An alarm logic optimization method for intelligent power plant thermal control

Abstract

This invention relates to the field of smart power plants and discloses an alarm logic optimization method for thermal control in smart power plants, used for intelligent early warning of dynamic health status in thermal control. The method includes: first, preprocessing the time series of key process parameters to generate a dimensionless data sequence; then, reconstructing the phase space to construct a multi-dimensional phase space point set; next, analyzing the complexity at multiple time scales and calculating the sample entropy value; obtaining an anomaly index by comparing the sample entropy value with a benchmark entropy value; and triggering early warning, alarm, or normal signals in a graded manner based on the index and the original measured values. This invention can also generate control command sequences to adjust the equipment operating status, select characteristic entropy values ​​by analyzing the response characteristics of sample entropy values, and call benchmark entropy values ​​according to operating conditions, thereby achieving accurate and reliable alarms.

Description

Technical Field

[0001] This invention relates to the field of smart power plants, and more particularly to an alarm logic optimization method for thermal control in smart power plants. Background Technology

[0002] In the construction and development of smart power plants, the thermal control system, as the core component ensuring the safe, stable, and efficient operation of the power plant, directly affects the overall production efficiency and safety of the plant. The thermal system involves numerous complex process parameters that are interconnected and mutually influential, collectively constituting the dynamic characteristics of the power plant's thermal processes. Accurate and timely monitoring of the thermal system's operating status, and the prompt issuance of effective alarm signals when anomalies occur, are of paramount importance for preventing accidents, reducing equipment damage, and ensuring personnel safety.

[0003] Traditional thermal control alarm methods are mainly based on fixed threshold settings; that is, when process parameters exceed preset upper and lower limits, the system triggers an alarm. However, this method has many limitations in practical applications.

[0004] Thermal systems are highly nonlinear and dynamically changing complex systems whose operating states are affected by various factors, such as load variations, changes in environmental conditions, and equipment aging. Fixed thresholds are difficult to adapt to the dynamic changes of the system under different operating conditions, which may lead to false alarms during fluctuations in normal operating conditions, and fail to issue timely warnings when the system has potential faults but the parameters have not yet reached the fixed thresholds, thus missing the best opportunity for handling.

[0005] Traditional methods focus only on the threshold exceedance of a single parameter, ignoring the interrelationships between parameters and the overall dynamic characteristics of the system. This makes it difficult to comprehensively and accurately assess the system's operating status. For some complex fault modes, such as gradual faults and coupled faults, they often cannot be effectively identified and warned.

[0006] With the continuous development of information technology, control theory and artificial intelligence, how to use advanced data analysis methods and intelligent algorithms to deeply mine and analyze the operating data of thermal systems in order to achieve more accurate and reliable alarm logic optimization has become a research hotspot and urgent need in the field of thermal control of smart power plants.

[0007] Therefore, we propose an alarm logic optimization method for thermal control in smart power plants to solve the above problems. Summary of the Invention

[0008] This invention provides an alarm logic optimization method for thermal control in smart power plants, which is used for intelligent early warning of dynamic health status in thermal control of smart power plants.

[0009] The first aspect of this invention provides an alarm logic optimization method for thermal control of a smart power plant, the method comprising: preprocessing the time series of at least one key process parameter of the thermal system of the smart power plant to generate a dimensionless data sequence; The dimensionless data sequence is reconstructed in phase space to construct a multidimensional phase space point set; Complexity analysis is performed on the sequence corresponding to the multidimensional phase space point set at multiple time scales, and the sample entropy value at at least one selected scale is calculated and output. By comparing the sample entropy value under the selected scale with the baseline entropy value established under historical normal operating conditions, an anomaly index characterizing the degree of deviation from the current operating state is calculated. Based on the different numerical ranges of the anomaly index and in combination with the original process measurements of the key process parameters, a state signal is obtained.

