A real-time scheduling method and system for production operation of a three-dimensional visualized gas turbine power plant

Abstract

The application discloses a three-dimensional visual gas turbine power plant production operation real-time scheduling method and system, and belongs to the technical field of intelligent scheduling of gas turbine power plants. The method establishes a full-plant three-dimensional virtual scene model containing equipment hierarchical topology, pipeline connection path and monitoring node spatial coordinates; collects load flow, temperature field distribution and pressure fluctuation time sequence through a real-time data monitoring unit; performs fluid dynamic balance analysis based on three-dimensional spatial coordinates on operation data to generate real-time load distribution coefficients, thermal efficiency compensation factors and pressure stability indicators; calls a steady-state coordinator to perform multi-objective constraint matching to generate an initial scheduling strategy sequence; maps the strategy to a three-dimensional model for visual simulation deduction to generate simulation operation trajectory data. The scheme realizes precise scheduling of production operation of a gas turbine power plant through spatial fluid dynamic balance analysis and three-dimensional simulation closed-loop correction.

Description

Technical Field

[0001] This invention relates to the field of gas turbine power plant scheduling technology, specifically a three-dimensional visualized real-time scheduling method and system for the production and operation of gas turbine power plants. Background Technology

[0002] During the operation of gas turbine power plants, it is necessary to dynamically adjust the load distribution, valve opening, and start-up / shutdown sequence of each equipment unit based on the real-time status of the units and the fluid conditions of the pipeline network to maintain the safe and efficient operation of the system. Existing scheduling methods mostly rely on distributed control systems to collect data from various measuring points, with operators issuing scheduling instructions based on preset performance curves or simplified lumped parameter models. Such methods do not consider the differences in pipeline friction resistance caused by the three-dimensional spatial layout of the power plant, the impact of temperature gradients between nodes on thermal efficiency, and the non-uniformity of pressure fluctuations propagating in space. The resulting load distribution and valve adjustments are difficult to match the actual dynamic equilibrium state of the fluid, often leading to problems such as local pipeline overpressure and accumulated thermal deviations.

[0003] Existing dispatching systems lack visualization and pre-execution methods for dispatching commands. After dispatching strategies are issued, it is impossible to observe their dynamic impact on the fluid flow, temperature distribution, and pressure levels of the entire plant's pipeline network in advance. Once the command contradicts the field constraints, it is very easy to trigger load overruns or pressure overshoots, and there is a significant lag in post-event correction. Existing technologies have also attempted to solve dispatching schemes using steady-state optimization algorithms, but they are usually based on single-objective or multi-objective static models and do not integrate the real-time fluid flow and thermodynamic state in the three-dimensional space of the gas turbine power plant. They cannot calculate the load distribution coefficient, thermal efficiency compensation factor, and pressure stability index that reflect spatial distribution differences based on the pipeline topology in real time, resulting in a coarse-grained dispatching strategy in the spatial dimension.

[0004] The challenges this solution addresses are: how to combine three-dimensional spatial coordinates with real-time operational data in a digital gas turbine power plant scenario to perform fluid dynamic balance analysis, obtain indicators characterizing the actual load proportion, thermal efficiency degradation, and pressure stability of each equipment unit, and generate scheduling strategies based on these indicators. Another challenge is how to dynamically simulate and extrapolate the original scheduling strategy in a three-dimensional virtual environment, quickly generating correction values ​​based on the extrapolated operational trajectory and safety constraint deviations, thus avoiding the direct issuance of scheduling commands that have not been verified in both spatial and temporal dimensions. Summary of the Invention

[0005] This invention provides a three-dimensional visualization method and system for real-time scheduling of gas turbine power plant production and operation. It aims to overcome the problem that existing scheduling methods lack three-dimensional spatial fluid dynamic balance analysis and visualization simulation deduction closed-loop correction. By establishing a three-dimensional virtual model of the entire plant and associating it with real-time data, it performs fluid dynamic balance analysis and multi-objective matching based on spatial coordinates. Combined with three-dimensional visualization simulation deduction and real-time deviation correction, it generates safe and economical scheduling instructions.

[0006] To achieve the above objectives, the present invention provides the following technical solution: The present invention provides a three-dimensional visualized real-time scheduling method for the production and operation of a gas turbine power plant, comprising: A three-dimensional virtual scene model of the entire gas turbine power plant is established. The three-dimensional virtual scene model of the entire plant includes the equipment hierarchical topology, pipeline connection paths and monitoring node spatial coordinates, thereby forming a digital mirror that is highly consistent with the physical power plant, providing a spatial visualization basis for scheduling decisions. The real-time operating data set of each equipment unit of the gas turbine power plant is collected at the current moment through the real-time data monitoring unit associated with the three-dimensional virtual scene model of the whole plant. The real-time operating data set includes load flow, temperature field distribution and pressure fluctuation time sequence to ensure the spatiotemporal consistency and integrity of the data. The real-time operating data set is subjected to fluid dynamic balance analysis based on three-dimensional spatial coordinates to generate real-time load distribution coefficients, thermal efficiency compensation factors and pressure stability indices for each equipment unit, so as to accurately characterize the thermal and hydraulic dynamic state of the entire plant's fluid network. The pre-configured steady-state coordinator is invoked to perform multi-objective constraint matching processing on the real-time load allocation coefficient, the thermal efficiency compensation factor, and the pressure stability index to generate an initial scheduling strategy sequence. The initial scheduling strategy sequence can coordinate the relationship between load demand, thermal efficiency, and pressure stability. The initial scheduling strategy sequence is mapped to the plant-wide 3D virtual scene model for 3D visualization simulation and deduction processing, generating simulation operation trajectory data that includes dynamic fluid flow trajectory, equipment color grading changes and valve opening animation. Potential over-limit risks are exposed in advance through pre-verification in the virtual environment. Based on the deviation analysis results between the simulated running trajectory data and the preset real-time scheduling constraints, the initial scheduling strategy sequence is corrected in real time to generate the final scheduling instruction and distribute it to the actuator of the gas turbine power plant, thereby realizing safe closed-loop scheduling of the production process and effectively preventing overload and pressure exceeding limits.

[0007] As a technical solution of the present invention, a three-dimensional virtual scene model of the entire gas turbine power plant is established, including: obtaining the construction engineering drawings and pipeline design specifications of the gas turbine power plant, extracting the geometric dimension parameters, pipeline direction coordinates, and connection node numbers of each equipment; constructing a proportional three-dimensional geometry of each equipment based on the geometric dimension parameters, and spatially assembling the proportional three-dimensional geometry according to the pipeline direction coordinates and connection node numbers to generate an initial three-dimensional scene framework; embedding the historical operating performance feature vectors and real-time monitoring point location labels of each equipment into the initial three-dimensional scene framework, wherein the real-time monitoring point location labels include the coordinates of temperature sensitive points, the installation coordinates of pressure transmitters, and the positioning coordinates of flow meters; and performing spatial coordinate normalization and texture mapping processing on the initial three-dimensional scene framework to generate a high-fidelity three-dimensional virtual scene model of the entire plant.

[0008] Furthermore, a real-time operational data set is collected through the real-time data monitoring unit associated with the plant's three-dimensional virtual scene model. This includes: using the load flow subunit of the real-time data monitoring unit to read the instantaneous flow values ​​of each main steam pipeline and feedwater pipeline according to a preset sampling period, generating a load flow data stream; using the temperature field subunit of the real-time data monitoring unit to obtain the real-time temperature values ​​of each temperature-sensitive point based on the monitoring point location labels, generating a temperature field distribution data matrix; using the pressure field subunit of the real-time data monitoring unit to obtain the pressure values ​​of each node based on the pressure transmitter installation coordinates, generating a pressure fluctuation time series data sequence; and performing timestamp alignment processing on the load flow data stream, the temperature field distribution data matrix, and the pressure fluctuation time series data sequence to obtain a synchronized real-time operational data set. Preferably, the preset sampling period is adaptively adjusted according to the fluctuation frequency of the load flow data stream. When the fluctuation frequency exceeds a set threshold, the sampling period is shortened to improve the tracking capability for rapid load changes.

[0009] Furthermore, the real-time operating data set is subjected to fluid dynamic balance analysis based on three-dimensional spatial coordinates, including: constructing a fluid network topology diagram in the three-dimensional virtual scene model of the entire plant, where the nodes of the fluid network topology diagram correspond to the inlets and outlets of each equipment unit, and the edges correspond to the connection paths of each pipeline; using the instantaneous flow rate value in the load flow data stream as the flow load parameter of each edge in the fluid network topology diagram, and calculating the real-time load distribution coefficient by traversing and searching the inflow and outflow flow difference of each node; using the temperature gradient in the temperature field distribution data matrix as the calculation input of the thermal efficiency compensation factor, and generating the thermal efficiency compensation factor based on the mapping relationship between the design thermal efficiency curve of each equipment unit and the temperature gradient; incorporating the pressure difference characteristics of adjacent sampling points in the pressure fluctuation time series data sequence into the analysis, and generating the pressure stability index by statistically analyzing the ratio of the standard deviation to the average value of the pressure fluctuation time series, thereby obtaining key parameters reflecting the dynamic balance of energy and working fluid in the entire plant.

