A method and system for automatic load regulation in a gas-fired power plant
By constructing a locally grounded Laplace matrix and a decentralized control method in a gas-fired power plant, the dependence on centralized control architecture and the problem of fixed parameter adjustment are solved, achieving adaptive and reliable automatic load adjustment.
Patent Information
- Application Number
- CN202611352096.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-02
- Publication Date
- 2026-11-17
AI Technical Summary
The existing centralized control architecture of gas-fired power plants is highly dependent on the central controller, has poor system reliability, limited scalability, and fixed parameter control methods are difficult to balance rapid response to large deviations and smooth adjustment to small deviations, requiring repeated manual parameter tuning.
By acquiring the operating status information of multiple units in a gas-fired power plant, a local grounding Laplace matrix is constructed to determine the neighbor set and the decentralized control order, decentralized sliding modulus and decentralized control gain are generated, decentralized coordinated control commands are generated, and a global convergence index is constructed to achieve automatic load adjustment.
It enables the generation of control commands without a central controller, automatically adapts to the magnitude of deviation by switching the control order, balances fast response and smooth adjustment, reduces manual parameter tuning, and improves system reliability and scalability.
Smart Images

Figure CN123113537A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of automatic control technology for gas-fired power plants, and specifically to a method and system for automatic load regulation in gas-fired power plants. Background Technology
[0002] With the continuous increase in the number and installed capacity of gas-fired power plant units, automatic load distribution and coordinated control when multiple units are operating in parallel has become a key link in ensuring the safe and economical operation of the power grid.
[0003] In existing technologies, multi-unit load regulation typically employs a centralized control architecture, where a central controller collects information from all units, calculates control commands for each unit, and then issues them for execution. However, with the increasing number of units and longer communication distances, the demands on the central controller's computing power and communication bandwidth rise dramatically. A failure in the central controller or the backbone communication network can paralyze the entire load regulation system. Furthermore, adding or removing units in a centralized architecture requires reconfiguring the entire system's control parameters, resulting in poor scalability. Therefore, existing centralized control architectures suffer from strong dependence on the central controller, poor system reliability, and limited scalability.
[0004] Furthermore, the load regulation process of gas turbines exhibits strong nonlinearity and multiple constraints. The power and speed deviations of each unit fluctuate significantly with changes in external load commands. Existing fixed-parameter control methods struggle to maintain excellent control performance across different deviation amplitudes, often requiring a trade-off between speed and stability during tuning, resulting in substantial commissioning work. Therefore, existing fixed-parameter control methods suffer from the technical challenge of balancing rapid response to large deviations with smooth adjustment to small deviations, and the need for repeated manual parameter tuning. Summary of the Invention
[0005] The main objective of this invention is to provide an automatic load adjustment method and system for gas-fired power plants, aiming to solve the technical problems mentioned in the background art, such as the strong dependence on the central controller, poor system reliability, and limited scalability of the existing centralized control architecture, as well as the difficulty of the existing fixed parameter control method in balancing rapid response to large deviations and smooth adjustment to small deviations, and the need for repeated manual parameter tuning.
[0006] To achieve the above objectives, the present invention provides an automatic load regulation method for a gas-fired power plant, comprising: S1: Obtain operating status information of multiple units in a gas-fired power plant; S2: Determine the neighbor set of each unit based on the operating status information, and construct the local grounding Laplace matrix of each unit based on the neighbor set; S3: Determine the per-unit active power deviation and per-unit speed deviation of each unit based on the operating status information, and determine the distributed control order of each unit in combination with the neighbor set; S4: Determine the distributed sliding modulus and distributed control gain of each unit based on the local grounding Laplace matrix, the per-unit active power deviation, the per-unit speed deviation, and the distributed control order; S5: Generate distributed coordinated control commands for each unit based on the distributed sliding modulus and distributed control gain; S6: Determine the fuel valve position command for each unit based on the distributed coordinated control command, and construct a global convergence index based on the distributed sliding modulus. Use the global convergence index to determine whether the unit load deviation has converged. If so, use the fuel valve position command as the automatic load adjustment command; otherwise, update the operating status information and return to S3.
[0007] Optionally, the operating status information includes the inter-unit communication topology connection relationship, real-time operating data, and safety constraint parameters.
[0008] Optionally, S2 specifically includes: S201: Determine the neighbor set of each unit based on the communication topology connection relationship between units; S202: Determine whether the neighbor set of each unit is empty; if so, treat the unit corresponding to the empty set as an isolated unit, connect the isolated unit to the communication link of any neighboring unit, redetermine the neighbor set, and obtain the final neighbor set; otherwise, use the neighbor set as the final neighbor set. S203: Construct a local communication subgraph for each unit based on each unit and its corresponding final neighbor set of neighbor units; S204: Based on the unit connection relationships in the local communication subgraphs, construct the graph Laplace matrix corresponding to each local communication subgraph; S205: Delete the rows and columns of the corresponding units in each graph Laplace matrix to obtain the local grounding Laplace matrix of each unit.
[0009] Optionally, S3 specifically includes: S301: Based on real-time operating data, determine the per-unit active power deviation and per-unit speed deviation of each unit; S302: Determine the double factorial scaling factor for each unit based on the number of neighbor units in the neighbor set; S303: Based on the double factorial scaling factor, the per-unit active power deviation and per-unit speed deviation are fused and scaled to obtain the composite deviation of each unit. S304: Compare the composite deviation with the interval threshold constructed based on the double factorial ratio, and determine the decentralized control order of each unit based on the comparison result.
[0010] Optionally, S4 specifically includes: S401: Determine the local topology factor for each unit based on the local grounding Laplace matrix, double factorial scaling factor, and decentralized control order; S402: Perform topological weighting on the per-unit active power deviation based on the local topological factor, and then fuse the per-unit active power deviation and the per-unit speed deviation after topological weighting to obtain the dispersed sliding modulus of each unit. S403: Determine the distributed control gain of each unit based on the local grounding Laplace matrix and the distributed control order.
