Urban Multi-Heat Source Steam Heating Collaborative Scheduling and Management Platform Based on Load Analysis

By modifying the thermal-hydraulic model and performing dynamic pipe inventory quality analysis, combined with load forecasting and multi-constraint optimization, the problems of inaccurate pipe network status perception and economic dispatch in the existing steam heating system have been solved, achieving high-precision and efficient economic operation of the heating system.

CN123117597APending Publication Date: 2026-11-17HUANENG SUZHOU THERMAL POWER CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611170145.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-04
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

The existing steam heating management and control platform suffers from inaccurate perception of pipeline status and distorted assessment of leakage due to the aging of static models. Furthermore, it lacks the ability to utilize the heat storage and release characteristics of the pipeline network for economic optimization scheduling of multiple heat sources, and thus cannot proactively adjust unit output while ensuring heating safety.

Method used

The urban multi-heat source steam heating collaborative scheduling and control platform based on load analysis uses real-time operation data to correct the thermal hydraulic process mechanism model, calculates the demand load sequence, heat storage adjustable amount and heat release adjustable amount by combining dynamic pipe storage mass and external parameters, generates the temperature control lower limit, performs optimization solution of multiple constraints, and outputs control commands to optimize the heating system.

Benefits of technology

It improves the accuracy of the control platform in perceiving the actual operating boundary of the pipeline network, reduces the probability of false alarms, realizes comprehensive profit optimization control of multi-heat source systems, and enhances the flexibility and economy of the heating system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123117597A_ABST
    Figure CN123117597A_ABST
Patent Text Reader

Abstract

This invention relates to the field of centralized heating dispatching and control technology, and discloses a collaborative dispatching and management platform for urban multi-heat source steam heating based on load analysis. The platform includes a data acquisition module to obtain real-time operating data; a modeling module to correct the thermal-hydraulic mechanism model using the data and output soft-sensor status data; an analysis module to calculate the demand load sequence, dynamic pipe storage mass, and adjustable heat storage / release capacity; a diagnostic module to isolate transient imbalances based on dynamic pipe storage mass to calculate actual leakage and generate a lower limit for temperature control; and a dispatching module to use the above parameters and economic settlement boundary parameters as joint constraints for optimization, obtaining output allocation control commands and issuing them for execution. This invention achieves dynamic correction and soft-sensoring of pipe network status parameters, accurately separates actual pipe network leakage, and transforms the pipe network into a flexible energy storage space, achieving optimal control of the comprehensive profit of multi-source systems while ensuring the safety of the heating network operation.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of centralized heating dispatch and control technology, specifically to a collaborative dispatch and management platform for urban multi-heat source steam heating based on load analysis. Background Technology

[0002] Urban multi-heat-source combined steam heating systems are a crucial infrastructure for ensuring heat supply in industrial production. Due to the compressibility of steam and the continuous heat loss during transmission, the status perception and coordinated scheduling of heating systems are highly complex.

[0003] Existing heating management and control platforms have certain limitations in actual operation. Regarding pipeline status monitoring, these platforms typically incorporate static thermal-hydraulic models. After long-term operation, scaling on the pipe walls and damage to external insulation materials can cause deviations in the actual hydraulic resistance and heat transfer coefficients of the pipelines. Furthermore, due to the limited number of physical sensor points on-site, using fixed-parameter models without dynamic correction makes it difficult to accurately extrapolate the true pressure and temperature status of unmonitored nodes across the entire network.

[0004] In terms of pipeline loss assessment, conventional diagnostic mechanisms often directly calculate the actual leakage by subtracting the total steam supply from each heat source from the total steam consumption at the terminal. This method ignores the impact of dynamic pressure fluctuations in the heating pipeline network on the density of the medium. When changes in system operating pressure cause the steam stored inside the pipeline to expand or compress, the resulting transient flow difference is directly misjudged as medium loss, leading to inaccurate pipeline loss calculations and false alarms in abnormal warnings. Regarding overall system scheduling, conventional control logic primarily passively follows fluctuations in the steam load of end users to allocate heat source output.

[0005] This operating mode does not quantitatively analyze the heat storage and release regulation margin of the physical volume of the steam heating network, and the system cannot coordinate the network as a buffer. When multi-heat source units face fluctuations in external spot market electricity prices and grid auxiliary frequency regulation requirements, the dispatching platform lacks an optimization control mechanism based on the network's flexible boundaries and multiple economic parameters. It cannot proactively adjust the high-pressure cylinder steam extraction and power generation load of the units while ensuring heating safety, thus limiting the improvement of the overall plant profit of the multi-heat source heating system. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a collaborative scheduling and management platform for urban multi-heat source steam heating based on load analysis. This platform solves the problems of inaccurate perception of pipeline status due to aging static models, distorted leakage assessment due to failure to separate pipeline storage fluctuations, and lack of utilization of pipeline heat storage and release characteristics for economic optimization scheduling of multiple heat sources in existing steam heating management platforms.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A city-wide multi-heat-source steam heating collaborative scheduling and management platform based on load analysis includes: The acquisition module obtains real-time running data and transmits it to the modeling module; The modeling module uses the real-time operating data to correct the pre-established thermal-hydraulic process mechanism model, and outputs soft-sensor state data through the corrected thermal-hydraulic process mechanism model. The analysis module combines the soft-sensor status data with external parameters to calculate the demand load sequence, dynamic pipe inventory mass, instantaneous maximum heat storage adjustable amount, and maximum heat release adjustable amount; The diagnostic module uses the soft measurement status data and the dynamic pipe inventory quality to calculate the actual leakage amount and calculates the node superheat parameters to generate the lower limit of temperature control. The scheduling module uses the demand load sequence, the instantaneous maximum adjustable heat storage capacity, the maximum adjustable heat release capacity, the temperature control lower limit, and the economic settlement boundary parameters as constraints to perform optimization and solve for the output allocation control sequence. The output allocation control sequence is then converted into control commands and transmitted to the actuator.

[0008] Preferably, the pre-established thermal-hydraulic process mechanism model is established through the following steps: Construct a topological model of a multi-source steam heating network, and abstract the existing steam network system into a directed graph containing a set of nodes and a set of branches; Construct nodal continuity equations based on the law of conservation of fluid mass; Construct a branch hydraulic pressure drop model that characterizes the mapping relationship between the absolute pressure parameters at both ends of the branch and the steam mass flow rate parameters flowing through the branch; Based on the law of conservation of energy, a thermodynamic friction decay model of the branch inlet and outlet temperatures is constructed to show the temperature decay along the branch under the influence of external environmental parameters. The topology model, the node continuity equation, the branch hydraulic pressure drop model, and the branch thermodynamic friction loss model are integrated to form a set of multivariate nonlinear equations describing the characteristics of the steam pipeline network. The set of multivariate nonlinear equations constitutes the pre-established thermal-hydraulic process mechanism model.

[0009] Preferably, the modeling module uses the real-time operating data to correct the pre-established thermal-hydraulic process mechanism model, and outputs soft-sensor state data through the corrected thermal-hydraulic process mechanism model, including: Determine the spatial mapping relationship of the real-time operating data in the pipeline network topology and define the observable set; Extract the actual measurement values ​​within the observable set from the real-time operating data, and construct a reverse optimization objective function based on the residual between the actual measurement values ​​and the theoretical values ​​obtained from the forward simulation calculation of the thermal-hydraulic process mechanism model; Substitute the actual measured values ​​into the reverse optimization objective function, and iteratively adjust the comprehensive hydraulic resistance coefficient and the comprehensive heat transfer coefficient of the pipeline in the thermal-hydraulic process mechanism model until the residual target value meets the convergence requirement. The thermal-hydraulic process mechanism model, after updating the comprehensive hydraulic resistance coefficient and the comprehensive heat transfer coefficient of the pipeline, will be re-executed with forward fluid network simultaneous calculation to output the soft-sensor state data.

[0010] Preferably, the analysis module combines the soft sensor state data with external parameters to calculate the demand load sequence, including: Extract the continuous historical steam flow sequence of each industrial heat user to construct a historical load feature vector, obtain external meteorological parameters as the external parameters to construct a meteorological environment feature vector, and extract the time node of the production shift to construct the user's process production cycle feature identifier. The machine learning algorithm model is invoked to input the historical load feature vector, the meteorological environment feature vector, and the user process production cycle feature identifier into the prediction mapping network to output a single-point steam demand prediction value; A linear summation operation is performed on the single-point steam demand forecast values ​​along the spatial dimension to generate the overall demand load sequence; The overall demand load sequence is generated by continuously recursively calculating the overall demand load sequence according to the set time window length and time resolution.