[0010] Further, a pre-configured steady-state coordinator is invoked to perform multi-objective constraint matching processing, generating an initial scheduling strategy sequence, including: the initial scheduling strategy sequence containing the load adjustment amount, valve opening correction value, and unit start-up / shutdown priority identifier for each equipment unit; comparing the real-time load allocation coefficient, thermal efficiency compensation factor, and pressure stability index with the preset load allocation benchmark interval, thermal efficiency benchmark range, and pressure stability benchmark threshold stored in the steady-state coordinator to calculate the load deviation value, thermal efficiency deviation, and pressure fluctuation difference; constructing a three-dimensional constraint space based on the load deviation value, thermal efficiency deviation, and pressure fluctuation difference, searching for coordinate regions that satisfy the upper limit constraint of load, lower limit constraint of thermal efficiency, and upper limit constraint of pressure fluctuation in the three-dimensional constraint space to obtain the load adjustment amount; then calculating the valve opening correction value and the unit start-up / shutdown priority identifier based on the coupling correlation coefficient between the load adjustment amount and adjacent equipment units; combining the load adjustment amount, valve opening correction value, and unit start-up / shutdown priority identifier according to the time sequence to generate an initial scheduling strategy sequence for coordinating multiple objectives.

[0011] Furthermore, the initial scheduling strategy sequence is mapped to a plant-wide 3D virtual scene model for 3D visualization simulation and deduction processing. This includes: assigning each load adjustment, valve opening correction value, and unit start-up / shutdown priority identifier in the initial scheduling strategy sequence to the corresponding equipment unit virtual model and valve component virtual model in the plant-wide 3D virtual scene model; starting the fluid dynamic simulation engine in the plant-wide 3D virtual scene model, setting the simulation time step, iteratively calculating the operating state of the equipment unit virtual model and the valve component virtual model within the simulation time step, generating virtual flow values, virtual temperature values, and virtual pressure values ​​for each node; recording the virtual flow values, virtual temperature values, and virtual pressure values ​​of each node in chronological order as a virtual operating data stream, and mapping it back to the 3D spatial coordinate system of the plant-wide 3D virtual scene model to generate simulation operating trajectory data containing dynamic fluid flow trajectories, equipment color grading changes, and valve opening animations, thus intuitively displaying the global dynamic effect of the scheduling scheme in a 3D visualization manner.

[0012] Furthermore, the initial scheduling strategy sequence is corrected in real time based on the deviation analysis results, including: extracting the simulated load and simulated pressure values ​​of key equipment nodes from the simulated operation trajectory data; calculating the deviation between the simulated load value and the rated load upper limit value in the preset real-time scheduling constraints to obtain the load over-limit deviation; calculating the deviation between the simulated pressure value and the safe pressure threshold in the preset real-time scheduling constraints to obtain the pressure over-limit deviation; when the load over-limit deviation or the pressure over-limit deviation is greater than zero, calling the correction algorithm of the steady-state coordinator, taking the load over-limit deviation and the pressure over-limit deviation as input, and outputting the compensation correction value of the load adjustment and the compensation correction value of the valve opening; replacing the corresponding values ​​in the initial scheduling strategy sequence with the compensation correction values ​​to form the final scheduling command, and sending the final scheduling command to the combustion controller, turbine regulating valve, and feedwater regulating valve of the gas turbine power plant through the communication bus to realize closed-loop control from simulation verification to physical execution. Preferably, the correction algorithm uses a proportional-integral-derivative controller, which can quickly and effectively eliminate the over-limit deviations found in the simulation and improve the safety and feasibility of the final scheduling instructions.

[0013] This invention also provides a three-dimensional visualization real-time scheduling system for gas turbine power plant production operation, including a processor and a memory. The memory and the processor are connected. The memory stores program instructions, and the processor executes the program instructions to implement the three-dimensional visualization real-time scheduling method for gas turbine power plant production operation described in this invention. This system achieves dynamic three-dimensional monitoring and scheduling optimization of the gas turbine power plant's operating status through this method, ensuring the stability and efficiency of production under complex operating conditions.

[0014] The technical effects and advantages provided by the present invention in the above technical solution are as follows: By acquiring construction drawings and pipeline design specifications for gas turbine power plants, a 3D virtual scene model of the entire plant is constructed, including equipment hierarchical topology, pipeline connection paths, and spatial coordinates of monitoring nodes. Historical operating performance feature vectors and real-time monitoring point location labels are embedded in the model, precisely binding the coordinates of temperature-sensitive points, pressure transmitter installation coordinates, and flow meter positioning coordinates to the geometry. The real-time data monitoring unit reads the instantaneous flow values ​​of the main steam pipeline and feedwater pipeline according to a preset sampling period. Based on the monitoring point location labels, it obtains the temperature field distribution data matrix and pressure fluctuation time-series data sequence, which are then timestamped to form a real-time operating data set. Within the 3D spatial coordinate framework, this data set is mapped to a fluid network topology diagram composed of equipment inlet and outlet nodes and pipeline paths. The inflow and outflow differences at each node are traversed to calculate the real-time load distribution coefficient, reflecting the unevenness of flow distribution across branches in real time. Using the temperature gradient in the temperature field distribution matrix as input, a thermal efficiency compensation factor is generated based on the mapping relationship between the design thermal efficiency curve of each equipment unit and the temperature gradient, accurately quantifying the impact of spatial temperature changes on equipment thermal efficiency. The pressure stability index is generated by comparing the ratio of the standard deviation to the mean of the differences between adjacent sampling points in a time-series data of statistical pressure fluctuations. This index assesses the pressure stability of each node in space based on the severity of fluctuations. This index comprehensively describes the dynamic equilibrium state of the fluid in three-dimensional space, replacing traditional methods that rely on single-point experience or simplified parameters. It enables scheduling strategies to perceive differences in pipeline friction resistance, temperature gradients, and pressure propagation, providing a more physically accurate quantitative basis for subsequent load adjustments.

[0015] After generating the initial scheduling strategy sequence, the load adjustment, valve opening correction, and unit start-up / shutdown priority flags included in it are assigned one by one to the corresponding equipment unit virtual models and valve component virtual models in the plant's 3D virtual scene model. The fluid dynamic simulation engine is started, the simulation time step is set, and the virtual flow rate, virtual temperature, and virtual pressure values ​​of each node within the model are iteratively calculated within the step, forming a virtual operation data stream arranged by time. This data stream is mapped back to the 3D spatial coordinate system to generate simulation operation trajectory data containing dynamic fluid flow trajectories, equipment color grading changes, and valve opening animations. The simulation load and pressure values ​​of key equipment nodes are extracted from the simulation operation trajectory. The deviation between the simulation load value and the preset rated load upper limit is calculated to obtain the load over-limit deviation, and the deviation between the simulation pressure value and the safe pressure threshold is calculated to obtain the pressure over-limit deviation. When the deviation is greater than zero, the correction algorithm of the steady-state coordinator is invoked. Using the load over-limit deviation and pressure over-limit deviation as input, the algorithm outputs compensation correction values ​​for the load adjustment and valve opening. These values ​​replace the corresponding parts in the initial scheduling strategy sequence to form the final scheduling command, which is then distributed to the combustion controller, turbine regulating valve, and feedwater regulating valve via the communication bus. Using three-dimensional visualization simulation, the fluid response dynamics of the entire plant's pipeline network can be observed in advance before the actual execution of the scheduling command. Over-limit deviation feedback directly drives the correction of scheduling parameters, realizing the transformation from open-loop issuance to closed-loop pre-simulation correction, effectively avoiding local overload and pressure over-limit conditions caused by improper scheduling. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 It is a 3D visualized flowchart of a real-time scheduling method for production and operation of a gas turbine power plant; Figure 2 This is a spatial distribution diagram of the connection nodes of the main steam, reheat steam, and condensate pipelines in a gas turbine power plant. Figure 3 These are the real-time load distribution coefficient and pressure stability variation curves at key nodes of a gas turbine power plant. Figure 4 This is a time series graph of equipment load adjustment and valve opening correction values ​​in a gas turbine power plant. Figure 5 These are virtual temperature simulation curves for key equipment nodes; Figure 6 It is a dynamic correction curve for the load over-limit deviation and the pressure over-limit deviation. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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 some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] See Figure 1 This invention provides a three-dimensional visualization method for real-time scheduling of gas turbine power plant production operations, comprising: establishing a three-dimensional virtual scene model of the entire gas turbine power plant, the model including equipment hierarchical topology, pipeline connection paths, and spatial coordinates of monitoring nodes; collecting real-time operating data sets of each equipment unit of the gas turbine power plant at the current moment through real-time data monitoring units associated with the three-dimensional virtual scene model, the real-time operating data sets including load flow, temperature field distribution, and pressure fluctuation time series; performing fluid dynamic balance analysis processing based on three-dimensional spatial coordinates on the real-time operating data sets to generate real-time load allocation coefficients, thermal efficiency compensation factors, and pressure stability indices corresponding to each equipment unit; calling a pre-configured steady-state coordinator to perform multi-objective constraint matching processing on the real-time load allocation coefficients, thermal efficiency compensation factors, and pressure stability indices to generate an initial scheduling strategy sequence; mapping the initial scheduling strategy sequence to the three-dimensional virtual scene model of the entire plant for three-dimensional visualization simulation and deduction processing to generate simulation operation trajectory data; and performing real-time correction processing on the initial scheduling strategy sequence based on the deviation analysis results between the simulation operation trajectory data and the preset real-time scheduling constraints to generate final scheduling instructions and distribute them to the execution mechanism of the gas turbine power plant.