[0011] Optionally, S5 specifically includes: S501: Determine the first fractional power exponent for local sliding mode feedback based on the distributed control order of each unit; S502: Determine the second fractional power exponent for inter-unit sliding modulus consistency control based on the decentralized control order of each unit and the decentralized control order of each neighbor unit in its corresponding neighbor set; S503: Based on the distributed sliding mode, distributed control gain, and the distributed sliding mode of each neighboring unit in the corresponding neighbor set of each unit, combined with the first fractional power exponent and the second fractional power exponent, generate the distributed coordinated control command for each unit.
[0012] Optionally, after S5 and before S6, it also includes: S7: Construct a multi-dimensional state safety function for each unit based on safety constraint parameters, per-unit active power deviation, per-unit speed deviation, and distributed coordinated control commands, and perform safety projection correction on the distributed coordinated control commands based on the multi-dimensional state safety function.
[0013] Optionally, the multi-dimensional state safety function includes the valve position opening upper and lower limit safety function, the valve position change rate safety function, and the thermal state margin safety function; S7 specifically includes: S701: Based on the safety constraint parameters, per-unit active power deviation, per-unit speed deviation and distributed coordinated control commands, construct the upper and lower limit safety functions of valve position opening, valve position change rate safety function and thermal state margin safety function for each unit respectively. S702: Based on the distributed coordinated control command, construct a control affine model to describe the relationship between fuel valve position changes and control commands; S703: Based on the safety functions of valve position opening upper and lower limits, valve position change rate, thermal state margin, and control affine model, combined with control obstacle function conditions, determine the safety control set for each unit; S704: Within the security control set, perform minimally invasive security projection corrections on distributed coordinated control commands.
[0014] Optionally, S6 specifically includes: S601: Determine the fuel valve position command for each unit based on the distributed sliding modulus, the distributed control order, and the distributed coordinated control command; S602: Construct a sliding mode column vector based on the distributed sliding mode of each unit; S603: Construct a global convergence index based on the sliding modulus column vector; S604: Determine whether the unit load deviation has converged through the global convergence index; if yes, use the fuel valve position command as the automatic load adjustment command and proceed to S605; otherwise, update the operating status information and return to S3. S605: Outputs automatic load adjustment commands to complete automatic load adjustment of gas-fired power plants.
[0015] This invention also proposes an automatic load regulation system for gas-fired power plants, comprising: The acquisition module is used to acquire the operating status information of multiple units in a gas-fired power plant; The module is used to determine the neighbor set of each unit based on the operating status information, and to construct the local grounding Laplace matrix of each unit based on the neighbor set; The first determination module is used to determine the per-unit active power deviation and per-unit speed deviation of each unit based on the operating status information, and to determine the distributed control order of each unit in combination with the neighbor set; The second determination module is used to determine the distributed sliding modulus and distributed control gain of each unit based on the local grounding Laplace matrix, the per-unit active power deviation, the per-unit speed deviation, and the distributed control order. The generation module is used to generate distributed coordinated control commands for each unit based on the distributed sliding modulus and distributed control gain. The judgment module is used to determine the fuel valve position command of each unit according to the distributed coordinated control command, and construct a global convergence index based on the distributed sliding modulus. The global convergence index is used to determine whether the unit load deviation has converged. If so, the fuel valve position command is used as the load automatic adjustment command; otherwise, the operating status information is updated and the execution of the first determination module is returned.
[0016] The technical solution of this invention determines the neighbor set of each unit based on the operating status information and constructs the local grounding Laplace matrix of each unit based on the neighbor set. It can generate control commands using only local and neighbor information, eliminating the dependence on the central controller and global communication. This solves the technical problems of strong dependence on the central controller, poor system reliability, and limited scalability of the existing centralized control architecture. Furthermore, by determining the per-unit active power deviation and per-unit speed deviation of each unit based on the operating status information and combining the neighbor set to determine the distributed control order of each unit, the control order automatically switches according to the magnitude of the deviation without manual tuning. This solves the technical problem of existing fixed parameter control methods that are difficult to balance fast response to large deviations and smooth adjustment to small deviations, and require repeated manual parameter tuning. Attached Figure Description
[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art 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 the structures shown in these drawings without creative effort.
[0018] Figure 1 A flowchart of an automatic load regulation method for a gas-fired power plant provided for embodiments of this application; Figure 2 A schematic diagram illustrating the process of generating distributed control parameters based on local topology information, provided for embodiments of this application; Figure 3 A schematic diagram of the safety constraint projection correction process provided for embodiments of this application; Figure 4 A schematic diagram of the structure of an automatic load regulation system for a gas-fired power plant provided for an embodiment of this application; The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation
[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.
[0020] Example 1: like Figure 1 and Figure 2 As shown, an automatic load regulation method for a gas-fired power plant includes the following steps: S1: Obtain the operating status information of multiple units in a gas-fired power plant.
[0021] The operational status information includes the communication topology between units, real-time operational data, and safety constraint parameters.
[0022] It should be noted that real-time operating data refers to the raw operating parameters directly collected from each unit under its current operating state without secondary processing. This includes: the per-unit values of real-time active power, real-time speed, current fuel valve position feedback, and the per-unit value of the target load command currently being executed. Inter-unit communication topology refers to the connection status of communication links between units in a gas-fired power plant, i.e., whether there is a direct communication link between any two units, used to subsequently determine the neighbor relationships of each unit. Safety constraint parameters refer to the inherent physical limitations of each unit that do not change with its operating state, including: the per-unit values of the upper limit and lower limit of fuel valve opening, the upper limit of the allowable valve position change rate, and the per-unit value of the rated thermal state margin threshold.
[0023] S2: Determine the neighbor set of each unit based on the operating status information, and construct the local grounding Laplace matrix of each unit based on the neighbor set.
[0024] S3: Determine the per-unit active power deviation and per-unit speed deviation of each unit based on the operating status information, and determine the distributed control order of each unit in combination with the neighbor set.
[0025] S4: Determine the distributed sliding modulus and distributed control gain of each unit based on the local grounding Laplace matrix, the per-unit active power deviation, the per-unit speed deviation, and the distributed control order.