[0010] Optionally, in a first implementation of the first aspect of the present invention, the method includes: Define a set of time scale factors for multiscale analysis of the system to obtain a set of scale factors; For each scale factor in the set of scale factors, the single-dimensional time series data corresponding to the multidimensional phase space point set is coarsened to generate a coarse-grained time series corresponding to each scale factor. For each of the coarse-grained time series, its sample entropy value is calculated based on the given template dimension and similarity tolerance, thereby obtaining a set of sample entropy values ​​corresponding to the set of scale factors; Based on preset selection rules, at least one sample entropy value is selected from the set of sample entropy values.

[0011] Optionally, in a second implementation of the first aspect of the present invention, the response characteristics of entropy values ​​under different scale factors in the sample entropy value set to historical typical fault events are analyzed to obtain the sensitivity analysis results of the changes in each scale factor. Based on the results of the change sensitivity analysis, the scale factors are sorted to generate a sorted list of scale factors arranged from high to low sensitivity. Based on the scale factor sorting list, the scale factor ranked first is selected as the optimal scale factor; The sample entropy value corresponding to the optimal scaling factor is selected as the feature entropy value.

[0012] Optionally, in a third implementation of the first aspect of the present invention, the method includes: Based on historical normal operation data, for the selected scale, a baseline entropy value characterizing the normal complexity of the system and the standard deviation of entropy fluctuation reflecting the normal fluctuation range of the entropy value are calculated and stored. During real-time operation, the current feature entropy value corresponding to the selected scale within the current time window is obtained; Calculate the absolute difference between the current feature entropy value and the baseline entropy value; The anomaly index is obtained by dividing the absolute difference by the standard deviation of the entropy fluctuation.

[0013] Optionally, in a fourth implementation of the first aspect of the present invention, the method includes: Obtain the calculated anomaly index and the current original process measurement values ​​of the key process parameters; The anomaly index is compared with a preset first-level threshold, and an early warning trigger judgment is generated based on the comparison result. The anomaly index is compared with a second-level threshold that is higher than the first-level threshold, and it is simultaneously determined whether the current original process measurement value exceeds its preset engineering safety threshold. The alarm trigger judgment is generated by combining the two judgment results. Based on the aforementioned warning trigger judgment and alarm trigger judgment, output the corresponding warning signal, alarm signal or normal operation indication signal.

[0014] Optionally, in a fifth implementation of the first aspect of the present invention, it further includes: Based on the output graded alarm signals, generate the corresponding control command sequence; The control command sequence is sent to the corresponding thermal process actuator to adjust the operating status of the relevant equipment.

[0015] Optionally, in a sixth implementation of the first aspect of the present invention, a reference value correction instruction for fine-tuning the relevant process control loop is generated based on the output warning signal. Based on the output alarm signal, generate protective control commands for switching or interlocking related equipment.

[0016] Beneficial effects: By performing autocorrelation analysis and pseudo-nearest neighbor analysis on dimensionless data sequences, the time delay parameter and embedding dimension parameter are determined, and then the multidimensional phase space point set is reconstructed. This can describe the dynamic characteristics of the system from a higher dimension, uncover the intrinsic information of the system that is difficult to obtain by traditional univariate analysis, and more comprehensively reflect the complex behavior of thermal systems. It fully considers the multi-scale characteristics of the operating state changes of thermal systems, and can capture the dynamic changes of the system at different time scales, providing rich feature quantities for accurate assessment of system state and overcoming the limitations of traditional single-scale analysis. By analyzing the response characteristics of entropy values ​​under different scale factors in the sample entropy value set to historical typical fault events, the sample entropy value corresponding to the optimal scale factor is selected as the feature entropy value, which can highlight fault-sensitive features and improve the accuracy and reliability of fault identification. The tiered alarm system can issue alarms of the appropriate level in a timely manner according to the severity of the fault, enabling operators to take targeted measures, avoid overreaction or delayed handling, and improve the effectiveness and practicality of the alarm. In real-time operation, the system calls the corresponding parameters based on the current operating conditions to calculate the anomaly index and assess the status, which has wider applicability and higher accuracy, and can meet the operating needs of smart power plants under different operating conditions. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of an embodiment of the alarm logic optimization method for thermal control of smart power plants in this invention. Figure 2 This is a schematic diagram of an embodiment of the alarm logic optimization device for thermal control of smart power plants in this invention. Detailed Implementation

[0018] This invention provides an alarm logic optimization method for thermal control in smart power plants, used for intelligent early warning of dynamic health status in thermal control of smart power plants. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0019] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 One embodiment of the alarm logic optimization method for thermal control of smart power plants in this invention includes: 101. Preprocess the time series of at least one key process parameter of the thermal system of a smart power plant to generate a dimensionless data series normalized based on a reference steady-state condition.