[0020] Example 1:

[0021] In practice, the construction drawings and pipeline design specifications of the gas turbine power plant are obtained. The construction drawings can be CAD files containing plan views, elevation views, and section views, while the pipeline design specifications can be Piping and Instrumentation Diagrams (P&IDs) and pipeline isometric drawings. By parsing the layers and recognizing the elements in the CAD files, the outline dimensions of each piece of equipment are extracted to obtain geometric dimensional parameters. The coordinates of the start point, end point, and inflection point of each pipe segment are parsed from the pipeline isometric drawings, and the connection node numbers of valves, flanges, tees, and other pipe fittings are extracted from the P&IDs to obtain the pipeline route coordinates and connection node numbers.

[0022] Based on geometric dimension parameters, proportional 3D geometries of each device are constructed. In specific implementation, 3D modeling kernel libraries (such as ACIS or Parasolid) can be used to generate corresponding 3D solid models for devices such as gas turbines, waste heat boilers, steam turbines, and condensers according to the extracted geometric dimension parameters. For example, for cylindrical devices, the outer wall structure is generated using parametric modeling of cylinders; for pipes with bends and diameter changes, the pipe geometry is generated by sweeping cross-sections along the path. The proportional 3D geometries of each device are then spatially assembled according to the pipeline direction coordinates and connection node numbers: local coordinate systems are created at the connection node numbers of all device 3D geometries, and the start and end points of the pipelines are constrained to the corresponding device connection node coordinate systems. At the same time, intermediate control points are generated according to the pipeline direction coordinates to form 3D paths connecting pipe segments, thereby generating the initial 3D scene framework of the entire plant's pipelines and equipment interconnected in space.

[0023] Historical operating performance feature vectors and real-time monitoring point location tags for each device are embedded into the initial 3D scene framework. The historical operating performance feature vectors are obtained from the power plant's historical operating database (such as the historical stations of the plant-level monitoring information system SIS), specifically including the average heat rate of the equipment over a complete operating cycle, the load rate corresponding to the highest efficiency point, and the parameters of the exhaust temperature-load relationship curve. These multi-dimensional data are used as a feature vector and attribute-bound to the corresponding 3D geometry of the equipment. The real-time monitoring point location tags are embedded as follows: temperature-sensitive point coordinates, pressure transmitter installation coordinates, and flow meter positioning coordinates are extracted from the field point cloud obtained from the instrumentation and control as-built drawings or a 3D laser scanner. Visualized monitoring point markers (such as small balls or arrows) are generated at the corresponding 3D coordinate positions in the initial 3D scene framework. Simultaneously, the equipment identifier, measurement point type, channel address, and other information for each monitoring point are stored in the extended attributes of this marker, ensuring that the real-time monitoring point location tags include temperature-sensitive point coordinates, pressure transmitter installation coordinates, and flow meter positioning coordinates.

[0024] The initial 3D scene framework undergoes spatial coordinate normalization and texture mapping to generate a plant-wide 3D virtual scene model. Spatial coordinate normalization transforms the coordinates of all geometric objects, pipelines, and monitoring points from the original drawing coordinate system to a unified plant-wide coordinate system with the turbine generator foundation center as the origin, east as the X-axis, vertically upward as the Y-axis, and north as the Z-axis. During the transformation, the unit of length is standardized to meters. Texture mapping uses on-site photographs of equipment and pipeline surfaces, employing UV mapping technology to attach the images to the corresponding 3D geometric surfaces. Different colors are assigned to pipelines in different systems (such as main steam, reheat steam, and condensate) for differentiation. The generated plant-wide 3D virtual scene model includes the equipment hierarchical topology, pipeline connection paths, and the spatial coordinates of monitoring nodes.

[0025] The real-time data collection unit, linked to a 3D virtual scene model of the entire plant, acquires real-time operating data from each equipment unit of the gas turbine power plant at the current moment. The load flow subunit within the real-time data monitoring unit connects to field flow transmitters or the DCS system via the OPCUA communication protocol, reading the instantaneous flow values ​​of each main steam pipeline and feedwater pipeline according to a preset sampling period. During data acquisition, the load flow subunit maintains a list of acquisition points, recording the pipeline name, point number, upper and lower limits of the measurement range, and unit of measurement for each point. The load flow subunit cyclically accesses the point addresses in the list, reads the analog values, and appends a GPS or NTP synchronized acquisition timestamp, generating a load flow data stream arranged in chronological order.

[0026] The temperature field subunit in the real-time data monitoring unit obtains the real-time temperature values ​​of each temperature-sensitive point based on the real-time monitoring point location tags. The temperature field subunit iterates through the monitoring point marker attributes in the plant-wide 3D virtual scene model to extract the coordinates and corresponding DCS tag names of all points of type "temperature-sensitive point." It then reads the current value of the corresponding tag from the DCS temperature acquisition card via MODBUSTCP or OPCDA protocol. The temperature value is in degrees Celsius, rounded to one decimal place. All the read temperature values ​​constitute a temperature field distribution data matrix with the measurement point identifier as the row index and the acquisition time as the column index.

[0027] The pressure field subunit in the real-time data monitoring unit obtains the pressure values ​​of each node based on the installation coordinates of the pressure transmitters. Similarly, the pressure field subunit obtains the tag name based on the pressure transmitter installation coordinates in the monitoring point attributes and reads the pressure value from the DCS. The pressure value unit is kilopascals, forming a one-dimensional pressure fluctuation time-series data sequence that increases with the sampling time. The load flow data stream, temperature field distribution data matrix, and pressure fluctuation time-series data sequence are then timestamped to generate a real-time operating data set. The specific implementation of the timestamp alignment process is as follows: the sampling time of the load flow data stream is selected as the alignment reference time sequence. For the temperature field distribution data matrix, if the difference between the acquisition time of the temperature measurement point and the reference time is less than a preset tolerance (e.g., 0.1 seconds), the temperature value at that acquisition time is directly used as the alignment value; otherwise, the estimated temperature value at the reference time is calculated using two consecutive temperature sampling values ​​through linear interpolation, forming the aligned temperature field distribution data matrix. For the pressure fluctuation time-series data sequence, the same nearest neighbor matching or linear interpolation method is used to obtain the pressure value aligned with the reference time. The aligned load flow data stream, temperature field distribution data matrix, and pressure fluctuation time series data sequence are combined into a multi-dimensional time series. Each row represents a moment in time and contains the flow vector, temperature matrix, and pressure vector at that moment. The set of all rows constitutes the real-time running data set.

[0028] In some embodiments, the preset sampling period of the load flow subunit is adaptively adjusted according to the fluctuation frequency of the load flow data stream. The fluctuation frequency is characterized by the rate of change of the standard deviation of the load flow data stream over a sliding window (e.g., 10 seconds) in real time. When the rate of change of the standard deviation exceeds a set threshold (e.g., the change per second exceeds 5% of the rated flow), the preset sampling period is shortened from the base value of 1 second to 0.2 seconds; when the rate of change of the standard deviation is below one-tenth of the threshold for several consecutive periods, the sampling period is restored to the base value of 1 second.