[0026] S5: Generate distributed coordinated control commands for each unit based on the distributed sliding modulus and distributed control gain.
[0027] S6: Determine the fuel valve position command for each unit based on the distributed coordinated control command, and construct a global convergence index based on the distributed sliding modulus. Use the global convergence index to determine whether the unit load deviation has converged. If so, use the fuel valve position command as the automatic load adjustment command. Otherwise, update the operating status information and return to S3.
[0028] Furthermore, after S5 and before S6, it also includes: S7: Construct a multi-dimensional state safety function for each unit based on safety constraint parameters, per-unit active power deviation, per-unit speed deviation, and distributed coordinated control commands, and perform safety projection correction on the distributed coordinated control commands based on the multi-dimensional state safety function.
[0029] Among them, the multi-dimensional state safety function includes the upper and lower limit safety function of valve position opening, the safety function of valve position change rate, and the thermal state margin safety function.
[0030] In this embodiment of the invention, real-time operating data, communication topology, and safety constraint parameters are acquired to provide a data foundation for distributed control. A Laplace matrix is constructed using topological relationships, allowing the control law to adapt to different topologies. The control order is adaptively switched based on power and speed deviations, balancing rapid response to large deviations with smooth adjustment to small deviations. The sliding modulus and gain are automatically matched to the unit's topology position, eliminating the need for manual tuning. The generated distributed control commands require only local and neighbor information to achieve multi-machine collaboration and finite-time convergence. Simultaneously, a safety function is constructed to perform minimally invasive projection corrections on the control commands, ensuring that valve positions and thermal states do not exceed limits, adjusting only when necessary to preserve the original control performance. Finally, a convergence index automatically determines whether the global load deviation has converged synchronously, forming a fully automatic closed-loop regulation.
[0031] Example 2: In one possible implementation, S2 specifically includes the following steps: S201: Determine the neighbor set of each unit based on the communication topology connection relationship between units.
[0032] Specifically, based on the communication topology between units, for each unit, all other unit numbers that have direct communication links with the unit are recorded as the unit's neighbors, and the neighbor set of the unit is composed of all other unit numbers.
[0033] S202: Determine whether the neighbor set of each unit is empty. If so, treat the unit corresponding to the empty set as an isolated unit, connect the isolated unit to the communication link of any neighboring unit, and redetermine the neighbor set to obtain the final neighbor set. Otherwise, use the previous neighbor set as the final neighbor set.
[0034] Specifically, for each unit, it is determined whether the unit's neighbor set is empty. If the unit's neighbor set is empty, the unit corresponding to the empty set is marked as an isolated unit, and the isolated unit is connected to the communication link of any neighboring unit. Then, the neighbor set of the isolated unit is re-determined based on the updated communication topology, and the re-determined neighbor set is used as the unit's final neighbor set. If the unit's neighbor set is not empty, the neighbor set is directly used as the unit's final neighbor set.
[0035] S203: Construct a local communication subgraph for each unit based on each unit and its corresponding final neighbor set of neighbor units.
[0036] Specifically, for each unit, a local communication subgraph is constructed using the unit itself and all neighbor units in the final neighbor set corresponding to the unit as nodes, and the communication links between nodes as edges.
[0037] S204: Based on the unit connection relationships in the local communication subgraphs, construct the graph Laplace matrix corresponding to each local communication subgraph.
[0038] Specifically, for each unit's corresponding local communication subgraph, the degree of each node in the local communication subgraph is counted, where the degree is the number of edges directly connected to that node. A diagonal matrix is constructed using the degree values as diagonal elements to serve as the degree matrix of the local communication subgraph. An adjacency matrix is constructed based on the connection relationships between nodes in the local communication subgraph. In the adjacency matrix, if there is an edge between two nodes, the corresponding element has a value of 1; otherwise, it has a value of 0. Subtracting the adjacency matrix from the degree matrix yields the graph Laplacian matrix corresponding to the local communication subgraph.
[0039] S205: Delete the rows and columns of the corresponding units in each graph Laplace matrix to obtain the local grounding Laplace matrix of each unit.
[0040] Specifically, for each unit's corresponding graph Laplace matrix, the rows and columns in the graph Laplace matrix that correspond to the unit itself are determined. After deleting the rows and columns, the remaining matrix is used as the unit's local grounding Laplace matrix.
[0041] In this embodiment of the invention, by defining directly connected units in the communication topology as neighbors, a clear local communication relationship is established for each unit. By detecting and processing isolated units and forcing them to connect to nearby links, it is ensured that all units in the system have communicable neighbors, avoiding control failures caused by communication islands. By constructing a local subgraph with each unit and its neighbors as nodes, the globally complex network is decomposed into multiple independently processable small-scale structures. By constructing a degree matrix and an adjacency matrix by counting node degrees and subtracting them, the local subgraph is transformed into an algebraic Laplace matrix, allowing topological information to participate in matrix operations. Furthermore, by deleting corresponding rows and columns, a locally grounded Laplace matrix is obtained, making it a symmetric positive definite matrix, providing a structurally invertible basis for the subsequent endogenous calculation of control parameters.
[0042] Example 3: In one possible implementation, S3 specifically includes the following steps: S301: Based on real-time operating data, determine the per-unit active power deviation and per-unit speed deviation of each unit.
[0043] Specifically, based on real-time operating data, the real-time active power per-unit value, target load command per-unit value, and rated active power per-unit base value are obtained for each unit. The ratio of the real-time active power per-unit value minus the target load command per-unit value to the rated active power per-unit base value is used as the per-unit active power deviation for that unit. Simultaneously, based on real-time operating data, the real-time speed per-unit value, rated speed setpoint per-unit value, and rated speed per-unit base value are obtained for each unit. The ratio of the real-time speed per-unit value minus the rated speed setpoint per-unit value to the rated speed per-unit base value is used as the per-unit speed deviation for that unit.
[0044] S302: Determine the double factorial scaling factor for each unit based on the number of neighbor units in the neighbor set.