[0020] It is understood that the executing entity of this invention can be an alarm logic optimization device for thermal control in smart power plants, or it can be a terminal or a server; no specific limitation is made here. This embodiment of the invention will be described using a server as an example of the executing entity.

[0021] Specifically, from the thermal system of the smart power plant, at least one key process parameter reflecting the state of the core thermal process is selected; the original time series data of the key process parameter is preprocessed, including removing outlier data points and filling in missing data points, to obtain a continuous and complete process data sequence; based on historical operating data, a reference steady-state operating condition period is determined for the key process parameter, and the statistical mean and statistical standard deviation of the process data sequence within this reference steady-state operating condition period are calculated; using the statistical mean and statistical standard deviation, the continuous and complete process data sequence is standardized to generate a dimensionless data sequence.

[0022] It should be noted that the boiler's "main steam temperature" is used as an example for explanation. Main steam temperature is a key process parameter of the thermal system, and its stability directly reflects the boiler's thermal state. The original time-series data comes from the distributed control system (DCS), with a sampling interval of 1 minute, collecting a total of 1000 data points, in degrees Celsius. Under normal operating conditions, the main steam temperature should fluctuate between 500℃ and 550℃. Anomalies and missing data exist in the original data: the 200th data point was recorded as 800℃ due to sensor interference, significantly exceeding the reasonable range; the 500th data point is missing due to communication interruption.

[0023] Based on engineering safety thresholds, the effective range for the main steam temperature is set to 400℃ to 600℃; values ​​exceeding this range are considered abnormal. Therefore, the 800℃ value at point 200 is automatically identified and removed. Next, missing data points are filled in: linear interpolation is used, calculated using adjacent valid data. Point 499 is 523℃, point 501 is 524℃, therefore the missing point 500 is filled in as 523.5℃. After processing, a continuous and complete sequence of 1000 process data points is obtained, with all temperature values ​​within the normal range.

[0024] A typical period of stable low-load operation of the system was selected, with 60 data points chosen from historical records between 2:00 AM and 3:00 AM. During this period, the load was constant and undisturbed, and the main steam temperature was stable, representing steady-state conditions. The statistical mean and standard deviation of the data for this period were calculated: the mean μ was 525℃, and the standard deviation σ was 5℃. This reflects the central trend and fluctuation level of the temperature under steady-state conditions.

[0025] Standardization calculations were performed to generate a dimensionless data sequence. Using a mean μ = 525℃ and a standard deviation σ = 5℃, the standardized value z was calculated point-by-point for the entire preprocessed data sequence, using the formula z = (x - μ) / σ, where x is the original temperature value. For one original data point x = 530℃, the standardized value z = 1.0; for another point x = 520℃, the standardized value z = -1.0. The entire sequence was transformed into dimensionless data with a mean of 0 and a standard deviation of 1, eliminating dimensional and baseline differences and providing standardized input for subsequent steps.

[0026] 102. Reconstruct the phase space of the dimensionless data sequence to construct a multidimensional phase space point set to describe the dynamic characteristics of the system.

[0027] Specifically, autocorrelation analysis is performed on the dimensionless data sequence, and a time delay parameter is determined based on the decay characteristics of the autocorrelation function; pseudo-nearest neighbor analysis or saturated correlation dimension analysis is performed on the dimensionless data sequence to determine an embedding dimension parameter; using the time delay parameter and the embedding dimension parameter, the dimensionless data sequence is reconstructed to construct a multidimensional phase space point set composed of multiple multidimensional vectors.

[0028] It should be noted that in step 101, we have obtained a dimensionless data sequence of the main steam temperature, denoted as Z. This sequence is 1000 units long, with a sampling interval of 1 minute, and the data has been standardized (mean 0, standard deviation 1). The goal of this step is to reconstruct its phase space to reveal the inherent dynamic characteristics of the system.