[0029] See Figure 2 The figure shows the spatial distribution of pipeline connection nodes for different systems in a 3D virtual scene model of the entire gas turbine power plant. The horizontal axis is the X-coordinate, in meters, ranging from approximately 0 to 55 meters; the vertical axis is the Y-coordinate, in meters, ranging from approximately 0 to 14 meters. The legend identifies three types of pipeline connection nodes: main steam pipeline connection nodes are represented by blue dots, reheat steam pipeline connection nodes by orange squares, and condensate pipeline connection nodes by green triangles.

[0030] The main steam pipeline connection nodes extend from approximately 0 meters to 50 meters along the X-axis, while the Y-coordinate remains generally within the range of 7 to 13 meters, exhibiting a slow, fluctuating trend of first rising, then falling, and then rising again. This reflects the spatial hierarchical topology and pipeline routing characteristics of the main steam pipeline. The X-coordinate distribution of the reheat steam pipeline connection nodes is similar to that of the main steam pipeline, ranging from approximately 5 to 45 meters. However, its Y-coordinate shows a non-linear change, first decreasing to a low of approximately 1 meter and then rising to approximately 8 meters, indicating a significant tortuous path and spatial reversal in the reheat steam pipeline. The condensate pipeline connection nodes are mainly distributed within the X-axis range of 0 to 55 meters, with the Y-coordinate linearly rising from approximately 1 meter to approximately 9 meters, indicating a steadily rising spatial layout characteristic of the condensate pipeline. There are intersection areas between the three types of pipeline connection nodes within the X-axis range of 5 to 40 meters. Particularly at approximately 35 meters on the X-axis, the Y-coordinates of the main steam and reheat steam pipeline nodes converge, indicating spatial overlap of key connection nodes.

[0031] Example 2:

[0032] In practical implementation, a fluid network topology diagram is constructed within the 3D virtual scene model of the entire plant. Nodes in the fluid network topology diagram correspond to the inlets and outlets of each equipment unit. Nodes are stored in a data structure as records containing a unique node identifier (e.g., "GT_IN_01" represents gas turbine inlet 1), the equipment number to which the node belongs, and the node's spatial coordinates (extracted from the flange locations of the equipment models in the 3D virtual scene model of the entire plant). Edges in the fluid network topology diagram correspond to the connection paths of each pipeline. The edge data structure includes an edge identifier, a starting node identifier, an ending node identifier, the pipe segment inner diameter (extracted from the pipeline design specification document), the pipe segment length (calculated based on the Euclidean distance between the spatial coordinates of the two endpoints), and the pipe segment friction resistance coefficient (extracted from the pipeline design specification document). The fluid network topology diagram is stored using an adjacency list approach. Each node maintains a linked list of edges originating from it, thus comprehensively describing the flowable paths of the working fluid among all equipment in the plant.

[0033] The instantaneous flow rate values ​​in the load flow data stream are used as the flow load parameters for each edge in the fluid network topology diagram. In specific implementation, each instantaneous flow rate value in the load flow data stream carries a measurement point identifier, which has a pre-configured mapping relationship with the identifier of a certain edge in the fluid network topology diagram. In each processing cycle, the latest instantaneous flow rate value of each measurement point is assigned to the flow load parameter field of the corresponding edge. For edges without installed flow measurement points, the flow load parameters are calculated using the flow load parameters of adjacent edges with measurement points, based on the mass conservation relationship of the pipe segment where the edge is located: when the flow load parameters of all inflow edges of a node in the fluid network topology diagram are known, the flow load parameters of each outflow edge of that node are allocated according to the proportion of the pipe segment cross-sectional area of ​​the outflow edge.

[0034] The real-time load distribution coefficient is calculated by traversing and searching the inflow and outflow differences of each node in the fluid network topology. For the [node name missing] node in the fluid network topology... Each node defines the inflow node. The sum of the flow load parameters of all edges is Define outflow nodes The sum of the flow load parameters of all edges is Real-time load distribution factor The calculation formula is:

[0035] in, For nodes The real-time load distribution coefficient is dimensionless and ranges from -1 to 1. A negative value indicates that the equipment unit where the node is located is in a heat storage or energy storage state, while a positive value indicates that the equipment unit where the node is located is in an additional energy release state. For inflow node Total flow rate, expressed in kilograms per second; outflow node Total flow rate, expressed in kilograms per second; It is a minimal positive constant to prevent the denominator from being zero, set as kilograms per second; Indicates taking and The maximum value in.

[0036] The temperature gradient in the temperature field distribution data matrix is ​​used as the input for calculating the thermal efficiency compensation factor. In specific implementation, one element in the temperature field distribution data matrix represents the temperature-sensitive point at the [missing element] position. Temperature values ​​at each sampling time (Unit: degrees Celsius). The temperature gradient is calculated by selecting two adjacent temperature-sensitive points on the same equipment unit. The temperature difference at each sampling time is calculated by dividing the spatial distance between the two points. For the first... The temperature gradient along the working fluid flow direction of each equipment unit is denoted as . The unit is degrees Celsius per meter.

[0037] A thermal efficiency compensation factor is generated based on the mapping relationship between the design thermal efficiency curve and temperature gradient of each equipment unit. In specific implementation, the factor is obtained from the design documents provided by the equipment manufacturer. The design thermal efficiency curve for each equipment unit is expressed as the thermal efficiency under rated operating conditions. A table showing the relationship between intake air temperature and load rate is created and stored as a two-dimensional lookup table. When generating the thermal efficiency compensation factor, the first step is to... For each equipment unit, the design thermal efficiency value corresponding to the intake temperature sensitive point is obtained by linear interpolation from a two-dimensional lookup table at that temperature value. Then, based on the temperature gradient... With respect to the preset temperature gradient standard limit value (Compare with the maximum allowable temperature gradient provided by the equipment manufacturer, such as 20 degrees Celsius per meter): When Absolute value less than or equal to At that time, thermal efficiency compensation factor Set to 1.0; when Absolute value greater than At that time, thermal efficiency compensation factor according to Calculate the thermal efficiency compensation factor. The lower limit is truncated to 0.7. is a dimensionless coefficient with a value range of [0.7, 1.0].

[0038] This study correlates the pressure difference between adjacent sampling points in a pressure fluctuation time series with pressure stability indices. The pressure fluctuation time series is an array of pressure values ​​arranged in chronological order. ,in For sequence length, Indicates the first The pressure values ​​at each sampling point are expressed in kilopascals (kPa). The method for calculating the pressure difference between adjacent sampling points is as follows: ,in The value range is from 1 to .

[0039] A pressure stability index is generated by calculating the ratio of the standard deviation to the mean of a time-series pressure fluctuation data series. In practice, this involves first calculating the pressure values ​​in the time-series pressure fluctuation data series. average The unit is kilopascal; then calculate the standard deviation of the pressure fluctuation time series data. The unit is kilopascal (kPa), and the calculation formula is: Pressure stability index Defined as the absolute value of the ratio of the standard deviation to the mean multiplied by 100%, i.e. . This is a dimensionless percentage, with values ​​greater than or equal to 0. The closer to 0, the more stable the pressure.

[0040] In some embodiments, when the working fluid in the pipeline corresponding to the pressure measurement point is a compressible gas, the pressure stability index is calculated using data within a sliding window. The window length is the most recent 60 sampling points (corresponding to a 60-second duration). Each time a new sampling point is added, the window slides forward once, thereby continuously generating a pressure stability index that updates over time.

[0041] See Figure 3 In the figure, the horizontal axis represents the sampling time (seconds), the left side of the vertical axis represents the real-time load distribution coefficient (dimensionless), and the right side of the vertical axis represents the pressure stability index (percentage). The figure shows the real-time load distribution coefficient curves of the gas turbine inlet node, steam turbine inlet node, and waste heat boiler inlet node, as well as the pressure stability index change curves. All curves change dynamically over time.

[0042] Within the sampling time range of 0 to 200 seconds, the real-time load distribution coefficient of the gas turbine inlet node remained at approximately 0.70, while the load distribution coefficients of the steam turbine inlet and waste heat boiler inlet nodes remained at approximately 0.65 and 0.60, respectively. The pressure stability index remained stable between approximately 1.8% and 2.1%, showing a stable state, indicating that the equipment load distribution and pressure fluctuations were within the normal operating range.

[0043] At approximately 200 seconds, all three real-time load allocation coefficient curves showed a significant upward trend. The load allocation coefficient at the gas turbine inlet node rapidly increased from approximately 0.70 to approximately 0.85, the steam turbine inlet node from approximately 0.65 to approximately 0.75, and the waste heat boiler inlet node from approximately 0.60 to approximately 0.73. This increase in load allocation coefficients during this phase reflects that the equipment load adjustment was executed as expected, and the overall load allocation level was improved.