[0045] Specifically, for each generator set, the number of neighboring generator sets is determined based on the number of neighboring generator sets in its corresponding neighbor set. Twice the number of neighboring generator sets is used as the starting value for the double factorial operation. From this starting value, the product is successively subtracted by 2 and multiplied until it reaches 2 or 1, and the resulting product is the first double factorial value. Twice the number of neighboring generator sets minus 1 is used as the starting value for the double factorial operation. From this starting value, the product is successively subtracted by 2 and multiplied until it reaches 1, and the resulting product is the second double factorial value. The second double factorial value is divided by the first double factorial value, and the resulting ratio is used as the double factorial scaling factor for that generator set.
[0046] S303: Based on the double factorial scaling factor, the per-unit active power deviation and per-unit speed deviation are fused and scaled to obtain the composite deviation of each unit.
[0047] Specifically, for each generator unit, the absolute sum of the unit's per-unit speed deviation and per-unit active power deviation is obtained by adding the absolute deviations of the unit and taking the absolute value. This absolute sum of deviations is then divided by the corresponding double-factor scaling factor for that unit, and the resulting ratio is used as the composite deviation for that unit. This composite deviation is subsequently compared with an interval threshold to determine the distributed control order of the unit.
[0048] S304: Compare the composite deviation with the interval threshold constructed based on the double factorial ratio, and determine the decentralized control order of each unit based on the comparison result.
[0049] Specifically, for each generator set, the range of values for the interval threshold is determined based on the number of its neighboring generator sets, with the index starting from 0 and increasing to the number of neighboring generator sets minus 1. For each index, twice the index is used as the starting value for the double factorial operation. From this starting value, the product is successively reduced by 2 and multiplied until 2, and the result is used as the third double factorial value. Simultaneously, twice the index minus 1 is used as the starting value for the double factorial operation. From this starting value, the product is successively reduced by 2 and multiplied until 1, and the result is used as the fourth double factorial value. The third double factorial value is divided by the fourth double factorial value to obtain the interval threshold corresponding to that index. When the index is 0, the interval threshold corresponding to that index is 1. In this way, multiple interval thresholds are constructed, equal to the number of neighboring generator sets for that generator set, with each interval threshold increasing sequentially in the direction of index increase. The composite deviation is compared sequentially with each interval threshold to determine the interval range into which the composite deviation falls, and the distributed control order of the generator set is determined based on the interval range into which it falls.
[0050] In this embodiment of the invention, by normalizing the deviations of the actual power and speed values relative to the command values, the differences in the dimensions of capacity and speed of different units are eliminated, allowing subsequent control decisions to be based on a unified, standardized scale. A double-factor scaling factor is constructed using the number of neighbors to scale the composite deviation, enabling the composite to adaptively adjust with the number of neighbors. Furthermore, a range threshold constructed using the double-factor ratio is used to perform graded matching of the composite deviation, allowing the control order to automatically switch according to the magnitude of the deviation. For small deviations, low-order control is used to maintain stable adjustment, while for large deviations, high-order control is used to accelerate the response speed. The entire threshold system is uniquely determined by the topology and the deviation itself, requiring no manual setting of any adjustment parameters.
[0051] Example 4: In one possible implementation, S4 specifically includes the following steps: S401: Determine the local topology factor for each unit based on the local grounding Laplace matrix, double factorial scaling factor, and distributed control order.
[0052] S402: Perform topological weighting on the normalized active power deviation based on the local topological factor, and then fuse the normalized active power deviation and normalized speed deviation after topological weighting to obtain the distributed sliding modulus of each unit.
[0053] Specifically, based on the properties of determinant expansion, the locally grounded Laplace matrix is... With double factorial scaling factor Composition The determinant value of the matrix, divided by the matrix from Select from neighboring units Number of permutations The resulting ratio is then multiplied by the double factorial ratio. These factors together constitute the local topology factor of the unit at the current moment. The local topology factor is used to calculate the per-unit active power deviation. Topology weighting is performed, which involves calculating the product of the local topology factor and the normalized active power deviation to obtain the topology-weighted active power deviation contribution. The normalized speed deviation is then... The active power deviation contribution is then fused with the topology-weighted contribution, i.e., the two are added together to obtain the dispersed sliding modulus of the unit at the current moment. Specifically, it is expressed as: in, Indicates the first Taiwanese crew at all times Dispersion sliding modulus, Indicates the first Taiwanese crew at all times The per-unit speed deviation, Indicates the first Taiwanese crew at all times Local topological factors, Indicates the first Taiwanese crew at all times The order of distributed control, This represents the double factorial operation. Represents determinant operations. Indicates the first The local grounding Laplace matrix of the Taiwanese generator set. Indicates the first The double factorial scaling factor corresponding to the Taiwanese unit. express An identity matrix of order 1. Indicates the first The number of neighboring units of the unit. This represents permutation operations. Indicates the first Taiwanese crew at all times The per-unit active power deviation.
[0054] S403: Determine the distributed control gain of each unit based on the local grounding Laplace matrix and the distributed control order.
[0055] Specifically, based on the permutation number normalization relation, from Select from neighboring units Number of permutations As a molecule, the locally grounded Laplace matrix Its transpose The product plus the identity matrix The square root of the determinant of the resulting matrix is used as the denominator. The ratio of the numerator to the denominator is then calculated, and the resulting ratio is multiplied by the double factorial ratio. The distributed control gain of the unit at the current moment is obtained. Specifically, it is expressed as: in, Indicates the first Taiwanese crew at all times Distributed control gain, superscript This indicates the transpose operation.
[0056] In this embodiment of the invention, a local topology factor is constructed by combining the local grounding Laplace matrix with the double factorial scaling factor and performing a determinant operation. This, along with the number of permutations and the double factorial ratio, integrates topology information and adaptive order into a quantifiable weight. This topology factor is used to weight power deviations, mapping the connectivity strength of the local subgraph to the contribution weight of the power deviation. This weight is then superimposed with the speed deviation to construct a sliding mode, allowing the sliding surface to reflect the unit's position and connectivity tightness within the local topology. Simultaneously, a control gain is constructed by combining the ratio of the number of permutations to the square root of the determinant containing the local grounding Laplace matrix with the double factorial ratio. This allows the gain to automatically match the local topology structure, enabling adaptive adjustment of the control strength based on connectivity strength without manual trial and error, thereby improving the consistency and robustness of multi-unit coordinated regulation.
