Fire-fighting installation project fault prediction and life evaluation method and system

Abstract

The invention relates to the technical field of fire-fighting installation engineering, in particular to a fire-fighting installation engineering fault prediction and service life evaluation method and system.The fire-fighting installation engineering fault prediction and service life evaluation method comprises the steps that fire pump pressure flow and power signals are collected to construct a load delay sequence, and a pipeline strain recession sequence is formed by combining weld joint strain and current data; comparing pressure-stabilized power heat flow differences to extract an energy flux dissipation offset sequence, associating multi-source signals to calculate synchronism indexes to generate a residual life curve, constructing a full-period recession map according to curve inflection points, and finally obtaining an equipment life evaluation result. According to the invention, energy load and structure degradation mapping is established through multi-source signal coupling analysis, adaptive correlation and trend fitting are realized, energy flux dissipation and synchronization attenuation laws are revealed, a continuous and traceable life trajectory is constructed, energy loss and material response are correlated, potential life turning symptoms are captured in advance, and evaluation stage and predictive force are enhanced. And the operation reliability and the health management precision are improved.

Description

Technical Field

[0001] This invention relates to the field of fire protection installation engineering technology, and in particular to a method and system for fire protection installation engineering fault prediction and life assessment. Background Technology

[0002] The field of fire protection installation engineering technology encompasses the design, installation, maintenance, and operational testing of fire protection facilities within buildings. The core of this technology lies in ensuring that the fire protection system possesses reliable fire alarm and protection capabilities throughout its service life. This involves the piping layout of automatic sprinkler systems, the wiring installation of automatic fire alarm systems, the matching configuration of fire water sources and supply equipment, and the testing and commissioning of terminal devices. Furthermore, this technology also covers construction quality inspection, equipment operation monitoring, and maintenance. Through comprehensive management of pipe connections, electrical wiring, equipment linkage testing, and compliance with safety regulations, a complete engineering implementation and technical support system is formed, thereby supporting the stable operation of the building's fire safety system.

[0003] Among them, the method for fault prediction and life assessment of fire protection installation engineering refers to the method for analyzing and assessing the performance degradation trend of key components in a completed or operating fire protection installation system by collecting operational data such as pipeline pressure, water flow, valve opening and closing status, alarm device response time, and power consumption characteristics, combined with equipment operating cycle and usage environment conditions. It typically uses sensors to collect operating parameters, establishes component health assessment models through time series comparison, threshold judgment, and comparison with historical operating records, and calculates the time nodes and remaining service life of each component based on statistical laws, thereby realizing the quantitative assessment and fault prediction of the internal equipment status of the fire protection installation project, and providing a basis for the formulation and updating of subsequent maintenance cycles.

[0004] Existing technologies for fire protection system condition assessment mostly rely on single-dimensional signal detection and statistical regularity analysis, lacking dynamic interaction and correlation between multi-source data. This results in a lack of coupling mapping between changes in operational data and material degradation behavior, failing to capture energy transfer anomalies and structural strain hysteresis characteristics during start-up and shutdown cycles. This easily leads to delays in degradation identification and ambiguity in lifespan trend differentiation. When the system experiences frequent load fluctuations or voltage imbalances, the assessment model's ability to synchronously respond to energy attenuation and structural degradation is insufficient, making it difficult to form a complete lifespan trend evolution chain. This results in biased maintenance timing judgments and untimely response to fault risks. Summary of the Invention

[0005] To address the technical problems existing in the prior art, embodiments of the present invention provide a method and system for fault prediction and life assessment of fire protection installation engineering. The technical solution is as follows: Methods for predicting failures and assessing lifespan in fire protection installation projects include the following steps: S1: Acquire pressure, flow and power signals when the pump unit starts and stops, align and compare the response according to the time sequence, identify time difference and hysteresis characteristics to construct load records, analyze the continuity and fluctuation segments of load signals in the continuous start and stop cycle and form a chain to obtain the load hysteresis change sequence. S2: Based on the load fluctuation segment in the load hysteresis change sequence, adjust the strain and current signals at the anti-epidemic pipeline weld, identify the response difference time and analyze the direction, reflect the material degradation trend according to the frequency of the repeating segment, screen the continuous segment, and generate the pipeline strain decay change sequence. S3: Based on the strain decay time segment in the pipeline strain decay change sequence, locate the operating stage of the pressure stabilizing equipment, retrieve inlet and outlet power and heat flow data, compare the signal trends and identify segments with inconsistent change directions, correlate offset segments to extract energy difference segments, and obtain the energy flux dissipation offset sequence. S4: Based on the time interval of energy shift in the energy flux dissipation shift sequence, compare the monitoring component signals to determine the trend, construct the relationship between load response and energy change, calculate the proportion of multi-signal unidirectional fluctuation as a synchronization index, analyze the continuous downward trend, and generate the equipment residual life change curve.

[0006] As a further aspect of the present invention, the load hysteresis change sequence includes pressure and flow response time difference, power response hysteresis coefficient, and load combination signal fluctuation interval identifier; the pipeline strain decay change sequence specifically includes strain current response difference amplitude, difference change monotonicity duration, and weld material degradation frequency characteristic value; the energy flux dissipation offset sequence specifically refers to the input and output power deviation value of the voltage regulator, the correspondence between heat flow change and power difference, and the energy transmission asynchronous period marker; the equipment residual life change curve includes the consistency ratio of multiple signal fluctuation directions, the attenuation rate of operating synchronization index, and the signal response mode difference curve.

