High-low pressure bypass heat supply equipment of steam turbine

Abstract

The invention discloses steam turbine high-low pressure bypass heat supply equipment, which relates to the technical field of heat energy engineering and power machinery, and comprises a signal acquisition module, an offset early warning module, a correction control module and a stable operation module, and constructing a time anchor point set, and generating a phase residual frequency spectrum for identifying a steam flow fluctuation source. According to the method, signals are collected through a unified time base line, a steam fluctuation source is identified in combination with a phase residual frequency spectrum, an impact track is reconstructed through phase conjugate inversion, and an overtemperature risk window is identified. Feedforward correction control is generated through track playback and predictive mapping, real-time rearrangement of valve behaviors is achieved, and the operation stability and the overtemperature resistance of the steam turbine under the deep peak regulation and disturbance conditions are improved.

Description

Technical Field

[0001] This invention relates to the field of thermal energy engineering and power machinery technology, specifically to a high and low pressure bypass heating device for a steam turbine. Background Technology

[0002] "High and low pressure bypass heating equipment for steam turbines" refers to a type of heating device installed during the operation of steam turbines in thermal power plants to achieve flexible distribution and efficient utilization of steam. It adds bypass channels, valves, and desuperheating and pressure-reducing devices between the high-pressure and low-pressure sections of the steam turbine, allowing some steam to bypass the turbine's power output and directly enter the heating system network or heat exchanger, thus achieving thermoelectric decoupling. The core significance of this equipment is that it can still guarantee stable heating demand when the power plant is operating under low power load or deep peak-shaving conditions, not only improving the unit's heating efficiency and flexibility but also reducing coal consumption and operating costs. Based on the analysis of the materials, this equipment, through the coordinated control of high and low bypass systems, a reasonable thermoelectric decoupling strategy, and optimized design of the heating network, can effectively improve the peak-shaving capacity of the cogeneration unit by 20%, reduce the minimum output of the unit to 40%-50% of the rated capacity, and achieve a peak-shaving improvement of 15%-20% under pure condensing conditions. Thus, it takes into account heating capacity, energy conservation and emission reduction, and environmental protection, and provides advanced technical support for power plants to achieve energy transformation and industrial upgrading.

[0003] The existing technology has the following shortcomings: In existing technologies, during dynamic peak-shaving operation of steam turbines, the intermediate-pressure regulating valve typically relies on control feedback signals to coordinate its operation with the high-pressure and low-pressure bypasses, ensuring that steam flow and temperature remain within a stable range. However, under rapid load fluctuations, the feedback signal is prone to delays or deviations, causing the intermediate-pressure regulating valve to deviate in the opposite direction. Valves that should gradually contract may open abnormally, resulting in a sudden surge in steam flow within a very short time. This nonlinear disturbance causes a rapid increase in the internal temperature of the intermediate-pressure cylinder, leading to an overheating condition. Due to the transient nature of temperature shocks, the intermediate-pressure cylinder blade material can easily experience strength decay and instantaneous softening within seconds, potentially causing complete blade failure or even blade ejection.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a high and low pressure bypass heating device for steam turbines to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a high and low pressure bypass heating device for a steam turbine, comprising a signal acquisition module, an offset early warning module, a correction control module, and a stable operation module: The signal acquisition module collects the time series of steam pressure and regulating valve opening under a unified time baseline, constructs a set of time anchor points, and generates a phase residual spectrum to identify the source of steam flow fluctuations. The offset early warning module identifies the precursors of valve opening offset based on the phase residual spectrum, performs phase conjugate inversion to reconstruct the steam flow impact trajectory, and marks the dynamic time window that triggers the risk of overheating in the intermediate pressure cylinder. The calibration control module generates a calibration vector based on a dynamic time window, injects it into a counterfactual playback chain to execute multi-level trajectory playback, and constructs a predictive calibration mapping to form a control surface constraining the valve opening, thereby realizing feedforward control of steam flow. The stable operation module promotes the convergence of predictive calibration mapping based on the control surface constraining the valve opening, constructs a time-reversal conformal closed loop, applies frequency misalignment traction and time-scale traction strategies, and performs amplitude-limited write-back to rearrange the valve opening distribution in real time, suppressing the over-temperature instability of the intermediate-pressure cylinder and maintaining the stable operation of the control system.

[0007] Preferably, the phase residual spectrum generation steps are as follows: Under a unified time baseline, steam pressure time series and regulating valve opening time series are collected synchronously, and a set of time anchor points is constructed based on the synchronously collected data. Based on the time anchor set, a sliding window traversal analysis is performed to extract non-stationary feature points and construct a high-resolution time anchor set. Segmented frequency domain analysis is performed based on time anchor point set to generate phase residual spectrum and identify the phase difference distribution between steam pressure change signal and regulating valve opening change signal; Clustering is used to identify high residual segments in the phase residual spectrum. Back mapping is performed in combination with the original acquired data to mark high-risk time periods and extract valve response features before the steam flow surge, which serve as the input basis for subsequent trajectory reconstruction and control strategy formulation.