[0011] Preferably, the analysis module combines the soft-sensor status data with external parameters to calculate the dynamic tube mass, the instantaneous maximum adjustable heat storage capacity, and the maximum adjustable heat release capacity, including: The local steam storage mass is calculated by multiplying the pipe volume by the average steam density determined from the soft-sensor state data; The dynamic pipeline storage quality is obtained by performing an integral accumulation procedure on the local steam storage quality along the entire network spatial dimension. By using the fluid state equation in conjunction with the upper limit of the highest operating pressure and the lower limit of the lowest steam supply pressure, which are the external parameters, the theoretical maximum and minimum pipe mass are solved. The instantaneous maximum adjustable heat storage capacity is obtained by subtracting the dynamic tube storage mass from the theoretical maximum tube storage mass. The maximum adjustable heat release is obtained by subtracting the theoretical minimum tank mass from the dynamic tank mass.

[0012] Preferably, the diagnostic module calculates the actual leakage amount using the soft measurement status data and the dynamic pipe quality, including: The total input steam flow rate of the entire network is obtained by summing the real-time steam mass flow rate of each heat source injection node extracted from the soft measurement status data, and the total output steam flow rate of the entire network is obtained by summing the real-time steam consumption mass flow rate of each heat user extraction node extracted from the soft measurement status data. The difference between the two is used to obtain the apparent mass loss. Extract the dynamic storage quality corresponding to adjacent time steps and perform time differentiation to obtain the transient quality imbalance. The actual leakage is obtained by algebraically subtracting the transient mass imbalance from the apparent mass loss.

[0013] Preferably, the diagnostic module calculates the node superheat parameter to generate a lower limit for temperature control, including: The absolute pressure values ​​of each node in the soft-sensor state data are extracted to calculate the corresponding saturation temperature; Extract the actual operating steam temperature from the soft measurement status data, and subtract the corresponding saturation temperature from the actual operating steam temperature to obtain the node superheat parameter; The saturation temperature of each node is algebraically added to the safe superheat margin threshold set in the system database to generate the minimum anti-condensation temperature baseline required to maintain a pure vapor phase state. The minimum anti-condensation temperature baseline is used as the lower limit of temperature control for each intersection node and user extraction node.

[0014] Preferably, the diagnostic module is further used to calculate the overall heat decay rate of the entire network, including: The steam specific enthalpy and specific internal energy values ​​are calculated based on the absolute pressure and steam temperature parameters of each node in the soft-sensor state data. The specific internal energy value is combined with the internal steam mass of each branch to obtain the total internal energy value of the internal steam in the pipeline network, and time differential calculation is performed to obtain the rate of change of the total internal energy of the internal steam with time. The total extracted heat power and the total input heat power of the system are calculated based on the soft measurement status data, and the leakage heat power is calculated based on the actual leakage amount. The total heat attenuation rate of the entire network is obtained by subtracting the total extracted heat power, the leaked heat power, and the transient heat power fluctuation deviation caused by the rate of change of the total internal energy of the internal steam over time from the total input heat power of the system in sequence.

[0015] Preferably, the scheduling module uses the demand load sequence, the instantaneous maximum adjustable heat storage capacity, the maximum adjustable heat release capacity, the temperature control lower limit, and the economic settlement boundary parameters as constraints to optimize and solve for the output allocation control sequence, including: The total system revenue and total system cost within a set time window are calculated using the aforementioned economic settlement boundary parameters. By integrating the revenue model and the cost model, the difference between the total revenue value and the total cost value of the system is calculated by time integration, and a global comprehensive profit index value reflecting the overall economic performance of the entire plant is constructed. Maximizing the global comprehensive profit index is used as the sole evaluation index of the multivariate nonlinear programming solution program, and the demand load sequence, the instantaneous maximum adjustable heat storage, the maximum adjustable heat release, and the lower limit of temperature control are used as the boundary constraints of the multivariate nonlinear programming solution program to obtain the output allocation control sequence.

[0016] Preferably, the scheduling module converts the output allocation control sequence into control commands and transmits them to the actuator, including: The first control vector in the output allocation control sequence is extracted as the initial action value of the current cycle, and the target value of active power and the target value of extraction steam mass flow rate for the generator set are separated. A load adjustment signal, which serves as the control command, is sent to the generator set to drive the servo hydraulic motor to change the opening of the high-pressure cylinder speed regulating valve to match the target value of active power. The extraction steam pressure and flow rate offset signals, which serve as control commands, are sent to the heating control subsystem to drive the heating butterfly valve or rotary baffle to change the flow area to match the target value of the extraction steam mass flow rate. The control commands are transmitted to the generator set and the heating control subsystem, which serve as the actuators.

[0017] This invention provides a collaborative scheduling and management platform for urban multi-heat source steam heating based on load analysis. It has the following beneficial effects: 1. This invention constructs a reverse optimization objective function by extracting real-time operational data from measuring points, dynamically corrects the comprehensive hydraulic resistance coefficient and pipeline comprehensive heat transfer coefficient in the thermal-hydraulic mechanism model, and re-executes the forward network solution based on the updated parameters. This solves the problem of parameter distortion in the static mechanism model caused by aging, scaling, or insulation layer damage in heating pipelines. The system can output accurate soft-sensor status data covering all unmeasured locations in the network without the need for densely adding physical sensors, improving the perception accuracy of the control platform regarding the actual operational boundaries of the underlying physical pipeline network.

[0018] 2. This invention calculates the dynamic pipe inventory mass of the entire network by combining pipeline geometric parameters and fluid state equations, performs time differential calculations to obtain the transient mass imbalance, and algebraically subtracts this transient mass imbalance from the apparent mass loss between input and output nodes. In the material balance calculation process, it eliminates the interference of inventory changes caused by local compression or expansion of the steam medium due to system pressure fluctuations, decouples and separates the actual pipeline leakage, reduces the probability of false alarms caused by fluctuations in pipeline operating conditions, and improves the reliability of steam pipe loss assessment.

[0019] 3. This invention calculates the instantaneous maximum adjustable heat storage and maximum adjustable heat release based on dynamic pipe storage quality and system operating pressure boundaries. These, along with the load prediction sequence, superheat control lower limit, and multiple external market economic settlement boundary parameters, serve as joint constraints for the solver. The rigid steam heating network is equivalently transformed into an energy storage and release space with buffering and adjustment capabilities. This allows the management platform to actively adjust the generator power and extraction valve opening using the pipe network's heat storage margin, responding to time-of-use pricing and frequency regulation compensation, while ensuring the thermal and hydraulic safety of the pipe network and the heating demand of end users. This achieves optimal control of the overall plant profit for multi-heat source systems. Attached Figure Description

[0020] Figure 1 This is a diagram of the overall architecture of the urban multi-heat source steam heating collaborative scheduling and control platform of the present invention.

[0021] Figure 2 This is a flowchart of the method of the present invention.

[0022] Figure 3 This is an abstract schematic diagram illustrating the topology and technological mechanism of the multi-source steam heating pipeline network of the present invention.

[0023] Figure 4 This is a flowchart of the parameter identification and soft measurement calculation based on reverse optimization of the present invention.

[0024] Figure 5 This is a flowchart of the pipeline spatial load prediction based on multi-dimensional feature fusion according to the present invention.

[0025] Figure 6 The following is a comparison chart of the experimental verification and effects of the control platform of the present invention, wherein (a) is a bar chart comparing the accuracy of state estimation and leakage diagnosis, and (b) is a line chart comparing the cumulative benefits of multi-heat source joint economic dispatch over 30 consecutive days. Detailed Implementation

[0026] 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, and 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.

[0027] Reference Figure 1 This invention provides a collaborative scheduling and management platform for urban multi-heat source steam heating based on load analysis, including: a data acquisition module, a modeling module, an analysis module, a diagnostic module, and a scheduling module.