[0057] Example 5: In one possible implementation, S5 specifically includes the following steps: S501: Determine the first fractional power exponent for local sliding mode feedback based on the distributed control order of each unit.
[0058] Specifically, for each unit, the current distributed control order of the unit is obtained. The distributed control order is increased by 2 and used as the denominator. The constant 2 is used as the numerator. The ratio of the numerator to the denominator is calculated. The resulting ratio is used as the first fractional power exponent for the local sliding mode feedback of the unit at the current moment. The first fractional power exponent is greater than 0 and not greater than 1.
[0059] S502: Determine the second fractional power exponent for inter-unit sliding modulus consistency control based on the decentralized control order of each unit and the decentralized control order of each neighboring unit in its corresponding neighbor set.
[0060] Specifically, for each generating unit and each neighboring unit in its corresponding neighbor set, the current distributed control order of the generating unit and the current distributed control order of its neighboring units are obtained. The distributed control order of the neighboring unit is incremented by 2 to obtain the numerator. The distributed control order of the generating unit is added to the distributed control order of its neighboring units, and then 2 is added to obtain the denominator. The ratio of the numerator to the denominator is calculated, and this ratio is used as the second fractional power exponent for the inter-unit sliding modulus consistency control between the generating unit and its neighboring units at the current moment. The second fractional power exponent is greater than 0 and not greater than 1.
[0061] S503: Based on the distributed sliding mode, distributed control gain, and the distributed sliding mode of each neighboring unit in the corresponding neighbor set of each unit, combined with the first fractional power exponent and the second fractional power exponent, generate the distributed coordinated control command for each unit.
[0062] Specifically, based on the terminal sliding mode control principle, a fractional power exponent is used to construct a nonlinear feedback term to ensure that the system state converges to the equilibrium point within a finite time. For each unit, the distributed control gain of the unit at the current moment is multiplied by the sign function value of the distributed sliding mode of the unit, then multiplied by the absolute value of the distributed sliding mode of the unit and the first fractional power exponent used for local sliding mode feedback at the current moment, and the result is inverted to obtain the local sliding mode feedback term of the unit. Simultaneously, for each neighbor unit in the neighbor set corresponding to the unit, the difference between the distributed sliding mode of the unit and its neighbor units is calculated to obtain the connection weight between the unit and its neighbor units, and the maximum value of the distributed control order of the unit and its neighbor units at time t is obtained. According to the maximum value Calculate the ratio of double factorials The connection weight is multiplied by a double factorial ratio, then by the sign function value of the difference in distributed sliding modulus between the unit and its neighboring units, and by the absolute value of this difference in distributed sliding modulus and the second fractional power exponent used for inter-unit sliding modulus consistency control between the unit and its neighboring units at the current moment, to obtain the neighbor consistency term corresponding to the neighboring unit. The neighbor consistency terms corresponding to all neighboring units of the unit are summed and added to the unit's local sliding modulus feedback term to obtain the distributed coordination control command for the unit at the current moment. Specifically, it is expressed as: in, Indicates the first Taiwanese crew at all times Distributed coordination control commands. Represents a symbolic function. Indicates the first Taiwanese crew at all times Dispersion sliding modulus, This represents the absolute value operation. Indicates the first Taiwanese crew at all times The first fractional power exponent used for local sliding mode feedback. Indicates the first The neighbors of the Taiwanese unit gathered. Indicates the first Taiwan unit and the first Connection weights between neighboring units (It should be noted that those skilled in the art can set these weights according to actual needs). The size is not limited herein. Optionally, The value can be 1 or 0. Indicates the first Taiwan unit and the first Taiwan neighboring crew at time The maximum value of the decentralized control order, Indicates the first Taiwan neighboring crew at time Dispersion sliding modulus, Indicates the first Taiwan unit and the first Taiwan neighboring crew at time The second fractional power exponent is used for consistency control of sliding modulus between units.
[0063] In this embodiment of the invention, the sliding modulus is nonlinearly scaled using a fractional power exponent constructed by the control order, enabling the system state to converge to the equilibrium point within a finite time. This provides a faster dynamic response compared to asymptotic convergence, while avoiding excessive control shocks that may occur with large deviations in linear feedback. The generated distributed control command includes a local sliding modulus feedback term and a neighbor sliding modulus consistency term. The local term drives the individual unit's deviation to quickly return to zero, while the neighbor term ensures that the sliding modulus of adjacent units tends to synchronize by connecting weights and an adaptive double factorial ratio. Together, they achieve multi-unit coordinated regulation. The entire control command generation only requires each unit to obtain its own and its neighbors' sliding modulus information, eliminating the need for global communication and centralized computation. This ensures finite-time convergence performance while providing plug-and-play flexibility and adaptability to changes in communication topology.
[0064] Example 6: In one possible implementation, S6 specifically includes the following steps: S601: Determine the fuel valve position command for each unit based on the distributed sliding modulus, distributed control order, and distributed coordinated control command.
[0065] Specifically, based on the discretized control principle, for each unit, the fuel valve position feedback at the current sampling moment is obtained, along with the distributed coordinated control command, distributed sliding mode, and per-unit sampling period at the current moment. The double factorial ratio is calculated based on the distributed control order of the unit at the current moment. This double factorial ratio is multiplied by the distributed sliding mode of the unit to obtain the sliding mode damping term. The sliding mode damping term is subtracted from the distributed coordinated control command at the current moment to obtain the valve position deviation correction. The valve position deviation correction is multiplied by the per-unit sampling period to obtain the valve position change within the sampling period. The fuel valve position feedback and the valve position change are added together to obtain the fuel valve position command at the current sampling moment of the unit. Specifically, it is expressed as: in, Indicates the first Taiwanese crew at all times Fuel valve position command, Indicates the first Taiwanese crew at all times Fuel valve position feedback, This indicates the per-unit sampling period.
