A method for setting PID parameters of a steam generator liquid level control system

Abstract

The application relates to the technical field of industrial process automatic control, and particularly discloses a PID parameter setting method of a steam generator liquid level control system, which comprises the following steps: first, collecting high-frequency pressure fluctuation signals of a feedwater pipeline, and extracting energy data of a characteristic frequency band through frequency spectrum conversion; then, generating a real-time confidence index representing a small leakage probability by analyzing the deviation and variation trend of the energy data relative to a reference value; and finally, dynamically adjusting parameters and logic of integral action in a liquid level PID controller according to the index: when the confidence level is low, conservative parameters are used to suppress false water level; when the confidence level is medium, integral accumulation is frozen and an attenuation water supply bias is applied; and when the confidence level is high, the integrator input is switched to a flow deviation signal based on material balance and active parameters are used; the application realizes early intelligent identification and differentiated control of interference sources, and improves the self-adaptive ability and intrinsic safety level of the liquid level control system.

Description

Technical Field

[0001] This invention relates to the field of industrial process automatic control technology, and specifically to a PID parameter tuning method for a steam generator liquid level control system. Background Technology

[0002] Liquid level control in steam generators (such as boilers and nuclear power plant steam generators) is a core element in ensuring their safe and stable operation. Excessive liquid level can lead to water carryover in the steam, damaging downstream turbine equipment; insufficient liquid level can expose the heating tube bundle, causing overheating or even tube rupture. Therefore, maintaining the liquid level near the set value is a crucial control objective.

[0003] In existing steam generator level control, the initial signal characteristics of "false water level" and "real leakage" are similar, which makes it impossible for integral-dominant PID controllers with fixed conservative parameters to effectively distinguish between them. As a result, when a minor leak occurs, the controller will misjudge it as a "false water level" that needs to be suppressed and continuously generate erroneous control commands in the opposite direction (reducing water supply). This creates a dangerous positive feedback loop that exacerbates leakage and delays response, seriously threatening the safe operation of the system. Summary of the Invention

[0004] The purpose of this invention is to provide a PID parameter tuning method for a steam generator level control system to solve the problems mentioned above.

[0005] The objective of this invention can be achieved through the following technical solutions: A PID parameter tuning method for a steam generator level control system, S1: Acquire the original high-frequency pressure fluctuation signal in the steam generator feedwater pipeline; S2: Perform real-time spectrum transformation and feature separation on the original high-frequency pressure fluctuation signal to extract the spectrum energy data of the predetermined frequency band related to abnormal fluid disturbance; S3: Based on the continuous deviation and time-domain change trend of the spectrum energy data and the preset benchmark energy data under the corresponding operating conditions, a real-time confidence index characterizing the probability of minor pipeline leakage is generated. S4: Based on the magnitude of the real-time confidence index, dynamically adjust the parameters of the integral action in the liquid level control loop. The adjustment methods include: when the real-time confidence index is lower than the first threshold, use the first set of preset conservative integral parameters to suppress false water levels; when the real-time confidence index is higher than the first threshold but lower than the second threshold, freeze the accumulation of the integral action and apply a positive feedwater flow bias; when the real-time confidence index is higher than the second threshold, switch the input signal of the integral action to the flow deviation signal calculated based on the material balance, and use the second set of preset active integral parameters to deal with leakage. S5: Using the integral parameter dynamically adjusted by S4, the deviation between the liquid level measurement signal and the liquid level set value is calculated, and combined with proportional and derivative actions, the final control signal is generated to regulate the water supply flow. The integral parameters are conservative integral parameters and active integral parameters.

[0006] As a further aspect of the present invention: S2 specifically includes: The original high-frequency pressure fluctuation signal is preprocessed, including low-pass filtering and downsampling based on a preset cutoff frequency, to obtain the signal to be analyzed. Perform time-frequency domain transformation on the signal to be analyzed to obtain the transformation coefficient sequence corresponding to different frequency sub-bands; Based on the transform coefficient sequence, calculate the signal energy of each sub-band within a predetermined frequency band; The signal energy of each sub-band is weighted and summed to obtain the spectral energy data.

[0007] As a further aspect of the present invention: the calculation of the signal energy of each sub-band within a predetermined frequency band based on the transform coefficient sequence specifically includes: For each transformation coefficient in the transformation coefficient sequence, calculate the product of the absolute value of the transformation coefficient and the preset energy weighting factor to obtain the absolute value of the weighted coefficient; For each subband, the absolute values ​​of all weighting coefficients within the subband are summed to obtain the preliminary energy value of the subband; Determine whether the initial energy value of each sub-band is greater than the historical background noise energy baseline value of the sub-band. If so, take the difference of the excess as the final signal energy of the sub-band; otherwise, set the final signal energy of the sub-band to 0.