[0008] Preferably, in the process of generating the phase residual spectrum, the steam pressure change signal is used as a reference term and the valve opening change signal is used as a response term. The phase difference between the two in each time anchor point interval is calculated, and the phase residual is extracted through frequency correlation to identify the coupling relationship between the abnormal fluctuation source of steam flow and the control response lag.

[0009] The preferred method for dynamic time window annotation is as follows: Under the constraint of phase residual spectrum, identify the high residual segment where there is phase misalignment between steam pressure signal and control valve opening signal, and extract signal feature groups with causal sequence relationship in the high residual segment to identify the precursor signal of control valve opening deviation; Based on the identified high residual section, a phase conjugate inversion operation is performed. A mirror evolution path is constructed based on the historically acquired steam pressure signal and the control valve opening signal to reconstruct the impact trajectory of steam disturbance from the control valve to the inlet of the intermediate pressure cylinder. In the reconstructed impact trajectory, the period of violent steam flow fluctuation was identified, and the overlapping time interval was formed by matching the position of the precursor signal, which was marked as the dynamic time window that triggered the temperature shock of the intermediate pressure cylinder. Labels are added to the parameters of valve opening offset, steam flow impact peak, time delay interval, and intermediate pressure cylinder temperature rise slope in the dynamic time window as the basis for control strategy correction and early warning model training.

[0010] Preferably, the intermediate-pressure cylinder temperature rise slope parameter in the dynamic time window is determined by the time delay and amplitude correlation between the intermediate-pressure cylinder wall temperature change curve and the steam flow impact trajectory under historical operating conditions, and is used to determine the physical coupling strength of steam disturbance on the thermal response of the intermediate-pressure cylinder.

[0011] Preferably, the steps for forming the control surface for regulating valve opening are as follows: Extract the time series of regulating valve opening and steam pressure in the dynamic time window, analyze their changing trends and calculate the deviation from the ideal trajectory, and generate a correction vector that includes amplitude correction and time advance adjustment. After generating the correction vector, the correction vector is gradually injected into different historical steam flow scenarios to construct a multi-level trajectory playback path, simulate the feedback response of valve behavior to changes in steam pressure, and collect key indicators. During trajectory playback, feedback response data is used to construct a predictive calibration mapping relationship, establish a quantitative relationship between the correction vector parameters and the playback results, and introduce operating load conditions and environmental boundaries to enhance the robustness of the mapping. Based on the predicted calibration mapping relationship, a control surface constraining the valve opening is generated, mapping the input and output to a continuous surface, and limiting the minimum allowable value, maximum response speed and fluctuation range of the control valve opening; After completing the construction of the control surface constraining the opening of the regulating valve, the operation phase begins. The early fluctuation characteristics of steam pressure are matched with the dynamic time window characteristics, and the corresponding control surface segment is selected based on the prediction calibration mapping to limit the response behavior of the regulating valve and realize the feedforward control of steam flow. Preferably, the control surface constraining the valve opening is constructed by interpolation to limit the minimum allowable value, maximum rate of change, and fluctuation range of the valve opening under the combined action of steam pressure input and intermediate pressure cylinder temperature rise rate. This is used to predictively limit the valve behavior before disturbances occur.

[0012] Preferably, based on the control surface constrained by the valve opening, the predictive calibration mapping is dynamically converged, a time-reversal conformal closed loop is constructed, and frequency misalignment and time-scale traction strategies are superimposed. The following steps are performed to execute the amplitude-limited write-back: Based on the control surface constrained by the valve opening, the dynamic convergence process of the prediction calibration mapping relationship is promoted. Real-time operating status data is used to compare the prediction results with the actual control behavior, and incremental correction is made according to the deviation trend. Based on the convergence of the prediction calibration mapping, a time-reversal conformal closed-loop control structure is constructed, which integrates the historical backtracking path and the future trend path to form a two-way control convergence to determine the optimal control adjustment. By superimposing the golden ratio frequency misalignment traction strategy and the dual mirror time scale traction strategy in the time reversal conformal closed loop, the response behavior of the regulating valve is guided into a rhythmic, high-frequency, and highly symmetrical perturbation regulation state. Before executing the regulation command, a pulse-level limiting write-back mechanism is introduced. The control quantity is segmented according to the steam flow slope and the temperature rise rate of the intermediate pressure cylinder, and written back to the regulation path step by step to realize the real-time rearrangement and dynamic smooth control of the valve opening.

[0013] Preferably, in the process of promoting the dynamic convergence of the prediction calibration mapping relationship, the convergence parameter adjustment adopts a step-by-step approach, allowing only slight adjustments to the mapping value within each sampling period, and using the convergence speed and deviation reduction rate as evaluation criteria to ensure convergence stability and directionality under multi-dimensional operating conditions.

[0014] Preferably, during the pulse-level amplitude limiting write-back process, the adjustment amount of the regulating valve opening is set to the maximum allowable amplitude in stages according to the slope of the steam flow change and the temperature rise rate of the intermediate pressure cylinder, and the control amount is written back step by step in each pulse cycle to avoid violent valve action and achieve smoothing of the regulation path.