[0029] Autocorrelation analysis was performed on sequence Z to calculate its autocorrelation function values ​​at different time delays. The analysis revealed that when the time delay was 5 sampling intervals (i.e., 5 minutes), the autocorrelation function value decayed from an initial 1.0 to approximately 0.37 (i.e., 1 / e, the reciprocal of the natural logarithm). This point is generally considered a critical point for significant decay of correlation; therefore, we selected the appropriate time delay parameter. =5.

[0030] A pseudo-nearest neighbor analysis method was employed. Embedding began with a low-dimensional (2-dimensional) approach, gradually increasing the dimension, and the proportion of "pseudo-nearest neighbors" in the phase space was calculated for each dimension. The results showed that when the embedding dimension m=2, the proportion of pseudo-nearest neighbors reached as high as 25%; when m=3, this proportion rapidly decreased to 5%; and when m=4, the proportion further decreased to approximately 1.8%, with little change. This indicates that increasing the dimension to 3 effectively unfolds the geometric structure of the dynamical system, avoiding major information folding. Therefore, we selected an embedding dimension parameter of m=3.

[0031] Using the determined parameters Given m=5 and m=3, reconstruct the original one-dimensional dimensionless sequence Z. The construction method is as follows: starting from the first data point of the sequence, reconstruct the sequence at every... The data from 5 points are combined into a vector of m=3 dimensions. Specifically: The first phase space point Y1 is composed of three scalar data: Z(1), Z(1+5)=Z(6), and Z(1+2*5)=Z(11), i.e., Y1=[Z(1),Z(6),Z(11)].

[0032] The second phase space point Y2 is composed of Z(2), Z(7), and Z(12), i.e., Y2=[Z(2),Z(7),Z(12)].

[0033] Similarly, the i-th point is Yi = [Z(i), Z(i+5), Z(i+10)].

[0034] From a one-dimensional time series Z of length 1000, a multi-dimensional phase space point set {Y1,Y2,...,YN} consisting of N three-dimensional vectors was reconstructed. Calculations show that the effective number of points is N=990 (because the index of the last point must satisfy i+10≤1000). The trajectory formed by this point set in three-dimensional space can more comprehensively describe the evolution pattern and state of the main steam temperature dynamics system, laying the foundation for further multi-scale complexity analysis.

[0035] 103. Perform complexity analysis on the sequences corresponding to multidimensional phase space point sets at multiple time scales, and calculate and output the sample entropy value at at least one selected scale.

[0036] Specifically, a set of time scale factors for system multi-scale analysis is defined, resulting in a scale factor set. For each scale factor in the scale factor set, the single-dimensional time series data corresponding to the multi-dimensional phase space point set is coarse-grained to generate a coarse-grained time series corresponding to each scale factor. For each coarse-grained time series, its sample entropy value is calculated based on a given template dimension and similarity tolerance, thus obtaining a sample entropy value set corresponding to the scale factor set. Based on a preset selection rule, at least one sample entropy value is selected from the sample entropy value set as a feature entropy value for subsequent state assessment. Further, the response characteristics of entropy values ​​under different scale factors in the sample entropy value set to historical typical fault events are analyzed to obtain the change sensitivity analysis results of each scale factor. Based on the change sensitivity analysis results, each scale factor is sorted to generate a scale factor sorting list arranged from high to low sensitivity. According to the scale factor sorting list, the scale factor ranked first is selected as the optimal scale factor. The sample entropy value corresponding to the optimal scale factor is selected as the feature entropy value.

[0037] It should be noted that, following the phase space point set obtained in step 102, this step aims to perform multi-scale sample entropy analysis on the dynamic complexity of the main steam temperature system.

[0038] A set of time scale factors, s, is selected to analyze the system's complexity at different time resolutions. Based on engineering experience and common analytical ranges, the set of scale factors is set to {1, 2, 3, 4, 5}. Scale 1 corresponds to the raw, uncoarse-grained data.