[0028] The data acquisition module connects to monitoring equipment in multi-heat source unit systems and steam heating pipelines. It acquires real-time operational data from each heat source node, pipeline junction node, and user terminal. This real-time operational data includes node pressure, node temperature, pipeline flow rate, and ambient temperature. The acquired real-time operational data is then transmitted to the modeling module.

[0029] The modeling module pre-establishes a thermal-hydraulic process mechanism model incorporating multiple heat sources and a steam pipeline network topology. Using received real-time operational data, the module constructs a reverse optimization objective function. Based on the residuals between the measured values ​​in the real-time operational data and the forward simulation values ​​from the process mechanism model, it corrects the drag coefficient and heat transfer coefficient in the process mechanism model. The modeling module performs simulation calculations using the corrected process mechanism model, outputting soft-sensor status data covering all pipeline nodes. This soft-sensor status data includes flow rate, pressure, and temperature values ​​at locations where monitoring equipment is not deployed.

[0030] The analysis module receives soft-sensor status data. It combines historical load data, environmental meteorological parameters, and user production process cycles to build a predictive model, calculating the demand load sequence for a future specified time period. Based on pipeline network geometric parameters and fluid state equations, the analysis module calculates the current dynamic pipeline inventory mass of the entire network. Finally, it incorporates the safety pressure boundary parameters of the pipeline network design to quantify the instantaneous maximum adjustable heat storage and maximum adjustable heat release of the network.

[0031] The diagnostic module uses soft-sensor status data and dynamic pipe inventory mass to calculate the pipeline network's mass loss and heat loss. It separates transient imbalances caused by changes in pipe inventory mass over time from the input-output mass difference, thus calculating the actual leakage. The module compares the node temperatures in the soft-sensor status data with the saturation temperatures at the corresponding operating pressures to calculate node superheat parameters and generate lower temperature control limits to suppress hydrophobic condensation.

[0032] The scheduling module acquires electricity spot market transaction price parameters and ancillary service frequency regulation compensation parameters. The module constructs an objective function with the goal of maximizing the overall plant profit, using the demand load sequence, instantaneous maximum adjustable heat storage, maximum adjustable heat release, unit thermal-electric decoupling operation range limitations, and safety boundaries of each network node as constraints. Within a set time window, the module solves for the output allocation control sequence of each heat source node. The scheduling module extracts the initial action values ​​from the output allocation control sequence, converts them into control commands, and transmits them to the actuators.

[0033] Reference Figure 2 The workflow of the multi-heat source steam heating collaborative scheduling and control platform in the aforementioned cities is executed according to the following operating logic.

[0034] The acquisition module continuously collects system operating status data and passes it as input to the modeling module. The modeling module executes a parameter identification algorithm to update the thermal-hydraulic state estimation matrix, achieving full network hydrodynamic and thermodynamic state perception.

[0035] The analysis module performs load extrapolation based on the status perception results and external environmental conditions, simultaneously evaluating the compression and expansion characteristics of the steam medium to quantify the pipeline network's regulation margin. The diagnostic module performs material and energy balance calculations for the steam-water system, outputting leakage indicators and superheat control boundary conditions reflecting the pipeline network's health status.

[0036] The scheduling module aggregates the network regulation margin, superheat control boundary conditions, and real-time electricity market dispatch parameters. Within the model predictive control architecture, the scheduling module performs multi-constraint optimization steps to generate heat source production parameters and valve regulation commands that meet the safety requirements of the heating network and the principle of supply and demand balance, completing a closed-loop operation for one control cycle. The management and control platform triggers the operations of the above modules cyclically according to a fixed time step.

[0037] Reference Figure 3 The modeling module establishes a thermal-hydraulic process mechanism model that includes multiple heat sources and a steam pipe network topology. This can be divided into the following steps: Construct a topological model of a multi-source steam heating network. Abstract the actual steam network system as a directed graph consisting of nodes and connections. .in, This represents a set of nodes, specifically including heat source injection nodes, heat user extraction nodes, and pipeline junction and branch nodes in the actual pipeline network. This represents the set of branches, specifically including straight-through heating pipes, regulating valves, and desuperheaters / pressure reducers in the actual pipe network. The modeling module, based on the pipe network design drawings, defines the geometric connections and baseline fluid flow direction between the aforementioned set of nodes and the set of branches using an adjacency matrix.

[0038] The nodal continuity equation is constructed based on the law of conservation of fluid mass. For any node in the topological model, the sum of the inflow and outflow mass flows of steam at the same moment, along with the sum of the source and sink terms of that node, are kept at zero by algebraic constraints.

[0039] Specifically, when a node is set as a heat source injection node, the source and sink items are extracted as positive steam supply values; when a node is set as a heat user extraction node, the source and sink items are extracted as negative steam consumption values; and when a node is set as a pipeline junction branch node that only serves a connection function, the source and sink items are set to zero.

[0040] The modeling module constructs a branch hydraulic pressure drop model. Steam exhibits significant fluid compressibility during pipeline transport. For any branch in the topology model, the modeling module establishes a hydraulic resistance equation characterizing the direct proportional relationship between the squared absolute pressure difference between its two endpoints and the squared mass flow rate of steam flowing through that branch. The comprehensive hydraulic resistance coefficient in the equation is determined by the corresponding pipe's inner diameter, designed pipe length, and the absolute roughness of the pipe's inner wall, among other geometric and material properties.

[0041] Furthermore, the modeling module constructs a branch thermodynamic friction decay model based on the law of conservation of energy. Since steam continuously transfers heat with the external environment as it flows through the pipe, the modeling module establishes an equation for temperature decay at the branch inlet and outlet, which is exponentially reduced by the combined effects of the external ambient temperature, the overall heat transfer coefficient of the pipe, and the specific heat capacity of the fluid. For the temperature state of each converging steam stream at the junction node, those skilled in the art can use an enthalpy-weighted average algorithm to establish a node heat balance equation for inverse calculation; the specific process is well-known in the field and will not be elaborated here.

[0042] The modeling module integrates the above topological structure model, node continuity logic, branch hydraulic pressure drop mapping relationship, and thermodynamic friction loss equation to form a set of multivariate nonlinear equations describing the characteristics of the multi-source steam pipeline network.

[0043] This set of multivariate nonlinear equations forms the core of the pre-established thermal-hydraulic process mechanism model.

[0044] Reference Figure 4 The modeling module uses the received real-time running data to construct the inverse optimization objective function and output soft measurement state data. This process can be divided into the following steps: Determine the spatial mapping relationship of the real-time operational data acquired by the acquisition module within the pipeline network topology. Define the nodes or branches equipped with monitoring devices as the observable set. Due to limitations in project cost and on-site construction conditions, flow meters, pressure transmitters, and temperature sensors are only installed at specific locations in actual steam pipeline systems.

[0045] Increased pipeline operating time leads to oxidation and scaling on the inner walls of the pipes, altering surface roughness. Delamination and water leakage from the external insulation layer reduce thermal insulation performance. These changes cause the combined hydraulic resistance coefficient and the combined pipeline heat transfer coefficient in the established thermal-hydraulic process mechanism model to gradually deviate from their initial design values. The modeling module sets the combined hydraulic resistance coefficient and the combined pipeline heat transfer coefficient as dynamic parameter vectors to be identified.

[0046] The modeling module constructs a reverse optimization objective function for real-time correction of pipeline network parameters. This reverse optimization objective function is constructed based on the sum of squared residuals between the actual measured values ​​at each measuring point within the observable set and the theoretical values ​​obtained from forward simulation calculations of the thermal-hydraulic process mechanism model. The mathematical expression of the reverse optimization objective function is: ; in, This represents the target residual value for the overall condition assessment; This represents the vector of comprehensive hydraulic resistance coefficients to be identified across the entire network; This represents the vector of the overall heat transfer coefficients of the pipelines to be identified across the entire network. This represents a set of nodes or branches that have measurement points; Indicates the measuring point The actual pressure measurement value at the location; This represents the measurement points obtained from the solution of the thermal-hydraulic process mechanism model within the current calculation period. Simulated pressure values ​​at the location; Indicates the measuring point The actual temperature measurement value at the location; The measuring points obtained from the solution of the thermal-hydraulic process mechanism model Simulated temperature values ​​at the location; This represents the numerical weighting coefficient for the pressure residual term; This represents the numerical weighting coefficient of the temperature residual term. and The specific value is determined based on the range and calibration error limit of the field sensor instrument.