[0015] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention synchronously acquires multi-source operating signals based on a unified time baseline, accurately identifies the source of abnormal steam flow fluctuations by combining phase residual spectrum, and reconstructs the steam impact path through phase conjugate inversion, effectively capturing the risk window that may lead to over-temperature instability of the intermediate-pressure cylinder. Based on this, a feedforward correction control path is constructed, generating dynamic valve opening control boundaries through counterfactual trajectory playback and predictive calibration mapping, thereby achieving active regulation of steam flow. Especially during dynamic operation, the equipment employs a time-inversion conformal closed-loop and amplitude-limiting write-back mechanism, combined with golden ratio frequency misalignment traction and dual-mirror time-scale guidance, to achieve real-time rearrangement and optimized control of the regulating valve behavior. This improves the operational stability and control robustness of the turbine system under deep peak shaving, high-frequency disturbances, and complex boundary conditions, effectively preventing strength attenuation and structural failure of the intermediate-pressure cylinder blades due to temperature shocks. 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 This is a schematic diagram of a high and low pressure bypass heating device for a steam turbine according to the present invention. Detailed Implementation

[0018] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0019] This invention provides, for example Figure 1 The steam turbine high and low pressure bypass heating equipment shown includes a signal acquisition module, an offset early warning module, a correction control module, and a stable operation module. The signal acquisition module simultaneously acquires the steam pressure time series and the regulating valve opening time series under a unified time baseline. Based on the acquisition results, it constructs a time anchor point set and generates a phase residual spectrum based on the time anchor point set to identify the source of steam flow fluctuation. To address the issues of sudden steam flow surges and intermediate-pressure cylinder over-temperature instability caused by valve control feedback lag during dynamic peak-shaving operation of steam turbines, a joint analysis method based on a unified time baseline and steam pressure and valve opening sequence is proposed. By utilizing a high-resolution time anchor point set and phase residual spectrum extraction, the method identifies and predicts steam fluctuation sources, providing accurate data for subsequent control actions. The specific steps are as follows: During the actual operation of the steam turbine, two core operating parameters are simultaneously collected under a high-frequency sampling mechanism: the steam pressure time series in the steam channel and the opening time series of the intermediate-pressure regulating valve. To ensure a strictly consistent time reference point for both sets of data, the master clock of the turbine control center is used as the unified benchmark, and a timestamp is assigned to each sampling point to ensure complete alignment of data across different physical measurement paths in the time dimension. Steam pressure data is acquired in real time by piezoelectric pressure sensors located at multiple key nodes such as the main steam channel inlet, high-pressure cylinder outlet, and intermediate-pressure cylinder inlet. The sampling frequency should be controlled at the millisecond level, prioritizing industrial-grade sensor equipment with high stability and fast response capabilities. Valve opening is acquired in real time through the actuator feedback signal of the intermediate-pressure regulating valve, typically a continuous proportional opening value between 0% and 100%. The sampling frequency should be consistent with the pressure data to avoid phase drift errors during data fusion. All sampled data must be transmitted to the data processing unit in real time, and the raw data must be paired on the time axis using a unified time label to form a one-to-one time series pair, ensuring data consistency and comparability in subsequent processing.

[0020] After completing synchronous data acquisition and constructing basic time series pairs, the acquired steam pressure time series and valve opening time series are correlated and identified to build a high-density set of time anchor points. This set is not an equidistant set of time points, but rather a set of key time points with abrupt changes, fluctuation boundaries, or inflection points extracted by analyzing non-stationary change segments in the two types of series. These key time points serve as the time anchoring basis for potential disturbance sources. For example, when the valve opening changes by a large amplitude in a very short time, or when the steam pressure curve shows abnormal peaks, dips, or oscillations, it can be determined that there is a potential disturbance response process at such locations, which should be recorded as highly sensitive time anchor points. To improve the time-series resolution of the anchor point set, this implementation method uses a sliding window traversal approach to analyze the original sequence segment by segment, evaluating the signal fluctuation amplitude, rise slope, fall rate, and local extreme value change amplitude within different time periods. Based on the analysis results, a set of time points that conform to the characteristics of non-steady-state changes are extracted. Furthermore, by combining the causal relationship that steam pressure changes precede valve behavior or valve changes precede pressure changes, the verification and screening of key time points are completed, constructing a time anchor point set that highly reflects the system disturbance characteristics, providing a time input basis for subsequent spectrum construction.

[0021] Based on the established time anchor point set, a joint frequency domain analysis is performed on the steam pressure time series and valve opening time series within the time period defined by this set. This constructs a phase residual spectrum to reveal the causal relationship between abnormal fluctuations in steam flow and control lag. In practice, the time segment between each pair of time anchor points is used as the base interval. The corresponding steam pressure and valve opening subsequences within this interval are extracted, and the two sets of subsequences are synchronized to avoid phase misalignment or time drift. During spectrum generation, global Fourier analysis is not used. Instead, local frequency components are extracted segment by segment based on each time anchor point, forming multiple micro-spectral segments reflecting the system state response in the local frequency domain. Within each segment, the phase difference and frequency correlation between the two signals are calculated by using the pressure change signal as the reference term and the valve opening change as the response term, thus obtaining the phase residual distribution. Ideally, the opening change of the regulating valve should exhibit a high degree of phase coordination with the steam pressure response. However, in reality, due to control delays, actuator inertia, and nonlinear effects introduced by load disturbances, abnormal characteristics such as phase misalignment, response shift, and even reverse coupling may occur during certain time periods. By systematically collecting the phase residuals from different time anchor points, the regular location of abnormal steam flow fluctuations can be gradually revealed, and the time window before the problem occurs and the degree of anomaly in the associated signals can be clearly identified, providing a basis for problem localization and prediction.