[0007] As a further aspect of the present invention, the step of obtaining the load hysteresis change sequence is as follows: S101: Acquire continuous signal data from pressure sensors, flow sensors, and motor power measurement devices during the start-up and shutdown phases of the fire pump unit; arrange each signal according to timestamps; compare the pressure, flow, and power values ​​at the same time point; identify the response time of flow before and after pressure signal changes; calculate the delay interval of motor power relative to pressure changes; and generate the start-up and shutdown response time difference interval. S102: Based on the start-stop response time difference interval, call the time series data of pressure, flow and power signals, align them according to time sequence, calculate the corresponding offset values ​​of the three types of signals at adjacent time points, connect the offset values ​​in time sequence to form a continuous combination sequence, reflect the correlation between signals, and obtain the load coupling offset sequence. S103: Based on the load coupling offset sequence, calculate the amplitude change rate of pressure, flow and power signals within a continuous time period, identify time segments with consistent change rate directions, integrate adjacent segments according to time sequence, construct a trend chain reflecting signal hysteresis characteristics, and obtain the load hysteresis change sequence.

[0008] As a further aspect of the present invention, the step of obtaining the pipeline strain decay change sequence is as follows: S201: Based on the load fluctuation segment in the load hysteresis change sequence, retrieve the strain signal and current density signal data of the fire pipeline weld at the corresponding time period, align the two sets of signals according to the time axis, calculate the difference between the strain value and the current density value at each time point, extract the time position where the difference exceeds the set deviation threshold, and generate the strain-current difference interval. S202: Call the strain current difference interval, identify the direction and duration of the difference change within the continuous start-stop cycle, calculate the duration of the direction value in different time periods, determine the time segments in which the direction remains consistent, merge adjacent segments with the same direction in chronological order, and obtain a sequence of consistent direction duration. S203: Based on the consistent direction duration sequence, count the number of times the same direction segment appears in each start-stop cycle, calculate the recurrence frequency and arrange the time periods with continuously increasing frequency, integrate the time periods into a continuous trend line, establish a time series reflecting the strain decay trend of the weld, and obtain the pipeline strain decay change sequence.

[0009] As a further aspect of the present invention, the step of obtaining the energy flux dissipation offset sequence is as follows: S301: Based on the time segments in the pipeline strain decay change sequence where the strain decay rate increases or the fluctuations are concentrated, retrieve the operating signal data of the pressure stabilizing equipment within the same time period, compare the start and end times of each segment with the time marker of the equipment's operating cycle, calculate the time difference between the segment start point and the cycle stage marker, determine the correspondence between strain fluctuations and operating stages, and obtain the corresponding interval of the operating stage. S302: Based on the corresponding interval of the operation stage, call the inlet and outlet power signals and heat flow change data of the voltage regulator, calculate the difference curve between the input power value and the output power value, compare the change value of the heat flow change amplitude with the power difference curve, extract the time period where the input and output power change directions are inconsistent, establish the time sorted offset segment sequence, and generate the energy offset segment interval sequence. S303: Based on the energy offset segment interval sequence, calculate the time interval between adjacent offset segments, filter sequences with continuous intervals and maintained difference directions, merge them into a continuous energy difference segment set in chronological order, accumulate the time lag between energy output and input within the set, establish an energy difference chain for continuous operation, and obtain the energy flux dissipation offset sequence.

[0010] As a further aspect of the present invention, the step of obtaining the residual lifespan variation curve of the equipment is as follows: S401: Based on the energy offset time segment in the energy flux dissipation offset sequence, determine its position in the overall operating time axis, retrieve the pressure signal, flow signal and strain signal recorded by the fire network monitoring component and the pressure stabilizing pump control component during the same time period, synchronize and align them with the energy offset segment according to the timestamp, calculate the time offset and trend change rate of each signal, establish the correspondence between signal and energy change on the time axis, and obtain the energy load corresponding time chain; S402: Based on the time chain corresponding to the energy load, select the continuous data segment of the pressure signal and flow signal, calculate the ratio of the change amplitude between the pressure value and the flow value at each time point, construct the load change curve, compare the fluctuation direction of the strain signal and the power signal in the same time segment, count the number of time points with the same fluctuation direction and the proportion of the total count, establish a numerical index reflecting the consistency of operation, and obtain the synchronization ratio index. S403: Based on the synchronicity ratio index, identify the time period during which the index continues to decline, calculate the rate of change within the declining segment and draw a time trend sequence, compare the direction of change of each segment within the continuous operating cycle, filter the interval where the signal response difference expands, extract the attenuation trend of synchronicity changes within the cycle, establish a time curve reflecting the decline relationship of equipment operating status, and generate the equipment residual life change curve.

[0011] As a further aspect of the present invention, the method further includes: S5: Based on the time intervals in the equipment residual life change curve where the life change rate decreases significantly or an inflection point appears, extract life change characteristics, combine the fire pump unit start-up and shutdown and the operating cycle information of the pressure stabilizing equipment to divide the life change trend, track the decline and stagnation segments of each stage and identify the inflection points, form a decay law map according to the time sequence, and obtain the equipment life assessment result. The specific equipment life assessment results include life trend inflection points, full-cycle decline pattern graphs, and operation phase failure probability mapping tables.

