocks the active power flow optimization path across sections to complete independent partitioned solutions. To ensure that the independent partitioned solution results do not violate Kirchhoff's current law under the actual physical connectivity of the network topology, the independent solution of the above sub-matrices is achieved by introducing virtual boundary equivalent nodes at the network connection. The zero-value constraint of cross-section power increment forces the interactive power increment of the physical nodes at both ends of the key transmission section to be locked to zero. At this time, the upstream and downstream sub-matrices perform power flow convergence calculations under independent equivalent network boundaries, ensuring that when the calculated power correction commands of each node are actually executed in the physical network, the total power flow of the section remains unchanged in its original ground state, thereby satisfying the continuity and balance constraints of the physical power flow of the entire network. Instead, independent local offset calculations of power deviations are organized within the upstream and downstream regions. Through spatial dimension partition control, the deviation correction action is prevented from adding additional transmission pressure to the transmission section near the limit, ensuring the thermal stability safety of the transmission section of the distribution network when there are drastic power fluctuations at multiple nodes.

[0035] Example 3: In this embodiment, within the power grid operation environment of a real-time digital simulation system, the test system is configured with an electromagnetic transient simulation environment with a step size of 50μs. It integrates a converter control logic simulation model to provide millisecond-level power command response capability. Simultaneously, its AC bus voltage monitoring module has a measurement resolution of 0.1V to acquire real-time data on voltage change rate. The experiment uses a three-point support method to construct the present invention's sample group, the lower limit sample group, and the upper limit sample group under the baseline operating condition. The active power correction command step for each group is set to 10MW, and the test running time is set to 5.0s. The active power physical ramp-up limit slope of the present invention's sample group is... The slope limit for the lower limit sample group was set to 20 MW / s, while the slope limit for the upper limit sample group was tightened to 5 MW / s, in order to verify the regulation capability of the active power rigid cutoff mechanism in suppressing transient voltage instability.

[0036] After the simulation test was started and running, the system recorded the converter grid connection point voltage in real time through the phasor measurement unit. In the test operation record of the lower limit sample group, due to the excessive time derivative of the converter's response to the active step command, the grid connection point voltage dropped sharply from 1.0 pu to 0.82 pu within 0.5s, which triggered the low voltage ride-through logic and caused the converter protection to lock out, making the power regulation target fail. However, in the test operation record of the sample group of this invention, because the system detected that the time derivative of the primary power correction command exceeded the ramp limit slope of 20MW / s, the controller triggered the rigid cutoff logic within 0.05s, reducing the output power change slope. The rate was clamped at the 20MW / s boundary, keeping the voltage fluctuation amplitude at the converter grid connection point within 3% of the rated value. This ensured that the converter could complete power correction under continuous and stable operation. In the test operation record of the sample group exceeding the upper limit, although the voltage stability was ensured, the time for the system to complete the power correction command was extended to 2.0s due to the excessively small ramp limit slope setting, resulting in power tracking lag. This confirms that the setting of the active power physical ramp limit slope provides an optimal working window for the system to achieve a better balance between ensuring voltage stability and improving power tracking real-time performance. Regarding the parameter relationships involved in the above simulation test, the specific active power correction command is as follows: The physical ramp slope limit for active power is The transient voltage change rate is The formula for calculating the transient voltage change rate is: ,in, AC bus voltage, For time, the system measures the transient voltage change rate. Residual reactive power margin of the underlying converter The unit is A joint solution is performed to determine the physical slope limit of active power ramp. , so that The control unit automatically reduces the height when lowering the height. This achieves a physical fit between the closed-loop dynamic behavior of the control system and its real-time electrical support capabilities.