[0022] Based on the extraction results of the phase residual spectrum, high residual segments in multiple spectral bands are clustered and identified, and a group of high-risk time periods with significant phase mismatch characteristics are selected as potential disturbance source time periods. Simultaneously, these markers are back-mapped with the original valve opening data to identify the initial valve response behavior and related steam pressure mutation events that trigger sudden increases in steam flow. Through this process, disturbance points can be identified not only in the time domain but also the causes of response lag can be revealed in the frequency domain, achieving composite identification of flow fluctuation sources in both time and frequency spaces. Furthermore, based on these high-risk time periods and their associated signal characteristics, precise time entry points and disturbance templates are provided for the next stage of trajectory reconstruction, correction vector generation, and predictive control mapping. This ensures that the subsequent control strategy formulation process can be directly based on engineering-significant physical disturbance phenomena, avoiding the error accumulation caused by relying on empirical judgment or theoretical approximation in traditional modeling processes, thereby significantly improving the regulation accuracy and operational stability during turbine heating operation.

[0023] The offset early warning module identifies the precursor signal of the valve opening offset under the phase residual spectrum constraint, performs phase conjugate inversion operation to reconstruct the steam flow impact trajectory, and marks the dynamic time window that causes the risk of overheating of the intermediate pressure cylinder of the steam turbine in the reconstruction result. To further achieve early identification and dynamic risk warning of nonlinear disturbances in steam flow, a steam flow impact trajectory reconstruction method based on the phase inversion principle is proposed, building upon the existing phase residual spectrum. This method aims to effectively capture precursors of regulating valve opening deviations and accurately label the dynamic range that triggers overheating in the intermediate-pressure cylinder through key signal conjugate analysis and time window identification. The specific steps are as follows: The generated phase residual spectrum has revealed significant phase misalignment and phase drift between steam pressure and regulating valve opening signals across multiple time periods. This step focuses on these high residual segments to construct a steam flow disturbance prediction model. To ensure the effectiveness of early warning identification, all spectral segments with phase residuals exceeding a set sensitivity threshold are selected as analysis targets. The pressure fluctuation curves and valve opening curves within these segments are cross-matched to extract signal feature groups with causal relationships. For example, if a pressure curve shows leading fluctuations while the valve opening response is delayed in a certain spectral segment, and the phase difference exceeds the critical angle change value, it can be determined that there may be a precursor to valve control mismatch within this segment. To further confirm the effectiveness of this precursor signal, a differential dynamic comparison sequence is constructed using pressure change as the leading signal and opening change as the response signal within this time period. This sequence quantifies the development speed and duration of the reverse offset trend, thereby extracting high-potential-risk disturbance trigger points as the basis for identifying offset precursor signals.

[0024] After identifying the precursors of the control valve offset, this step performs phase conjugate inversion processing on the identified high-risk time periods to reconstruct the steam flow impact trajectory and simulate the influence path of the steam disturbance transmission process inside the turbine. Unlike traditional mathematical modeling or simulation fitting methods, this invention employs the signal conjugate mapping principle. Based on the actual acquired pressure-opening signal pairs, it performs a mirror inversion operation in the time dimension to construct an equidistant signal evolution path, used to deduce the energy delivery characteristics of the steam disturbance from the control valve output to the intermediate-pressure cylinder inlet. During this process, to ensure the physical rationality of the impact trajectory reconstruction, a series of physical boundary conditions are introduced, such as the inertial response delay of the intermediate-pressure cylinder steam channel, the thermal inertia of the metal inner wall, and the propagation speed of the pressure wave in the steam medium. These physical quantities are dynamically corrected based on real-time operating data, resulting in a steam impact path curve that closely approximates the actual thermodynamic process. Such curves typically exhibit a multi-peak fluctuation shape, with a rapid rise and slow decay energy release characteristic, which is more pronounced under conditions of severe load disturbance.

[0025] After reconstructing the steam flow impact trajectory, this step continues to identify dynamic time windows that could trigger overheating risks in the turbine's intermediate-pressure cylinder based on the trajectory results. Specifically, the time period in the reconstructed curve where the steam flow slope changes most drastically and the fluctuation exceeds a set threshold is selected as the candidate window range. A secondary intersection screening is performed using the precursor points of the previous stage to retain overlapping time intervals that are supported by the phase residual spectrum and exhibit a sudden energy increase trend in the impact trajectory. Furthermore, to assess the impact of these candidate windows on the intermediate-pressure cylinder temperature change, the response curve of the intermediate-pressure cylinder wall temperature from historical operating data is introduced to match the time delay and response amplitude between the reconstructed results and the measured temperature rise curve. When the time difference between the peak value of a reconstructed trajectory and the inflection point of the intermediate-pressure cylinder temperature rise falls within a reasonable range, and the pressure-temperature correlation is significant, this segment can be confirmed as a typical dangerous window for steam disturbance-induced temperature shocks. The dynamic time window identified in this way is no longer based on static threshold judgment set by rules, but is a physical causal window extracted by combining the measured coupling results of multi-dimensional signals such as pressure, flow, and temperature, which is more adaptable and engineering-oriented.