[0012] As a further aspect of the present invention, the step of obtaining the equipment life assessment result is as follows: S501: Based on the time segment in the residual life change curve of the equipment where the life change rate decreases or an inflection point occurs, extract the corresponding time segment data, calculate the life value change rate within the segment and identify the rate decrease interval, establish the life step value sequence of each segment, and determine the affiliation of each life change segment in the operating cycle according to the time axis position of the fire pump unit start-up and shutdown and the operating cycle signal of the pressure stabilizing equipment, thereby obtaining the life stage positioning interval. S502: Based on the life stage positioning interval, calculate the average rate of decline and duration of life value in each operating stage, arrange the declining stage and the stable stage in chronological order, compare the ratio of decline magnitude to dwell time, extract the difference interval between the steady state segment and the decay segment, establish a persistent distribution sequence of life change trend, and obtain the stage trend ratio sequence. S503: Based on the aforementioned stage trend ratio sequence, identify the trend reversal positions and continuous decline segments in the lifespan curve, calculate the time interval between the turning points of each segment and arrange them sequentially to form a lifespan change time chain covering the entire operating cycle. Merge sequences with consistent trend decline directions in the time chain into continuous regular lines, establish an equipment lifespan decline pattern map, and obtain the equipment lifespan assessment results.

[0013] As a further aspect of the present invention, the process of calculating the time offset and trend change rate of each signal is specifically as follows: The difference between the start time of each energy offset time segment in the energy flux dissipation offset sequence and the timestamp of the pressure signal is obtained. The obtained difference is used as the time offset reference value. The offset interval of the flow signal and strain signal relative to the pressure signal is calculated. The numerical changes of each signal within the offset interval are continuously differentiated to obtain the ratio of the change amplitude of adjacent time moments. The obtained ratio is defined as the trend change rate. The process of establishing the correspondence between signal and energy changes on the time axis is as follows: Within the period of continuous and stable trend change rate, the operating time corresponding to the energy offset time interval is determined. Based on the time offset reference value, the pressure signal, flow signal and strain signal are arranged on a unified time axis and sequentially associated according to the energy fluctuation direction. The consistency determination result of the trend change rate of each signal within the continuous time interval is recorded to form a correspondence table containing time offset, trend change rate and energy offset direction. The process of calculating the proportion in a consistent direction is as follows: In the load change curve, the change direction marks of the pressure signal, flow signal, strain signal and power signal are extracted respectively. The consistency of the four types of signal direction marks at the same time point is compared. The ratio of the number of consistent marks to the total number of marks is calculated. The ratio is used as the operation synchronization index value. When the index value is lower than the set synchronization threshold, it is marked as a continuous decline in the synchronization index. The rate of change of the continuously declining segment of the synchronicity index is calculated as follows: The difference between the synchronicity index values ​​in adjacent time intervals is divided by the interval length to obtain the synchronicity decay rate. A continuous time trend curve is plotted according to the change order of the decay rate over time to establish a time curve reflecting the degradation relationship of the equipment's operating status, thereby generating the equipment's residual life change curve.

[0014] A fire protection installation engineering fault prediction and life assessment system, the system comprising: The signal coupling analysis module acquires pressure, flow and power signals when the pump unit starts and stops, aligns and compares the responses according to the time sequence, identifies time difference and hysteresis characteristics to construct load records, analyzes the continuity and fluctuation segments of load signals in continuous start-stop cycles and forms a chain to obtain the load hysteresis change sequence. The weld degradation identification module, based on the load fluctuation segment in the load hysteresis change sequence, adjusts the strain and current signals at the weld of the anti-corrosion pipeline, identifies the response difference time and analyzes the direction, reflects the material degradation trend according to the frequency of the repeating segment, filters continuous segments, and generates the pipeline strain decay change sequence. The energy difference correlation extraction module locates the operating stage of the pressure stabilizing equipment based on the strain decay time segment in the pipeline strain decay change sequence, retrieves inlet and outlet power and heat flow data, compares the signal trend and identifies segments with inconsistent change directions, correlates the offset segments to extract energy difference segments, and obtains the energy flux dissipation offset sequence. The multi-parameter synchronization module compares the monitoring component signals to trend based on the time segment of energy shift in the energy flux dissipation offset sequence, constructs the relationship between load response and energy change, calculates the proportion of multi-signal unidirectional fluctuation as a synchronization index, analyzes the continuous downward trend, and generates the equipment residual life change curve. The lifespan trend analysis module extracts lifespan change characteristics based on the time segments in the equipment's residual lifespan change curve where the lifespan change rate significantly decreases or inflection points appear. It then combines the start-up and shutdown information of the fire pump unit and the operating cycle information of the pressure stabilizing equipment to divide the lifespan change trend, tracks the decline and stagnation segments in each stage and identifies the inflection points. By correlating these segments in time sequence, it forms a decay pattern map and obtains the equipment lifespan assessment results.