[0044] The pressure stability index rose synchronously after 200 seconds, rapidly increasing from about 2% to a peak of about 3.4%, and then fluctuated between 3.2% and 3.4% between 300 and 500 seconds, indicating that the system pressure fluctuation amplitude increased, but still remained below the baseline threshold of 5%, indicating that although the pressure stability was affected, it had not exceeded the allowable range.

[0045] Example 3:

[0046] In practical implementation, the initial scheduling strategy sequence includes the load adjustment amount, valve opening correction value, and unit start-up / shutdown priority identifier for each equipment unit. The steady-state coordinator is a pre-configured multivariate coordinated control module. Internally, the steady-state coordinator stores preset load allocation reference ranges, thermal efficiency reference ranges, and pressure stability reference thresholds. The preset load allocation reference range is the operating range that the load allocation coefficient should maintain for each equipment unit. For example, for a gas turbine, the preset load allocation reference range is [0.6, ...]. [0.9]; The thermal efficiency benchmark range is the allowable range of the thermal efficiency compensation factor set for each equipment unit, with a lower limit of 0.85; the pressure stability benchmark threshold is the upper limit that the pressure stability index should not exceed, set to 5%. The preset load distribution benchmark interval, thermal efficiency benchmark range, and pressure stability benchmark threshold are obtained by looking up the benchmark parameter table compiled from the equipment design conditions and operating procedures, and stored in the non-volatile memory of the steady-state coordinator.

[0047] The real-time load allocation coefficient, thermal efficiency compensation factor, and pressure stability index are compared with the preset load allocation reference range, thermal efficiency reference range, and pressure stability reference threshold stored in the steady-state coordinator, respectively. For the k-th equipment unit, the real-time load allocation coefficient is denoted as... The preset load distribution benchmark interval is denoted as ,like Within the range, the load deviation value Set to 0; if Below Then the load deviation value ;like Higher than Then the load deviation value The thermal efficiency compensation factor is denoted as... (From Example 2), the lower limit of the thermal efficiency reference range is denoted as... thermal efficiency deviation Calculated as ,when A positive value indicates that the thermal efficiency compensation factor is below the lower limit of the reference value, requiring compensation. Pressure stability is indicated by... (Originally obtained from Example 2), the pressure stability reference threshold is denoted as... Pressure fluctuation difference Calculated as ,when A positive value indicates that the pressure fluctuation exceeds the allowable limit.

[0048] Based on load deviation value thermal efficiency deviation and pressure fluctuation difference A three-dimensional constrained space is constructed. The three coordinate axes of the three-dimensional constrained space represent the load adjustment, thermal efficiency improvement, and pressure fluctuation suppression, respectively, but the deviation is directly used as the coordinate. In actual processing, the load deviation value is... thermal efficiency deviation and pressure fluctuation difference As a starting point in three-dimensional space, the load ceiling constraint stipulates that the load adjustment amount must not exceed the maximum allowable increment. Its value is determined based on the difference between the equipment's rated capacity and its current output; the lower limit constraint on thermal efficiency stipulates that the improvement in thermal efficiency must reach at least [a certain percentage]. When it is a positive value The adjustment ensures that the thermal efficiency compensation factor returns to the baseline range; the upper limit constraint on pressure fluctuation stipulates that the pressure fluctuation difference after adjustment must be less than or equal to 0. These three constraints form three half-spaces in three-dimensional space, and the intersection of the three half-spaces constitutes a feasible polyhedral region.

[0049] The load adjustment is obtained by searching for coordinate regions in a three-dimensional constrained space that satisfy the upper limit of load constraints, the lower limit of thermal efficiency constraints, and the upper limit of pressure fluctuation constraints. The search process employs a step-by-step local search method: starting from the initial point... Starting from the beginning, move in steps of 0.01 in the directions of decreasing load adjustment, increasing thermal efficiency improvement, and decreasing pressure fluctuation difference. After each move, check whether the new point falls within the feasible polyhedron region. Select the point that minimizes the overall Euclidean distance within the feasible polyhedron region, and use the load adjustment coordinates corresponding to that point as the load adjustment amount. .

[0050] Based on load adjustment amount The coupling correlation coefficient between adjacent equipment units is used to calculate the valve opening correction value and the unit start-up / shutdown priority identifier. The coupling correlation coefficient between adjacent equipment units is derived from the fluid network topology diagram in the plant-wide 3D virtual scene model. Specifically, for the k-th equipment unit and the j-th equipment unit, the coupling correlation coefficient... Defined as the reciprocal of the pipe segment resistance coefficient of the shortest path from the outlet node of the k-th equipment unit to the inlet node of the j-th equipment unit, and normalized to make... Valve opening correction value The calculation method is as follows: ,in The gain coefficient of the valve opening of the k-th equipment unit with respect to the load is obtained by looking up a table from the flow characteristic curve of the corresponding regulating valve of the equipment unit; simultaneously, a coupling correction is applied to the valve opening of each adjacent j-th equipment unit, with the correction amount being... Finally, the valve opening correction value of the j-th device unit is the sum of the above coupling correction value and the valve opening correction value of the j-th device unit itself.

[0051] The unit start-up / shutdown priority identifier is an integer value, indicating the order in which each equipment unit is started or stopped during scheduling. When the load adjustment... When the value is positive and exceeds 10% of the rated capacity of the equipment unit, if the equipment unit is currently shut down, the unit start / stop priority flag is set to 1 (priority start); when the load adjustment amount When the value is negative and its absolute value exceeds 10% of the rated capacity, if the equipment unit is currently in operation, the unit start-stop priority flag is set to -1 (priority shutdown); otherwise, the unit start-stop priority flag is set to 0 (maintain status). The priority flags of multiple equipment units can be combined into a one-dimensional array.

[0052] Load adjustment amount Valve opening correction value The unit start-up and shutdown priority identifiers are combined according to the time sequence to generate an initial scheduling strategy sequence. The time sequence starts from the current time and takes a preset control cycle (e.g., 1 minute) as the step size to generate the initial scheduling strategy sequence for the next 10 cycles. The strategy vector for each cycle includes the load adjustment amount, valve opening correction value and start-up and shutdown priority identifier of all equipment units. The strategy vectors of different cycles are arranged in chronological order to form a complete initial scheduling strategy sequence.

[0053] See Figure 4The graph uses future time (in minutes) as the horizontal axis, the left vertical axis as a percentage representing load adjustment, and the right vertical axis as a degree representing valve opening correction. The curves include load adjustments (solid lines) and corresponding valve opening correction values ​​(dashed lines) for three types of equipment units: gas turbines, waste heat boilers, and steam turbines. The load adjustment ranges from -6% to 10%, and the valve opening correction ranges from -1° to 1.5°.

[0054] Within the time interval of 0 to 10 minutes, the load adjustment and valve opening correction values ​​showed a clear phased trend. In the first minute, the load adjustment for all three types of equipment increased from 0 to approximately 5% to 8%, with the valve opening correction values ​​increasing simultaneously, reflecting the initial load increase. In the second and third minutes, the load adjustment and valve opening correction values ​​remained stable at a relatively high level. In the fourth and fifth minutes, the load adjustment decreased significantly to approximately -4% to -6%, with the valve opening correction values ​​decreasing accordingly, reflecting the load reduction adjustment by the dispatching strategy. In the sixth and seventh minutes, the load increased again, with the load adjustments for the gas turbine, waste heat boiler, and steam turbine rising back to the 3% to 7% range, and the valve opening correction values ​​changing upwards simultaneously. In the eighth minute, a brief negative peak appeared in the load adjustment, with the gas turbine load adjustment dropping to approximately -3%, and the waste heat boiler and steam turbine showing corresponding negative adjustments, with the valve opening correction values ​​also decreasing accordingly. Subsequently, within 8 to 9 minutes, the load adjustment significantly increased to approximately 10%, 8%, and 6%, respectively, and the valve opening correction value reached its maximum positive range, reflecting the dispatch response to the rapid load increase. By 10 minutes, the load adjustment and valve opening correction values ​​had fallen back into the negative range, completing one load adjustment cycle.

[0055] Example 4:

[0056] In practice, each load adjustment, valve opening correction value, and unit start-stop priority identifier in the initial scheduling strategy sequence is assigned to the corresponding equipment unit virtual model and valve component virtual model in the plant-wide 3D virtual scene model. The plant-wide 3D virtual scene model maintains an object management list. Each equipment unit virtual model object in the list has an equipment identifier field. Each load adjustment in the initial scheduling strategy sequence carries a target equipment unit identifier. Through identifier matching, the load adjustment value is written to the "current load bias" attribute field of the equipment unit virtual model object. Similarly, the valve component virtual model object has a valve identifier field. Each valve opening correction value in the initial scheduling strategy sequence is written to the "opening correction" attribute field of the valve component virtual model object through valve identifier matching. This opening correction is expressed in degrees, with positive values ​​indicating an increased opening direction. The unit start-stop priority identifier is written to the "start-stop state machine" control field of the equipment unit virtual model object. When the identifier is 1, the state machine enters the start-up preparation state; when the identifier is -1, it enters the shutdown preparation state; and when the identifier is 0, the current operating state remains unchanged.

