Special equipment hidden danger active identification method based on multi-source operation and maintenance data fusion

Abstract

A special equipment hidden danger active identification method based on multi-source operation and maintenance data fusion, the method comprises the following steps: acquiring multi-source operation and maintenance data of the boiler during operation, and preprocessing the multi-source operation and maintenance data to construct a multi-source time series data set. Based on the multi-source time series data set and the boiler operation state, the boiler operation stage is identified to divide the multi-source time series data set into multiple working condition intervals. According to the correlation judgment criterion between the boiler water level and the multi-source operation and maintenance data in each working condition interval, the water level hidden danger candidate state of the boiler water level in each working condition interval is screened. The correlation between the water level hidden danger candidate state and the working condition interval is checked to exclude the water level hidden danger candidate state under the working condition transition, and the final water level hidden danger state is output. Through the separate management of the transition response of each working condition stage, the present application effectively avoids the large fluctuation of the determination result in the adjacent time window, and improves the accuracy and stability of the water level hidden danger identification result.

Description

Technical Field

[0001] This invention relates to the field of equipment management technology, and in particular to a method for proactively identifying potential hazards in special equipment based on the fusion of multi-source operation and maintenance data. Background Technology

[0002] In chemical plants, combined heat and power (CHP) systems, or centralized heating systems, boilers often experience various operating conditions, including ignition and start-up, rapid load ramp-up, intermittent steady-state operation, emergency load reduction, or start-up / shutdown switching. During these switching processes, the boiler water level is not a single control target but is affected by a combination of physical processes, such as feedwater regulation, changes in evaporation intensity, thermal expansion and contraction of the steam drum, and disturbances at the steam-water interface.

[0003] In engineering practice, significant rises and falls, amplified fluctuations, or brief instability in boiler water levels within a short period are often normal dynamic responses under changing operating conditions, rather than genuine hidden dangers. Traditional technical solutions determine whether water level anomalies constitute hidden dangers by constructing a consistency relationship among multiple sources of characteristics between boiler operating parameters, environmental conditions, and maintenance operations. However, this consistency judgment implicitly assumes that the correlation between boiler water level and feedwater flow, steam load, and operational behavior is comparable within adjacent time windows. However, in scenarios with frequent operating condition changes, this premise no longer holds, specifically manifested in the following ways: (1) During the load ramp-up phase, the sudden increase in evaporation intensity leads to a “false drop” in water level, but this change is consistent with the boiler thermodynamic response mechanism; (2) During emergency load reduction or start-up / shutdown, the coupling relationship between steam drum pressure and water level undergoes phased reconstruction, and the short-term fluctuation amplitude of water level increases significantly; (3) The “normal water level fluctuation range” varies significantly under different operating stages.

[0004] Therefore, the same water level change characteristic may represent both a reasonable operating condition response and a genuine potential water supply imbalance within different operating window periods. The judgment mechanism based on unified consistency constraints cannot distinguish between these two fundamentally different states. From an engineering mechanism perspective, the essence of this deficiency lies in the fact that boiler water level is a comprehensive response quantity strongly dependent on operating conditions, and its stable correlation with multi-source operation and maintenance characteristics only has engineering significance within a relatively stable operating range. However, when the boiler operating state frequently crosses the boundaries of operating conditions such as ignition, steady state, variable load, and start-up / shutdown, the normal correlation between multi-source characteristics exhibits segmentation and stages. Consequently, the original consistency constraints are prone to structural failure at the moment of operating condition switching, and the hazard judgment results are prone to significant fluctuations within adjacent time windows, leading to inaccurate water level hazard identification results. Summary of the Invention

[0005] Therefore, it is necessary to propose a method for proactive identification of potential hazards in special equipment based on the fusion of multi-source operation and maintenance data to address the aforementioned technical issues.

[0006] The present invention adopts the following technical solution.

[0007] The first aspect of this invention discloses a method for proactive identification of potential hazards in special equipment based on multi-source operation and maintenance data fusion, the method comprising: Acquire multi-source operation and maintenance data during boiler operation, and preprocess the multi-source operation and maintenance data to construct a multi-source time series dataset; Based on the multi-source time series dataset and boiler operating status, the boiler operating stage is identified to divide the multi-source time series dataset into multiple operating condition intervals. Based on the correlation judgment criteria between boiler water level and multi-source operation and maintenance data in each operating condition interval, candidate states of potential water level hazards in boiler water level in each operating condition interval are screened. The correlation between the candidate states of water level hazards and the operating condition intervals is verified to exclude candidate states of water level hazards under the transition of operating conditions, and the final state of water level hazards is output.

[0008] Furthermore, the step of acquiring multi-source operation and maintenance data during boiler operation and preprocessing the multi-source operation and maintenance data to construct a multi-source time series dataset includes: The system acquires water level status data, load status data, and operation and maintenance data during boiler operation, and marks the water level status data, load status data, and operation and maintenance data with equipment identifiers and timestamps, and integrates them to obtain the multi-source operation and maintenance data. A standard time axis is constructed by setting a unified time step, and various types of operation and maintenance data in the multi-source operation and maintenance data are mapped and linearly interpolated on the standard time axis to obtain a multi-source data sequence with a unified time axis.