[0015] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: In this invention, by coupling analysis of pressure, flow, power, and strain signals, a mapping relationship between energy transfer, load response, and structural degradation is established on the time axis. This enables adaptive correlation identification and dynamic trend fitting among multiple signals, obtaining the overall law of energy flux dissipation offset and synchronous decay, forming a continuous and traceable lifespan change trajectory. Energy loss and material response correspond to each other in the operating state, allowing for the capture of lifespan turning points before trend fluctuations, enhancing the ability to identify potential degradation risks in advance, and enabling lifespan assessment to have phased, coherent, and trend predictive attributes. This improves the operational reliability and health management accuracy of the fire protection system during cyclic operation. Attached Figure Description

[0016] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a flowchart illustrating the process of obtaining the load hysteresis change sequence according to the present invention. Figure 3 This is a flowchart illustrating the process of obtaining the pipeline strain decay change sequence according to the present invention. Figure 4 This is a flowchart illustrating the process of obtaining the energy flux dissipation offset sequence of the present invention. Figure 5 This is a flowchart illustrating the process of obtaining the residual life variation curve of the device according to the present invention. Figure 6 This is a flowchart illustrating the process of obtaining the equipment life assessment results of this invention. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0019] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0020] Please see Figure 1 This invention provides a technical solution: a method for fault prediction and life assessment of fire protection installation engineering, comprising the following steps: S1: Acquire continuous signal data recorded by pressure sensors, flow sensors, and motor power measurement equipment during the start-up and shutdown phases of the fire pump unit. Arrange the signals according to timestamps and align them in time sequence. Compare the pressure response, flow changes, and power fluctuations at the same time. Identify the time difference of flow response and the lag characteristics of power response before and after pressure changes. Combine them in time sequence to form a load combination signal reflecting the operating status. Integrate the time difference and lag data of pressure, flow, and power to construct a continuous operating load change record. Identify the continuity and fluctuation segments of load combination signal changes in the continuous start-up and shutdown cycle of the fire pump unit. Form a signal trend chain characterizing the time delay characteristics to obtain the load delay change sequence. S2: Based on the load fluctuation segment in the load hysteresis change sequence, retrieve the strain signal data and current density signal data of the fire pipeline weld within the load fluctuation segment and match them on the time axis to identify the time segment where the strain and current response differ. Analyze the direction and duration of the difference within the continuous start-stop cycle, aggregate the time segments where the difference direction remains consistent, and use the frequency of recurrence of the time segments to reflect the degradation trend of the weld metal material under different pressure fluctuations. Screen the time segments where the change trend continues to extend and integrate them to form a time series, generating a pipeline strain decay change sequence. S3: Based on the time segments in the pipeline strain decay change sequence where the strain decay rate increases significantly or fluctuations are concentrated, time correspondence is performed with the operating signal data of the pressure stabilizing equipment. The position of each time segment in the equipment's operating cycle is identified by comparison, the corresponding operating stage is located, the inlet and outlet power signals and heat flow change data of the pressure stabilizing equipment are retrieved, the change trends of input and output power signals are compared, the correspondence between heat flow change and power difference is analyzed, time segments where the change directions of input and output signals are inconsistent are identified, the offset segments are associated in chronological order, continuous energy difference segments are extracted and archived, and segments where the energy output change delay is not synchronized with the input change are connected in series to form a continuous offset information chain of the complete operating stage, thus obtaining the energy flux dissipation offset sequence. S4: Based on the time segment of energy offset in the energy flux dissipation offset sequence, determine the corresponding time axis position. Retrieve pressure, flow and strain signals recorded by the fire pipeline monitoring component and the pressure stabilizing pump control component in the same time period, and perform time alignment and trend comparison with the energy flux offset time segment. Construct load change curve with pressure and flow signals, establish the correspondence between load response and energy change, characterize the synchronization characteristics of energy change and load change on a unified time line, compare the fluctuation direction of pressure, flow, strain and power signals in the same segment, calculate the proportion that maintains the same direction, use the proportion as the operation synchronization index, identify the time segment of continuous decline of synchronization index, analyze the change trend of synchronization index and signal response difference in continuous operation cycle, extract the time law of synchronization decay, and generate the equipment residual life change curve. S5: Based on the time intervals in the equipment residual life change curve where the life change rate significantly decreases or inflection points appear, extract the corresponding life change characteristics. Combined with the start-up and shutdown information of the fire pump unit and the operating cycle information of the pressure stabilizing equipment, divide the life change trends of different operating stages, track the decline process of the life trend in each stage one by one, analyze the time intervals of the trend decline and the relatively stable periods, identify the turning points of the life trend according to the time sequence of the operating cycle, and mark the turning points in sequence on the timeline. Correlate the life change trends of each stage in time sequence to form an equipment life degradation pattern map covering the entire operating cycle. Use the degradation pattern to distinguish the correspondence between the operating stages and the occurrence of potential failures to obtain the equipment life assessment results.

[0021] After the above-mentioned lifespan change characteristics are identified, a lifespan prediction model is established based on the lifespan decay patterns and multi-cycle signal data of each operating stage. The feature sequences of pressure, flow, power, strain and energy flux are used as training input samples. The network structure is used to adaptively learn the temporal dependencies between features. The lifespan trend is fitted after multi-feature fusion through model parameter optimization. The real-time signals of subsequent operating stages are used for inference calculations to output the equipment lifespan evolution trend and potential failure probability, thereby realizing intelligent prediction and lifespan forecasting of the operating health status of the fire protection installation system.

[0022] The load hysteresis change sequence includes pressure and flow response time difference, power response hysteresis coefficient, and load combination signal fluctuation range identifier. The pipeline strain decay change sequence specifically includes strain current response difference amplitude, difference change monotonicity duration, and weld material degradation frequency characteristic value. The energy flux dissipation offset sequence specifically refers to the input and output power deviation of the voltage regulator, the correspondence between heat flow change and power difference, and the energy transmission asynchronous period marker. The equipment residual life change curve includes the consistency ratio of multiple signal fluctuation directions, the attenuation rate of operating synchronicity index, and the signal response mode difference curve. The equipment life assessment results specifically include life trend turning point, full-cycle decay pattern map, and operation phase failure probability mapping table.