[0039] For each s value in the scale factor set, the one-dimensional dimensionless original sequence Z (length 1000) used to reconstruct the phase space in step 102 is processed. The method is to divide the original sequence into multiple non-overlapping segments of length s and calculate the arithmetic mean of the data in each segment, thereby generating a coarse-grained time series of length 1000 / s. When the scale factor s=2, the first data point of the new sequence is the average of the original sequences Z(1) and Z(2), the second point is the average of Z(3) and Z(4), and so on, finally obtaining a coarse-grained sequence of length 500. This operation is performed for s=1, 2, 3, 4, 5 respectively to obtain coarse-grained sequences at 5 different time scales.

[0040] For each coarse-grained sequence, its sample entropy is calculated based on a given template dimension m=2 (different from the concept of phase space embedding dimension) and a similarity tolerance r=0.2 (usually taken as 0.1 to 0.25 times the standard deviation). Sample entropy reflects the regularity and complexity of the sequence at a specific scale; a higher entropy value indicates greater complexity and irregularity. The calculated set of sample entropy values ​​is assumed to be: Scale 1: 1.05, Scale 2: 0.82, Scale 3: 0.88, Scale 4: 0.45, Scale 5: 0.30. This shows a trend of overall system complexity decreasing as the time scale increases, but the entropy value at scale 3 is relatively higher than that at scale 2, possibly reflecting the dynamic characteristics of the system at this specific time scale.

[0041] According to preset rules, the response characteristics of the sample entropy value set to typical faults in historical data were analyzed. In the initial stage of a historical "minor water-cooled wall leakage" fault, the entropy values ​​at all scales increased, but the entropy value at scale 2 jumped from the normal baseline of 0.82 to 1.50, with a change of approximately 83%, which was significantly higher than other scales (scale 1 changed by approximately 30%, and scale 3 changed by approximately 50%), indicating that scale 2 was the most sensitive to the dynamic changes of this fault. Therefore, based on the sensitivity to change, scale 2 is the optimal scale factor. We selected the sample entropy value corresponding to scale 2 (which is the current calculated value in real-time analysis) as the "characteristic entropy value" for subsequent state assessment. This value will be compared with the baseline entropy value corresponding to the specific operating condition established in step 104.

[0042] 104. By comparing the sample entropy value under the selected scale with the baseline entropy value established under historical normal operating conditions, an anomaly index characterizing the degree of deviation from the current operating state is calculated.

[0043] Specifically, based on historical normal operation data, for a selected scale, a baseline entropy value characterizing the normal complexity of the system and the standard deviation of entropy fluctuation reflecting the normal fluctuation range of this entropy value are calculated and stored. During real-time operation, the current feature entropy value corresponding to the selected scale under the current time window is obtained; the absolute difference between the current feature entropy value and the baseline entropy value is calculated; the absolute difference is divided by the standard deviation of entropy fluctuation to obtain the anomaly index. Further, according to different typical operating conditions in the historical normal operation data, the data is divided into multiple data subsets; for each data subset, its baseline entropy value and standard deviation of entropy fluctuation at the selected scale are calculated, forming a set of baseline entropy value sets and standard deviation sets corresponding to different operating conditions; during real-time operation, the current typical operating condition is monitored and determined, and the corresponding baseline entropy value and standard deviation of entropy fluctuation are retrieved from the baseline entropy value set and standard deviation set according to the determined condition.

[0044] It should be noted that step 104, based on the feature entropy value (sample entropy at scale 2) selected in step 103, calculates a quantified anomaly index by comparing the real-time entropy value with the historical benchmark, to characterize the degree to which the system's operating state deviates from normal operating conditions. A specific implementation example is as follows: Normal operating data of the main steam temperature of the smart power plant was collected over a relatively long historical period (three months) under various typical stable operating conditions. For this data, steps 101 to 103 were repeated to calculate a large number of sample entropy values ​​at scale 2. It is assumed that, statistically, the overall statistical mean of these entropy values ​​(i.e., the general baseline entropy value) is 0.82, and its standard deviation is 0.12. This means that, under normal circumstances, the sample entropy value at scale 2 typically fluctuates around 0.82.