[0008] As a further aspect of the present invention: the generation of the real-time confidence index characterizing the probability of minor pipeline leaks specifically includes: Using the spectral energy data within the sliding time window, calculate the difference between the spectral energy data and the baseline energy data under the same operating conditions to obtain the current energy deviation. Calculate the rate of change of the current energy deviation within the sliding time window, and use it as the energy change rate. The current energy deviation is compared with the first deviation threshold, and the energy change rate is compared with the pre-trained positive energy change rate threshold. Only when both exceed their respective thresholds for a preset duration, the weighted accumulation of an internal state variable is initiated. The accumulated results of the internal state variables are mapped through a preset nonlinear function to output a real-time confidence index.

[0009] As a further aspect of the present invention: the step of mapping the accumulated result of the internal state variables through a preset nonlinear function specifically includes: The accumulated result of the internal state variables is obtained as the input value for the nonlinear mapping; The input value is input into a preset nonlinear function. The nonlinear function is divided into three consecutive intervals within its domain: the output of the nonlinear function grows slowly in the first interval, the output of the nonlinear function grows faster in the second interval, and the output of the nonlinear function grows slower again and approaches the saturation value in the third interval. The output of the nonlinear function is normalized so that it falls within the numerical range of 0 to 1. The normalized values ​​are output as the real-time confidence index.

[0010] As a further aspect of the present invention: the adoption of a first set of preset conservative integral parameters for suppressing false water levels specifically includes: Based on the current operating point of the steam generator, the corresponding set of conservative integral parameters is queried from a preset operating point-parameter mapping table; Based on the magnitude of the real-time confidence index, calculate the interpolation weights between the set of conservative integral parameters and a set of more conservative alternative integral parameters. The two sets of parameters are linearly interpolated using interpolation weights to obtain the integral parameter values ​​for the current practical application.

[0011] As a further aspect of the present invention: the accumulation of the freezing integral effect and the application of a positive feedwater flow bias specifically includes: The historical accumulated value of the integration effect is locked and saved; The internal weighting coefficients are reduced to 0 from their initial values ​​according to a preset decay rate. The portion of the real-time confidence index that exceeds the first threshold is multiplied by a preset bias baseline value. Multiply by the internal weighting coefficient to obtain the dynamic positive bias value; The dynamic positive bias value is superimposed on the water supply flow setpoint.

[0012] As a further aspect of the present invention: the adoption of a second set of preset active integral parameters for responding to leakage specifically includes: Switching the input signal for integral action and using the second set of preset active integral parameters to address leakage specifically includes: Initiate a signal transition process, during which the liquid level deviation signal and the flow rate deviation signal are mixed according to a preset time-varying ratio to form a transition input signal; In the first control cycle after the signal transition process is completed, the historical cumulative value of the integral action is reset to 0; Based on the extent to which the real-time confidence index exceeds the second threshold, the corresponding positive integral parameter value is dynamically read from the preset curve and applied.

[0013] As a further aspect of the present invention: S5 specifically includes: The deviation between the liquid level measurement signal and the liquid level setpoint is calculated with the selected proportional action parameter to obtain the proportional action amount, with the integral parameter adjusted by S4 to obtain the integral action amount, and with the selected differential action parameter to obtain the differential action amount. The proportional action, integral action, and differential action are combined according to a preset superposition rule; The total amount synthesized is used as the final control signal output.

[0014] As a further aspect of the present invention: the synthesis of the proportional action, integral action, and differential action according to a preset superposition rule specifically includes: Obtain real-time confidence metrics; Based on the value of the real-time confidence index, the real-time dynamic composite weights corresponding to the proportional action, integral action, and differential action are retrieved from the preset weight mapping relationship. The proportional action, integral action, and differential action are multiplied by their respective real-time dynamic composite weights to obtain the weighted proportional component, integral component, and differential component. The weighted proportional components, integral components, and differential components are algebraically added together to obtain the total composite.

[0015] The beneficial effects of this invention are: (1) This invention introduces a novel detection dimension—high-frequency pressure pulsation spectrum analysis—to extract leakage-related feature fingerprints from the microscopic dynamics of fluids and generate real-time confidence indices accordingly. This enables the control system to make probabilistic judgments about the nature of disturbances at the initial stage of their occurrence and dynamically switch to drastically different control logics and parameter sets for two fundamentally different disturbances: "false water level" and "minor leakage." This effect resolves the control paradox caused by disturbance confusion in existing technologies and enhances the adaptive fault tolerance of the control system.