[0023] Please see Figure 2 The steps for obtaining the load hysteresis change sequence are as follows: S101: Acquire continuous signal data from pressure sensors, flow sensors, and motor power measurement devices during the start-up and shutdown phases of the fire pump unit; arrange each signal according to timestamps; compare the pressure, flow, and power values ​​at the same time point; identify the response time of flow before and after pressure signal changes; calculate the delay interval of motor power relative to pressure changes; and generate the start-up and shutdown response time difference interval. Continuous signal data from pressure sensors, flow sensors, and motor power measurement devices are acquired during the start-up and shutdown phases of the fire pump unit. An industrial-grade high-frequency pressure transmitter (range 0-2.5MPa, accuracy 0.5%) is used to collect pipeline pressure, and an electromagnetic flow meter (range 0-100L / s) is used to collect main pipeline flow. The 110kW rated power motor in the fire pump unit is used as the monitoring object; a three-phase power analyzer is connected to the motor terminal, and the sampling frequency is set to 1000Hz to obtain high-precision discrete-time series data. A multi-channel data acquisition card (DAQ) converts the analog voltage signal read from the pressure sensor into a pressure value sequence in megapascals (MPa) and the pulse signal fed back from the flow sensor into liters per second (L / s). The system collects a sequence of flow rate values ​​and an instantaneous active power (kW) value sequence of the motor using a power analyzer. A unified nanosecond-level time axis is established, and the three dimensions of the numerical sequence are mapped onto this time axis. A timestamp alignment operation is performed. The aligned time axis is traversed, and the time point marked as the start-up state is selected as the starting anchor point. A pressure threshold parameter is set to 5% of the rated working pressure, i.e., 0.05 MPa, as the pressure signal start-up criterion. A flow rate threshold parameter is set to 2% of the rated flow rate, i.e., 1.5 L / s, as the flow response criterion. A power threshold parameter is set to 110% of the no-load power, i.e., 5.5 kW, as the power response criterion. The time point on the time axis where the pressure value first exceeds the pressure threshold parameter is recorded as... Continue searching along the time axis until the time point when the flow rate first exceeds the flow rate threshold parameter is recorded as... Calculate the time difference between two time points. Similarly, the search point is the time when the power value first exceeds the power threshold parameter. And calculate the lag time relative to the pressure start-up time: If the calculated It is 0.8 seconds and If the delay is 0.3 seconds, the flow response is determined to lag behind the pressure change, and the power response is also delayed but faster than the flow. The closed interval from 0.3 seconds to 0.8 seconds is defined as the delay interval. For the stop phase, reverse logic search is performed, and the stop point is set when the pressure drops below 0.05 MPa. The corresponding delay time is calculated, and the delay interval of the start phase is merged with the delay interval of the stop phase to construct a numerical range including upper and lower limits, generating the start and stop response time difference interval.

[0024] S102: Based on the start-stop response time difference interval, call the time series data of pressure, flow and power signals, align them according to time sequence, calculate the corresponding offset values ​​of the three types of signals at adjacent time points, connect the offset values ​​in time sequence to form a continuous combination sequence, reflect the correlation between signals, and obtain the load coupling offset sequence. Based on the start-stop response time difference interval, the start-stop response time difference interval obtained in the previous steps is extracted by calling the time series data of pressure, flow and power signals. The pressure signal sequence is used as a reference sequence while keeping the time axis unchanged. The flow signal sequence is shifted left by a time axis shift step of 800 sampling points (based on a 1000Hz sampling rate), corresponding to the upper limit of the interval (0.8 seconds). The power signal sequence is also shifted left by a time axis shift step of 300 sampling points (corresponding to the lower limit of the interval (0.3 seconds)). This achieves physical time alignment. To eliminate numerical differences caused by different physical units (MPa, L / s, kW), a normalization process is introduced. Historical data from the past 100 operating cycles are selected to calculate the historical maximum values ​​of pressure, flow rate, and power. , , and minimum value , , Using the range transformation formula Map the currently aligned three types of signal values ​​to Within the dimensionless interval, for each aligned sampling time... Read the normalized pressure value Flow value after translation and normalization and the power value after translation and normalization Perform vector subtraction operation; Calculate the dimensionless deviation between pressure and flow rate. Calculate the dimensionless deviation between pressure and power. ,Will and As dual-channel feature data, according to time The calculated deviation values ​​are arranged in ascending order to form a shape. A two-dimensional matrix sequence, where The total number of sampling points represents the matrix sequence, which directly characterizes the instantaneous coupling degree between physical quantities after eliminating inherent mechanical hysteresis. If at a certain moment... The value is 0.15 and A value of 0.05 indicates that the coupling deviation between flow rate and pressure is relatively large at this moment, while the power following performance is relatively good. This two-dimensional matrix sequence is used as the input feature tensor of the deep neural network model to obtain the load coupling offset sequence.