[0009] Furthermore, the step of acquiring multi-source operation and maintenance data during boiler operation and preprocessing the multi-source operation and maintenance data to construct a multi-source time series dataset also includes: Key operational quantities are extracted from the multi-source data sequence, and derived features of the key operational quantities are calculated. The key operational quantities are boiler water level and steam flow at different time points, and the derived features are water level change rate and boiler load change intensity. Using the standard time axis as an index, the bit change rate and boiler load change intensity at each time point are correlated with the operation and maintenance data to construct the multi-source time series dataset.

[0010] Furthermore, based on the multi-source time series dataset and boiler operating status, the boiler operating stage is identified to divide the multi-source time series dataset into multiple operating condition intervals, including: Based on the multi-source time series dataset, the rate of change of boiler drum pressure during boiler operation is calculated according to the set sampling period. The rate of change of boiler drum pressure is the rate of change of boiler drum pressure in adjacent sampling periods. Based on the steam drum pressure change rate and derived characteristics, a comprehensive operating condition change intensity index is defined, and the boiler operation stage is identified based on the comprehensive operating condition change intensity index and the boiler load change intensity to obtain the preliminary judgment result of the operation stage. If the duration of any operating phase in the preliminary judgment of the operating phase is not less than the preset minimum duration constraint, the corresponding adjacent operating phases will be merged into the same operating condition interval; otherwise, they will be merged into the operating condition transition phase.

[0011] Furthermore, the step of screening candidate states of potential water level hazards in each operating condition interval based on the correlation judgment criteria between boiler water level and multi-source operation and maintenance data includes: According to the set sampling period, multiple sets of operation and maintenance data variables corresponding to each working condition interval are extracted from the multi-source time series dataset, and the operation and maintenance data variables are normalized within the corresponding working condition interval to construct a mechanism variable set sequence. Based on the sequence of mechanistic variables, the expected increment of the boiler water level response is established within the operating range using a physical conservation framework, and the expected boiler water level is determined based on the expected increment.

[0012] Furthermore, the step of screening candidate states of potential water level hazards in each operating condition interval based on the correlation judgment criteria between boiler water level and multi-source operation and maintenance data in each operating condition interval also includes: Obtain the measured rate of change of boiler water level and the expected rate of change of boiler water level during different sampling periods, and calculate the deviation of the measured rate of change of boiler water level from the expected rate of change of boiler water level. The water level deviation within the operating range is defined based on the water level change rate deviation, and a deviation persistence count is generated based on the water level deviation. Within the operating range, a deviation threshold and a minimum persistence threshold are set. When the water level deviation exceeds the deviation threshold and the deviation persistence count is not lower than the minimum persistence threshold, a candidate state for water level hazard is determined.

[0013] Furthermore, the step of verifying the correlation between the candidate states of water level hazards and the operating condition intervals to exclude candidate states of water level hazards under transitional operating conditions and outputting the final state of water level hazards includes: Obtain the boundary time between different operating conditions, and construct a boundary window based on the boundary time, so as to calculate the transition weight corresponding to the operating time when any running time is within the boundary window; The running time within the boundary window and the transition weight of the running time are marked as transition response events, and the water level deviation corresponding to the transition response events is suppressed by a set suppression coefficient.

[0014] Furthermore, the step of verifying the correlation between the candidate states of water level hazards and the operating condition intervals to exclude candidate states of water level hazards under transitional operating conditions and outputting the final state of water level hazards also includes: The effective water level deviation after the transition response event is suppressed is continuously verified according to the set time window length, and when the effective water level deviation reaches the hidden danger judgment rule, the transition response event is marked as a water level hidden danger state. The water level hazard status is standardized and classified according to the preset hazard intensity classification threshold, so as to output the classified water level hazard status.

[0015] The second aspect of this invention discloses a special equipment hazard proactive identification device based on multi-source operation and maintenance data fusion, used to implement the special equipment hazard proactive identification method based on multi-source operation and maintenance data fusion as described in any one of the first aspects, the device comprising: The data preprocessing module is used to acquire multi-source operation and maintenance data during boiler operation and to preprocess the multi-source operation and maintenance data to construct a multi-source time series dataset. The operating condition interval division module is used to identify the boiler operating stage based on the multi-source time series dataset and the boiler operating status, so as to divide the multi-source time series dataset into multiple operating condition intervals. The candidate hazard screening module is used to screen the candidate status of boiler water level hazards in each operating condition range based on the correlation judgment benchmark between boiler water level and multi-source operation and maintenance data in each operating condition range. The hidden danger verification output module is used to verify the correlation between the candidate states of water level hidden dangers and the working condition range, so as to exclude the candidate states of water level hidden dangers under the transition of working conditions and output the final state of water level hidden dangers.

[0016] A third aspect of the present invention discloses a terminal, including a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method described in the first aspect.

[0017] A fourth aspect of the present invention discloses a computer-readable storage medium having a computer program stored thereon, characterized in that the program, when executed by a processor, implements the steps of the method described in the first aspect.