[0026] To improve the efficiency of dynamic time windows in subsequent control strategy formulation, this step structures the dynamic time windows marked above and adds key metadata tags to each window, including parameters such as the precursor amplitude of the corresponding regulating valve opening deviation, the peak value of steam flow impact, the time delay interval, and the incremental value of the intermediate-pressure cylinder temperature rise slope. These structured window tags can not only be used as a basis for selecting correction points for subsequent control curves, but also constitute a source of training samples for the early warning model of faults. In actual operation, once it is found that the trend of the phase residual spectrum in the current operating data is highly similar to the characteristics of historical windows, the early warning process can be triggered in advance, or the valve opening change rate can be actively adjusted according to the window position, thereby completing the intervention operation before the disturbance forms a significant temperature impact.

[0027] The calibration control module generates a calibration vector for adjusting valve trajectory correction based on a dynamic time window. The calibration vector is input into the counterfactual playback chain for multi-level trajectory playback. During the trajectory playback process, a predictive calibration mapping relationship is constructed to generate a control surface for valve opening constraint to achieve predictive feedforward control of steam flow. To achieve feedforward control of intermediate-pressure cylinder temperature anomalies caused by steam disturbances, a trajectory correction and control constraint construction method is proposed based on the identified dynamic time window. This method can generate a correction vector based on disturbance precursors and achieve dynamic optimization of the regulating valve behavior through playback and prediction mapping. The specific steps are as follows: The dynamic time window clearly marks the steam flow impact range caused by the valve opening deviation and its corresponding intermediate-pressure cylinder temperature rise risk point. To proactively correct control behavior, starting from this dynamic time window, the valve opening time series and steam pressure response series before and after the window are extracted, and their changing trends, slope distributions, and fluctuation amplitudes are analyzed to identify the dominant driving factors of steam flow fluctuation trends. Based on this, according to the known correlation structure between valve opening response and steam impact behavior, a set of reference trajectories representing the ideal control target is constructed, and these trajectories are used as a benchmark to calculate the deviation between the current actual trajectory and the target trajectory in both time and amplitude dimensions. These deviations will serve as the basis for generating a correction vector, which includes not only corrections to the deviation amplitude but also adjustments to the response delay time advance. This correction vector possesses both temporal continuity and behavioral consistency, ensuring a smooth transition during subsequent control curve corrections, thereby avoiding secondary disturbances during control implementation.

[0028] After generating the correction vector, to verify its dynamic repair capability for regulating valve behavior and evaluate its adaptability under different operating boundaries, a response analysis mechanism based on multi-level trajectory playback is proposed. This involves progressively injecting the correction vector into different historical steam flow scenarios to construct multiple virtual playback paths, simulating the feedback response of valve behavior to steam pressure changes under different control intensities and regulation starting points. This process is similar to conducting comparative experiments on the same correction strategy under different operating boundary conditions and observing its actual contribution to mitigating steam shock in typical scenarios. Each playback path is based on historical measured data, while introducing the correction vector as a disturbance correction term, maintaining its time anchor point aligned with the previous dynamic window to ensure the targeted nature of the control behavior. During playback, key indicators under each path are continuously collected, including the maximum steam flow fluctuation, response time length, and intermediate-pressure cylinder temperature rise slope, to form an evaluation system for the correction strategy.

[0029] Based on the feedback response data collected during trajectory playback, a predictive calibration mapping relationship is further constructed. The core of this mapping relationship lies in establishing a quantitative connection between the correction vector parameters and the playback results, thereby achieving a forward-looking data-driven predictive capability. When constructing the mapping, not only are the response differences of the input vector under different playback scenarios considered, but also multiple sets of operating load conditions, ambient temperature boundaries, and initial steam states are introduced as disturbance factors to ensure the robustness of the mapping relationship. The predictive calibration mapping relationship can be regarded as a dynamic response reference map combining historical behavior, current operating conditions, and control objectives, enabling real-time prediction and trend correction of valve behavior during future regulation. Especially under nonlinear disturbance conditions, this mapping relationship effectively compensates for the insufficient control accuracy of traditional linear regulation strategies under complex boundary conditions by recording the state transition characteristics presented by multi-factor response combinations within different time intervals.