[0047] The modeling module executes an optimization procedure to minimize the inverse optimization objective function. The modeling module substitutes actual physical measurement data into the inverse optimization objective function, iteratively adjusting the comprehensive hydraulic resistance coefficient and comprehensive heat transfer coefficient of each pipe segment until the residual target value meets the convergence requirement.

[0048] For specific algorithms to solve nonlinear least squares optimization problems with nonlinear equality constraints, those skilled in the art can use the Gauss-Newton iteration method or the trust region algorithm for numerical solutions. The underlying data processing procedures for constructing the Jacobian matrix and calculating the descent iteration step size are well-known techniques in the field and will not be elaborated upon here. After the convergence condition is met, the modeling module outputs the corrected comprehensive hydraulic resistance coefficient matrix and the corrected comprehensive pipeline heat transfer coefficient matrix, reflecting the current actual operating state of the pipeline network.

[0049] The modeling module executes a soft-sensor calculation program covering the state of all network nodes. The modeling module updates the corrected integrated hydraulic resistance coefficient matrix and the corrected integrated pipeline heat transfer coefficient matrix to the underlying database of the thermal-hydraulic process mechanism model.

[0050] The modeling module utilizes the updated thermal-hydraulic process mechanism model, using the steam supply pressure and temperature values ​​at each heat source injection node as known boundary conditions, to re-execute the forward fluid network simultaneous calculation for the entire pipeline network. The system solves the multivariate nonlinear equation system to obtain numerical solutions for all unknown nodes within the pipeline network. The modeling module outputs soft-sensor status data for all nodes in the pipeline network.

[0051] The soft-sensor status data supplemented the steam mass flow rate, operating absolute pressure, and in-pipe steam temperature data for locations where monitoring equipment was not installed in the topology model. The above-mentioned mechanism of superimposing data correction avoids the divergence problem of purely data-driven models under varying operating conditions.

[0052] Reference Figure 5 The analysis module combines historical load data, environmental meteorological parameters, and user production process cycles to establish a prediction model and calculate the demand load sequence. This process can be broken down into the following steps: The analysis module extracts multi-dimensional load characteristic vectors for each industrial heat user. Industrial users' steam consumption behavior is jointly constrained by their own production schedules and changes in the external environment. The analysis module extracts the first... The historical steam flow sequence of each industrial heat user up to the current moment is used to construct a historical load feature vector. The analysis module obtains air temperature, relative humidity, and wind speed parameters of the area through an external meteorological interface to construct a meteorological environment feature vector.

[0053] The analysis module extracts production shift time nodes, intermittent batch heat use start-up and shutdown signals, and holiday rest indicators from the industrial heat user's production schedule in the plant-level manufacturing execution system or manually entered. This constructs a user process production cycle feature identifier that reflects the specific industrial production rhythm. These feature vectors together form the input dataset for demand load sequence prediction.

[0054] The analysis module calls a pre-trained machine learning algorithm model to execute a single-point load forecasting program. The analysis module inputs a multi-dimensional feature vector, including historical load, meteorological environment, and user process production cycle identifiers, into a prediction mapping network containing bias weight parameters, and calculates and outputs a single-point steam demand forecast value for a single user within a set future time step.

[0055] For the operation of constructing and training a prediction network model based on multidimensional time feature sequences, those skilled in the art can perform forward and backward error propagation operations based on historical sample datasets, which are well-known techniques in the field and will not be elaborated here. Subsequently, the analysis module executes a spatial aggregation calculation program for the global load of the pipeline network.

[0056] After obtaining the single-point load forecast values ​​of all independent heat user extraction nodes within the pipeline network topology, the analysis module performs a linear summation operation along the spatial dimension, and finally outputs and generates an overall demand load sequence covering the corresponding future time nodes at the entire network level.

[0057] The analysis module iteratively applies the above calculation process according to the set time window length and time resolution, generating a continuous network-wide demand load sequence on a time scale. This network-wide demand load sequence constitutes the baseline demand constraint boundary for subsequent source-grid dynamic energy matching.

[0058] The analysis module calculates the dynamic pipe storage mass based on the pipe network geometric parameters and fluid state equations, and quantifies the instantaneous maximum adjustable heat storage and maximum adjustable heat release of the pipe network. This can be broken down into the following steps: The analysis module calculates the local steam storage mass of each pipeline branch. Based on geometric parameters including pipe inner diameter and pipe segment design length, the analysis module calculates the internal space pipe volume corresponding to each branch in the entire network topology.

[0059] The analysis module extracts the absolute pressure and steam temperature at both ends of each branch from the soft-sensor state data, and takes the arithmetic mean as the average operating pressure and average operating temperature of that branch. The analysis module substitutes the average operating pressure and average operating temperature into the fluid state equation to obtain the average steam density value at the corresponding state point. Multiplying the pipe volume of the local branch by this average steam density value yields the local steam storage mass retained within that branch at a specific moment.

[0060] For the specific operation of deriving the physical property density of water vapor from pressure and temperature parameters, those skilled in the art can call the standard library functions for calculating the thermodynamic properties of water and water vapor to execute table lookup or calculation instructions. The internal process of deriving physical property parameters based on specific free energy equations is a well-known technology in the field and will not be described in detail here.

[0061] The analysis module executes an integral accumulation procedure for the local steam storage mass along the entire network spatial dimension. The analysis module sums the local steam storage masses within all pipeline branches to obtain the current dynamic pipeline storage mass of the entire network. The dynamic pipeline storage mass represents the total mass of steam medium actually contained within the internal space of the steam heating pipeline network at the current moment. The mathematical expression for calculating the dynamic pipeline storage mass is: ; in, Indicates at time The overall dynamic pipe inventory quality of the steam heating network; This represents the set of branches in the pipeline network topology; Indicates from node Pointing to node Directed branch; This indicates the volume of the pipe corresponding to that branch. Indicates time-based Average operating pressure of branch lines With average operating temperature The average density of the steam is obtained by solving the problem.

[0062] The analysis module establishes the system's pipeline extreme boundary based on safety constraints. It also obtains the upper limit of the maximum permissible operating pressure in the pipeline design, a parameter limited by pipe material and valve pressure rating. Finally, the analysis module obtains the lower limit of the minimum steam supply pressure required to ensure the normal operation of process steam equipment at the user end.

[0063] The analysis module substitutes the upper limit of the maximum operating pressure and the lower limit of the minimum steam supply pressure into the fluid state equation to solve for the average steam density under the corresponding ultimate pressure state. It then combines the pipe volume of each branch of the pipeline network to perform a cumulative summation calculation along the entire network spatial dimension, thereby obtaining the theoretical maximum and minimum pipe storage mass that the system can accommodate under the condition of not exceeding the limits.

[0064] The analysis module quantifies the instantaneous heat storage adjustable amount and the maximum heat release adjustable amount based on the compressibility of the steam medium. Since increasing the overall operating pressure of the pipeline network will cause the input steam to be converted and retained inside the pipeline network, and decreasing the operating pressure will cause the stored steam to expand and be discharged to provide medium replenishment, the analysis module directly uses the theoretical maximum pipeline storage mass calculated beforehand to subtract the current actual dynamic pipeline storage mass to obtain the maximum medium mass that the system can be forced into the pipeline network during this period, that is, the instantaneous maximum heat storage adjustable amount; Simultaneously, by subtracting the theoretical minimum pipe storage mass from the current actual dynamic pipe storage mass, the maximum mass of medium that can be released to the outside of the pipe network during this period can be obtained, which is the instantaneous maximum adjustable heat release.

[0065] The analysis module outputs the above-mentioned instantaneous maximum adjustable heat storage and maximum adjustable heat release data sequences, which convert the original rigid pipe network into a boundary constraint set with flexible energy storage and release capabilities, and transmits it to the scheduling link for source-network coordinated control boundary definition.

[0066] The diagnostic module separates the transient imbalance caused by the change in pipe quality over time from the input-output quality difference and calculates the actual leakage amount. This process can be divided into the following steps: The diagnostic module calculates the input-output apparent quality difference at the macroscopic boundary of the steam heating network within a set calculation period. It extracts real-time steam mass flow rate data for each heat source injection node from the soft-sensor status data and accumulates this data along the spatial dimension to obtain the total input steam flow rate for the entire network.