[0018] The present invention has the following advantages: (1) This invention collects multi-source operation and maintenance data generated during boiler operation, including operating parameters reflecting water level status, operating status quantities reflecting load changes, and operation and maintenance records reflecting human adjustments. The collected multi-source operation and maintenance data are aligned along a unified time axis to construct a multi-source time series dataset reflecting the continuous operation of the boiler. Then, combining the characteristics of state changes during boiler operation, different operating stages such as ignition, load ramp-up, steady-state operation, load reduction, or start-up / shutdown are identified, and the continuous time series is divided into several operating condition intervals. Simultaneously, a correlation judgment benchmark is established only within each operating condition interval between boiler water level and feedwater regulation status, evaporation load changes, and operation and maintenance behavior. Water level changes within each operating condition interval are continuously evaluated. When the water level characteristics deviate from the normal response range within that operating condition interval, it is judged as a candidate state for water level hazard within the operating condition. This method of first identifying the operating condition and then evaluating water level hazards within the operating condition achieves separate management of the transition response at each operating condition stage, effectively avoiding significant fluctuations in the judgment results within adjacent time windows, and improving the accuracy and stability of water level hazard identification results.

[0019] (2) Based on the determination of the candidate states of water level hazards in each working condition interval, the present invention performs working condition correlation verification on the obtained candidate states of water level hazards. For water level fluctuations that occur within the working condition switching boundary, they are not directly determined as hazards, but are managed separately as working condition transition responses. For deviation states that continue to exist or gradually worsen within the same working condition interval, the output is the final active identification result of water level hazards, thereby completing the stable identification of special equipment hazards and further improving the accuracy and stability of water level hazard identification results. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating the proactive identification method for potential hazards in special equipment based on multi-source operation and maintenance data fusion provided by the present invention.

[0022] Figure 2This is a schematic diagram of the structure of the special equipment hidden danger active identification device based on multi-source operation and maintenance data fusion provided by the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] like Figure 1 As shown in one embodiment, a method for proactive identification of potential hazards in special equipment based on multi-source operation and maintenance data fusion includes the following steps: Step S110: Obtain multi-source operation and maintenance data during boiler operation and preprocess the multi-source operation and maintenance data to construct a multi-source time series dataset.

[0025] In some embodiments, the special equipment hidden danger proactive identification method based on multi-source operation and maintenance data fusion provided by the present invention includes the following steps in step S110: Step S111: Obtain water level status data, load status data, and operation and maintenance data during boiler operation, and mark the water level status data, load status data, and operation and maintenance data with equipment identification and timestamps, and integrate them to obtain multi-source operation and maintenance data.

[0026] Step S112: Construct a standard time axis by setting a unified time step, and perform mapping and linear interpolation on various types of operation and maintenance data in the multi-source operation and maintenance data on the standard time axis to obtain a multi-source data sequence with a unified time axis.

[0027] In some embodiments, the special equipment hidden danger proactive identification method based on multi-source operation and maintenance data fusion provided by the present invention further includes the following steps in step S110: Step S113: Extract key operating quantities from the multi-source data sequence and calculate the derived features of the key operating quantities. The key operating quantities are the boiler water level and steam flow at different time points, and the derived features are the water level change rate and the boiler load change intensity.

[0028] Step S114: Using the standard time axis as an index, correlate the rate of change of position and the intensity of change of boiler load at each time point with the operation and maintenance data to construct a multi-source time series dataset.

[0029] In a specific embodiment, the special equipment hidden danger proactive identification method based on multi-source operation and maintenance data fusion provided by the present invention includes steps 1 to 4: Step 1: Multi-source operation and maintenance data collection and unified time series construction.

[0030] Collect multi-source operation and maintenance data generated during boiler operation, including operating parameters reflecting water level status, operating status quantities reflecting load changes, and operation and maintenance operation records reflecting human adjustments. Align these multi-source data according to a unified time axis to construct a multi-source time series dataset reflecting the continuous operation of the boiler. This includes the following sub-steps: Sub-step 1.1: Classification, collection, and solidification of the physical meaning of multi-source boiler operation and maintenance data.

[0031] Specifically, during boiler operation, the following three types of data with clear engineering implications are collected: The first category is data reflecting boiler water level status, used to characterize the dynamic changes in the steam-water interface within the steam drum, denoted as the water level status data set. Its basic quantities include: boiler steam drum water level value (in millimeters); water level change rate (in millimeters per second). The second category is data reflecting boiler load and evaporation intensity, used to characterize the trend of operating conditions, denoted as the load status data set. Its basic quantities include: steam flow rate (in tons per hour); feedwater flow rate (in tons per hour); steam drum pressure (in megapascals). The third category is data reflecting human intervention behavior, used to characterize the operational intervention process, denoted as the operation and maintenance data set. Its basic quantities include: feedwater valve opening degree (in percentage); start / stop operation status (value 0 or 1); and the time of the adjustment action. Subsequently, to ensure the feasibility of subsequent fusion processing, each of the above data categories is appended with a unique device identifier and a collection timestamp, forming the original multi-source operation and maintenance data set. This set consists of the water level status dataset, load status dataset, and operation and maintenance data set.