[0016] (2) This invention not only switches integral parameters but also performs crucial "integral input signal switching" and "integrator reset" operations. This immediately corrects the integrator's control objective from tracking potentially false liquid level signals to tracking real material imbalances (deviation between steam flow and feedwater flow), thereby eliminating the possibility of the integrator continuously accumulating in the wrong direction (reducing feedwater) in the early stages of leakage. It breaks the dangerous positive feedback loop of traditional integral control under fault conditions, transforming passive defense into active fault tolerance, and can inject the correct control trend in the early stages of a fault, avoiding the risk of amplifying minor equipment defects into serious operational accidents, and improving the inherent safety level of the system. Attached Figure Description

[0017] The invention will now be further described with reference to the accompanying drawings.

[0018] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only 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.

[0020] Please see Figure 1 As shown, this invention provides a PID parameter tuning method for a steam generator level control system, comprising the following steps: S1: Acquire the original high-frequency pressure fluctuation signal in the steam generator feedwater pipeline; S2: Perform real-time spectrum transformation and feature separation on the original high-frequency pressure fluctuation signal to extract the spectrum energy data of the predetermined frequency band related to abnormal fluid disturbance; S3: Based on the continuous deviation and time-domain change trend of the spectrum energy data and the preset benchmark energy data under the corresponding operating conditions, a real-time confidence index characterizing the probability of minor pipeline leakage is generated. S4: Based on the magnitude of the real-time confidence index, dynamically adjust the parameters of the integral action in the liquid level control loop. The adjustment methods include: when the real-time confidence index is lower than the first threshold, use the first set of preset conservative integral parameters to suppress false water levels; when the real-time confidence index is higher than the first threshold but lower than the second threshold, freeze the accumulation of the integral action and apply a positive feedwater flow bias; when the real-time confidence index is higher than the second threshold, switch the input signal of the integral action to the flow deviation signal calculated based on the material balance, and use the second set of preset active integral parameters to deal with leakage. S5: Using the integral parameter dynamically adjusted by S4, the deviation between the liquid level measurement signal and the liquid level set value is calculated, and combined with proportional and derivative actions, the final control signal is generated to regulate the water supply flow. The integral parameters are conservative integral parameters and active integral parameters.

[0021] In S1, the acquisition of the raw high-frequency pressure fluctuation signal is accomplished by a dedicated sensing device deployed at a specific location in the steam generator feedwater pipeline. The core of this sensing device is a high-frequency dynamic pressure sensor. The sensor is installed on a straight pipe section after the feedwater regulating valve and before the steam generator feedwater inlet. This location effectively reflects the true dynamic characteristics of the pipeline fluid while avoiding direct mechanical vibration interference caused by valve operation.

[0022] In actual operation, the high-frequency dynamic pressure sensor continuously senses pressure pulsations within the pipeline and converts them into a continuous analog electrical signal. This analog signal is first processed by a hardware anti-aliasing filter to eliminate signal components above the Nyquist frequency, preventing spectral aliasing during subsequent digitization. The filtered analog signal is then converted from analog to digital by the data acquisition unit to form a discrete-time series of original high-frequency pressure fluctuation digital signals. This digital signal serves as the raw data input for subsequent processing steps.

[0023] In step S2, the preprocessing process first performs low-pass filtering on the original high-frequency pressure fluctuation digital signal. The cutoff frequency used for low-pass filtering is determined comprehensively based on the actual acoustic characteristics of the water supply pipeline and the effective frequency response range of the sensor, and is usually set between 1 / 2 and 2 / 3 of the sensor's nominal upper limit frequency. The purpose is to retain the effective frequency band related to fluid dynamics while suppressing higher-frequency irrelevant noise. The filtered signal is then downsampled. The downsampling ratio is determined based on the relationship between the cutoff frequency and the upper limit of the frequency to be analyzed, ensuring that the sampling rate satisfies the Nyquist sampling theorem and reducing the amount of data for subsequent calculations, thereby obtaining the signal to be analyzed.

[0024] After obtaining the signal to be analyzed, a time-frequency domain transformation is performed on it. This transformation process converts the signal from the time domain to a domain that can simultaneously characterize frequency components and energy distribution. The output is a sequence of transformation coefficients, where each coefficient corresponds to a specific frequency sub-band and its energy contribution within a specific time window. Specifically, a sliding window processing method based on a fixed-length time window is used to perform orthogonal basis function decomposition on the signal data points within the window, thereby obtaining a series of transformation coefficients characterizing the energy distribution of different frequency sub-bands.