[0066] S602: Construct a sliding mode column vector based on the distributed sliding mode of each unit.
[0067] Specifically, the distributed sliding modulus of each unit at the current moment is obtained, and the distributed sliding modulus of each unit is arranged in the order of their numbers. The distributed sliding modulus of each unit is used as the elements in the column vector to construct the sliding modulus column vector.
[0068] S603: Construct a global convergence index based on the sliding modulus column vector.
[0069] Specifically, the distributed sliding modulus quantities of all units are arranged in order of unit number, constructing a sliding modulus column vector. Using the matrix determinant lemma, the determinant of the matrix obtained by multiplying the identity matrix, the sliding modulus column vector, and their transpose, by the total number of units, is expanded. The expanded determinant equals 1 plus the sum of the squares of the distributed sliding modulus quantities of each unit divided by the total number of units. This expanded scalar expression is used as the global convergence index. Specifically, it is expressed as: in, Indicates time The global convergence index, express An identity matrix of order 1. Indicates the total number of generating units. Indicates time The sliding modulus column vector.
[0070] S604: Determine whether the unit load deviation has converged using the global convergence index. If yes, use the fuel valve position command as the automatic load adjustment command and proceed to S605. Otherwise, update the operating status information and return to S3.
[0071] Specifically, based on the global convergence index, the difference between the index and the value 1 is calculated to obtain the convergence deviation value. The total number of units participating in the regulation is obtained. Based on the total number of units, the first double factorial value and the second double factorial value are calculated. The first double factorial value is twice the total number of units as the starting value, and is multiplied by 2 successively until the product is 2. The second double factorial value is twice the total number of units minus 1 as the starting value, and is multiplied by 1 successively until the product is 1. The first double factorial value is divided by the second double factorial value to obtain the double factorial ratio. The double factorial ratio is divided by the total number of units to obtain the convergence judgment threshold. It is determined whether the convergence deviation value is less than or equal to the convergence judgment threshold. If yes, the sliding mode of all units is determined to have converged, i.e., the unit load deviation has converged synchronously, and the fuel valve position command is output as the automatic load adjustment command, proceeding to the next step. If not, the unit load deviation is determined not to have converged, the real-time operating data of each unit is updated, and the process returns to the step of determining the decentralized control order to continue the next round of iterative regulation. The convergence criterion is specifically expressed as follows: .
[0072] S605: Outputs automatic load adjustment commands to complete automatic load adjustment of gas-fired power plants.
[0073] In this embodiment of the invention, a sliding mode damping term is introduced when converting distributed control commands into fuel valve position commands, effectively suppressing oscillations during valve position adjustment and improving the stability of the actuator. A column vector is constructed using the global sliding mode and aggregated into a quantifiable scalar index based on the determinant lemma, reflecting the overall convergence state of all units in a concise form. This index is then compared with a double factorial threshold uniquely determined by the total number of units, achieving precise automatic determination of the global convergence state. Commands are output immediately upon convergence, and if convergence fails, the system automatically returns to the corresponding step for further iteration, forming a fully automatic closed-loop control process without manual intervention.
[0074] Example 7: In one possible implementation, S7 specifically includes the following steps: S701: Based on the safety constraint parameters, per-unit active power deviation, per-unit speed deviation, and distributed coordinated control commands, construct the upper and lower limits of valve opening safety function, valve position change rate safety function, and thermal state margin safety function for each unit.
[0075] Specifically, for each unit, based on the per-unit values of the upper and lower limits of the fuel valve position opening in the safety constraint parameters, and the per-unit value of the fuel valve position feedback at the current moment, the first difference between the per-unit value of the upper limit of the fuel valve position opening and the per-unit value of the fuel valve position feedback is calculated as the safety function for the upper limit of the valve position opening at the current moment. Similarly, the second difference between the per-unit value of the fuel valve position feedback and the per-unit value of the lower limit of the fuel valve position opening is calculated as the safety function for the lower limit of the valve position opening at the current moment. Based on the per-unit value of the upper limit of the allowable valve position change rate in the safety constraint parameters, and the absolute value of the distributed coordinated control command at the current moment, the third difference between the per-unit value of the upper limit of the allowable valve position change rate and the absolute value of the distributed coordinated control command is calculated as the safety function for the valve position change rate at the current moment. Based on the rated thermal state margin threshold per unit value in the safety constraint parameters, and the absolute value of the sum of the normalized active power deviation and the normalized speed deviation at the current moment, the fourth difference between the rated thermal state margin threshold per unit value and the absolute value of the sum of the active power deviation and the speed deviation is calculated as the thermal state margin safety function of the unit at the current moment, specifically expressed as: in, and They represent the first Taiwanese crew at all times The upper limit safety function and the lower limit safety function for valve opening. and They represent the first The upper and lower per-unit values of the fuel valve opening of the generator unit. Indicates the first Taiwanese crew at all times The per-unit value of the fuel valve position feedback. Indicates the first Taiwanese crew at all times The safety function for the rate of change of valve position. Indicates the first The per-unit upper limit of the allowable valve position change rate for the generator set. Indicates the first Taiwanese crew at all times thermal state margin safety function Indicates the first The per-unit value of the rated thermal state margin threshold for the unit.
[0076] S702: Based on the distributed coordinated control command, construct a control affine model to describe the relationship between fuel valve position changes and control commands.
[0077] Specifically, for each unit, within a single sampling period, the current distributed coordinated control command of the unit is used as the rate of change of the unit's fuel valve position. A control affine model is then constructed to describe the relationship between the fuel valve position change and the control command. In the control affine model, the rate of change of the fuel valve position is equal to the current distributed coordinated control command of the unit. That is, the control affine system does not contain any natural dynamic terms related to the current valve position state, and the action coefficient of the control input is 1.