[0045] Furthermore, to improve accuracy, historical data can be divided into subsets based on operating conditions to establish separate benchmarks, distinguishing between two typical operating conditions: "high load steady state" (unit load > 90%) and "low load steady state" (unit load between 60% and 70%). Calculations show: The baseline entropy value for the "high load steady state" condition is 0.85, and the standard deviation of entropy fluctuation is 0.09.

[0046] The baseline entropy value for the "low load steady state" condition is 0.80, and the standard deviation of entropy fluctuation is 0.08.

[0047] This resulted in a set of benchmarks corresponding to the operating conditions.

[0048] In real-time operation, the system first determines the current operating condition. The current unit load is 92%, which is identified as a "high load steady state" condition. The system then automatically calls the corresponding baseline entropy value of 0.85 and standard deviation of 0.09. The system processes the data in the current time window (the most recent 30 minutes) to obtain the feature entropy value at the current scale 2, assuming a calculated value of 1.20.

[0049] The absolute difference between the current feature entropy value of 1.20 and the baseline entropy value of 0.85 is calculated to be 0.35. Then, this absolute difference is divided by the standard deviation of the entropy value fluctuation under this operating condition, which is 0.09, to obtain the final anomaly index of approximately 3.9.

[0050] The calculated anomaly index (3.9 in this example) is a dimensionless value. Its physical meaning is clear: if the value is less than 1, it indicates that the current state fluctuation is within the normal range; if it is much greater than 1, it indicates that the deviation of the system's dynamic complexity has exceeded the normal fluctuation range, which may indicate an anomaly. This index provides accurate and quantitative input for the graded alarm judgment in step 105.

[0051] 105. Based on the different numerical ranges of the anomaly index and in conjunction with the original process measurement values ​​of key process parameters, trigger early warning, alarm, or normal status signals respectively.

[0052] Specifically, the system acquires the calculated anomaly index and the current original process measurement values ​​of key process parameters; compares the anomaly index with a preset first-level threshold, and generates a warning trigger judgment based on the comparison result; compares the anomaly index with a second-level threshold that is higher than the first-level threshold, and simultaneously determines whether the current original process measurement value exceeds its preset engineering safety threshold, and generates an alarm trigger judgment based on the combined results of the two judgments; and outputs the corresponding warning signal, alarm signal, or normal operation indication signal based on the warning trigger judgment and the alarm trigger judgment.

[0053] It should be noted that the following is a specific implementation example of step 105. This step receives the output from step 104: the anomaly index (AI) is 3.9, and the current raw process measurement value of the key process parameter (main steam temperature) is 538°C. The system's preset thresholds are as follows: Level 1 Threshold (Warning Threshold): AI=2.0 Second-level threshold (alarm threshold): AI=3.0 Level 3 threshold (Severe abnormal condition threshold): AI=4.5 Engineering safety threshold: The safe operating range for main steam temperature is 500℃ to 550℃.

[0054] The system first identifies the current unit load (92%) as a "high load steady state" condition and automatically calls the corresponding benchmark established in step 104. The specific classification judgment logic and signal generation process are shown in Table 1 below: Table 1 Three judgment channels are executed in parallel. Because AI=3.9 exceeds the warning threshold of 2.0, the system generates a warning. In alarm channel A, although AI exceeds the alarm threshold of 3.0 (meeting A1), the current main steam temperature of 538℃ is within the safe range (not meeting A2), so the parameter over-limit alarm is not triggered. In the newly added alarm channel B, AI=3.9 does not reach the severe abnormal state threshold of 4.5, so it is not triggered. Therefore, the system ultimately outputs a level-one warning signal. This signal indicates to operators that the system's inherent dynamic characteristics have shown significant abnormalities, posing a potential risk, but key parameters are still within safe limits. Close monitoring and preparation for preventative checks or adjustments are necessary. The optimized logic, while retaining the original rigor, avoids overlooking extremely chaotic states due to sensors or other factors causing the system to be stuck within safe limits, thanks to the independent channel B.

[0055] 106. Generate the corresponding control command sequence based on the output hierarchical alarm signals; The control command sequence is sent to the corresponding thermal process actuators to adjust the operating status of the relevant equipment. Furthermore, based on the output warning signal, reference value correction commands are generated for fine-tuning the relevant process control loops; based on the output alarm signal, protective control commands are generated for switching or interlocking the relevant equipment.