[0025] The signal energy of each sub-band within a predetermined frequency band is calculated based on the transform coefficient sequence. First, for each transform coefficient in the sequence, its absolute value is taken, and this absolute value is multiplied by a weighting factor pre-assigned to the frequency sub-band corresponding to that coefficient. This weighting factor reflects the sensitivity of that frequency component to abnormal fluid disturbances, thus obtaining the absolute value of the weighted coefficient for each coefficient. Next, for each frequency sub-band, the absolute values ​​of the weighted coefficients of all transform coefficients contained in that sub-band are summed to obtain the preliminary energy value of that sub-band.

[0026] The method for determining the historical background noise energy baseline value is as follows: Under multiple stable load conditions with no steam generator leakage, the preliminary energy values ​​of each sub-band are collected and calculated over a long period. The high percentile value (e.g., the 95th percentile) of the statistical distribution under each operating condition is taken as the baseline value of that sub-band. In real-time calculation, it is determined whether the preliminary energy value of each sub-band is greater than its corresponding historical background noise energy baseline value. If it is greater, the baseline value is subtracted from the preliminary energy value, and the difference is taken as the final signal energy of that sub-band; if it is not greater, the final signal energy of that sub-band is set to zero. This step aims to filter out the inherent background noise of the system and highlight the characteristic energy generated by abnormal disturbances.

[0027] Finally, the final signal energies of all sub-bands within the predetermined frequency band are weighted and summed to obtain comprehensive spectral energy data. The predetermined frequency band is pre-defined by summarizing and analyzing recurring characteristic frequency components in historical leakage event data, typically including a low-frequency band and a high-frequency band. The weights used in the summation are allocated based on the reliability of each sub-band's representation of the leakage characteristics; sub-bands with higher representation reliability are given greater weight. The result of the weighted summation is the spectral energy data representing the intensity of abnormal fluid disturbance within the current time window.

[0028] In S3, the process of generating a real-time confidence index characterizing the probability of minor pipeline leaks begins with a time-series analysis of spectral energy data. This is achieved using a length of... A sliding time window of one sampling period is used, which slides forward over time, each time covering one sampling period. At each moment... Extract all spectral energy data falling within the window. For the latest data within the window... (i.e., the current spectral energy data), it is necessary to calculate its relationship with the preset reference energy data under the corresponding operating conditions. The difference is used to obtain the current energy deviation. The reference energy data It is preset by long-term monitoring and statistical analysis of spectral energy data under various stable operating conditions during the normal operation of the steam generator without leakage. Its value is typically taken as a long-term statistical characteristic value of the spectral energy data under that operating condition, such as the 95th percentile. Current energy deviation. The calculation expression is: Energy deviation characterizes the absolute degree of deviation of the current fluid state from the normal background state.

[0029] Based on the current energy deviation, its changing trend is further analyzed. The current energy deviation is calculated. The rate of change within the sliding time window is taken as the rate of energy change. The specific calculation method is as follows: take the initial value of the current energy deviation sequence within the sliding time window. and the ending value (i.e., the current value). Calculate the difference between the two, and then divide by the length of time covered by the sliding window. (In seconds). Its calculation expression is: ; The rate of change of energy It reflects whether the abnormal energy deviation is in a state of growth, stability or decay, with a positive value indicating that the deviation is increasing.

[0030] To distinguish between random fluctuations and persistent abnormal events, the two indicators mentioned above need to be compared with preset thresholds and their persistence needs to be assessed. The first deviation threshold... Typically based on the current energy deviation under normal operating conditions The statistical distribution is set, for example, its mean plus twice the standard deviation. The positive threshold for the rate of energy change. Based on the energy change rate under normal operating conditions The statistical distribution is set, for example, its mean plus three standard deviations, and is usually set to a positive number to focus on abnormal growth trends. The judgment logic is: only if the current energy deviation... Continuously greater than the first deviation threshold And simultaneously the rate of energy change Continuously greater than the positive threshold of the rate of change of energy The state is maintained for a preset duration. The subsequent weighted accumulation process only begins after (for example, 5 consecutive sampling periods). If any condition is met within the duration... If the condition is not met consecutively, the judgment logic is reset and the timing restarts.

[0031] When the above dual threshold conditions are met, an action is initiated on an internal state variable. The weighted summation. The internal state quantity... The initial value is zero. An accumulation operation is performed each time the condition is met. The increment for each accumulation is determined by the current energy deviation. And a dynamic weighting coefficient The dynamic weighting coefficients are jointly determined. It is not a fixed value; its magnitude depends on the current rate of energy change. A positive correlation indicates a higher rate of change and a larger weighting coefficient, signifying a greater cumulative weighting for anomalous signs of rapid growth. Each cumulative increment... This can be expressed as the product of energy deviation and weighting coefficients. Internal state quantity. Starting from the first time the condition is met, the increment is continuously accumulated in each subsequent sampling period when the condition is met, until the double threshold condition is no longer met. The accumulation process can be described as follows: for each sampling period when the condition is met, the new internal state quantity is equal to the internal state quantity of the previous sampling period plus the accumulated increment calculated in this period.