[0078] S703: Based on the safety functions of valve position opening upper and lower limits, valve position change rate, thermal state margin, and control affine model, combined with control obstacle function conditions, determine the safety control set for each unit.
[0079] Specifically, for each unit, the values of the valve position opening upper and lower limits safety functions, the valve position change rate safety function, and the thermal state margin safety function are obtained at the current moment, and the rate of change of each safety function over time is also obtained. For each safety function, the current value of the safety function is multiplied by the preset positive-zero extended class K function coefficient, and then the rate of change of the safety function over time is added to construct the control obstacle function condition corresponding to the safety function. The control affine model is substituted into the control obstacle function condition of each safety function, and the relationship between the fuel valve position change rate and the distributed coordinated control command in the control affine model is used to transform each control obstacle function condition into the control command allowable interval corresponding to each safety function. The intersection of the control command allowable intervals corresponding to all safety functions is taken to obtain the safe control set of the unit at the current moment.
[0080] S704: Within the security control set, perform minimally invasive security projection corrections on distributed coordinated control commands.
[0081] Specifically, for each unit, using the distributed coordinated control command of the unit at the current moment as the reference value, within the safety control set of the unit at the current moment, half of the square of the deviation between the control command variable and the distributed coordinated control command is used as the objective function. The control command variable that minimizes the objective function is then used as the control command of the unit at the current moment after minimum intrusive safety projection correction. Specifically, it is expressed as: in, Indicates the first Taiwanese crew at all times Control commands corrected by minimally invasive safety projection This refers to finding the control quantity within the set of safety controls that minimizes the objective function. , Indicates the first Taiwanese crew at all times A set of security controls, This represents the control command variable to be solved.
[0082] In this embodiment of the invention, three types of safety functions—upper and lower limits of valve opening, rate of change, and thermal state margin—are constructed respectively to quantify the hard constraints of the unit's actuators and thermodynamics into unified mathematical expressions. Using a control affine model, the safety functions are substituted into the control obstacle function conditions, transforming them into allowable intervals for control commands corresponding to each safety constraint, and then the intersection is taken, giving the solution of the safety control set a rigorous mathematical foundation. Then, through minimum intrusive projection, the correction value with the smallest difference from the original command is found within this set. This ensures that the corrected control command strictly satisfies all safety constraints while preserving the synergistic performance of the original distributed sliding mode control to the greatest extent possible, intervening only with minimal amplitude when necessary, avoiding excessive conservatism that could lead to a decline in control quality.
[0083] like Figure 3 As shown in the figure, the distributed coordinated control command projection correction process based on multi-dimensional safety constraints is illustrated. First, a safety control boundary is constructed based on safety constraints regarding valve opening limits, valve position change rate, and thermal state margin. The feasible control regions corresponding to each safety constraint are then merged to form a safety control set. When a distributed coordinated control command is located outside this set, it is mapped into the set using a minimum intrusive safety projection method, resulting in a corrected control command that satisfies the safety constraints. Through this process, the control command satisfies fuel valve position and unit thermal state constraints while maintaining the original distributed coordinated control strategy's adjustment direction and control performance as much as possible.
[0084] The foregoing primarily describes the solutions provided by the embodiments of this application from a methodological perspective. To achieve the aforementioned functions, it includes corresponding hardware structures and / or software modules for executing each function. Those skilled in the art should readily recognize that, in conjunction with the method steps of the various examples described in the embodiments disclosed herein, the embodiments of this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.
[0085] This application embodiment can divide the system into functional modules based on the above method example. For example, each function can be divided into its own functional module, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware or as a software functional module. Optionally, the module division in this application embodiment is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.
[0086] Figure 4 This is a schematic diagram of an automatic load regulation system for a gas-fired power plant, provided as an embodiment of this application. The system includes an acquisition module 110, a construction module 111, a first determination module 112, a second determination module 113, a generation module 114, and a judgment module 115. Wherein: The acquisition module 110 is used to acquire the operating status information of multiple units in a gas-fired power plant; Module 111 is used to determine the neighbor set of each unit based on the operating status information, and to construct the local grounding Laplace matrix of each unit based on the neighbor set. The first determining module 112 is used to determine the per-unit active power deviation and per-unit speed deviation of each unit based on the operating status information, and to determine the distributed control order of each unit in combination with the neighbor set; The second determining module 113 is used to determine the distributed sliding modulus and distributed control gain of each unit based on the local grounding Laplace matrix, the per-unit active power deviation, the per-unit speed deviation and the distributed control order. The generation module 114 is used to generate distributed coordinated control commands for each unit based on the distributed sliding mode and the distributed control gain. The judgment module 115 is used to determine the fuel valve position command of each unit according to the distributed coordinated control command, and construct a global convergence index according to the distributed sliding modulus. The global convergence index is used to determine whether the unit load deviation has converged. If so, the fuel valve position command is used as the load automatic adjustment command; otherwise, the operating status information is updated and the execution of the first determination module is returned.
[0087] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural transformations made using the contents of the specification and drawings of the present invention under the inventive concept of the present invention, or direct / indirect applications in other related technical fields, are included within the scope of patent protection of the present invention.
Claims
1. A method for automatic load adjustment of a gas power plant, characterized in that, include: S1: Obtain operating status information of multiple units in a gas-fired power plant; S2: Determine the neighbor set of each unit based on the operating status information, and construct the local grounding Laplace matrix of each unit based on the neighbor set; S3: Determine the per-unit active power deviation and per-unit speed deviation of each unit based on the operating status information, and determine the distributed control order of each unit in conjunction with the neighbor set; S4: Determine the distributed sliding modulus and distributed control gain of each of the units based on the local grounding Laplace matrix, the per-unit active power deviation, the per-unit speed deviation, and the distributed control order; S5: Generate distributed coordinated control commands for each of the generator units based on the distributed sliding modulus and the distributed control gain; S6: Determine the fuel valve position command for each of the units according to the distributed coordination control command, and construct a global convergence index according to the distributed sliding modulus. Use the global convergence index to determine whether the unit load deviation has converged. If so, use the fuel valve position command as the load automatic adjustment command; otherwise, update the operating status information and return to S3.