[0030] After establishing the predictive calibration mapping relationship, a control surface constraining the valve opening is further constructed to achieve the physical implementation of the real-time control strategy. This control surface is a dynamic boundary structure generated based on the predictive calibration mapping, defining the behavioral range that the valve opening should follow under different steam pressure inputs and operating conditions. Specifically, during the construction of the control surface, each input-output pair in the mapping relationship is mapped to three-dimensional coordinate points, and a continuous surface covering the control area is constructed through interpolation. This surface defines the minimum allowable value, maximum response speed, and fluctuation limit range of the valve opening under specific steam disturbance trends. Through this control surface, a predictive behavioral boundary can be provided before the valve responds, guiding it to adjust along the optimal path, thereby avoiding secondary shocks caused by blind adjustment. At the same time, this control surface also has adaptive evolution capabilities, and can be iteratively optimized based on continuously injected playback data and operating condition feedback, thereby achieving real-time adaptation to dynamic changes in the operating boundary.

[0031] After constructing the control surface constraining the valve opening, the entire control strategy enters the feedforward control process during operation. In actual operation, once early fluctuations in the steam pressure time series are detected and highly matched with the identified dynamic window characteristics, the control surface segment corresponding to the current steam disturbance can be quickly matched using predictive calibration mapping, thereby obtaining a real-time constraint boundary for the current operating condition. Under this boundary, the response behavior of the control valve will be limited to a preset reasonable range, avoiding adverse behaviors such as response lag, sudden increase in opening, or control reversal. In this way, the ability to respond early to steam disturbances can be effectively improved, preventing the intermediate-pressure cylinder from entering a thermal shock state in a short time, thus ensuring the stability and safety of the entire heating equipment operation.

[0032] The stable operation module promotes the dynamic convergence process of the prediction calibration mapping relationship based on the control surface of the regulating valve opening constraint, constructs a time-reversal conformal closed-loop control structure, superimposes the golden ratio frequency misalignment traction and dual mirror time-scale traction control strategy in the closed-loop structure, and performs pulse-level amplitude limiting back-write operation to rearrange the distribution of regulating valve opening in real time, dynamically suppressing the over-temperature instability of the intermediate pressure cylinder and maintaining the operating stability of the steam turbine system. After constructing the control surface constraining the valve opening, a closed-loop control structure with time-reversal characteristics is proposed to achieve adaptive evolution and accuracy convergence of the control strategy under dynamic operating conditions. Multiple traction mechanisms and a limit-rewrite mechanism are introduced to realize dynamic reconstruction and real-time optimized control of the valve behavior. The specific steps are as follows: After the control surface constraining the valve opening is constructed, to prevent mismatch or response lag due to external disturbances, load fluctuations, or changes in the steam source during actual operation, this implementation introduces a dynamic convergence mechanism for the predictive calibration mapping relationship. This convergence mechanism takes real-time acquired operating status data as input, including key variables such as current steam pressure, valve opening response speed, and the rate of change of the intermediate-pressure cylinder wall temperature, and compares it with the predictive calibration mapping constructed in the previous stage to calculate its deviation trend under the same operating conditions. When a persistent deviation or significant inconsistency in the rate of change between the actual control behavior and the predicted mapping result is detected, an iterative correction process for the mapping relationship is triggered. To ensure the convergence process has directionality and stability under multi-dimensional operating conditions, this implementation adopts an incremental parameter adjustment method, allowing only minor adjustments to the mapping value within each sampling period. Simultaneously, the convergence speed and deviation reduction rate are used as evaluation criteria to gradually promote the dynamic convergence behavior of the predictive calibration mapping on the control surface. The essence of this process is a data-driven self-integration process, which differs from the traditional static surface model and can continuously improve the adaptability of valve control strategies to actual operating conditions.

[0033] To improve the response speed and control sensitivity of the convergence mechanism, this invention constructs a conformal closed-loop control structure based on the time-reversal principle. The key feature of this structure is that it no longer relies solely on the current state error for closed-loop adjustment, but instead introduces a mirrored backtracking path of historical state information, combined with a future trend prediction path, to achieve bidirectional time-domain control intervention. Specifically, before each control cycle, the control structure first constructs a backward time-reversal path based on historical operating data to trace the contribution of previous valve behavior to the current system state; simultaneously, combined with a predictive calibration mapping, it constructs a forward future trend path at the current moment to predict the system's response state under the current control input. The two paths converge at the current moment, forming a time-conformal structure, and an error minimization rule is introduced into this structure to find the optimal control adjustment. Through this bidirectional control mechanism, not only is the ability to offset control lag improved, but the ability to proactively intervene in future disturbances is also enhanced, thus effectively overcoming the problem of insufficient response of traditional closed-loop control under sudden disturbances.

[0034] Building upon the time-reversal conformal closed-loop structure, this implementation superimposes a dual-traction mechanism to further enhance the rhythmic adjustment and dynamic guidance capabilities of the control strategy: a golden ratio frequency misalignment traction strategy and a dual-mirror time-scale traction strategy. The golden ratio frequency misalignment traction strategy uses the valve response cycle as a benchmark, introducing an asynchronous traction cycle constructed according to the golden ratio. This applies a non-periodic rhythmic disturbance to the valve behavior, breaking the system resonance or response inertia that may be formed by the inherent adjustment frequency. This asynchronous adjustment cycle can guide the valve behavior into a higher-frequency micro-perturbation response state without changing the control boundary, improving the valve's adaptability to short-period disturbances. Simultaneously, the dual-mirror time-scale traction strategy constructs a symmetrical time reference structure before and after the current moment, considering the symmetry of past and expected behaviors in the control timing arrangement, giving the valve adjustment behavior stronger inertia suppression and path stability. The synergistic effect of these traction mechanisms allows the adjustment process to no longer be limited by a linear timing response, but to complete high-frequency fine-tuning within a dynamic rhythm, effectively improving the real-time performance and anti-interference capability of the entire control process.