[0032] Sub-step 1.2: Unify the sampling frequency of multi-source operation and maintenance data and map it to the time axis.

[0033] Specifically, due to the different sampling frequencies of various maintenance data, a standardized time step (ranging from 1 to 10 seconds, determined based on the boiler control system refresh cycle) is first established to construct a standard time axis. For any type of maintenance data, mapping is performed on the standard time axis. If a data point has a sampled value at the corresponding time point, it is directly mapped; otherwise, linear interpolation between adjacent time points is used to fill in the gaps, ultimately ensuring that all maintenance data have a one-to-one time series representation on the same time axis.

[0034] Sub-step 1.3: Derivation calculation of water level change characteristics and load change characteristics.

[0035] Specifically, under a unified standard timeline, derived characteristics are calculated for key operational quantities, including: The rate of change of boiler water level is calculated using the following expression: ; In the formula, for The rate of change of boiler water level at any given time; , These are the boiler water level values ​​at two adjacent time points; The set uniform time step.

[0036] Similarly, the expression for calculating the intensity of boiler load changes is: ; In the formula, for The intensity of boiler steam load changes at any given time; , The steam flow rate at two adjacent time points; The set uniform time step.

[0037] Therefore, the key operational variables are boiler water level and steam flow rate, with derived features being the rate of change of boiler water level and the intensity of change of boiler steam load. Through the above calculations, the multi-source time series dataset includes not only the original operational status variables but also derived variables reflecting dynamic trends.

[0038] Sub-step 1.4: Structured encapsulation of multi-source time series datasets.

[0039] Specifically, using the aforementioned unified standard time axis as an index, the boiler water level change rate, boiler steam load change intensity, and operation and maintenance data at each moment are combined to obtain multiple data groups. Then, each data group is arranged in chronological order to obtain the final multi-source time series dataset.

[0040] Step S120: Based on the multi-source time series dataset and boiler operating status, identify the boiler operating stage to divide the multi-source time series dataset into multiple operating condition intervals.

[0041] In some embodiments, the special equipment hidden danger proactive identification method based on multi-source operation and maintenance data fusion provided by the present invention includes the following steps in step S120: Step S121: Based on the multi-source time series dataset, calculate the rate of change of boiler drum pressure during boiler operation according to the set sampling period. The rate of change of boiler drum pressure is the rate of change of boiler drum pressure in adjacent sampling periods.

[0042] Step S122: Define a comprehensive operating condition change intensity index based on the steam drum pressure change rate and derived characteristics, and identify the boiler operation stage based on the comprehensive operating condition change intensity index and the boiler load change intensity to obtain the preliminary judgment result of the operation stage.

[0043] Step S123: If the duration of any operating phase in the preliminary judgment result of the operating phase is not less than the preset minimum duration constraint, the corresponding adjacent operating phases are merged into the same operating condition interval; otherwise, they are merged into the operating condition transition phase.

[0044] In a specific embodiment, the special equipment hidden danger proactive identification method based on multi-source operation and maintenance data fusion provided by the present invention includes step 2, operating condition interval identification and operating condition label generation. Based on the multi-source time series dataset constructed in step 1, and combined with the state change characteristics during boiler operation, the method identifies different operating stages of the boiler, such as ignition, load ramp-up, steady-state operation, load reduction, or start-up and shutdown, and divides the continuous time series into several operating condition intervals. This step generates a unique operating condition label for each operating condition interval, ensuring that the water level response mechanism within the same operating condition interval remains relatively consistent. This includes the following sub-steps: Sub-step 2.1: Extraction of key operating condition discrimination quantities and calculation of change rate.

[0045] Specifically, firstly, the rate of change of the multi-source time series is calculated under a unified sampling period to construct basic quantities reflecting the intensity of changes in the boiler's operating state. This includes calculating the rate of change of the steam drum pressure, i.e., the difference in steam drum pressure between adjacent sampling periods is used as the ratio of the numerator to the sampling period, which is used to reflect whether the boiler's thermal state is in a rapid change phase. Then, the rate of change of the steam drum pressure is integrated with the aforementioned boiler water level change rate and boiler steam load change intensity to form a set of basic quantities for preliminary judgment of operating conditions.

[0046] Sub-step 2.2: Preliminary assessment of the operating stage based on thermal mechanisms.

[0047] Specifically, based on the aforementioned preliminary set of fundamental quantities for determining operating conditions, and combined with the boiler's thermal operation mechanism, the typical changing characteristics of different operating stages are described in a standardized manner. First, a comprehensive operating condition change intensity index is defined, with the following expression: ; In the formula, , These are all weighting coefficients, ranging from 0.3 to 1.5, used to balance the impact of load and pressure on changes in operating conditions; for The intensity index of comprehensive operating condition changes at any given time; for The rate of change of steam drum pressure at any given time; for Steam flow rate at any given time; for The boiler water level at any given time.