[0025] S103: Based on the load coupling offset sequence, calculate the rate of change of the amplitude of pressure, flow and power signals in a continuous period, identify time segments with consistent rate of change direction, integrate adjacent segments according to time sequence, construct a trend chain reflecting signal hysteresis characteristics, and obtain the load hysteresis change sequence. Based on the load coupling offset sequence, the rate of change of the amplitude of pressure, flow, and power signals within a continuous time period is calculated. A time window length of 50 milliseconds is set, and the rate of change of the pressure signal is calculated using the first-order difference method within this window. Similarly, calculate the rate of change of flow. With power change rate Introducing symbolic functions Extracting the directional features of the rate of change, if Marked as +1 (increase), if Marked as -1 (decreasing), if Marked as 0 (keep); Constructing triplet feature vectors Traverse the entire time series and compare adjacent time points. and eigenvectors ,like and If they are completely equal, it is determined that the direction of change is consistent, that is, the three physical quantities change in coordination in the same direction. A continuous scan program is then initiated to count the longest continuous time segment in which the characteristic vector remains unchanged. For example, from the 2.5-second interval to the 3.1-second interval, the direction of the rate of change of the three quantities always remains the same. If all trends are in an upward trend, then the 0.6-second time span is treated as an independent trend unit. The start time, end time, and common direction of change of this unit are recorded. If the direction of change of any signal is reversed (e.g., from +1 to -1), the current cooperative state is determined to be disrupted. The current trend unit is immediately truncated and a new unit is started. For trend units with a duration of less than 0.1 seconds, they are treated as transient noise and are removed. The remaining valid trend units are linked in chronological order to form a chain structure describing the cooperative motion state of physical quantities. The mean of the load coupling offset sequence segment in each trend unit is extracted as the strength attribute of the unit. Finally, structured sequence data containing trend direction, duration, and coupling strength attributes is output for subsequent temporal pattern recognition of deep learning models to obtain the load hysteresis change sequence.

[0026] Please see Figure 3 The steps for obtaining the pipeline strain decay change sequence are as follows: S201: Based on the load fluctuation segment in the load hysteresis change sequence, retrieve the strain signal and current density signal data of the fire pipeline weld at the corresponding time period, align the two sets of signals according to the time axis, calculate the difference between the strain value and the current density value at each time point, extract the time position where the difference exceeds the set deviation threshold, and generate the strain-current difference interval. Based on the load fluctuation segment in the load hysteresis change sequence, the time window in which pressure and flow rate violently couple and oscillate is locked. The starting point of this window is set to time zero to construct a relative time axis. The micro-strain value sequence is read at a frequency of 1000Hz from a metal foil resistance strain gauge (sensitivity coefficient 2.0) installed at the critical weld of the fire pipeline. At the same time, the current density value sequence is read from an electrochemical corrosion monitoring probe (using linear polarized resistance technology) installed adjacent to the weld. Max-min normalization is performed, and the maximum strain value within the past 50 complete operating cycles is selected. and minimum value and the maximum current density and minimum value As a benchmark, using the formula The measured strain signal and current density signal were respectively mapped to... Dimensionless interval, for each normalized sampling point ; Perform subtraction operation Calculate the instantaneous deviation between the two and set the deviation threshold parameter. This parameter is set to 3 times the average deviation between the two under historical stable operating conditions. If the average historical deviation is 0.05, then it is set to... The result of traversal calculation Sequence, filter out all that satisfy Abnormal time points, such as consecutive deviation values ​​of 0.16, 0.18, and up to 0.22 detected within the relative time range of 0.52 seconds to 0.88 seconds, are merged into a closed interval. Furthermore, extremely short gaps with fewer than 5 sampling points within the interval are filled and connected to capture the hysteresis phenomenon of stress concentration and corrosion current response mismatch in welds under alternating loads, thereby generating strain current difference intervals.

[0027] S202: Call the strain current difference interval, identify the direction and duration of the difference change within the continuous start-stop cycle, calculate the duration of the direction value in different time periods, determine the time segment in which the direction is consistent, merge adjacent segments with the same direction in chronological order, and obtain the sequence of consistent direction duration. Call the strain current difference interval and extract the normalized difference value sequence within the interval. The first derivative of the sequence is calculated using the five-point central difference method. To characterize the instantaneous rate and direction of the difference in change, if This is then judged as positive divergence, meaning the asynchrony between strain and corrosion current is intensifying. This is considered negative convergence, meaning the two tend to synchronize. A duration threshold of 20 milliseconds is set. The difference sequence is scanned, and the number of consecutive sampling points with unchanged statistical signs is counted. Multiplying this number by the sampling period of 1 millisecond yields the hold duration. For example, within the interval... Within that timeframe, the period from 0.52s to 0.72s was identified. Since the value is always greater than 0, the duration of this positive divergence state is calculated to be 0.2 seconds, ranging from 0.72s to 0.88s. The value is always less than 0, and the duration is 0.16 seconds. For brief reverse fluctuations with an interval of less than 10 milliseconds, they are treated as measurement noise and forcibly merged into the segment of the preceding and following dominant directions. A binary sequence consisting of a direction identifier and the corresponding duration value is constructed, such as... This process is repeated for all the difference intervals in sequence, splicing the scattered local features in the time dimension to form a continuous description reflecting the physical field coupling instability characteristics during the weld damage evolution process, resulting in a consistent direction maintenance duration sequence.