[0035] To ensure that the dynamic adjustment results in the control mechanism are ultimately translated into physical control commands, a pulse-level limiting write-back mechanism is introduced to achieve real-time rearrangement of the valve opening distribution. Specifically, after each control output is generated, this write-back mechanism does not directly act on the execution level. Instead, it first performs multi-level pulse limiting processing. Based on constraints such as the steam flow rate change slope, the intermediate-pressure cylinder temperature rise rate, and the valve's mechanical response characteristics, the upcoming opening adjustment is segmented and compressed, with different maximum allowable adjustment ranges set. Within each pulse cycle, the control quantity is written back to the valve adjustment command level by level according to the layered limiting results, forming a series of rhythmic, amplitude-limited fine-tuning signals. This method effectively avoids drastic valve movements caused by overly aggressive control strategies, thus preventing the recurrence of steam shock. Simultaneously, during the limit-down and write-back process, the spatial distribution of valve openings is dynamically rearranged, allocating the opening adjustments of highly sensitive areas to relatively slower response time periods. This smooths and structures the overall valve opening adjustment path, further suppressing the temperature surge trend of the intermediate-pressure cylinder and ultimately achieving comprehensive maintenance of the turbine system's operational stability. The innovation of this limit-down and write-back process lies in its nonlinear compression mechanism and dynamic allocation logic.

[0036] This invention synchronously acquires multi-source operating signals based on a unified time baseline, accurately identifies the source of abnormal steam flow fluctuations by combining phase residual spectrum, and reconstructs the steam impact path through phase conjugate inversion, effectively capturing the risk window that may lead to over-temperature instability of the intermediate-pressure cylinder. Based on this, a feedforward correction control path is constructed, generating dynamic valve opening control boundaries through counterfactual trajectory playback and predictive calibration mapping, thereby achieving active regulation of steam flow. Especially during dynamic operation, the equipment employs a time-inversion conformal closed-loop and amplitude-limiting write-back mechanism, combined with golden ratio frequency misalignment traction and dual-mirror time-scale guidance, to achieve real-time rearrangement and optimized control of the regulating valve behavior. This improves the operational stability and control robustness of the turbine system under deep peak shaving, high-frequency disturbances, and complex boundary conditions, effectively preventing strength attenuation and structural failure of the intermediate-pressure cylinder blades due to temperature shocks.

[0037] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A high and low pressure bypass heating device for a steam turbine, characterized in that, It includes a signal acquisition module, an offset early warning module, a correction control module, and a stable operation module: The signal acquisition module collects the time series of steam pressure and regulating valve opening under a unified time baseline, constructs a set of time anchor points, and generates a phase residual spectrum to identify the source of steam flow fluctuations. The offset early warning module identifies the precursors of valve opening offset based on the phase residual spectrum, performs phase conjugate inversion to reconstruct the steam flow impact trajectory, and marks the dynamic time window that triggers the risk of overheating in the intermediate pressure cylinder. The calibration control module generates a calibration vector based on a dynamic time window, injects it into a counterfactual playback chain to execute multi-level trajectory playback, and constructs a predictive calibration mapping to form a control surface constraining the valve opening, thereby realizing feedforward control of steam flow. The stable operation module promotes the convergence of predictive calibration mapping based on the control surface constraining the valve opening, constructs a time-reversal conformal closed loop, applies frequency misalignment traction and time-scale traction strategies, and performs amplitude-limited write-back to rearrange the valve opening distribution in real time, suppressing the over-temperature instability of the intermediate-pressure cylinder and maintaining the stable operation of the control system.

2. The steam turbine high and low pressure bypass heating equipment according to claim 1, characterized in that, The steps for generating the phase residual spectrum are as follows: Under a unified time baseline, steam pressure time series and regulating valve opening time series are collected synchronously, and a set of time anchor points is constructed based on the synchronously collected data. Based on the time anchor set, a sliding window traversal analysis is performed to extract non-stationary feature points and construct a high-resolution time anchor set. Segmented frequency domain analysis is performed based on time anchor point set to generate phase residual spectrum and identify the phase difference distribution between steam pressure change signal and regulating valve opening change signal; Clustering is used to identify high residual segments in the phase residual spectrum. Back mapping is performed in combination with the original acquired data to mark high-risk time periods and extract valve response features before the steam flow surge, which serve as the input basis for subsequent trajectory reconstruction and control strategy formulation.

3. A steam turbine high and low pressure bypass heating device according to claim 2, characterized in that, In the process of generating the phase residual spectrum, the steam pressure change signal is used as the reference term and the valve opening change signal is used as the response term. The phase difference between the two in each time anchor point interval is calculated, and the phase residual is extracted by frequency correlation.