[0048] Subsequently, based on the comprehensive operating condition change intensity index and the sign and amplitude of the steam flow rate, a preliminary judgment is made on the operating phase. When the comprehensive operating condition change intensity index is less than the preset steady-state threshold and When it approaches zero, it is determined to be in a steady-state operation phase; when When the value is significantly greater than zero and the duration exceeds the set window, it is determined to be in the load ramp-up phase; when When the value is significantly less than zero, it is determined to be in the load reduction phase; when the comprehensive operating condition change intensity index remains high for a long time and is accompanied by drastic changes in water level and pressure, it is determined to be in the start-up, shutdown, or ignition phase. The final preliminary judgment result of the operation phase is obtained.

[0049] Sub-step 2.3: Division of working condition intervals under continuity constraints.

[0050] Specifically, to avoid frequent changes in operating conditions caused by instantaneous disturbances, a minimum duration constraint is introduced. A candidate operating condition is defined as being valid continuously within a certain time interval. If and only if the length of the time interval is greater than or equal to the introduced minimum duration constraint (range 30 seconds to 5 minutes, set according to boiler capacity and control response speed), adjacent similar operating phases that meet the duration constraint are merged into a continuous operating condition interval. For short-term changes that do not meet the continuity condition, they are uniformly merged into operating condition transition segments and no separate operating condition intervals are generated.

[0051] Step S130: Based on the correlation judgment benchmark between boiler water level and multi-source operation and maintenance data in each operating condition interval, candidate states of potential water level hazards in boiler water level in each operating condition interval are screened.

[0052] In some embodiments, the special equipment hidden danger proactive identification method based on multi-source operation and maintenance data fusion provided by the present invention includes the following steps in step S130: Step S131: Extract multiple sets of operation and maintenance data variables corresponding to each operating condition interval from the multi-source time series dataset according to the set sampling period, and normalize each set of operation and maintenance data variables within the corresponding operating condition interval to construct a mechanism variable set sequence.

[0053] Step S132: Based on the sequence of mechanistic variables, establish the expected increment of boiler water level response within the operating range through the physical conservation framework, and determine the expected boiler water level based on the expected increment.

[0054] In some embodiments, the special equipment hidden danger proactive identification method based on multi-source operation and maintenance data fusion provided by the present invention further includes the following steps in step S130: Step S133: Obtain the measured boiler water level change rate and the expected boiler water level change rate during different sampling periods, and calculate the deviation of the measured boiler water level change rate from the expected boiler water level change rate.

[0055] Step S134: Define the water level deviation within the operating range based on the deviation of the water level change rate, and generate a deviation persistence count based on the water level deviation.

[0056] Step S135: Set a deviation threshold and a minimum persistence threshold within the operating range, and determine the candidate state of water level hazard when the water level deviation exceeds the deviation threshold and the deviation persistence count is not lower than the minimum persistence threshold.

[0057] In a specific embodiment, the special equipment hidden danger proactive identification method based on multi-source operation and maintenance data fusion provided by the present invention includes step 3, construction and deviation judgment of water level characteristic relationships within the operating condition. For each operating condition interval identified in step 2, a correlation judgment benchmark is established only within that operating condition interval between boiler water level and feedwater regulation status, evaporation load changes, and operation and maintenance behavior. Water level changes within the operating condition interval are continuously evaluated, and when the water level characteristics deviate from the normal response range within that operating condition interval, it is judged as a candidate state of water level hidden danger within the operating condition. This includes the following sub-steps: Sub-step 3.1: Construction and standardized mapping of the "water level response mechanism variable group" within the working condition.

[0058] Specifically, within each operating condition interval, a set of mechanistic variables is extracted and constructed using the same sampling period. This set of variables includes drum water level, feedwater flow rate, steam flow rate, drum pressure, feedwater regulation, and operational event intensity. To mitigate the influence of different dimensions on the same judgment criterion, each variable is standardized within the operating condition interval to obtain normalized variables. The operational event intensity is converted from "valve / pump start / stop, manual / automatic switching, and setpoint adjustment" records into intensity coefficients ranging from 0 to 1; the sampling period is set to 1-10 seconds, determined by the boiler DCS sampling capability.

[0059] Sub-step 3.2: Construct the "water level-water supply-evaporation-pressure" correlation benchmark within the operating condition.

[0060] Specifically, within the operating range, the expected increment of the water level response is established using a physical conservation framework. First, the water level increment and the net inflow-outflow ratio are defined to characterize the relative relationship between feedwater and evaporation, and the pressure term is used to describe the impact of the "flooding / shrinking" effect on the apparent water level.

[0061] The expression for the net inflow-outflow ratio is: ; In the formula, for Net inflow-outflow ratio at any given time; for The water flow rate at any given time is expressed in tons per hour or kilograms per second. for Steam flow rate at any given time; To prevent constants with zero denominators, we take values ​​between 0.01 and 0.10.