[0037] Example 4: In this embodiment, under the application scenario of multi-regional power trading deviation collaborative correction in a regional distribution network, the dispatch control center deploys a deviation control algorithm model to eliminate control oscillations caused by the opacity of algorithm execution logic. The algorithm model includes an input layer, a regional collaborative filtering layer, and a physical execution mapping layer. The system obtains the power deviation sequence of each node in the regional distribution network and maps the sequence to an input feature vector. ,in ,in, Represents a node The real-time power deviation value, Given the total number of nodes, the system constructs an association matrix in the regional collaborative filtering layer based on the spatial electrical distance between nodes. Calculate the smoothing weights of the correction instructions. The smoothing weight Power transfer distribution factor of nodes relative to critical transmission sections After normalization, the result is determined by the following formula: ,in, Index the node space and apply smooth weights in the physical execution mapping layer. The primary power correction command is decomposed and allocated to generate the target power adjustment amount for each power supply and consumption node. Specifically, the matrix elements of the correlation matrix A are... According to the node With nodes The absolute values ​​of the mutual impedances between nodes are normalized by reciprocal calculation to quantitatively characterize the spatial electrical distance density between nodes; the input feature vector is received at the input layer. Subsequently, the regional collaborative filtering layer uses the correlation matrix A to perform spatial smoothing filtering on the initial power deviation of each node, eliminating high-frequency noise components of local electrical islands; the calculated correction command smoothing weights are then applied. The command is passed to the physical execution mapping layer, which multiplies the primary power correction command by the smoothing weights of each node. The direct mapping is converted into the actual power active power regulation target of each converter's underlying actuator, thereby establishing a clear hierarchical data flow and logical mapping path.

[0038] To prevent algorithmic interference from implicit negative event recording injection, the control unit adds a data preprocessing and verification step before the primary power correction instruction is calculated, utilizing the previous sliding time window. A baseline probability density function is constructed using historical power deviation data within the range, and the Mahalanobis distance of the current input deviation sequence is calculated. and at Mahalanobis distance Greater than the preset abnormal threshold If the deviation is identified as a hidden abnormal trading pulse signal, it is automatically reset to the historical average through control logic. This utilizes objective and quantitative statistical characteristics to avoid the impact of unsteady-state signal injection on the power flow control algorithm. The control unit drives the underlying converter to perform active power regulation based on the verified and corrected deviation target, and monitors the voltage at the node's grid connection point in real time. With power The dynamic response is achieved, and proportional-integral-derivative (PID) control logic is used to eliminate active power regulation errors. Simultaneously, the control unit introduces a time delay compensation operator to calculate the time derivative of the output power in real time, and ensures that the time derivative is greater than... By adjusting the duty cycle of the PWM trigger pulse, the rate of change of active power is forced to converge to within the physical safety boundary, thus realizing a stable physical mapping from power trading plan deviation to the control behavior of the distribution network structure.

[0039] Example 5: In this embodiment, under the baseline calibration conditions before the formal commissioning of a regional distribution network with a high proportion of distributed power sources, the control center constructs an offline characteristic calibration environment in the background and injects simulated atypical deviation sequences. By collecting and analyzing the statistical characteristics of the converter's underlying response data, the Mahalanobis distance threshold for the normal deviation distribution is established. Mahalanobis distance threshold The calculation formula is ,in, The mean Mahalanobis distance of the active power deviation sequence over the historical normal operating period. The standard deviation of the Mahalanobis distance of the active power deviation sequence over a historical normal operating period. As an anomaly detection threshold, this feature calibration program provides initial criteria for identifying data anomalies in online operation, thereby eliminating the risk of insufficient disclosure due to ambiguous state conditions.

[0040] The online monitoring system collects data in real time during actual operation of the current sliding time window. Power deviation data within the range is used to calculate the input feature vector. Real-time Mahalanobis distance When the real-time Mahalanobis distance Greater than the anomaly detection threshold Furthermore, when the number of sampling points exceeds the preset number five times consecutively, the system triggers the implicit negative event defense mechanism and automatically clears the batch of input deviation data as invalid abnormal pulse records. The system then uses the remaining historical valid feature vector after clearing the abnormal records as the input parameter of the multi-constraint optimal power flow model to perform power deviation correction calculation, thereby using the preset quantization limit to shield the abnormal power transaction declaration information flow from the steady-state intervention of the power flow control algorithm.