[0032] Finally, the accumulated internal state quantities Through a pre-defined nonlinear function Perform mapping to output the final real-time confidence metric. This nonlinear function Its domain is divided into three consecutive intervals to achieve different sensitivity responses to the accumulated result: in the first interval (e.g. From 0 to ), function output The growth is very slow, primarily used to suppress low-level accumulations triggered by accidental disturbances or noise; in the second interval (e.g.) from arrive ), function output Growth accelerated significantly in response to persistent and intensifying evidence of anomalous activity; in the third interval (e.g.) Greater than ), function output The growth rate slows down again and gradually approaches a saturation value (e.g., 1.0), which ensures the boundedness of the confidence index and prevents the output from getting out of control due to excessive accumulation in extreme cases. The specific form of this function can be determined by a preset mathematical function (e.g., ...). The mathematical function is implemented using a function of type [type]. The calculation expression for the mathematical function is: Among them, parameters This represents the accumulated result of the internal state variables, i.e., the input of the function; This is the saturation value of the function, usually set to 1, corresponding to the maximum output value of the confidence index; parameter The center point of the function determines the inflection point where the function output transitions from slow growth to rapid growth, and its value corresponds to the preset starting point of the second interval. ;parameter ( >0) is the steepness coefficient of the curve, and the control function is at the center point. The larger the value of λ, the steeper the curve near the center point, and the faster the transition from slow to accelerated growth. To obtain a standardized probabilistic representation, the nonlinear function... The output value is normalized to fall within the range of 0 to 1. If the function's output is already within this range, no further processing is needed. Finally, this normalized value serves as the real-time confidence index. The closer the output value is to 1, the higher the probability of a minor leak in the pipeline. The entire process transforms raw spectral energy data into a leak confidence index with clear physical meaning and probabilistic interpretation.

[0033] In S4, the first and second thresholds are two pre-set judgment thresholds based on a large amount of historical operating data and engineering experience. The first threshold is set based on the statistical maximum value reached by the real-time confidence index when the steam generator has been operating without failure for a long period of time and has experienced various typical load changes. This statistical maximum value is multiplied by a safety factor less than 1, such as 0.8, and the resulting value is set as the first threshold. The second threshold is set as a fixed value that is significantly higher than the first threshold. Its specific value is usually determined based on the confidence level that can clearly characterize the occurrence of minor leaks, such as 0.75. These two thresholds divide the range of real-time confidence index values ​​into three continuous intervals, each corresponding to a different control strategy.

[0034] When the real-time confidence index is lower than the first threshold, it indicates that the system is under normal operating conditions or only slightly disturbed. In this case, the first set of preset conservative integral parameters for suppressing false water levels is used. First, the parameter index is determined based on the current operating point of the steam generator, which is determined by key operating parameters, such as the combination of the current steam flow rate and feedwater pressure. Using a preset operating point-parameter mapping table, the corresponding set of conservative integral parameters is retrieved based on the specific values ​​of the current steam flow rate and feedwater pressure. This set contains at least one integration time parameter. Second, the interpolation weight is calculated dynamically based on the magnitude of the real-time confidence index: the current value of the real-time confidence index is divided by the first threshold to obtain a ratio; then, 1 is subtracted from this ratio, and the result is the interpolation weight. When the real-time confidence index is close to 0, the weight is close to 1; when the index is close to the first threshold, the weight is close to 0. Finally, linear interpolation is performed. A set of backup integral parameters that are more conservative than the set of conservative integral parameters, such as parameters with longer integration times, is also preset. Each parameter in the retrieved conservative integration parameter set is weighted and summed with the corresponding parameter in the backup integration parameter set using the calculated interpolation weights. Specifically, the conservative integration parameter value is multiplied by the interpolation weight, and then the backup integration parameter value is multiplied by (1 minus the interpolation weight). The result is the integration parameter value for the current practical application. This allows the conservatism of the integration action to be smoothly adjusted as disturbance signs slightly increase within the low confidence interval.