2. The method of automatic load regulation of a gas power plant as claimed in claim 1, characterized in that, The operational status information includes the inter-unit communication topology connection relationship, real-time operational data, and safety constraint parameters.
3. The method of automatic load regulation of a gas power plant according to claim 2, characterized in that, S2 specifically includes: S201: Determine the neighbor set of each of the units based on the communication topology connection relationship between the units; S202: Determine whether the neighbor set of each of the aforementioned units is an empty set; if so, treat the unit corresponding to the empty set as an isolated unit, connect the isolated unit to the communication link of any neighboring unit, re-determine the neighbor set, and obtain the final neighbor set; otherwise, use the neighbor set as the final neighbor set. S203: Construct a local communication subgraph for each of the aforementioned units based on the neighboring units in the final neighbor set of each unit; S204: Based on the unit connection relationships in the local communication subgraphs, construct the graph Laplace matrix corresponding to each local communication subgraph; S205: Delete the rows and columns of the corresponding units in each of the aforementioned Laplace matrices to obtain the local grounding Laplace matrices of each of the aforementioned units.
4. The method of automatic load regulation of a gas power plant according to claim 2, characterized in that, S3 specifically includes: S301: Based on the real-time operating data, determine the per-unit active power deviation and per-unit speed deviation of each of the aforementioned units; S302: Determine the double factorial scaling factor for each of the neighboring units based on the number of neighboring units in the neighbor set; S303: Based on the double factorial scaling factor, the per-unit active power deviation and the per-unit speed deviation are fused and scaled to obtain the composite deviation of each of the units; S304: Compare the composite deviation with the interval threshold constructed based on the double factorial ratio, and determine the decentralized control order of each of the units based on the comparison result.
5. The method of automatic load regulation of a gas power plant according to claim 4, characterized in that, S4 specifically includes: S401: Determine the local topology factor of each of the units based on the local grounding Laplace matrix, the double factorial scaling factor, and the distributed control order; S402: Perform topological weighting on the per-unit active power deviation according to the local topological factor, and fuse the per-unit active power deviation after topological weighting with the per-unit speed deviation to obtain the dispersed sliding modulus of each of the units; S403: Determine the distributed control gain of each of the units based on the local grounding Laplace matrix and the distributed control order.
6. The method of automatic load regulation of a gas power plant according to claim 1, characterized in that, S5 specifically includes: S501: Determine the first fractional power exponent for local sliding mode feedback based on the decentralized control order of each of the aforementioned units; S502: Determine the second fractional power exponent for inter-unit sliding modulus consistency control based on the decentralized control order of each unit and the decentralized control order of each neighbor unit in its corresponding neighbor set; S503: Based on the distributed sliding mode, distributed control gain, and the distributed sliding mode of each neighboring unit in the corresponding neighbor set of each unit, combined with the first fractional power exponent and the second fractional power exponent, generate the distributed coordinated control command for each unit.
7. The method of automatic load regulation of a gas power plant as claimed in claim 2, characterized in that, After S5 and before S6, it also includes: S7: Construct a multi-dimensional state safety function for each of the generator units based on the safety constraint parameters, the per-unit active power deviation, the per-unit speed deviation, and the distributed coordinated control command, and perform a safety projection correction on the distributed coordinated control command based on the multi-dimensional state safety function.
8. A method of automatic load regulation of a gas power plant according to claim 7, characterised in that, The multi-dimensional state safety function includes a valve position opening upper and lower limit safety function, a valve position change rate safety function, and a thermal state margin safety function; S7 specifically includes: S701: Based on the safety constraint parameters, the per-unit active power deviation, the per-unit speed deviation, and the distributed coordinated control command, construct the upper and lower limit safety functions of valve position opening, the safety function of valve position change rate, and the safety function of thermal state margin for each of the aforementioned units. S702: Based on the distributed coordinated control command, construct a control affine model to describe the relationship between fuel valve position changes and control commands; S703: Based on the valve position opening upper and lower limit safety function, the valve position change rate safety function, the thermal state margin safety function, and the control affine model, combined with the control obstacle function condition, determine the safety control set for each of the units; S704: Within the security control set, perform minimal intrusive security projection correction on the distributed coordination control command.
9. The method of automatic load regulation of a gas power plant as claimed in claim 1, characterized in that, S6 specifically includes: S601: Determine the fuel valve position command for each of the units based on the dispersed sliding modulus, the dispersed control order, and the dispersed coordinated control command; S602: Construct a sliding mode column vector based on the distributed sliding mode of each unit; S603: Construct the global convergence index based on the sliding modulus column vector; S604: Determine whether the unit load deviation has converged based on the global convergence index; if yes, use the fuel valve position command as the automatic load adjustment command and proceed to S605; otherwise, update the operating status information and return to S3. S605: Output the automatic load adjustment command to complete the automatic load adjustment of the gas-fired power plant.
10. A gas power plant load automatic regulating system, characterized in that, include: The acquisition module is used to acquire the operating status information of multiple units in a gas-fired power plant; A construction module is used to determine the neighbor set of each of the units based on the operating status information, and to construct the local grounding Laplace matrix of each of the units based on the neighbor set; The first determining module is used to determine the per-unit active power deviation and per-unit speed deviation of each of the units based on the operating status information, and to determine the distributed control order of each of the units in combination with the neighbor set; The second determining module is used to determine the distributed sliding modulus and distributed control gain of each of the generator units based on the local grounding Laplace matrix, the per-unit active power deviation, the per-unit speed deviation, and the distributed control order. The generation module is used to generate distributed coordinated control commands for each of the generator units based on the distributed sliding modulus and the distributed control gain. The judgment module is used to determine the fuel valve position command of each of the units according to the distributed coordination control command, and construct a global convergence index according to the distributed sliding modulus. The global convergence index is used to determine whether the unit load deviation has converged. If so, the fuel valve position command is used as the load automatic adjustment command; otherwise, the operating status information is updated and the execution of the first determination module is returned.