[0041] During the on-site commissioning phase, the system performs a pre-calibration procedure, adjusting the ramp limit slope based on the actual voltage fluctuation response of distribution network nodes under different load levels. Initial calculation parameters To make corrections, the field commissioning unit performed a standardized active step disturbance test under three typical operating conditions of low load, medium load, and high load in the distribution network to record the voltage sag response data at the grid connection point, and used the measured response data to fit parameters. The correlation curve between the real-time node load rate and the actual node load rate is stored in the non-volatile memory sector of the control unit. This allows the system to accurately and dynamically retrieve the corresponding parameters based on the real-time node load rate during actual online operation. Numerical values ​​eliminate implementation obstacles caused by the floating of control parameters in different network topology environments.

[0042] Example 6: In this embodiment, under the pre-deployment calibration condition before the commissioning of the distribution network control equipment, the dispatch control unit inputs a sinusoidal power disturbance signal with an amplitude ranging from 1.0MW to 15.0MW and a frequency range of 0.1Hz to 5.0Hz through the converter control interface. Simultaneously, the bus voltage response sequence is acquired through the AC bus voltage transformer at the grid connection point. The voltage stability sensitivity coefficient is calculated by fitting the transfer function between the AC bus voltage transient drop amplitude and the power disturbance frequency and amplitude using the least squares method. The scheduling and control unit discretizes the fitted transfer function curve based on historical measured active and reactive power output data under different node load rates and stores it in a non-volatile storage sector of local memory. This is used to retrieve the matching curve based on the real-time node load rate during actual operation. Numerical values ​​are used to establish the closure and reproducibility of the initial parameter determination process for the control system, and further, to correlate with parameters. The solidification and discretization process is coordinated, and the voltage change rate adjustment constant is adjusted accordingly. Calibration was also performed based on the aforementioned disturbance test; using the measured maximum transient sag rate boundary value of the AC bus voltage, the corresponding regulation constants under various load rate conditions were calculated according to the inverse sensitivity law. Its value range is limited to to Between; when the dynamically monitored node load rate increases, in order to prevent the bus voltage from experiencing an avalanche-like drop under heavy load, the control unit automatically increases the corresponding constant through a discrete lookup table. Numerical values ​​to enhance the understanding of the maximum slope for active climbing. The downward clamping damping effect ensures a clear engineering boundary basis in the micro-dynamic evolution.

[0043] During the on-site deployment delay calibration condition of converter grid connection initialization, the control unit sends an active power step test signal with an amplitude of 1.0MW to the control chip of the underlying converter actuator. Simultaneously, the power calculation circuit samples the current transformer and voltage transformer at the grid connection point and records the time from the issuance of the test signal. When the active power at the converter outlet changes by 5% The time difference is used to obtain the time delay deviation compensation operator. Its calculation formula is ,in, The time when the power test signal is emitted. This refers to the moment when the active power at the converter output changes. As a time delay offset compensation operator, the control unit will use the time delay offset compensation operator. The dynamic compensation register written into the actual control loop adjusts the transmission timing of the primary power correction command in advance during the subsequent active power command issuance process to offset system delay, and the time delay deviation compensation operator... When the sampling period boundary value is greater than 10ms, the control unit automatically calls the first-order forward differential prediction loop to predict the active power lead value of the current period based on the primary power correction command of the previous two sampling periods, and uses the lead value as the active power adjustment input of the converter outlet to offset the dynamic response loss of the actuator.

[0044] Under the dynamic operating conditions of the power trading deviation control system in actual operation and facing interference from abnormal trading data entry, the control unit reads the sliding time window. eigenvectors within And calculate its Mahalanobis distance. Markovnikov distance The exception detection threshold stored in the local register Perform quantitative numerical comparisons, including anomaly detection thresholds. The calculation formula is ,in, The mean Mahalanobis distance of the active power deviation sequence over the historical normal operating period. The standard deviation of the Mahalanobis distance of the active power deviation sequence over a historical normal operating period. As an anomaly detection threshold, the control unit updates the sliding time window in real time during each sampling period. The statistical characteristics of the data within, and when the calculated Mahalanobis distance is obtained. Continuously exceeding the anomaly detection threshold When the number of sampling points reaches 5, the system outputs an abnormal transaction data warning work order to the local network control equipment through the internal communication interface, simultaneously clears the current input feature vector and loads the historical effective feature vector of the previous control cycle as the input parameter of the multi-constraint optimal power flow model, so that the power flow balance solution loop can still correct the power flow distribution of the distribution network based on the effective physical property benchmark when it suffers malicious data entry.