[0067] The diagnostic module extracts real-time steam mass flow rate data from each heat user's extraction node and accumulates it along the spatial dimension to obtain the total output steam flow rate of the entire network. The value of the total input steam flow rate minus the total output steam flow rate represents the apparent mass loss of the pipeline system. This apparent mass loss includes the actual leakage of the medium leaving the pipeline and the transient inventory changes caused by pressure fluctuations in the pipeline system.

[0068] The diagnostic module calculates the transient quality imbalance caused by dynamic memory fluctuations. It extracts the dynamic memory quality values ​​corresponding to adjacent time steps, performs differentiation operations on the time variable, and obtains the derivative value of the memory quality as a function of time.

[0069] This derivative value characterizes the transient steam flow rate absorbed or released by the entire pipeline network due to the compression or expansion of the medium. When the rate of change of the pipeline mass over time is positive, it indicates that the pipeline network is in the steam accumulation stage, and some of the medium injected by the heat source remains inside the pipeline and is converted into an increase in the pipeline mass. When this value is negative, it indicates that the pipeline network is in the steam release stage, and the system consumes the original stock to supplement the user load. For the derivative solution operation after discretization of continuous dynamic physical quantities, those skilled in the art can use the forward difference algorithm or the central difference algorithm to obtain the rate of change of the state variables. The specific construction of the difference equation and the derivation process of the discretization truncation error are well known techniques in this field and will not be elaborated here.

[0070] The diagnostic module isolates transient mass imbalance and decouples the actual leakage rate. Based on the principle of mass conservation at fluid network nodes, the module algebraically subtracts the transient mass imbalance obtained by performing time differentiation on the dynamic pipe mass from the apparent mass loss at the macroscopic boundary. This filters out spurious loss data caused by pressure fluctuations, revealing the true leakage rate reflecting the health of the pipeline network. The mathematical expression for decoupled calculation of mass loss is: ; in, Indicates time The actual leakage mass flow rate of the steam heating network, i.e., the mass pipe loss value; This represents the set of all heat source injection nodes contained in the pipeline network topology; Indicates input node The corresponding heat source steam mass flow rate; This represents the set of all heat user extraction nodes contained in the pipeline network topology; Indicates the extraction node The corresponding load consumes steam mass flow rate; Indicates the overall dynamic pipeline inventory quality of the pipeline network system. The rate of change over time.

[0071] The diagnostic module executes early warning actions for abnormal leaks in the pipeline network based on the actual leakage amount. The diagnostic module sets permissible thresholds for normal pipe loss in the system database, corresponding to the standard sealing parameters of the pipeline and the measurement accuracy of the on-site flow meters.

[0072] The diagnostic module compares the actual leakage mass flow rate sequence of the continuous calculation period with the normal pipe loss allowable threshold.

[0073] When the actual leakage mass flow rate continuously exceeds the time window set by the normal pipe loss allowable threshold, the diagnostic module determines that there is pipe flange leakage or abnormal behavior of user meters in the pipeline network, and outputs a mass pipe loss exceeding the limit warning signal containing the time node of the abnormal occurrence and the amount of abnormal loss.

[0074] The diagnostic module compares the node temperature in the soft-sensor status data with the saturation temperature under the corresponding operating pressure, calculates the node superheat parameters, and generates a lower limit for temperature control to suppress hydrophobic condensation. Simultaneously, it calculates the heat loss of the entire network. This process can be broken down into the following steps: The diagnostic module acquires the operating specific enthalpy values ​​of each node in the pipeline network and the rate of change of the total internal energy of steam in the entire network. It extracts the absolute pressure and steam temperature parameters of each heat source injection node and each heat user extraction node from the soft-sensor status data, and estimates the absolute pressure and steam temperature parameters of the equivalent leak location by combining these parameters with the average operating status parameters of the pipeline network.

[0075] Based on the above absolute pressure and steam temperature parameters, the diagnostic module calculates and obtains the steam specific enthalpy and specific internal energy values ​​at the corresponding state points.

[0076] The diagnostic module combines the pipe volume and average steam density of each branch of the pipeline network, multiplies the internal steam mass of each branch by its specific internal energy, and sums the results along the entire network's spatial dimensions to obtain the total internal energy of the steam within the pipeline network at the current moment. The diagnostic module then performs a differential operation on this total internal energy value with respect to the time variable to obtain the rate of change of the total internal energy of the steam within the pipeline network over time.

[0077] For the specific operation of deriving the specific enthalpy and specific internal energy values ​​of water vapor from pressure and temperature parameters, those skilled in the art can call the standard library functions for calculating the thermodynamic properties of water and water vapor to execute the state parameter conversion instructions. The internal process of solving the physical property parameters based on the basic thermodynamic equations is a well-known technology in this field and will not be described in detail here.

[0078] The diagnostic module isolates the heat lost due to transient energy changes and mass leakage, and calculates the overall heat decay rate of the entire network. Under actual operating conditions, the heat storage and release behavior of the pipeline network itself will cause dynamic fluctuations in the total energy within the system, and leaks will also cause some energy to be lost along with mass leakage.

[0079] To accurately assess the true thermal insulation performance of the pipeline insulation structure, the diagnostic module constructs energy calculation rules based on the law of conservation of energy. The diagnostic module sums the products of the injection flow rate and the enthalpy of the steam supply ratio at each heat source node to calculate the total input heat power of the system.

[0080] Subsequently, the diagnostic module subtracts three interference powers sequentially from the total input thermal power: First, the total extracted heat power is composed of the product of the user-side extracted flow rate and the corresponding consumption enthalpy. Second, the leakage heat power is calculated by multiplying the actual leakage mass by the estimated enthalpy of the leakage location by the product of the calculated actual leakage mass and the leakage location. Third, the transient thermal power fluctuation deviation introduced by the rate of change of the overall internal energy of the system network over time.

[0081] After the above energy stripping calculation, the diagnostic module separates the overall heat decay rate of the entire network caused purely by convective heat dissipation along the pipeline, and uses this continuously exceeding heat loss data as the key judgment basis for triggering subsequent pipeline insulation layer repair decisions.

[0082] The diagnostic module executes a program to calculate superheat parameters for the operating phase state of each pipeline node. During the transportation of heat through the pipeline, the temperature of the steam medium gradually decreases due to heat dissipation along the way. When the steam temperature drops to the saturation temperature baseline corresponding to the operating pressure, the steam medium will undergo a phase change and condensate. The diagnostic module calculates the corresponding saturation temperature under that pressure condition based on the absolute pressure values ​​of each node in the soft-sensor status data.

[0083] The diagnostic module subtracts the calculated corresponding saturation temperature from the actual operating steam temperature of the node to obtain the node superheat parameter, which characterizes the degree to which the steam state deviates from the saturation line.

[0084] The diagnostic module generates lower limits for node temperature control to suppress hydrophobic condensation. Condensation causes the drain valves at the bottom of the pipe network to automatically open and discharge the medium, leading to both fluid and heat loss and potentially triggering water hammer.

[0085] The diagnostic module sets a safe superheat margin threshold in the system database. The diagnostic module algebraically adds the saturation temperature of each node to this safe superheat margin threshold to generate the minimum anti-condensation temperature baseline required for each node to maintain a pure vapor phase state under the current operating pressure.

[0086] The diagnostic module uses the minimum anti-condensation temperature baseline as the lower limit parameter for temperature control at each intersection node and user extraction node, and inputs this boundary condition into the scheduling module to constrain the superheat setting value of the steam supply at each heat source outlet, ensuring that the entire network operates within the pure superheated steam range.

[0087] The scheduling module obtains electricity spot market transaction price parameters and ancillary service frequency regulation compensation parameters, and constructs an objective function with the goal of maximizing the overall plant profit. This process can be divided into the following steps: The scheduling module acquires the sequence of external economic settlement boundary parameters for the multi-source coupled system within a set time window. The scheduling module connects to the external trading center and the fuel procurement system via a communication interface to extract the economic parameters for the current control period.

[0088] The aforementioned economic parameters specifically include the time-of-use electricity price in the electricity spot market, the unit price for frequency regulation mileage compensation for automatic generation control ancillary services, the unit price for industrial steam heating contract settlement, the unit price for fuel procurement, and the real-time carbon price in the carbon emission trading market. The obtained sequence of external economic settlement boundary parameters provides economic incentive signals for the system's optimization calculation.