[0035] When the real-time confidence index is higher than the first threshold but lower than the second threshold, it indicates some abnormal signs but is not yet sufficient to confirm a leak. At this time, the operation of freezing the integral action accumulation and applying a positive feedwater flow bias is performed. First, the historical accumulated value of the integral action is locked and saved, that is, the accumulated amount in the current integrator is recorded, and the update of the accumulated amount based on the new deviation is temporarily stopped in subsequent control cycles. At the same time, an internal weighting coefficient is initialized with an initial value of 1. In each subsequent control cycle, the internal weighting coefficient decreases according to a preset decay rate, for example, multiplied by a fixed decay factor of 0.95 less than 1 in each control cycle, so that it gradually approaches 0. Next, the dynamic positive bias value is calculated. First, the portion of the real-time confidence index that exceeds the first threshold is calculated, that is, the current index value is subtracted from the first threshold; the difference is multiplied by a preset bias reference value, which is determined according to a percentage of the nominal variation range of steam flow, for example, 2% of the full scale of steam flow; a preliminary bias amount is obtained; then this preliminary bias amount is multiplied by the internal weighting coefficient of the current cycle to obtain the final dynamic positive bias value. Finally, this dynamic positive bias value is superimposed on the setpoint of the water supply flow rate. This series of operations provides a temporary and gradually decaying positive water supply excitation without disturbing the steady state of the integrator.

[0036] When the real-time confidence index exceeds the second threshold, it indicates a high probability of a minor leak in the pipeline, necessitating a switch to an active leak-response control mode. The core of this mode is to switch the input signal from integral action to a flow deviation signal calculated based on material balance, employing a second set of preset active leak-response integral parameters. First, a signal transition process is initiated, lasting for a fixed number of control cycles, e.g., 10 cycles. During this transition period, the integrator's input signal is a mixture of the level deviation signal and the flow deviation signal according to a preset time-varying ratio, which changes linearly with time: at the start of the transition, the level deviation signal ratio is 100%, and the flow deviation signal ratio is 0%; at the end of the transition, the level deviation signal ratio becomes 0%, and the flow deviation signal ratio becomes 100%. The integrator input at any given moment during the transition period is the signal mixed according to the current ratio. Second, in the first control cycle after the signal transition process is completed, an integrator reset is performed, forcibly setting the historical accumulated value of integral action to 0 to clear any potentially misguided integral amounts accumulated. Finally, the positive integration parameters are dynamically determined based on the extent to which the real-time confidence index exceeds the second threshold. A parameter curve is predefined, with the horizontal axis representing the exceedance magnitude (current index value minus the second threshold) and the vertical axis representing the corresponding integration parameter value, such as the reciprocal of the integration time. Based on the currently calculated exceedance magnitude, the corresponding parameter value is read from this curve and applied immediately. This curve is typically configured so that the parameter value monotonically changes in a more positive direction as the exceedance magnitude increases.

[0037] In step S5, the integral parameters dynamically adjusted in step S4 are used, combined with proportional and derivative actions, to generate the final control signal for regulating the water supply flow. The integral parameters specifically refer to the parameter values ​​actually applied in step S4 based on the interval of the real-time confidence index, including conservative or aggressive integral parameters. First, the action quantities of the three basic control actions in the level control loop are calculated. The level deviation is obtained by subtracting the level setpoint from the level measurement signal. The proportional action quantity is calculated by multiplying the level deviation by a proportional gain coefficient, which is selected from a pre-defined set of proportional parameters for each interval based on the interval of the current real-time confidence index (below the first threshold, between the first and second thresholds, or above the second threshold). The integral action is calculated as follows: the liquid level deviation is multiplied by an integral gain coefficient, which is a function of the actual integral parameter output in step S4 (for example, when the integral parameter is the integral time, the integral gain coefficient is equal to the reciprocal of the integral time). The product is then added to the cumulative integral value of the previous cycle in each control cycle. If the "freeze integral action accumulation" state is in place during step S4, this accumulation process is paused, and the cumulative integral value remains locked. The derivative action is calculated as follows: the difference between the liquid level deviation of the current control cycle and the liquid level deviation of the previous control cycle is calculated to obtain the deviation change rate. This deviation change rate is then multiplied by a derivative gain coefficient, which is also selected from a preset set of derivative parameters based on the current real-time confidence index interval.

[0038] After obtaining the proportional, integral, and derivative actions, they are synthesized according to a preset superposition rule. This superposition rule is not a simple addition of fixed weights, but a synthesis process that dynamically adjusts the weights of each component based on a real-time confidence index. First, the real-time confidence index is obtained. Second, based on the value of this real-time confidence index, the three real-time dynamic synthesis weights corresponding to the proportional, integral, and derivative actions are found from a preset weight mapping relationship. The weight mapping relationship is determined through prior simulations and experiments. Its construction method is as follows: under different real-time confidence index values ​​(from 0 to 1), the control system is optimized to achieve good performance in various test scenarios such as "false water level" and "real leakage," thereby deriving the optimal synthesis weights of the proportional, integral, and derivative actions. Finally, a data correspondence is formed with the real-time confidence index as input and the three weight values ​​as output, usually implemented by a lookup table.