Claims

1. A method for controlling deviations in electricity trading, characterized in that, Includes the following steps: Step S1: Obtain the active power sequence of the day-ahead transaction plan and the real-time injected active power of each power supply and consumption node from the real-time power dispatch control center of the distribution network. Based on the power deviation between the active power sequence of the day-ahead transaction plan and the real-time injected active power, and with the injected power balance and cross-sectional capacity of each node of the distribution network as the constraint boundary, solve the multi-constraint optimal power flow model to generate the primary power correction command. Step S2: Read the transient voltage change rate of the AC bus at the converter grid connection point in real time through the current transformer, and combine it with the remaining physical reactive capacity margin in the current network of the converter. Then, establish the active power ramping limit slope based on the transient voltage drop boundary of the AC bus through the transient voltage safety boundary model. Step S2 includes: acquiring the transient voltage of the AC bus and calculating the first derivative within a determined sampling time period to obtain the transient voltage change rate; subtracting the current real-time reactive power output from the maximum reactive power capacity to obtain the current remaining physical reactive power capacity margin; inputting the transient voltage change rate and the current remaining physical reactive power capacity margin into the determined transient voltage safety boundary model to output the active power ramping limit slope. The core processor substitutes the physical reactive power margin and transient voltage change rate obtained in the current sampling period into the following relationship: ,in, The maximum slope for active climbing. This is the voltage stability sensitivity coefficient. For physical reactive power capacity margin, The voltage change rate adjustment constant is For transient voltage change rate, the core processor retrieves the matching value from the lookup table based on the current node load rate. and Numericalize and calculate the output of the active ramp limit slope; Step S3: Monitor the time derivative of the primary power correction command in real time. When the time derivative is less than or equal to the active power ramping limit slope, issue the primary power correction command to the converter. When the time derivative is greater than the active power ramping limit slope, rigidly truncate the primary power correction command at the physical interface of the converter, filter out the power jump components that exceed the active power ramping limit slope to generate the target active power command, and use the target active power command to adjust the output active power of the converter. Specifically, the primary power correction command is rigidly truncated at the physical interface of the converter. The active power reference value is discretized and limited by the control chip in the real-time interrupt service routine. The control chip calculates the difference between the primary power correction command of the current sampling period and the target active power command sent in the previous sampling period, and divides it by the transient control period to obtain the time derivative. When the time derivative is greater than the active power ramping limit slope, the control chip adds the target active power command of the previous sampling period to the product of the active power ramping limit slope and the transient control period to generate the target active power command of the current sampling period. When the time derivative is less than or equal to the active power ramping limit slope and greater than or equal to the negative value of the active power ramping limit slope, the current primary power correction command is determined as the target active power command of the current sampling period, and the target active power command is converted into the active current setpoint of the current inner loop regulator to complete the converter output power regulation. Furthermore, the power trading deviation control method adopts a multi-timescale nested closed-network control mode, in which the update cycle of the active power sequence of the day-ahead trading plan is 24h, the optimization calculation cycle of the primary power correction command is 5min to 15min, and the transient control cycle of real-time monitoring of the time derivative of the primary power correction command and implementation of physical hard truncation of the active power command is 10ms to 50ms.

2. The method for controlling power trading deviations according to claim 1, characterized in that, The multi-constraint optimal power flow model solution in step S1 also includes congestion control of key transmission sections in the distribution network, including the following sub-steps derived from step S1: Step S11, real-time monitoring of the active power flow margin of key transmission sections in the distribution network; Step S12, when the active power flow margin is lower than the determined safety warning threshold, using the power transfer distribution factor of each power supply and consumption node to the key transmission section, dynamically dividing the distribution network into upstream and downstream regions of the key transmission section; Step S13: Modify the constraint logic of the multi-constraint optimal power flow model to prohibit the allocation of power correction commands across critical transmission sections, and instead organize independent local offset calculations of power deviations within the upstream and downstream regions respectively.