[0062] The expression for the expected increment based on water level increment and pressure correction is: ; In the formula, Operating range The equivalent coefficient for the change in net mass to the change in water level, with a value ranging from 0.0001 to 0.01 (determined by unit conversion). Operating range Below The expected increment of water level and pressure correction at any given time; The apparent influence coefficient of pressure on water level is taken as 0.00001 to 0.001; Operating range The reference steam drum pressure below; for The pressure in the steam drum at any given moment; The sampling period.

[0063] Therefore, the desired water level in the boiler is equal to the previous sampling time. Measured water level and expected increment By establishing a water level response benchmark through "conservation relationship within the working condition + pressure correction", the situation of mistaking normal dynamic responses such as start-up, shutdown, and ramping across working conditions as hidden dangers can be effectively avoided.

[0064] Sub-step 3.3: Calculation of deviation within the working condition and generation of the determination quantity for continuous enhancement.

[0065] Specifically, within the operating range, anomalies are evaluated simultaneously from two paths: "absolute deviation of water level" and "deviation of water level change rate". First, the measured water level change rate needs to be calculated, that is, the difference between the measured water levels at adjacent sampling times is used as the ratio of the numerator to the sampling period. Then, the measured water level change rate is compared with the expected water level change rate. The expected water level change rate is equal to the difference between the expected water level at the current sampling time and the measured water level at the previous sampling time, which is used as the ratio of the numerator to the sampling period.

[0066] Therefore, operating condition range The deviation within is defined as: ; In the formula, Operating range Inner Water level deviation at any given time; The water level deviation scale parameter is taken as the typical amplitude of the water level residual within the operating condition range; is the rate deviation scale parameter; is a small stability constant, ranging from 0.1 to 1.0 (selected according to water level units). Used to avoid the denominator being too small; , They are respectively The measured rate of water level change and the expected rate of water level change at any given time.

[0067] Then, based on the persistence count at the previous sampling time, the persistence count at the current time is determined to suppress single-point transient fluctuations. The persistence count at the current time is equal to the persistence count at the previous sampling time plus the indication of whether the deviation exceeds the limit (taken as 0 or 1).

[0068] Sub-step 3.4: Output candidate states of potential water level hazards within the working condition.

[0069] Specifically, a condition-specific threshold and a minimum persistence threshold are set within each operating condition interval. A potential hazard candidate status is output only when the deviation exceeds the limit and the persistence requirement is met. The condition-specific threshold is set to 2 to 6 (dimensionless), and the minimum persistence threshold is set to 3 to 20 (corresponding to 3 to 20 consecutive exceedances of the limit).

[0070] When the deviation exceeds the condition-specific threshold of the corresponding working condition interval and the persistence count is greater than or equal to the minimum persistence threshold, the corresponding time period is merged into candidate segments, and the maximum deviation and the main contributing item (from the water level item or the rate item) are output in each segment.

[0071] Step S140: Verify the correlation between the candidate states of water level hazards and the operating condition intervals to exclude candidate states of water level hazards under the transition of operating conditions, and output the final state of water level hazards.

[0072] In some embodiments, the special equipment hidden danger proactive identification method based on multi-source operation and maintenance data fusion provided by the present invention includes the following steps in step S140: Step S141: Obtain the boundary time between different operating condition intervals, and construct a boundary window based on the boundary time to calculate the transition weight corresponding to the operating time when any running time is within the boundary window.

[0073] Step S142: The running time and the transition weight of the running time located within the boundary window are marked as transition response events, and the water level deviation corresponding to the transition response events is suppressed by the set suppression coefficient.

[0074] In some embodiments, the special equipment hidden danger proactive identification method based on multi-source operation and maintenance data fusion provided by the present invention further includes the following steps in step S140: Step S143: Continuously verify the effective water level deviation after the transition response event is suppressed according to the set time window length, and mark the transition response event as a water level hazard state when the effective water level deviation reaches the hazard judgment rule.

[0075] Step S144: Standardize and classify the water level hazard status according to the preset hazard intensity classification threshold, so as to output the classified water level hazard status.

[0076] In a specific embodiment, the special equipment hidden danger proactive identification method based on multi-source operation and maintenance data fusion provided by the present invention includes step 4, cross-operating condition transition management and proactive hidden danger identification output. The candidate states of water level hidden dangers obtained in step 3 are subjected to operating condition correlation verification. Water level fluctuations occurring within the operating condition switching boundary are not directly identified as hidden dangers, but are managed separately as operating condition transition responses. For deviation states that persist or gradually worsen within the same operating condition range, the final proactive water level hidden danger identification result is output, thereby completing the stable identification of special equipment hidden dangers. This includes the following sub-steps: Sub-step 4.1: Extraction of boundary window for operating condition switching and transition response marking.

[0077] Specifically, for each moment when the operating condition label changes, a boundary window is constructed around that moment to cover transient switching events such as start-up, shutdown, ramp-up, and load reduction. This boundary window is set between 5 and 120 seconds, based on the boiler capacity and the response speed of the regulating system. For any given moment, it is determined whether that moment falls within the constructed boundary window, and its transition weight is calculated to suppress false judgments of "false water level fluctuations." The closer the current moment is to the center of the boundary window, the greater the transition weight, with a value of 0-1. Simultaneously, it is determined which side of the boundary window the current moment is on, and corresponding markings are applied to both sides.