[0057] Start the fluid dynamics simulation engine in the plant's 3D virtual scene model. The fluid dynamics simulation engine is a real-time hydraulic-thermal coupling solver based on the node-branch method, with its kernel employing an alternating direction implicit iterative format. At startup, set the simulation time step of the fluid dynamics simulation engine. The simulation time step is 0.1 seconds. The basis for setting the value to 0.1 seconds is that the working fluid velocity in the main steam pipeline of a gas turbine power plant does not exceed 60 meters per second, and the pressure wave propagation distance within 0.1 seconds is about 6 meters, which is smaller than the typical pipe section grid size (about 10 to 20 meters), thus satisfying the numerical stability requirement that the Courant condition number is less than 1.

[0058] The operating states of the virtual models of equipment units and valve components are iteratively calculated within the simulation time step. Starting from the simulation start time... Initially, at each simulation time step, the fluid dynamics simulation engine traverses all nodes and edges in the fluid network topology. For each edge containing a virtual valve component model, a virtual flow rate is calculated based on its current actual opening. The current actual opening of the valve component virtual model is obtained by superimposing its original opening base value with an opening correction amount. The opening base value is obtained from the valve feedback signal in the distributed control system. Virtual flow rate value. The calculation formula is:

[0059] in, The virtual flow rate value is expressed in kilograms per second through the edge of the virtual model of the valve assembly. The valve flow coefficient is dimensionless and obtained by linear interpolation from the flow coefficient table provided by the valve manufacturer, based on the valve type and diameter. The value ranges from 0.6 to 1.2; To match the current actual opening angle The corresponding effective circulation area, in square meters, is the current actual opening angle. This is the algebraic sum of the valve's original opening angle and the opening correction amount. Through valve characteristic curve function Calculation, where This represents the maximum flow area when the valve is fully open. This is the virtual pressure value of the upstream node of the valve, in Pascals; This is the virtual pressure value at the downstream node of the valve, in Pascals. This indicates taking the maximum value between the pressure difference and 0, avoiding invalid values ​​in the square root operation under negative pressure differences.

[0060] For pipe sections without valves, the virtual flow rate is determined by the pressure difference between adjacent nodes and the pipe resistance characteristics, and is approximated using a linearized Darcy-Weisbach formula. For each node corresponding to a virtual unit model, its virtual temperature is iteratively solved using the energy balance equation based on the current load level and upstream / downstream fluid enthalpy of the virtual unit model. The virtual unit model encapsulates a simplified thermodynamic model that receives the inlet virtual flow rate and inlet virtual temperature, adjusts the internal combustion intensity or heat transfer coefficient according to the current load adjustment, and calculates the outlet virtual temperature. The virtual pressure is updated by solving a system of simultaneous equations for node mass and momentum conservation; the relationship between node pressure and the inflow / outflow difference is iteratively performed using a compressibility model. Within each simulation time step, the hydraulic and thermodynamic calculations are iteratively performed until the maximum relative change in the virtual flow rate, virtual temperature, and virtual pressure values ​​at each node is less than [a certain value]. This completes the convergence calculation for one step.

[0061] The virtual flow rate, virtual temperature, and virtual pressure values ​​of each node are recorded in chronological order. A circular buffer is set up within the fluid dynamics simulation engine, with the number of rows equal to the total number of simulation steps. Each row records the calculation results of all nodes at a given simulation time. After convergence at each time step, the current time step is... All node identifiers, their corresponding virtual flow values, virtual temperature values, and virtual pressure values ​​are sequentially written into the buffer in the form of a structured array, forming a virtual running data stream that increases over time.

[0062] The virtual operation data stream is mapped back to its corresponding position in the 3D spatial coordinate system of the plant-wide 3D virtual scene model. Each node in the plant-wide 3D virtual scene model has pre-stored spatial coordinates. The virtual flow rate, virtual temperature, and virtual pressure values ​​of each node in each time frame of the virtual operation data stream are matched to the 3D spatial coordinate points using node identifiers. For the generation of dynamic fluid flow trajectories, particle system technology is used: tracer particles are released at preset intervals along the pipe segment axis at the starting point of each pipe segment. The particle's movement speed is proportional to the virtual flow rate value of that pipe segment at the current moment divided by the pipe segment's cross-sectional area. The particle position is updated with each frame, and the particle positions of consecutive frames are connected to form a flow trajectory line. For equipment color grading changes, a predefined color mapping function is used: the outer surface color of the equipment unit's virtual model gradually changes from blue (low temperature) to red (high temperature) according to the virtual temperature value of its outlet node, with a temperature grading interval of 10 degrees Celsius. If the virtual pressure value of a certain equipment unit exceeds a safety threshold, the outer surface color of the equipment is superimposed with a flashing effect. For valve opening animation, the rotation angle of the valve core component around the rotation axis in the virtual model of the valve assembly is directly derived from the current actual opening angle. The valve core is driven to rotate in real time. The dynamic fluid flow trajectory, equipment color grading changes, and valve opening animation together constitute the simulation operation trajectory data that is updated over time in a three-dimensional scene.

[0063] See Figure 5 In the graph, the horizontal axis represents the simulation time in seconds, ranging from 0 to 50 seconds; the vertical axis represents the virtual temperature value in degrees Celsius, ranging from approximately 0 to 500 degrees Celsius. In the legend, the solid red line represents the virtual temperature value at the gas turbine outlet, the dashed blue line represents the virtual temperature value at the steam turbine inlet, and the dotted green line represents the virtual temperature value of the high-pressure steam drum of the waste heat boiler.

[0064] As the simulation time increased, all three curves showed a gradual upward trend, exhibiting typical exponential heating characteristics. In the initial stage of the simulation (0-10 seconds), the virtual temperature value at the gas turbine outlet rose the fastest, rapidly increasing from approximately 30 degrees Celsius to around 300 degrees Celsius; the virtual temperature value at the steam turbine inlet followed, rising from 30 degrees Celsius to approximately 260 degrees Celsius; and the virtual temperature value of the high-pressure steam drum of the waste heat boiler rose the slowest, reaching approximately 150 degrees Celsius. This stage reflects the progressive process of the thermal response during the initial stage of equipment load adjustment.

[0065] During the intermediate stage (10-30 seconds), the three curves continued to rise slowly. The gas turbine outlet temperature gradually approached 450 degrees Celsius, the steam turbine inlet temperature approached 350 degrees Celsius, and the waste heat boiler high-pressure steam drum temperature approached 250 degrees Celsius. The rate of temperature increase slowed down significantly, indicating that the system was gradually reaching steady-state operation.

[0066] In the final stage (30-50 seconds), the three virtual temperature curves tend to stabilize with small fluctuations. The gas turbine outlet temperature eventually stabilizes at around 490 degrees Celsius, the steam turbine inlet temperature stabilizes at around 380 degrees Celsius, and the waste heat boiler high-pressure steam drum temperature stabilizes at around 280 degrees Celsius. This indicates that after iterative calculations by the fluid dynamics simulation engine, the temperature values ​​of the virtual models of the equipment units have converged to a reasonable and stable state.

[0067] Example 5:

[0068] In practice, the simulated load and pressure values ​​of key equipment nodes are extracted from the simulated operation trajectory data. The simulated operation trajectory data is a multidimensional array organized by simulation time frames. Each frame contains the virtual flow rate, virtual temperature, and virtual pressure values ​​of all nodes. Key equipment nodes are pre-marked monitoring nodes in the plant-wide 3D virtual scene model, covering the gas turbine outlet, steam turbine inlet, waste heat boiler high-pressure steam drum, and main steam header convergence point. The extraction operation involves traversing each simulation time frame of the simulated operation trajectory data, using the key equipment node identifier as an index, and reading the corresponding virtual flow rate value as the simulated load value and the corresponding virtual pressure value as the simulated pressure value from that frame of data. The unit for the simulated load value is kilograms per second, and the unit for the simulated pressure value is kilopascals.

[0069] The deviation between the simulated load value and the rated load upper limit value in the preset real-time scheduling constraints is calculated. The preset real-time scheduling constraints are stored in a constraint parameter configuration file, which is in Extensible Markup Language (EXPLAIN) format and contains the rated load upper limit value and safe pressure threshold for each critical equipment node. The rated load upper limit value is extracted from the equipment nameplate parameters and converted from the main steam flow rate corresponding to the maximum continuous output provided by the equipment manufacturer. For the b-th critical equipment node, its simulated load value is denoted as... Record its rated load upper limit as Overload deviation The calculation method is as follows: If ,but ;like ,but .