[0056] It should be noted that the system receives the status signal output from step 105 and generates a control command sequence matching the alarm level accordingly, thus achieving a closed loop from status perception to control execution. Following the example of step 105, the system currently outputs a level one warning signal (anomaly index 3.9, main steam temperature 538℃, under "high load steady state" condition). Based on this, preventative and diagnostic control commands will be generated, as shown in Table 2 below: Table 2 Command Sending and Execution: The above command sequence (commands 1, 2, and 4) is encapsulated into a standard communication protocol (OPCUA) data packet and sent by the alarm logic optimization server to the control station of the distributed control system (DCS). Commands 1 and 2 are received and executed by the analog control function block of the DCS, taking effect in the next control cycle (1 second later). Command 3 is sent to the plant-level monitoring information system (SIS) and auxiliary machine monitoring system, triggering a temporary change in their data acquisition strategy. Command 4 completes a self-test within the DCS.

[0057] Graded Response Explanation: This example generates preventative fine-tuning and enhanced monitoring commands for the "early warning signal." If step 105 outputs a "parameter limit exceeded alarm," the command sequence will escalate to stronger intervention measures, including significantly adjusting setpoints, drastically changing valve openings, or switching to standby equipment. If the output is a "severe abnormal status alarm," regardless of whether the parameter exceeds the limit, a diagnostic emergency operation will be triggered, forcibly switching the relevant control loop to manual mode and initiating an emergency diagnostic sequence for specific equipment. The entire design ensures that the control response and alarm level are strictly matched, forming a graded intelligent control closed loop.

[0058] Figure 2 This is a schematic diagram of the structure of an alarm logic optimization device for thermal control in a smart power plant, provided by an embodiment of the present invention. This alarm logic optimization device 200 for thermal control in a smart power plant can vary considerably due to differences in configuration or performance. The device 200 includes a transmitter 201, a receiver 202, and a processor 203. The processor 203 can also be a controller. Figure 2 The device is referred to as "controller / processor 203". Optionally, the device 200 may also include a modem processor 205, wherein the modem processor 205 may include an encoder 206, a modulator 207, a decoder 208, and a demodulator 209.

[0059] In one example, transmitter 201 modulates (e.g., analog-to-analog conversion, filtering, amplification, and up-conversion, etc.) the output sample and generates an uplink signal, which is transmitted via an antenna to an access network device. On the downlink, the antenna receives the downlink signal transmitted by the access network device. Receiver 202 modulates (e.g., filtering, amplification, down-conversion, and digitization, etc.) the signal received from the antenna and provides an input sample. In modem processor 205, encoder 206 receives traffic data and signaling messages to be transmitted on the uplink and processes (e.g., formatting, encoding, and interleaving) the traffic data and signaling messages. Modulator 207 further processes (e.g., symbol mapping and modulation) the encoded traffic data and signaling messages and provides an output sample. Demodulator 209 processes (e.g., demodulates) the input sample and provides a symbol estimate. Decoder 208 processes (e.g., deinterleaving and decoding) the symbol estimate and provides decoded data and signaling messages to device 200. Encoder 206, modulator 207, demodulator 209, and decoder 208 can be implemented by a combined modem processor 205. These units process data according to the radio access technology used by the radio access network (e.g., LTE and other evolved systems access technologies). It should be noted that when device 200 does not include modem processor 205, the aforementioned functions of modem processor 205 can also be performed by processor 203.

[0060] The processor 203 controls and manages the operation of the device 200, and is used to execute the processing procedures performed by the device 200 in the above embodiments of this disclosure. For example, the processor 203 is also used to execute various steps of the transmitting or receiving device in the above method embodiments, and / or other steps of the technical solutions described in the embodiments of this disclosure.

[0061] Furthermore, the device 200 may also include a memory 204 for storing program code and data for the device 200.

[0062] Understandable, Figure 2 Only a simplified design of device 200 is shown. In practical applications, device 200 can include any number of transmitters, receivers, processors, modem processors, memory, etc., and all devices that can implement the embodiments of this disclosure are within the protection scope of the embodiments of this disclosure.