[0039] Then, a weighted synthesis calculation is performed. The calculated proportional action is multiplied by the real-time dynamic synthesis weight corresponding to the proportional action obtained from a lookup table to obtain the weighted proportional component. The calculated integral action is multiplied by the real-time dynamic synthesis weight corresponding to the integral action obtained from a lookup table to obtain the weighted integral component. The calculated differential action is multiplied by the real-time dynamic synthesis weight corresponding to the differential action obtained from a lookup table to obtain the weighted differential component. This weighting process allows the contribution ratio of the three control actions to the final control signal to different leakage risk confidence levels to be dynamically adjusted. For example, in the case of high-confidence leakage, the weight of the integral action may be increased to quickly compensate for water loss, while the weight of the differential action, which is sensitive to measurement noise, may be appropriately reduced.

[0040] Finally, the weighted proportional component, weighted integral component, and weighted differential component are algebraically added together, that is, the three values ​​are directly summed, and the result is the synthesized total. This synthesized total is the final control signal that needs to be output in the current control cycle. This control signal is sent to the actuator to directly adjust the opening of the feedwater regulating valve or change the speed of the feedwater pump, thereby achieving precise control of the feedwater flow rate and ultimately maintaining the steam generator liquid level stably near the set value. The entire S5 step completes the entire process from the PID control law calculation after intelligent diagnosis and dynamic parameter adjustment to the generation of actual physical control action commands.

[0041] The working principle of this invention is as follows: First, the original high-frequency pressure fluctuation signal in the water supply pipeline is acquired, and real-time spectrum transformation and feature separation are performed to extract the spectrum energy data of the predetermined frequency band. Then, based on the continuous deviation and time-domain change trend of the spectrum energy data from the preset benchmark energy data under the corresponding operating conditions, a real-time confidence index characterizing the probability of minor pipeline leakage is generated. According to the magnitude of the real-time confidence index, the parameters and logic of the integral action in the liquid level control loop are dynamically adjusted, specifically including: when the index is below the first threshold, a conservative integral parameter to suppress false water levels is adopted; when the index is between the first and second thresholds, the integral accumulation is frozen and a gradually decaying positive water supply flow bias is applied; when the index is above the second threshold, the integral input signal is switched to a flow deviation signal based on material balance calculation and an active integral parameter to deal with leakage is adopted. Finally, using the dynamically adjusted integral parameters, combined with proportional and derivative actions, the liquid level deviation is calculated and the final control signal is synthesized to achieve precise regulation of water supply flow and stable control of liquid level.

[0042] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A method for tuning PID parameters in a steam generator level control system, characterized in that, Includes the following steps: S1: Acquire the original high-frequency pressure fluctuation signal in the steam generator feedwater pipeline; S2: Perform real-time spectrum transformation and feature separation on the original high-frequency pressure fluctuation signal to extract the spectrum energy data of the predetermined frequency band related to abnormal fluid disturbance; S3: Based on the continuous deviation and time-domain change trend of the spectrum energy data and the preset benchmark energy data under the corresponding operating conditions, a real-time confidence index characterizing the probability of minor pipeline leakage is generated. S4: Based on the magnitude of the real-time confidence index, dynamically adjust the parameters of the integral action in the liquid level control loop. The adjustment methods include: when the real-time confidence index is lower than the first threshold, use the first set of preset conservative integral parameters to suppress false water levels; when the real-time confidence index is higher than the first threshold but lower than the second threshold, freeze the accumulation of the integral action and apply a positive feedwater flow bias; when the real-time confidence index is higher than the second threshold, switch the input signal of the integral action to the flow deviation signal calculated based on the material balance, and use the second set of preset active integral parameters to deal with leakage. S5: Using the integral parameter dynamically adjusted by S4, the deviation between the liquid level measurement signal and the liquid level set value is calculated, and combined with proportional and derivative actions, the final control signal is generated to regulate the water supply flow. The integral parameters are conservative integral parameters and active integral parameters.

2. The PID parameter tuning method for a steam generator level control system according to claim 1, characterized in that, S2 specifically includes: The original high-frequency pressure fluctuation signal is preprocessed, including low-pass filtering and downsampling based on a preset cutoff frequency, to obtain the signal to be analyzed. Perform time-frequency domain transformation on the signal to be analyzed to obtain the transformation coefficient sequence corresponding to different frequency sub-bands; Based on the transform coefficient sequence, calculate the signal energy of each sub-band within a predetermined frequency band; The signal energy of each sub-band is weighted and summed to obtain the spectral energy data.