[0089] The scheduling module calculates the system's total revenue model based on the production schedule of each heat source unit.

[0090] The system's total revenue is a linear sum of three parts: revenue from electricity trading, revenue from ancillary service frequency regulation, and revenue from steam sales for heat. Specifically, the scheduling module multiplies the actual grid-connected power of each heat source unit by the time-of-use electricity price to derive the electricity trading revenue.

[0091] The revenue from ancillary service frequency regulation is quantified based on the dynamic heat storage and release capacity of the steam pipeline network. The scheduling module uses the instantaneous maximum adjustable heat storage and maximum adjustable heat release of the pipeline network obtained by the aforementioned analysis module as constraint boundaries. Under the premise of ensuring the pressure safety of the heating pipeline network, it changes the steam extraction flow by rapidly adjusting the opening of the unit's extraction steam valve, and uses the thermoelectric coupling mechanism to enable the unit's electrical output to quickly respond to the automatic power generation control commands issued by the grid.

[0092] The scheduling module multiplies the actual frequency regulation mileage provided by the unit with the frequency regulation mileage compensation unit price to obtain the ancillary service frequency regulation revenue. The scheduling module multiplies the steam mass flow rate of each heat source injection node with the steam heating contract settlement unit price to obtain the steam sales revenue.

[0093] The revenue from electricity trading, ancillary service frequency regulation, and steam heat sales is algebraically summed and extracted as the total revenue of the system in the current regulation cycle. Simultaneously, the scheduling module establishes a total system cost model that includes fuel consumption and carbon emission calculations. Based on the unit's heat consumption characteristic surface model, the scheduling module calculates the fuel consumption rate under the current operating conditions and multiplies it by the fuel purchase price to obtain the procurement cost.

[0094] Simultaneously, the scheduling module converts the fuel consumption rate into the corresponding actual carbon dioxide emissions, deducts the system's free emission allowance base, and multiplies it by the real-time carbon price to obtain the carbon allowance trading cost. The scheduling module combines the fuel procurement cost and the carbon allowance trading cost to form the total system cost.

[0095] Finally, the scheduling module integrates the revenue model and the cost model, and calculates the difference between the total revenue value and the total cost value of each distance within the set forward prediction time window by time integration, and constructs a global comprehensive profit index value that reflects the overall economic performance of the entire plant.

[0096] The scheduling module uses the maximization of the global comprehensive profit index as the sole evaluation metric for the optimization algorithm, driving the multi-source system to migrate towards the economically optimal operating point under complex electricity price conditions.

[0097] The scheduling module operates under constraints such as the demand load sequence, the instantaneous maximum adjustable heat storage capacity, the maximum adjustable heat release capacity, the unit's thermal-electric decoupling operation range limitations, and the safety boundaries of each pipeline node. This can be broken down into the following steps: The scheduling module integrates the upper and lower limits of the unit's output under the set extraction steam volume and the ramp-up and ramp-down restrictions of adjacent regulation cycles into a feasible domain constraint for decoupled thermal and electrical operation. The conditions that the total system heat source injection flow must be equal to the pipe inventory change rate, the sum of the predicted load extraction and the actual leakage are integrated into a global supply and demand balance equation constraint. The extracted nodal limit working pressure and minimum anti-condensation temperature baseline are integrated into the thermal hydraulic safety operation boundary constraints to prevent the risks of bursting and water hammer. At the same time, by combining the maximum heat storage and release index output by the analysis module, the current upper limit of the allowable medium filling and releasing capacity in the space is integrated into a flexible energy storage charging and discharging rate constraint.

[0098] The scheduling module compiles the four categories of constraint boundaries mentioned above into a constraint manifold space constraint set for the model predictive solver. Based on this, the scheduling module initiates a multi-constraint optimization solution process within the model predictive control architecture. The scheduling module selects the predictive line-of-sight and control line-of-sight for future system projections and assembles the extraction steam and electrical power control setpoints of each heat source unit into a time-based control action vector sequence capable of covering multiple future time spans. The set of adjustments constituted by these action vectors defines the multi-dimensional input variables for the feasible control solution in subsequent rolling calculations.

[0099] The scheduling module executes a multivariable nonlinear programming solver. It uses the comprehensive economic benefit objective function as the cost function for optimization calculations and multiple physical and operational boundary constraints as the constraint manifold space of the solver.

[0100] The scheduling module searches for the optimal control sequence within the predicted line of sight, maximizing the overall plant profit. The mathematical expression for solving the optimal control sequence is: ; ; ; in, This represents the optimal control sequence matrix obtained by solving the problem; This indicates the set predicted line-of-sight length; and They represent the times at time 1 and 2 respectively. Predicted future The system's total revenue and total cost at any given time; Indicates the discrete control step size; This represents the set of inequality constraints consisting of the feasible region of unit thermoelectric decoupling, ramp rate limitation, and the lower limit of pipeline operating pressure and temperature. This represents the set of equality constraints formed by the continuity equation of the steam system and the supply and demand flow balance; This represents the vector of state variables of the system at the corresponding future time.

[0101] For the specific operation of finding the global optimal solution of a nonlinear objective function in a feasible domain space with nonconvex constraints, those skilled in the art can use sequential quadratic programming algorithms or interior point methods to construct the Lagrangian function and iteratively solve the Carlow-Kun-Tucker conditions. The underlying numerical calculation process of using the Hessian matrix approximation and calculating the convergence search direction is a well-known technology in this field and will not be elaborated here.

[0102] The scheduling module extracts the initial action values ​​from the optimal control sequence matrix and initiates rolling time-domain feedback correction. After obtaining the optimal control sequence matrix covering the entire control line of sight, the scheduling module only extracts the first control vector in the matrix as the actual action command output for the current cycle, discarding the remaining future adjustment steps in the control sequence.

[0103] When the system reaches the next discrete time step, the acquisition module obtains new sensor measurement data, and the digital twin mechanism modeling module re-identifies the pipeline resistance and heat transfer parameters and updates the current real state of the system.

[0104] The scheduling module uses the updated real-time state as a new optimization starting point, shifts the entire prediction window one step forward along the future time axis, and re-executes the multivariate nonlinear programming solution to find the optimal solution for the overall plant profit within the prediction line of sight. This rolling correction mechanism, which overlays real-time data feedback, suppresses calculation errors caused by long-term weather forecasting errors and valve mechanical wear.

[0105] The scheduling module converts the initial action values ​​into control commands and transmits them to the actuators. The scheduling module parses the current control vector and separates the target active power and extraction steam mass flow rate for each specific generator set.

[0106] The scheduling module maps abstract numerical variables into underlying standard industrial control signals according to the hard-wiring specifications or industrial Ethernet communication protocols supported by the field distributed control system. The scheduling module sends load adjustment signals to the generator set's digital electro-hydraulic control system, driving the servo hydraulic motor to change the opening of the high-pressure cylinder speed regulating valve to match the target electric power value.

[0107] The scheduling module sends extraction steam pressure and flow bias signals to the heating control subsystem, driving the heating butterfly valve or rotating baffle on the low-pressure cylinder connecting pipe to change the flow area to match the target value of extraction steam mass flow rate. The scheduling module completes the signal response operation of the underlying actuators, ends the control loop of the current cycle, and waits for the triggering of the next scheduling optimization pulse according to the set fixed time step.

[0108] Specific application examples: The physical scenario in this embodiment is set as a centralized heating network in a chemical industrial park. The system includes two extraction-condensing combined heat and power units (heat source A and heat source B) as heat sources. The total length of the main pipeline is 15 kilometers, the pipe diameter is DN600, and the internal physical volume of the pipeline is approximately 4200 m³. 3 The terminal includes four continuous-production chemical plants and two intermittent-production pharmaceutical plants.

[0109] During a specific scheduling period (e.g., from 8:00 AM to 10:00 AM), the system operates according to the following specific parameters and logic.

[0110] Data Acquisition and Modeling State Awareness: The data acquisition module obtains real-time data from the field distributed control system. At this time, the steam supply pressure of heat source A is 1.05 MPa and the temperature is 285℃; the steam supply pressure of heat source B is 1.02 MPa and the temperature is 280℃.