[0028] S203: Based on the consistent direction duration sequence, count the number of times the same direction section appears in each start-stop cycle, calculate the recurrence frequency and arrange the time periods with continuously increasing frequency, integrate the time periods into a continuous trend line, establish a time series reflecting the strain decay trend of the weld, and obtain the pipeline strain decay change sequence. Based on the consistent direction of the duration sequence, the monitoring dimension is expanded to multiple start-stop cycles throughout the entire lifecycle. The statistical batch size is set to 10 start-stop cycles. The frequency of occurrence of the specific feature "positive divergence duration exceeding 0.15 seconds" within each batch is extracted. The proportion of this feature in the total number of features in the current batch is calculated as the recurrence frequency. For example, this feature appears twice in cycles 1-10 (frequency 0.2), while it appears eight times in cycles 91-100 (frequency increases to 0.8). Linear regression analysis is performed to analyze the slope of the frequency value as the batch number increases. Periods with a slope greater than 0.05 are selected, indicating that the strain-corrosion coupling performance of the weld is deteriorating rapidly. These periods of continuous frequency increase are used as key degradation windows. The center frequency value and growth rate corresponding to each window are extracted. These discrete degradation indicators are connected into a continuous trend curve according to the cycle number. Exponential smoothing is used to denoise the curve to highlight the long-term degradation pattern, constructing a curve with the shape of... A one-dimensional time series, where To assess the total number of batches, the monotonically increasing nature of this sequence directly quantifies the trajectory of the pipeline weld evolving from an initial healthy state to a fatigue failure state. As the time dimension input to the prediction model, this yields the pipeline strain decay sequence.

[0029] Please see Figure 4 The steps for obtaining the flux dissipation offset sequence are as follows: S301: Based on the time segments in the pipeline strain decay change sequence where the strain decay rate increases or fluctuations are concentrated, retrieve the operating signal data of the pressure stabilizing equipment within the same time period, compare the start and end times of each segment with the time marker of the equipment's operating cycle, calculate the time difference between the segment start point and the cycle stage marker, determine the correspondence between strain fluctuations and operating stages, and obtain the corresponding interval of the operating stage. Based on the time segments in the pipeline strain decay sequence where the strain decay rate increases or fluctuations are concentrated, the strain change rate threshold parameter is set to 0.05 microstrain per millisecond (µs). (This involves) traversing the entire sequence to filter out all discrete segments whose absolute rate of change exceeds a certain threshold, defining these segments as load anomaly fluctuation windows, for example, extracting relative time intervals. to The window retrieves the operating status code data of the pressure stabilizing pump and pressure tank within the specified time period via the PLC control system interface. The operating cycle of the pressure stabilizing equipment is divided into three discrete states: pressurization start-up, pressure maintenance, and pressure relief shutdown, marked as status codes 1, 2, and 3, respectively. The data is then retrieved from the system. When the status code of the device is 1, indicating the pressurization start-up phase, extract the start time marker of this pressurization phase. Perform subtraction operation Calculate the time delay of strain fluctuation relative to the start point of the operation phase, and obtain... This indicates that the pipeline strain anomaly occurred 0.2 seconds after the pressurization action began. This alignment and calculation process was repeated for all selected fluctuation windows to construct a system containing the fluctuation window ID, corresponding running status code, and relative delay time. The triplet dataset is mapped onto a continuous time axis to form a set of feature vectors describing the temporal correlation between pipeline physical deformation and equipment operating conditions. This serves as the input basis for subsequent causal analysis, yielding the corresponding intervals for the operating phase.

[0030] S302: Based on the corresponding interval of the operation phase, call the inlet and outlet power signals and heat flow change data of the voltage regulator, calculate the difference curve between the input power value and the output power value, compare the change value of the heat flow change with the power difference curve, extract the time period when the input and output power change directions are inconsistent, establish the time-sorted offset segment sequence, and generate the energy offset segment interval sequence. Based on the corresponding interval of the operation phase, the inlet and outlet power signals and heat flow change data of the voltage regulator are retrieved, and the input active power sequence of the voltage regulator pump motor is collected through a high-frequency power analyzer. Using the outlet pressure sensor value (Pa) and the flow sensor value (Pa) The product of the fluid output power sequence is used to calculate the fluid output power sequence. (Unit: W), perform point-by-point subtraction. Acquire the power loss difference curve and simultaneously read the heat flux sequence fed back by the thin-film heat flow meter attached to the pump casing surface. (unit: The sliding window algorithm was used to calculate the standardized variation of both, with a window width of 50 sampling points. The variation within the window was calculated separately. and slope and Introduce direction consistency determination logic, if If power loss increases significantly but heat flow does not increase synchronously (e.g., mechanical blockage causes electrical energy to be converted into internal energy rather than surface heat flow), or if power loss decreases but heat flow continues to accumulate, then it is judged as an abnormal energy conversion. For example in Calculate at any time and Since the two signs are opposite, this moment is identified as belonging to the energy dissipation offset state. All time points that meet the condition of inconsistent directions are connected into a continuous time period, and the start point, end point, and other parameters of each time period are recorded. and The cumulative deviation integral value is used to arrange these abnormal energy conversion segments in chronological order, generating an energy offset segment interval sequence.