3. The PID parameter tuning method for a steam generator level control system according to claim 2, characterized in that, The calculation of signal energy for each sub-band within a predetermined frequency band based on the transform coefficient sequence specifically includes: For each transformation coefficient in the transformation coefficient sequence, calculate the product of the absolute value of the transformation coefficient and the preset energy weighting factor to obtain the absolute value of the weighted coefficient; For each subband, the absolute values ​​of all weighting coefficients within the subband are summed to obtain the preliminary energy value of the subband; Determine whether the initial energy value of each sub-band is greater than the historical background noise energy baseline value of the sub-band. If so, take the difference of the excess as the final signal energy of the sub-band; otherwise, set the final signal energy of the sub-band to 0.

4. The PID parameter tuning method for a steam generator level control system according to claim 1, characterized in that, The real-time confidence index for generating the probability of minor pipeline leaks specifically includes: Using the spectral energy data within the sliding time window, calculate the difference between the spectral energy data and the baseline energy data under the same operating conditions to obtain the current energy deviation. Calculate the rate of change of the current energy deviation within the sliding time window, and use it as the energy change rate. The current energy deviation is compared with the first deviation threshold, and the energy change rate is compared with the pre-trained positive energy change rate threshold. Only when both exceed their respective thresholds for a preset duration, the weighted accumulation of an internal state variable is initiated. The accumulated results of the internal state variables are mapped through a preset nonlinear function to output a real-time confidence index.

5. The PID parameter tuning method for a steam generator level control system according to claim 4, characterized in that, The step of mapping the accumulated results of internal state variables through a preset nonlinear function specifically includes: The accumulated result of the internal state variables is obtained as the input value for the nonlinear mapping; The input value is input into a preset nonlinear function. The nonlinear function is divided into three consecutive intervals within its domain: the output of the nonlinear function grows slowly in the first interval, the output of the nonlinear function grows faster in the second interval, and the output of the nonlinear function grows slower again and approaches the saturation value in the third interval. The output of the nonlinear function is normalized so that it falls within the numerical range of 0 to 1. The normalized values ​​are output as the real-time confidence index.

6. The PID parameter tuning method for a steam generator level control system according to claim 1, characterized in that, The first set of preset conservative integral parameters for suppressing false water levels specifically includes: Based on the current operating point of the steam generator, the corresponding set of conservative integral parameters is queried from a preset operating point-parameter mapping table; Based on the magnitude of the real-time confidence index, calculate the interpolation weights between the set of conservative integral parameters and a set of more conservative alternative integral parameters. The two sets of parameters are linearly interpolated using interpolation weights to obtain the integral parameter values ​​for the current practical application.

7. The PID parameter tuning method for a steam generator level control system according to claim 1, characterized in that, The accumulation of the freezing integral effect and the application of a positive feedwater flow bias specifically include: The historical accumulated value of the integration effect is locked and saved; The internal weighting coefficients are reduced to 0 from their initial values ​​according to a preset decay rate. The portion of the real-time confidence index that exceeds the first threshold is multiplied by a preset bias baseline value. Multiply by the internal weighting coefficient to obtain the dynamic positive bias value; The dynamic positive bias value is superimposed on the water supply flow setpoint.

8. The PID parameter tuning method for a steam generator level control system according to claim 1, characterized in that, The second set of preset active integral parameters for responding to leaks specifically includes: Switching the input signal for integral action and using the second set of preset active integral parameters to address leakage specifically includes: Initiate a signal transition process, during which the liquid level deviation signal and the flow rate deviation signal are mixed according to a preset time-varying ratio to form a transition input signal; In the first control cycle after the signal transition process is completed, the historical cumulative value of the integral action is reset to 0; Based on the extent to which the real-time confidence index exceeds the second threshold, the corresponding positive integral parameter value is dynamically read from the preset curve and applied.

9. The PID parameter tuning method for a steam generator level control system according to claim 1, characterized in that, S5 specifically includes: The deviation between the liquid level measurement signal and the liquid level setpoint is calculated with the selected proportional action parameter to obtain the proportional action amount, with the integral parameter adjusted by S4 to obtain the integral action amount, and with the selected differential action parameter to obtain the differential action amount. The proportional action, integral action, and differential action are combined according to a preset superposition rule; The total amount synthesized is used as the final control signal output.

10. The PID parameter tuning method for a steam generator level control system according to claim 9, characterized in that, The synthesis of the proportional action, integral action, and differential action according to a preset superposition rule specifically includes: Obtain real-time confidence metrics; Based on the value of the real-time confidence index, the real-time dynamic composite weights corresponding to the proportional action, integral action, and differential action are retrieved from the preset weight mapping relationship. The proportional action, integral action, and differential action are multiplied by their respective real-time dynamic composite weights to obtain the weighted proportional component, integral component, and differential component. The weighted proportional components, integral components, and differential components are algebraically added together to obtain the total composite.

Images