[0070] The deviation between the simulated pressure value and the preset safe pressure threshold in the real-time scheduling constraints is calculated. For the b-th critical equipment node, its simulated pressure value is denoted as... Let its safe pressure threshold be . The safe pressure threshold is taken as 90% of the equipment design pressure and is determined by 0.9 times the water pressure test pressure in the equipment design specifications. Pressure exceeding the limit deviation... The calculation method is as follows: If ,but ;like ,but .

[0071] When the load exceeds the deviation Or pressure exceeding the limit deviation When the value is greater than zero, the correction algorithm of the steady-state coordinator is invoked. Load over-limit deviation. Or pressure exceeding the limit deviation The condition of being greater than zero is determined by a logical OR operation; as long as any deviation is positive, the correction process is triggered. The correction algorithm uses a proportional-integral-derivative (PID) controller. The core architecture of the PID controller consists of a proportional channel, an integral channel, and a derivative channel connected in parallel. The sum of the outputs of the three channels forms the total output of the controller. The input to the proportional channel is the deviation itself, the input to the integral channel is the cumulative integral of the deviation with respect to time, and the input to the derivative channel is the derivative of the deviation with respect to time.

[0072] The load over-limit deviation and pressure over-limit deviation are used as inputs to the proportional-integral-derivative (PID) controller. The input to the PID controller is set as a two-dimensional vector. The proportional-integral-derivative (PID) controller performs independent calculations on the two deviation channels and then weights them together. For the load over-limit deviation channel, the output increment of the PID controller in the nth control cycle is... It is obtained by superimposing the proportional, integral, and derivative terms; for the pressure over-limit deviation channel, the output increment of the proportional-integral-derivative controller in the nth control cycle. It is also obtained by superimposing three terms.

[0073] Output the calculated compensation correction values ​​for the load adjustment and valve opening. The compensation correction value for the load adjustment is denoted as... Its value is directly taken as the output increment of the load over-limit deviation channel. The opposite number, i.e. The compensation correction value for the valve opening correction value is denoted as... Its value is a weighted synthesis of the output increments of the load over-limit deviation channel and the pressure over-limit deviation channel according to the pipeline mechanical response coefficient. The specific calculation formula is as follows:

[0074] in, This is the compensation correction value for the valve opening correction value associated with the b-th critical equipment node, in degrees. The output increment of the proportional-integral-derivative controller for the load over-limit deviation channel in the nth control cycle is expressed in kilograms per second. The output increment of the proportional-integral-derivative controller for the pressure over-limit deviation channel in the nth control cycle is expressed in kilopascals. The conversion coefficient from the load channel to the valve opening is determined by the reciprocal of the slope of the valve flow characteristic curve at the operating point. The value ranges from 0.05 to 0.2 degrees per kilogram per second and is obtained by looking up the characteristic curve data table provided by the valve manufacturer through linear interpolation. The conversion coefficient from pressure channel to valve opening is obtained by inverting the pressure-flow sensitivity matrix of the pipe section. Its value ranges from 0.01 to 0.1 degrees per kPa. It is calculated offline from the linearized small perturbation model of the fluid network topology and stored in the steady-state coordinator parameter table.

[0075] In practical implementation, the proportional coefficient of the proportional-integral-derivative controller Set to 2.0, integration time constant Set to 15.0 seconds, differential time constant. Set to 0.5 seconds. Scale factor. The setting of 2.0 is based on: through step response testing, under an input load exceeding the limit by 10 kg / s, the controller output achieves an initial response amplitude of 20 kg / s while avoiding excessive overshoot; integral time constant. The setting of 15.0 seconds is based on the following: matching the response time constant of approximately 10 to 20 seconds for the combustion controller and turbine regulating valve in a gas turbine power plant, allowing the integral action to gradually eliminate steady-state residual error within the equipment response time; the derivative time constant... The basis for setting it to 0.5 seconds is to filter out high-frequency noise components with a period of less than 0.5 seconds in the pressure fluctuation time sequence, and to prevent the differential channel from introducing additional oscillations.

[0076] The load adjustment values ​​in the initial scheduling strategy sequence are replaced with their compensation correction values, and the valve opening correction values ​​are replaced with their compensation correction values, forming the final scheduling instruction. The replacement operation iterates through all equipment units and valve component entries associated with the b-th critical equipment node in the initial scheduling strategy sequence, overwriting the load adjustment value field with its compensation correction value. Overwrite the value of the valve opening correction field with For other equipment units that did not trigger over-limit correction in the initial scheduling strategy sequence, their load adjustment and valve opening correction values ​​remain unchanged. After the replacement is completed, the updated strategy sequence is obtained, which is the final scheduling instruction.

[0077] The final dispatch command is sent to the corresponding actuator in the gas turbine power plant via the communication bus. The communication bus uses redundant industrial Ethernet and runs the PROFINET protocol. The final dispatch command is encapsulated according to a preset message format, which includes the device address, function code, register start address, data length, and checksum of the target actuator. The actuator corresponding to the combustion controller is the fuel quantity regulator of the gas turbine, and the compensation correction value of the load adjustment included in the final dispatch command is written to the setpoint bias register of the fuel quantity regulator; the actuator corresponding to the turbine regulating valve is the electro-hydraulic servo valve, and the compensation correction value of the valve opening correction value included in the final dispatch command is written to the opening setpoint register of the electro-hydraulic servo valve; the actuator corresponding to the feedwater regulating valve is the electric actuator of the feedwater pump outlet regulating valve, and the compensation correction value of the valve opening correction value included in the final dispatch command is written to the position setpoint register of the electric actuator.

[0078] In some embodiments, the preset sampling period is adaptively adjusted based on the fluctuation frequency of the load flow data stream. The fluctuation frequency is calculated by performing a Fast Fourier Transform on the load flow data stream within a sliding time window, and the frequency component with the largest amplitude is taken as the dominant fluctuation frequency. When the dominant fluctuation frequency exceeds a set threshold of 0.5 Hz, the preset sampling period is shortened from its current value to half of its current value, and the shortened sampling period is not shorter than the lower limit of 0.05 seconds. When the dominant fluctuation frequency remains below the set threshold of 0.1 Hz for 10 window periods, the preset sampling period is restored to the default value of 1 second.

[0079] See Figure 6 In the graph, the horizontal axis represents the control cycle number, totaling 500 cycles; the left vertical axis represents the load over-limit deviation in kilograms per second, and the right vertical axis represents the pressure over-limit deviation in kilopascals. The solid brown curve represents the trend of load over-limit deviation, and the dashed pink curve represents the trend of pressure over-limit deviation.

[0080] As shown in the figure, in the initial stage (approximately the first 50 control cycles), both the load over-limit deviation and the pressure over-limit deviation exhibit significant oscillations and fluctuations. The maximum peak values ​​of the load deviation and pressure deviation are close to ±8 kg / s, and the fluctuations of both show a synchronous decaying trend. This stage corresponds to the initial response stage in Example 5, where the proportional-integral-derivative controller performs real-time correction of the over-limit deviations of the simulated load and simulated pressure values.

[0081] As the control cycle increases (approximately 50 to 200 cycles), the amplitude of the two curves gradually decreases, the oscillation amplitude decreases significantly, the load over-limit deviation gradually converges to near zero, and the pressure over-limit deviation also tends to zero, indicating that after continuous adjustment, both load and pressure deviations are effectively suppressed, and the system operation tends to be stable.

[0082] In the subsequent phase (200 to 500 control cycles), the load over-limit deviation remained within a small oscillation range close to zero, with an amplitude of less than 0.5 kg / s. The pressure over-limit deviation also remained near zero with minimal fluctuations, indicating that the controller continuously maintained the system in stable operation within safe constraints, with no obvious over-limit phenomena.

[0083] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A three-dimensional visualization method for real-time scheduling of production and operation in a gas turbine power plant, characterized in that, The method includes: A three-dimensional virtual scene model of the entire gas turbine power plant is established. The three-dimensional virtual scene model of the entire plant includes the equipment hierarchical topology, pipeline connection paths and spatial coordinates of monitoring nodes. The real-time operating data set of each equipment unit of the gas turbine power plant is collected at the current moment through the real-time data monitoring unit associated with the three-dimensional virtual scene model of the whole plant. The real-time operating data set includes load flow, temperature field distribution and pressure fluctuation time sequence. The real-time operating data set is subjected to fluid dynamic balance analysis based on three-dimensional spatial coordinates to generate real-time load distribution coefficients, thermal efficiency compensation factors and pressure stability indices for each equipment unit. The pre-configured steady-state coordinator is invoked to perform multi-objective constraint matching processing on the real-time load allocation coefficient, the thermal efficiency compensation factor, and the pressure stability index to generate an initial scheduling strategy sequence. The initial scheduling strategy sequence is mapped to the plant-wide 3D virtual scene model for 3D visualization simulation and deduction processing to generate simulation operation trajectory data; Based on the deviation analysis results between the simulated running trajectory data and the preset real-time scheduling constraints, the initial scheduling strategy sequence is corrected in real time to generate the final scheduling instruction and distribute it to the execution mechanism of the gas turbine power plant.