[0063] The present invention also provides an alarm logic optimization device for thermal control of smart power plants. The alarm logic optimization device for thermal control of smart power plants includes a memory and a processor. The memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor performs the steps of the alarm logic optimization method for thermal control of smart power plants in the above embodiments.

[0064] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the steps of the alarm logic optimization method for thermal control of a smart power plant.

[0065] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0066] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0067] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing alarm logic in the thermal control of a smart power plant, characterized in that, include: Preprocess the time series of at least one key process parameter of the thermal system of a smart power plant to generate a dimensionless data sequence. The dimensionless data sequence is reconstructed in phase space to construct a multidimensional phase space point set; Complexity analysis is performed on the sequence corresponding to the multidimensional phase space point set at multiple time scales, and the sample entropy value at at least one selected scale is calculated and output. By comparing the sample entropy value under the selected scale with the baseline entropy value established under historical normal operating conditions, an anomaly index characterizing the degree of deviation from the current operating state is calculated. Based on the different numerical ranges of the anomaly index and in combination with the original process measurements of the key process parameters, a state signal is obtained.

2. The alarm logic optimization method for intelligent power plant thermal control according to claim 1, characterized in that, include: Define a set of time scale factors for multiscale analysis of the system to obtain a set of scale factors; For each scale factor in the set of scale factors, the single-dimensional time series data corresponding to the multidimensional phase space point set is coarsened to generate a coarse-grained time series corresponding to each scale factor. For each of the coarse-grained time series, its sample entropy value is calculated based on the given template dimension and similarity tolerance, thereby obtaining a set of sample entropy values ​​corresponding to the set of scale factors; Based on preset selection rules, at least one sample entropy value is selected from the set of sample entropy values.

3. The alarm logic optimization method for intelligent power plant thermal control according to claim 2, characterized in that, The response characteristics of entropy values ​​under different scale factors in the sample entropy value set to historical typical failure events are analyzed, and the sensitivity analysis results of changes in each scale factor are obtained. Based on the results of the change sensitivity analysis, the scale factors are sorted to generate a sorted list of scale factors arranged from high to low sensitivity. Based on the scale factor sorting list, the scale factor ranked first is selected as the optimal scale factor; The sample entropy value corresponding to the optimal scaling factor is selected as the feature entropy value.

4. The alarm logic optimization method for intelligent power plant thermal control according to claim 1, characterized in that, include: Based on historical normal operation data, for the selected scale, a baseline entropy value characterizing the normal complexity of the system and the standard deviation of entropy fluctuation reflecting the normal fluctuation range of the entropy value are calculated and stored. During real-time operation, the current feature entropy value corresponding to the selected scale within the current time window is obtained; Calculate the absolute difference between the current feature entropy value and the baseline entropy value; The anomaly index is obtained by dividing the absolute difference by the standard deviation of the entropy fluctuation.

5. The alarm logic optimization method for intelligent power plant thermal control according to claim 1, characterized in that, include: Obtain the calculated anomaly index and the current original process measurement values ​​of the key process parameters; The anomaly index is compared with a preset first-level threshold, and an early warning trigger judgment is generated based on the comparison result. The anomaly index is compared with a second-level threshold that is higher than the first-level threshold, and it is simultaneously determined whether the current original process measurement value exceeds its preset engineering safety threshold. The alarm trigger judgment is generated by combining the two judgment results. Based on the aforementioned warning trigger judgment and alarm trigger judgment, output the corresponding warning signal, alarm signal or normal operation indication signal.

6. The alarm logic optimization method for intelligent power plant thermal control according to claim 1, characterized in that, Also includes: Based on the output graded alarm signals, generate the corresponding control command sequence; The control command sequence is sent to the corresponding thermal process actuator to adjust the operating status of the relevant equipment.

7. The alarm logic optimization method for intelligent power plant thermal control according to claim 5, characterized in that, Based on the output warning signal, a reference value correction command is generated for fine-tuning the relevant process control loop; Based on the output alarm signal, generate protective control commands for switching or interlocking related equipment.

Images