4. A steam turbine high and low pressure bypass heating device according to claim 2, characterized in that, The steps for dynamic time window annotation are as follows: Under the constraint of phase residual spectrum, identify the high residual segment where there is phase misalignment between steam pressure signal and control valve opening signal, and extract signal feature groups with causal sequence relationship in the high residual segment to identify the precursor signal of control valve opening deviation; Based on the identified high residual section, a phase conjugate inversion operation is performed. A mirror evolution path is constructed based on the historically acquired steam pressure signal and the control valve opening signal to reconstruct the impact trajectory of steam disturbance from the control valve to the inlet of the intermediate pressure cylinder. In the reconstructed impact trajectory, the period of violent steam flow fluctuation was identified, and the overlapping time interval was formed by matching the position of the precursor signal, which was marked as the dynamic time window that triggered the temperature shock of the intermediate pressure cylinder. Labels are added to the parameters of valve opening offset, steam flow impact peak, time delay interval, and intermediate pressure cylinder temperature rise slope in the dynamic time window as the basis for control strategy correction and early warning model training.

5. A steam turbine high and low pressure bypass heating device according to claim 4, characterized in that, The intermediate-pressure cylinder temperature rise slope parameter in the dynamic time window is determined by the time delay and amplitude correlation between the intermediate-pressure cylinder wall temperature change curve and the steam flow impact trajectory under historical operating conditions. It is used to determine the physical coupling strength of steam disturbance on the thermal response of the intermediate-pressure cylinder.

6. A steam turbine high and low pressure bypass heating device according to claim 4, characterized in that, The steps for forming the constraint control surface for regulating valve opening are as follows: Extract the time series of regulating valve opening and steam pressure in the dynamic time window, analyze their changing trends and calculate the deviation from the ideal trajectory, and generate a correction vector. After generating the correction vector, the correction vector is gradually injected into different historical steam flow scenarios to construct a multi-level trajectory playback path, simulate the feedback response of valve behavior to changes in steam pressure, and collect key indicators. During trajectory playback, feedback response data is used to construct a predictive calibration mapping relationship, establish a quantitative relationship between the correction vector parameters and the playback results, and introduce operating load conditions and environmental boundaries. Based on the predicted calibration mapping relationship, a control surface constraining the valve opening is generated, mapping the input and output to a continuous surface, and limiting the minimum allowable value, maximum response speed and fluctuation range of the control valve opening; After completing the construction of the control surface constraining the valve opening, the operation phase begins. The early fluctuation characteristics of steam pressure are matched with the dynamic time window characteristics, and the corresponding control surface segment is selected based on the prediction calibration mapping to limit the response behavior of the control valve.

7. A steam turbine high and low pressure bypass heating device according to claim 6, characterized in that, The control surface constraining the valve opening is constructed by interpolation to limit the minimum allowable value, maximum rate of change, and fluctuation range of the valve opening under the combined action of steam pressure input and intermediate pressure cylinder temperature rise rate. This is used to predictively limit the valve behavior before disturbances occur.

8. A steam turbine high and low pressure bypass heating device according to claim 6, characterized in that, Based on the control surface constrained by the valve opening, dynamic convergence of the predictive calibration mapping is promoted. A time-reversal conformal closed loop is constructed, and frequency misdirection traction and time-scale traction strategies are superimposed. The following steps are performed to execute the amplitude-limited write-back: Based on the control surface constrained by the valve opening, the dynamic convergence process of the prediction calibration mapping relationship is promoted. Real-time operating status data is used to compare the prediction results with the actual control behavior, and incremental correction is made according to the deviation trend. Based on the convergence of the prediction calibration mapping, a time-reversal conformal closed-loop control structure is constructed, which integrates the historical backtracking path and the future trend path to form a two-way control convergence to determine the optimal control adjustment. By superimposing the golden ratio frequency misalignment traction strategy and the dual mirror time scale traction strategy in the time reversal conformal closed loop, the response behavior of the regulating valve is guided into a rhythmic, high-frequency, and highly symmetrical perturbation regulation state. Before executing the regulation command, a pulse-level limiting write-back mechanism is introduced. The compression control quantity is segmented according to the steam flow slope and the intermediate pressure cylinder temperature rise rate, and then written back to the regulation path step by step.

9. A steam turbine high and low pressure bypass heating device according to claim 8, characterized in that, In the process of promoting the dynamic convergence of the prediction calibration mapping relationship, the convergence parameter adjustment adopts a step-by-step approach. Only a small adjustment of the mapping value is allowed in each sampling period. The convergence speed and the deviation reduction rate are used as evaluation criteria to ensure the convergence stability and directionality under multi-dimensional working conditions.

10. A steam turbine high and low pressure bypass heating device according to claim 8, characterized in that, During the pulse-level amplitude limiting write-back process, the adjustment amount of the regulating valve opening is set according to the slope of the steam flow change and the temperature rise rate of the intermediate pressure cylinder, and the control amount is written back step by step in each pulse cycle to avoid violent valve action and achieve smoothing of the regulation path.

Images