[0078] Sub-step 4.2, separate management of transition response and candidate suppression output.

[0079] Specifically, transition management is performed on candidate markers within the boundary window. At this time, potential risks are directly output and only recorded as transition response events. The deviation is suppressed by the suppression coefficient (ranging from 0.5 to 5) to prevent the deviation from being amplified during the switching instant. The suppressed deviation is equal to the ratio of the current deviation as the numerator to the product of the transition weight and the suppression coefficient plus the value 1.

[0080] Sub-step 4.3, verification of persistence and aggravation trend within the working condition.

[0081] Specifically, the persistence of the suppressed deviation is continuously verified only within the same operating condition range to determine whether the deviation that "continues to exist or gradually worsens" constitutes a real hidden danger. The persistence is evaluated with a set window length (ranging from 30 seconds to 600 seconds). If the deviation continues to exceed the limit or shows an aggravating trend within the non-boundary window, a hidden danger is output, and the final list of hidden danger events and the intensity of each hidden danger are obtained.

[0082] When determining whether the deviation within the non-boundary window continues to exceed the limit or shows an aggravating trend, it is necessary to first determine the number of potential hazards in the boundary window and the number of transition response events after deviation suppression. Then, by combining the absolute value of the difference between the average deviation at the end of the boundary window and the average deviation at the beginning of the boundary window, the intensity of the potential hazard in the transition section can be determined.

[0083] Sub-step 4.4: Output and closed-loop delivery of proactive hazard identification results.

[0084] Specifically, the list of potential hazards is graded and standardized according to the hazard intensity grading threshold. The grading threshold can be adjusted according to the boiler type. At the same time, the transition response events after the deviation of the boundary window transition stage is suppressed are output as "switching period observation records" but are not included in the hazard entries, thus forming a stable and usable active identification result delivery.

[0085] The aforementioned proactive identification method for special equipment hazards based on multi-source operation and maintenance data fusion addresses the problem of consistency constraint failure caused by frequent switching of water level characteristics with operating condition boundaries during boiler ignition, load ramping, load reduction, and start-up / shutdown. The core solution lies in first identifying the operating condition and then evaluating water level hazards within that condition. Specifically, this method no longer assumes a uniform correlation between water level and characteristics such as feedwater and evaporation load throughout the entire time period. Instead, it divides the operation process into several operating condition intervals based on the physical stages of the boiler's operation, ensuring the water level response mechanism remains relatively stable within each interval. Furthermore, it establishes a judgment benchmark between water level and relevant operation and maintenance characteristics only within the same operating condition interval. Water level fluctuations during cross-operating condition switching phases are not directly judged as hazards but are treated as transitional responses and managed separately. This effectively avoids significant fluctuations in judgment results within adjacent time windows, improving the accuracy and stability of water level hazard identification results.

[0086] The following describes the special equipment hidden danger proactive identification device based on multi-source operation and maintenance data fusion provided by the present invention. The special equipment hidden danger proactive identification device based on multi-source operation and maintenance data fusion described below can be referred to in correspondence with the special equipment hidden danger proactive identification method based on multi-source operation and maintenance data fusion described above.

[0087] like Figure 2As shown in one embodiment, a special equipment hidden danger active identification device based on multi-source operation and maintenance data fusion includes a data preprocessing module, an operating condition interval division module, a candidate hidden danger screening module, and a hidden danger verification output module.

[0088] The data preprocessing module is used to acquire multi-source operation and maintenance data during boiler operation and to preprocess the multi-source operation and maintenance data to construct a multi-source time series dataset.

[0089] The operating condition interval division module is used to identify the boiler operating stage based on the multi-source time series dataset and the boiler operating status, so as to divide the multi-source time series dataset into multiple operating condition intervals.

[0090] The candidate hazard screening module is used to screen candidate statuses of boiler water level hazards in each operating condition range based on the correlation judgment benchmark between boiler water level and multi-source operation and maintenance data in each operating condition range.

[0091] The hidden danger verification output module is used to verify the correlation between the candidate states of water level hidden dangers and the operating condition range, so as to eliminate the candidate states of water level hidden dangers under the transition of operating conditions and output the final state of water level hidden dangers.

[0092] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0093] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for proactive identification of potential hazards in special equipment based on multi-source operation and maintenance data fusion, characterized in that, The method includes: Acquire multi-source operation and maintenance data during boiler operation, and preprocess the multi-source operation and maintenance data to construct a multi-source time series dataset; Based on the multi-source time series dataset and boiler operating status, the boiler operating stage is identified to divide the multi-source time series dataset into multiple operating condition intervals. Based on the correlation judgment criteria between boiler water level and multi-source operation and maintenance data in each operating condition interval, candidate states of potential water level hazards in boiler water level in each operating condition interval are screened. The correlation between the candidate states of water level hazards and the operating condition intervals is verified to exclude candidate states of water level hazards under the transition of operating conditions, and the final state of water level hazards is output.