[0111] The modeling module identified that the performance of the pipeline insulation layer had recently deteriorated due to rainfall.

[0112] The module substitutes the measured values ​​obtained on-site into the inverse optimization objective function. After 12 iterations, it increases the comprehensive heat transfer coefficient of a specific pipe section in the thermal-hydraulic process mechanism model from the original design of 2.5 W / (m²) to... 2 ·K) Corrected identification is 3.2W / (m 2 ·K).

[0113] Using the updated model for forward simulation, the platform outputs soft measurement state data of intermediate intersection node C without installed measuring points. The actual pressure of node C is calculated to be 0.91 MPa and the temperature is 245℃, thus avoiding the problem of overestimation of the state caused by directly using the original design parameters.

[0114] Load forecasting and inventory quantification: The analysis module retrieves production schedules from pharmaceutical companies in the system database and identifies a batch of high-steam-consuming reactors that will be heated at 9:00 AM. Based on historical model projections, it generates a demand load sequence predicting that the overall network demand will gradually increase from 160 t / h to 185 t / h over the next two hours.

[0115] Synchronous analysis module based on fluid state equation and 4200m 3 Based on the volume data, the current dynamic pipe inventory mass of the entire network is calculated to be 18.5 tons. According to the minimum steam supply pressure of 0.8 MPa required by the end-user's process, the theoretical minimum pipe inventory mass is calculated to be 14.2 tons. Subtracting the two, we quantify that the pipeline network has a maximum adjustable heat release capacity of 4.3 tons at this time.

[0116] Leakage decoupling diagnosis: During the load ramp-up phase from 9:00 to 9:30, the total input steam flow rate is 165 t / h, and the total output steam consumption is 175 t / h. At this time, the apparent mass loss calculated by the traditional accounting method is -10 t / h (i.e., the system input is less than the output).

[0117] The diagnostic module extracts the dynamic changes in pipe inventory mass during this period, finding that the inventory mass decreased from 18.5 tons to 12.5 tons (the pipeline is depressurizing and releasing heat to make up for the shortfall). The time derivative yields a transient mass imbalance of -12 t / h. Substituting this into the diagnostic formula: Actual leakage = 165 - 175 - (-12) = 2t / h.

[0118] The platform successfully filtered out the apparent negative deviation caused by normal heat storage and release in the pipeline network, restoring the true 2t / h micro-flange leakage rate of the system, without triggering an abnormal leakage alarm.

[0119] Optimal economic scheduling: During this period, the external electricity spot market was in its morning peak, with the nodal electricity price reaching a high of 0.75 yuan / kWh.

[0120] When solving the multivariable nonlinear programming constraints, the scheduling module incorporates the predicted load gap and the 4.3-ton heat release margin of the pipeline into the optimization space.

[0121] The output distribution control sequence generated by the solver did not directly increase the steam extraction rate of the two units to meet the 185 t / h heat load. Instead, it issued instructions to maintain the total steam extraction rate at a lower level of 170 t / h. The short-term load gap of 15 t / h was buffered and filled by the active depressurization expansion of the heating network (consuming the 4.3 tons of pipe storage).

[0122] Meanwhile, servo control commands drive the heat source unit to generate electricity using the saved high-pressure cylinder steam. Without affecting the steam safety of the end-user chemical plants, the system achieves decoupled thermoelectric operation during periods of high electricity prices, enabling the unit to generate more electricity and extract less steam, thus maximizing the plant's overall profits.

[0123] Experimental verification and effect comparison: To verify the actual effectiveness of this solution, a playback test was conducted using the historical data set of the physical pipeline network operation established in the above embodiments. The control group was set with conventional control logic (using static hydraulic parameters, decoupling without pipe storage, and unit operation based solely on heat load). The verification period was 30 consecutive calendar days. The complete experimental verification and effect comparison data are as follows: During the 30-day verification period, a total of 45 instances of severe pressure fluctuations in the pipeline network occurred.

[0124] Comparison group (conventional platform): Since the alarm judgment is directly based on the difference in apparent flow rate, when the pipeline is rapidly pressurized and the medium is stored, a large amount of injected steam is converted into the increase in pipe storage, which causes the apparent loss to soar. The comparison group system output a total of 38 false high-level leakage alarms.

[0125] Experimental group (the platform of this invention): After using the diagnostic module to remove transient mass imbalance, the apparent flow deviation was mitigated. The system correctly output only two actual flange gasket damage alarms within 30 days. The false alarm rate decreased from 84.4% to 0%.

[0126] Table 1. Comparison of Condition Estimation and Leak Diagnosis Test Data

[0127] Joint revenue accounting is conducted for scenarios involving fluctuations in spot market electricity prices.

[0128] Comparison Group (Conventional Platform): Due to the rigid constraint of heat-driven power generation, the unit's output trajectory is completely tied to the terminal heat load curve. During off-peak hours with low electricity prices and peak hours with high electricity prices, the unit's power output cannot be flexibly shifted, and the 30-day cumulative plant-wide comprehensive profit benchmark is set at 100%.

[0129] Experimental Group (Platform of this Invention): Relying on the precise quantification of the instantaneous maximum adjustable heat storage and maximum adjustable heat release in the pipeline network by the scheduling module, a model predictive control strategy was implemented to extract more steam and pressurize for heat storage during low electricity price periods and extract less steam and pressurize for heat release and increase power generation during high electricity price periods. Test results show that the system's thermo-electric decoupling operation range was widened by approximately 12%, and while ensuring the safety of the minimum anti-condensation temperature limit at each node, the cumulative overall plant profit increased to 108.4% over 30 days.

[0130] Table 2. Comparison of Test Data on Benefits of Multi-Heat Source Joint Economic Dispatch

[0131] in conclusion:

[0132] Combination Figure 6 (a) The bar chart comparing the accuracy of state estimation and leak diagnosis clearly shows the intuitive mapping relationship of the above data. This group of bars uses grayscale to distinguish different data dimensions: light gray bars represent the actual number of flange failures, and dark gray bars represent the number of false alarms output by the system. The graph shows that the height of the light gray bars in the control group and the experimental group is the same, both being 2, indicating that the present invention does not cause the underreporting of actual leaks after removing transient quantities; The dark gray columns appeared 38 times in the control group, but dropped to 0 times in the experimental group. Figure 6 (a) and Table 1 intuitively demonstrate that the diagnostic module of the present invention can filter out false loss data and reduce the false alarm rate to zero when dealing with sharp fluctuations in pipeline pressure.

[0133] Combination Figure 6 (b) The cumulative profit comparison line chart of multi-heat source joint economic dispatch over 30 consecutive days shows the dynamic evolution of the overall profit. This line chart records the trajectory of cumulative profit changes during the verification period. Among them, the black dashed line marked with hollow squares represents the comparison group, which accumulates linearly over time and eventually converges to the 100% baseline horizontal line in the figure; The solid black line marked with a solid triangle represents the experimental group. As the number of days for verification on the horizontal axis progresses, the solid black line, after experiencing multiple time-of-use electricity price fluctuation cycles, gradually widens the gap between its vertical axis value and the dashed black line due to the excess revenue generated by flexible storage scheduling, reaching a final value of 108.4% on the 30th day.

[0134] The curve trend verifies that after the pipeline network is transformed into a flexible energy storage space and incorporated into the optimization constraints, the system has the ability to improve economic benefits in the long term and in a stable manner.

Claims

1. A city-wide multi-heat-source steam heating collaborative scheduling and control platform based on load analysis, characterized in that: include: The acquisition module obtains real-time running data and transmits it to the modeling module; The modeling module uses the real-time operating data to correct the pre-established thermal-hydraulic process mechanism model, and outputs soft-sensor state data through the corrected thermal-hydraulic process mechanism model. The analysis module combines the soft-sensor status data with external parameters to calculate the demand load sequence, dynamic pipe inventory mass, instantaneous maximum heat storage adjustable amount, and maximum heat release adjustable amount; The diagnostic module uses the soft measurement status data and the dynamic pipe inventory quality to calculate the actual leakage amount and calculates the node superheat parameters to generate the lower limit of temperature control. The scheduling module uses the demand load sequence, the instantaneous maximum adjustable heat storage capacity, the maximum adjustable heat release capacity, the temperature control lower limit, and the economic settlement boundary parameters as constraints to perform optimization and solve for the output allocation control sequence. The output allocation control sequence is then converted into control commands and transmitted to the actuator.