3. The method for controlling power trading deviations according to claim 1, characterized in that, It also includes the following sub-steps derived from step S1: Step S14, real-time acquisition of physical capacity boundary parameters of energy storage nodes with frequency regulation capability; Step S15, when the physical capacity boundary parameters approach the determined physical capacity warning boundary, reducing the participation weight of energy storage nodes in the multi-constraint optimal power flow model solution, and smoothly transferring the power correction task to the backup regulation node with sufficient capacity.

4. The method for controlling power trading deviations according to claim 3, characterized in that, The method for reducing the participation weight of energy storage nodes in the multi-constraint power flow model solution in step S15 is as follows: the participation weight decays according to the exponential function law. Specifically, the decayed real-time participation weight is equal to the determined initial participation weight multiplied by the negative exponent of the base of the natural logarithm. The exponent of the negative exponent is the negative value of the product of the determined decay coefficient and the capacity difference characterizing the physical capacity boundary parameter approaching the physical capacity warning boundary.

5. The method for controlling power trading deviations according to claim 1, characterized in that, The method for obtaining the active power sequence of the day-ahead trading plan and the real-time injected active power of each power supply and consumption node in step S1 includes: reading the power curve of the day-ahead trading plan through the power trading system interface as the active power sequence of the day-ahead trading plan; and obtaining the real-time injected power of each power generation device and load node in the distribution network in real time through the data acquisition and monitoring control system and phasor measurement unit as the real-time injected active power of each power supply and consumption node.

6. The method for controlling power trading deviations according to claim 2, characterized in that, Step S12, which dynamically divides the distribution network into upstream and downstream regions of the critical transmission section, includes the following sub-steps derived from step S12: Step S121, establishing a DC power flow model based on the base state reference power flow and network topology of the distribution network; Step S122, calculating the influence coefficient of the active power injection changes of each power supply node on the active power flow of the critical transmission section using the DC power flow model as the power transfer distribution factor; Step S123, assigning power supply nodes with a power transfer distribution factor greater than zero to the upstream region of the critical transmission section, and assigning power supply nodes with a power transfer distribution factor less than or equal to zero to the downstream region of the critical transmission section.

7. The method for controlling power trading deviations according to claim 1, characterized in that, The control constraints included in the multi-constraint optimal power flow model solution in step S1 include the nonlinear power flow balance constraint that the injected active power and injected reactive power of each node in the distribution network satisfy, the voltage safety constraint that the transient voltage deviation of each node is within the safe operation limit, and the cross-sectional capacity constraint that the transmission power of each transmission section does not exceed the thermal stability power limit.

8. The method for controlling power trading deviations according to claim 3, characterized in that, Backup regulation nodes include gas turbine generator sets, hydroelectric generator sets, and selected centralized energy storage power stations that are connected to the power distribution network and have the ability to quickly regulate active power.

9. A power trading deviation control system, used to implement the power trading deviation control method according to claim 1, characterized in that, include: The primary power correction instruction generation module is used to obtain the active power sequence of the day-ahead trading plan and the real-time injected active power of each power supply and consumption node from the real-time power dispatch control center of the distribution network. Based on the power deviation between the active power sequence of the day-ahead trading plan and the real-time injected active power, and with the injected power balance and cross-sectional capacity of each node of the distribution network as the constraint boundary, the module solves the multi-constraint optimal power flow model to generate the primary power correction instruction. The active power ramping limit slope determination module is used to read the transient voltage change rate of the AC bus at the converter grid connection point in real time through the current transformer, and combine it with the remaining physical reactive capacity margin in the current network of the converter to establish the active power ramping limit slope based on the transient voltage drop boundary of the AC bus through the transient voltage safety boundary model. The target active power command output module is used to monitor the time derivative of the primary power correction command generated by the primary power correction command generation module in real time. When the time derivative is less than or equal to the active power ramping limit slope established by the active power ramping limit slope establishment module, the primary power correction command is issued to the converter. When the time derivative is greater than the active power ramping limit slope, the primary power correction command is rigidly truncated at the physical interface of the converter to filter out the power jump components that exceed the active power ramping limit slope in order to generate the target active power command, and the output active power of the converter is adjusted using the target active power command.

Images