2. The three-dimensional visualization method for real-time scheduling of gas turbine power plant production and operation according to claim 1, characterized in that, The establishment of a full-plant 3D virtual scene model of the gas turbine power plant includes: Obtain the construction drawings and pipeline design specifications of the gas turbine power plant, and extract the geometric dimensions, pipeline routing coordinates, and connection node numbers of each piece of equipment; Based on the geometric dimension parameters, construct a proportional three-dimensional geometry of each device, and then perform spatial assembly processing on the proportional three-dimensional geometry according to the pipeline direction coordinates and connection node numbers to generate an initial three-dimensional scene framework. The historical operating performance feature vectors and real-time monitoring point location tags of each device are embedded in the initial three-dimensional scene framework. The real-time monitoring point location tags include the coordinates of temperature sensitive points, the installation coordinates of pressure transmitters, and the positioning coordinates of flow meters. The initial 3D scene framework is subjected to spatial coordinate normalization and texture mapping to generate the full-plant 3D virtual scene model.

3. The three-dimensional visualization method for real-time scheduling of gas turbine power plant production and operation according to claim 2, characterized in that, The real-time data collection unit, associated with the full-plant 3D virtual scene model, acquires the real-time operating data set of each equipment unit of the gas turbine power plant at the current moment, including: The load flow subunit in the real-time data monitoring unit reads the instantaneous flow values ​​of each main steam pipeline and water supply pipeline according to a preset sampling period, and generates a load flow data stream. The temperature field subunit in the real-time data monitoring unit obtains the real-time temperature value of each temperature-sensitive point based on the monitoring point location label, and generates a temperature field distribution data matrix. The pressure field subunit in the real-time data monitoring unit obtains the pressure values ​​of each node based on the installation coordinates of the pressure transmitter, and generates a pressure fluctuation time series data sequence. The load flow data stream, the temperature field distribution data matrix, and the pressure fluctuation time series data sequence are timestamped to generate the real-time operation data set.

4. The three-dimensional visualization method for real-time scheduling of gas turbine power plant production and operation according to claim 3, characterized in that, The process of performing fluid dynamic balance analysis based on three-dimensional spatial coordinates on the real-time operating data set generates real-time load distribution coefficients, thermal efficiency compensation factors, and pressure stability indices for each equipment unit, including: Construct a fluid network topology diagram in the three-dimensional virtual scene model of the entire plant. The nodes of the fluid network topology diagram correspond to the inlet and outlet of each equipment unit, and the edges of the fluid network topology diagram correspond to the connection paths of each pipeline. The instantaneous flow rate value in the load flow data stream is used as the flow load parameter of each edge in the fluid network topology graph. The real-time load allocation coefficient is calculated by traversing and searching the inflow and outflow difference of each node in the fluid network topology graph. The temperature gradient in the temperature field distribution data matrix is ​​used as the calculation input for the thermal efficiency compensation factor. The thermal efficiency compensation factor is generated based on the mapping relationship between the design thermal efficiency curve of each equipment unit and the temperature gradient. The pressure difference between adjacent sampling points in the pressure fluctuation time series data sequence is correlated with the pressure stability index. The pressure stability index is generated by statistically analyzing the ratio of the standard deviation to the mean of the pressure fluctuation time series data sequence.

5. The three-dimensional visualization method for real-time scheduling of gas turbine power plant production and operation according to claim 4, characterized in that, The invocation of the pre-configured steady-state coordinator performs multi-objective constraint matching processing on the real-time load allocation coefficient, the thermal efficiency compensation factor, and the pressure stability index to generate an initial scheduling strategy sequence, including: The initial scheduling strategy sequence includes the load adjustment amount, valve opening correction value and unit start-up / shutdown priority identifier for each equipment unit; The real-time load allocation coefficient, the thermal efficiency compensation factor, and the pressure stability index are compared with the preset load allocation benchmark range, thermal efficiency benchmark range, and pressure stability benchmark threshold stored in the steady-state coordinator to calculate the load deviation value, thermal efficiency deviation, and pressure fluctuation difference. A three-dimensional constraint space is constructed based on the load deviation value, the thermal efficiency deviation, and the pressure fluctuation difference. The load adjustment amount is obtained by searching the coordinate region that satisfies the upper limit constraint of the load, the lower limit constraint of the thermal efficiency, and the upper limit constraint of the pressure fluctuation in the three-dimensional constraint space. Based on the coupling correlation coefficient between the load adjustment amount and adjacent equipment units, the valve opening correction value and the unit start-up / shutdown priority identifier are calculated. The load adjustment amount, the valve opening correction value, and the unit start-up / shutdown priority identifier are combined according to the time sequence to generate the initial scheduling strategy sequence.

6. The three-dimensional visualization method for real-time scheduling of gas turbine power plant production and operation according to claim 5, characterized in that, The step of mapping the initial scheduling strategy sequence to the plant-wide 3D virtual scene model for 3D visualization simulation and deduction processing to generate simulation trajectory data includes: Each load adjustment, valve opening correction value, and unit start / stop priority identifier in the initial scheduling strategy sequence is assigned to the corresponding equipment unit virtual model and valve component virtual model in the whole plant three-dimensional virtual scene model. The fluid dynamics simulation engine in the full-plant 3D virtual scene model is started, the simulation time step is set, and the operating status of the equipment unit virtual model and the valve component virtual model is iteratively calculated within the simulation time step to generate virtual flow value, virtual temperature value and virtual pressure value of each node. The virtual flow rate, virtual temperature, and virtual pressure values ​​of each node are recorded in chronological order to form a virtual operation data stream; The virtual operation data stream is mapped back to the corresponding position in the three-dimensional spatial coordinate system of the whole plant three-dimensional virtual scene model to generate the simulation operation trajectory data containing dynamic fluid flow trajectory, equipment color grading changes and valve opening animation.

7. The three-dimensional visualization method for real-time scheduling of gas turbine power plant production and operation according to claim 6, characterized in that, The method of performing real-time correction processing on the initial scheduling strategy sequence based on the deviation analysis results between the simulated running trajectory data and the preset real-time scheduling constraints, generating final scheduling instructions and distributing them to the execution mechanism of the gas turbine power plant, includes: The simulated load and simulated pressure values ​​of key equipment nodes are extracted from the simulated operation trajectory data. The deviation between the simulated load value and the rated load upper limit value in the preset real-time scheduling constraints is calculated to obtain the load over-limit deviation. The deviation between the simulated pressure value and the safe pressure threshold in the preset real-time scheduling constraints is calculated to obtain the pressure over-limit deviation. When the load over-limit deviation or the pressure over-limit deviation is greater than zero, the correction algorithm of the steady-state coordinator is invoked, taking the load over-limit deviation and the pressure over-limit deviation as input, and outputting the calculated compensation correction value of the load adjustment and the compensation correction value of the valve opening. The load adjustment amount in the initial scheduling strategy sequence is replaced with the compensation correction value of the load adjustment amount, and the valve opening correction value is replaced with the compensation correction value of the valve opening correction value to form the final scheduling instruction. The final scheduling instruction is then sent to the combustion controller, turbine regulating valve and feedwater regulating valve of the gas turbine power plant via the communication bus.

8. The three-dimensional visualization method for real-time scheduling of gas turbine power plant production and operation according to claim 7, characterized in that, The correction algorithm uses a proportional-integral-derivative controller, taking the overload deviation and the pressure deviation as inputs, and outputs the compensation correction value.

9. A three-dimensional visualized real-time scheduling method for the production and operation of a gas turbine power plant according to claim 3, characterized in that, The preset sampling period is adaptively adjusted according to the fluctuation frequency of the load flow data stream, and the sampling period is shortened when the fluctuation frequency exceeds a set threshold.

10. A three-dimensional visualization real-time scheduling system for the production and operation of a gas turbine power plant, characterized in that, The device includes a processor and a memory, the memory and the processor being connected. The memory is used to store programs, instructions or code, and the processor is used to execute the programs, instructions or code in the memory to implement the three-dimensional visualization real-time scheduling method for production and operation of a gas turbine power plant as described in any one of claims 1-9.

Images