2. The method for proactive identification of potential hazards in special equipment based on multi-source operation and maintenance data fusion according to claim 1, characterized in that, The process of acquiring multi-source operation and maintenance data during boiler operation and preprocessing the multi-source operation and maintenance data to construct a multi-source time series dataset includes: The system acquires water level status data, load status data, and operation and maintenance data during boiler operation, and marks the water level status data, load status data, and operation and maintenance data with equipment identifiers and timestamps, and integrates them to obtain the multi-source operation and maintenance data. A standard time axis is constructed by setting a unified time step, and various types of operation and maintenance data in the multi-source operation and maintenance data are mapped and linearly interpolated on the standard time axis to obtain a multi-source data sequence with a unified time axis.

3. The method for proactive identification of potential hazards in special equipment based on multi-source operation and maintenance data fusion according to claim 2, characterized in that, The process of acquiring multi-source operation and maintenance data during boiler operation and preprocessing the multi-source operation and maintenance data to construct a multi-source time series dataset further includes: Key operational quantities are extracted from the multi-source data sequence, and derived features of the key operational quantities are calculated. The key operational quantities are boiler water level and steam flow at different time points, and the derived features are water level change rate and boiler load change intensity. Using the standard time axis as an index, the bit change rate and boiler load change intensity at each time point are correlated with the operation and maintenance data to construct the multi-source time series dataset.

4. The method for proactive identification of potential hazards in special equipment based on multi-source operation and maintenance data fusion according to claim 3, characterized in that, The step of identifying boiler operating stages based on the multi-source time series dataset and boiler operating status, thereby dividing the multi-source time series dataset into multiple operating condition intervals, includes: Based on the multi-source time series dataset, the rate of change of boiler drum pressure during boiler operation is calculated according to the set sampling period. The rate of change of boiler drum pressure is the rate of change of boiler drum pressure in adjacent sampling periods. Based on the steam drum pressure change rate and derived characteristics, a comprehensive operating condition change intensity index is defined, and the boiler operation stage is identified based on the comprehensive operating condition change intensity index and the boiler load change intensity to obtain the preliminary judgment result of the operation stage. If the duration of any operating phase in the preliminary judgment of the operating phase is not less than the preset minimum duration constraint, the corresponding adjacent operating phases will be merged into the same operating condition interval; otherwise, they will be merged into the operating condition transition phase.

5. The method for proactive identification of potential hazards in special equipment based on multi-source operation and maintenance data fusion according to claim 1, characterized in that, The method of screening candidate states of potential water level hazards in each operating condition range based on the correlation between boiler water level and multi-source operation and maintenance data within each operating condition range includes: According to the set sampling period, multiple sets of operation and maintenance data variables corresponding to each working condition interval are extracted from the multi-source time series dataset, and the operation and maintenance data variables are normalized within the corresponding working condition interval to construct a mechanism variable set sequence. Based on the sequence of mechanistic variables, the expected increment of the boiler water level response is established within the operating range using a physical conservation framework, and the expected boiler water level is determined based on the expected increment.

6. The method for proactive identification of potential hazards in special equipment based on multi-source operation and maintenance data fusion according to claim 5, characterized in that, The method of screening candidate states of potential water level hazards in each operating condition range based on the correlation judgment criteria between boiler water level and multi-source operation and maintenance data in each operating condition range also includes: Obtain the measured rate of change of boiler water level and the expected rate of change of boiler water level during different sampling periods, and calculate the deviation of the measured rate of change of boiler water level from the expected rate of change of boiler water level. The water level deviation within the operating range is defined based on the water level change rate deviation, and a deviation persistence count is generated based on the water level deviation. Within the operating range, a deviation threshold and a minimum persistence threshold are set. When the water level deviation exceeds the deviation threshold and the deviation persistence count is not lower than the minimum persistence threshold, a candidate state for water level hazard is determined.

7. The method for proactive identification of potential hazards in special equipment based on multi-source operation and maintenance data fusion according to claim 6, characterized in that, The process of verifying the correlation between the candidate states of water level hazards and the operating condition intervals to exclude candidate states of water level hazards under transitional operating conditions and outputting the final state of water level hazards includes: Obtain the boundary time between different operating conditions, and construct a boundary window based on the boundary time, so as to calculate the transition weight corresponding to the operating time when any running time is within the boundary window; The running time within the boundary window and the transition weight of the running time are marked as transition response events, and the water level deviation corresponding to the transition response events is suppressed by a set suppression coefficient.

8. The method for proactive identification of potential hazards in special equipment based on multi-source operation and maintenance data fusion according to claim 7, characterized in that, The step of verifying the correlation between the candidate states of water level hazards and the operating condition intervals to exclude candidate states of water level hazards under transitional operating conditions and outputting the final state of water level hazards also includes: The effective water level deviation after the transition response event is suppressed is continuously verified according to the set time window length, and when the effective water level deviation reaches the hidden danger judgment rule, the transition response event is marked as a water level hidden danger state. The water level hazard status is standardized and classified according to the preset hazard intensity classification threshold, so as to output the classified water level hazard status.

9. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1-8.

Images