2. The urban multi-heat source steam heating collaborative scheduling and control platform based on load analysis according to claim 1, characterized in that, The pre-established thermal-hydraulic process mechanism model is established through the following steps: Construct a topological model of a multi-source steam heating network, and abstract the existing steam network system into a directed graph containing a set of nodes and a set of branches; Construct nodal continuity equations based on the law of conservation of fluid mass; Construct a branch hydraulic pressure drop model that characterizes the mapping relationship between the absolute pressure parameters at both ends of the branch and the steam mass flow rate parameters flowing through the branch; Based on the law of conservation of energy, a thermodynamic friction decay model of a branch is constructed to show the temperature decay along the branch inlet and outlet ends under the influence of external environmental parameters. The topology model, the node continuity equation, the branch hydraulic pressure drop model, and the branch thermodynamic friction loss model are integrated to form a set of multivariate nonlinear equations describing the characteristics of the steam pipeline network. The set of multivariate nonlinear equations constitutes the pre-established thermal-hydraulic process mechanism model.

3. The urban multi-heat source steam heating collaborative scheduling and control platform based on load analysis according to claim 1, characterized in that, The modeling module uses the real-time operating data to correct the pre-established thermal-hydraulic process mechanism model, and outputs soft-sensor state data through the corrected thermal-hydraulic process mechanism model, including: Determine the spatial mapping relationship of the real-time operating data in the pipeline network topology and define the observable set; Extract the actual measurement values ​​within the observable set from the real-time operating data, and construct a reverse optimization objective function based on the residual between the actual measurement values ​​and the theoretical values ​​obtained from the forward simulation calculation of the thermal-hydraulic process mechanism model; Substitute the actual measured values ​​into the reverse optimization objective function, and iteratively adjust the comprehensive hydraulic resistance coefficient and the comprehensive heat transfer coefficient of the pipeline in the thermal-hydraulic process mechanism model until the residual target value meets the convergence requirement. The thermal-hydraulic process mechanism model, after updating the comprehensive hydraulic resistance coefficient and the comprehensive heat transfer coefficient of the pipeline, will be re-executed with forward fluid network simultaneous calculation to output the soft-sensor state data.

4. The urban multi-heat source steam heating collaborative scheduling and control platform based on load analysis according to claim 1, characterized in that, The analysis module combines the soft-sensor status data with external parameters to calculate the demand load sequence, including: Extract the continuous historical steam flow sequence of each industrial heat user to construct a historical load feature vector, obtain external meteorological parameters as the external parameters to construct a meteorological environment feature vector, and extract the time node of the production shift to construct the user's process production cycle feature identifier. The machine learning algorithm model is invoked to input the historical load feature vector, the meteorological environment feature vector, and the user process production cycle feature identifier into the prediction mapping network to output a single-point steam demand prediction value; A linear summation operation is performed on the single-point steam demand forecast values ​​along the spatial dimension to generate the overall demand load sequence; The overall demand load sequence is generated by continuously recursively calculating the overall demand load sequence according to the set time window length and time resolution.

5. The urban multi-heat source steam heating collaborative scheduling and control platform based on load analysis according to claim 1, characterized in that, The analysis module combines the soft-sensor status data with external parameters to calculate the dynamic tube storage mass, the instantaneous maximum adjustable heat storage capacity, and the maximum adjustable heat release capacity, including: The local steam storage mass is calculated by multiplying the pipe volume by the average steam density determined from the soft-sensor state data; The dynamic pipeline storage quality is obtained by performing an integral accumulation procedure on the local steam storage quality along the entire network spatial dimension. By using the fluid state equation in conjunction with the upper limit of the highest operating pressure and the lower limit of the lowest steam supply pressure, which are the external parameters, the theoretical maximum and minimum pipe mass are solved. The instantaneous maximum adjustable heat storage capacity is obtained by subtracting the dynamic tube storage mass from the theoretical maximum tube storage mass. The maximum adjustable heat release is obtained by subtracting the theoretical minimum tank mass from the dynamic tank mass.

6. The urban multi-heat source steam heating collaborative scheduling and control platform based on load analysis according to claim 1, characterized in that, The diagnostic module uses the soft-sensor status data and the dynamic pipe quality to calculate the actual leakage, including: The total input steam flow rate of the entire network is obtained by summing the real-time steam mass flow rate of each heat source injection node extracted from the soft measurement status data, and the total output steam flow rate of the entire network is obtained by summing the real-time steam consumption mass flow rate of each heat user extraction node extracted from the soft measurement status data. The difference between the two is used to obtain the apparent mass loss. Extract the dynamic storage quality corresponding to adjacent time steps and perform time differentiation to obtain the transient quality imbalance. The actual leakage is obtained by algebraically subtracting the transient mass imbalance from the apparent mass loss.

7. The urban multi-heat source steam heating collaborative scheduling and control platform based on load analysis according to claim 1, characterized in that, The diagnostic module calculates the node superheat parameters to generate the lower limit of temperature control, including: The absolute pressure values ​​of each node in the soft-sensor state data are extracted to calculate the corresponding saturation temperature; Extract the actual operating steam temperature from the soft measurement status data, and subtract the corresponding saturation temperature from the actual operating steam temperature to obtain the node superheat parameter; The saturation temperature of each node is algebraically added to the safe superheat margin threshold set in the system database to generate the minimum anti-condensation temperature baseline required to maintain a pure vapor phase state. The minimum anti-condensation temperature baseline is used as the lower limit of temperature control for each intersection node and user extraction node.

8. The urban multi-heat source steam heating collaborative scheduling and control platform based on load analysis according to claim 1, characterized in that, The diagnostic module is also used to calculate the overall heat decay rate of the entire network, including: The steam specific enthalpy and specific internal energy values ​​are calculated based on the absolute pressure and steam temperature parameters of each node in the soft-sensor state data. The specific internal energy value is combined with the internal steam mass of each branch to obtain the total internal energy value of the internal steam in the pipeline network, and time differential calculation is performed to obtain the rate of change of the total internal energy of the internal steam with time. The total extracted heat power and the total system input heat power are calculated based on the soft measurement status data, and the leakage heat power is calculated based on the actual leakage amount. The total heat attenuation rate of the entire network is obtained by subtracting the total extracted heat power, the leaked heat power, and the transient heat power fluctuation deviation caused by the rate of change of the total internal energy of the internal steam over time from the total input heat power of the system in sequence.

9. The urban multi-heat source steam heating collaborative scheduling and control platform based on load analysis according to claim 1, characterized in that, The scheduling module uses the demand load sequence, the instantaneous maximum adjustable heat storage capacity, the maximum adjustable heat release capacity, the temperature control lower limit, and the economic settlement boundary parameters as constraints to optimize and solve for the output allocation control sequence, including: The total system revenue and total system cost within a set time window are calculated using the aforementioned economic settlement boundary parameters. By integrating the revenue model and the cost model, the difference between the total revenue value and the total cost value of the system is calculated by time integration, and a global comprehensive profit index value reflecting the overall economic performance of the entire plant is constructed. Maximizing the global comprehensive profit index is used as the sole evaluation index of the multivariate nonlinear programming solution program, and the demand load sequence, the instantaneous maximum adjustable heat storage, the maximum adjustable heat release, and the lower limit of temperature control are used as the boundary constraints of the multivariate nonlinear programming solution program to obtain the output allocation control sequence.

10. The urban multi-heat source steam heating collaborative scheduling and control platform based on load analysis according to claim 1, characterized in that, The scheduling module converts the power allocation control sequence into control commands and transmits them to the actuator, including: The first control vector in the output allocation control sequence is extracted as the initial action value of the current cycle, and the target value of active power and the target value of extraction steam mass flow rate for the generator set are separated. A load adjustment signal, which serves as the control command, is sent to the generator set to drive the servo hydraulic motor to change the opening of the high-pressure cylinder speed regulating valve to match the target value of active power. The extraction steam pressure and flow rate offset signals, which serve as control commands, are sent to the heating control subsystem to drive the heating butterfly valve or rotary baffle to change the flow area to match the target value of the extraction steam mass flow rate. The control commands are transmitted to the generator set and the heating control subsystem, which are the actuators.