[0031] S303: Based on the energy offset segment interval sequence, calculate the time interval between adjacent offset segments, filter sequences with continuous intervals and maintained difference direction, merge them into a continuous energy difference segment set in time order, accumulate the time lag between energy output and input within the set, establish the energy difference chain of continuous operation stage, and obtain the energy flux dissipation offset sequence. Based on the energy offset segment interval sequence, calculate the time interval between adjacent offset segments, and iterate through the sequence to the th interval. End time of each section With the Start time of each segment Calculate the interval value The continuity threshold is set to 100 milliseconds. If the energy deviation direction indicators of the two segments are the same, i.e., both are of the "electrical energy input converted into non-thermal energy dissipation" type, then a merge operation is performed to combine the two segments into one extended energy difference segment, and the start time of the merged segment is updated. The end time is For each merged set, calculate the phase lag of the energy output waveform relative to the input waveform within the set. The total energy flux dissipation lag index for this continuous operation phase is obtained by summing the phase lags of all sampling points within the set. For example, if a merged segment lasts for 0.5 seconds and has a cumulative lag of 12.5 radians, this indicator is used as a key feature to measure the energy transmission efficiency degradation in the pipeline-equipment coupling system. According to the chronological order of occurrence, the individual or merged energy difference segments and their corresponding cumulative lag indicators are linked to construct a multi-dimensional time series feature matrix. Each row of this matrix represents an independent dissipation event, including the time of occurrence, duration, and dissipation intensity. This is directly used as the input tensor for deep neural networks to predict the health status of the pipeline network, resulting in an energy flux dissipation offset sequence.

[0032] Please see Figure 5 The steps for obtaining the equipment residual life change curve are as follows: S401: Based on the energy offset time segment in the energy flux dissipation offset sequence, determine its position in the overall operating time axis, retrieve the pressure signal, flow signal and strain signal recorded by the fire network monitoring component and the pressure stabilizing pump control component during the same period, synchronize and align them with the energy offset segment according to the timestamp, calculate the time offset and trend change rate of each signal, establish the correspondence between signal and energy change on the time axis, and obtain the energy load corresponding time chain; The specific process for calculating the time offset and trend change rate of each signal is as follows: The difference between the start time of each energy offset time segment in the energy flux dissipation offset sequence and the timestamp of the pressure signal is obtained. The difference is used as the time offset reference value. The offset intervals of the flow signal and strain signal relative to the pressure signal are calculated. The numerical changes of each signal within the offset interval are continuously differentiated to obtain the ratio of the change amplitudes of adjacent times. The obtained ratio is defined as the trend change rate. The process of establishing the correspondence between signal and energy changes on the time axis is as follows: Within a period of continuous and stable trend change rate, the operating time corresponding to the energy offset time interval is determined. Based on the time offset reference value, the pressure signal, flow signal, and strain signal are arranged on a unified time axis and sequentially correlated according to the energy fluctuation direction. The consistency determination results of the trend change rate of each signal within the continuous time interval are recorded, forming a correspondence table containing time offset, trend change rate, and energy offset direction. Based on the energy offset time segments in the energy flux dissipation offset sequence, traverse all discrete time windows marked as energy transfer anomalies in the sequence and extract the start timestamp of each window. With end timestamp Using this as the index key, numerical sequences of fire-fighting pipeline network pressure sensors within the same time span are retrieved in parallel from the distributed time-series database. Flow sensor numerical sequence and the strain count sequence at the pipe weld. Perform time base alignment operation and calculate the timestamp of the pressure signal. With energy shift start time The difference This difference is used as the inherent response delay benchmark of the system. Phase shift correction is applied to the flow and strain signals using this benchmark to ensure strict synchronization of multi-source signals at the moment of physical event occurrence. First-order difference operations are performed on each corrected signal sequence, with a sliding calculation step size of one sampling period. The ratio of numerical increments between adjacent time points is calculated, for example, for pressure signals. Similarly, the rate of change of flow can be derived. With strain rate of change Construct a shape as The trend feature matrix, where Given the number of sampling points within the time window, each vector element in the matrix is ​​associated with the offset direction identifier (positive dissipation or negative accumulation) of the energy offset segment, generating a structured dataset containing precise time offset, multidimensional signal trend change rate, and energy state label. This dataset serves as the input feature layer for a deep neural network model to capture physical field coupling anomalies, thus obtaining the time chain corresponding to the energy load.

[0033] S402: Based on the time chain corresponding to the energy load, select the continuous data segment of the pressure signal and flow signal, calculate the ratio of the change amplitude between the pressure value and the flow value at each time point, construct the load change curve, compare the fluctuation direction of the strain signal and the power signal in the same time segment, count the number of time points with the same fluctuation direction and the proportion of the total count, establish a numerical index reflecting the consistency of operation, and obtain the synchronization ratio index. The process of calculating the proportions that maintain a consistent direction is as follows: In the load change curve, the change direction marks of pressure signal, flow signal, strain signal and power signal are extracted respectively. The consistency of the four types of signal direction marks at the same time point is compared. The ratio of the number of consistent marks to the total number of marks is calculated. This ratio is used as the operation synchronization index value. When the index value is lower than the set synchronization threshold, it is marked as the synchronization index continuously decreasing segment. Based on the time chain corresponding to the energy load, continuous data segments of pressure and flow signals are selected, and the feature matrix is ​​sliced. For each aligned sampling time point... Extraction pressure variation range With the magnitude of flow change Perform division operation By quantifying the instantaneous fluctuations of fluid impedance characteristics, a load change curve reflecting the dynamic characteristics of the pipeline network load is constructed. A multi-physics collaborative evaluation mechanism is introduced to extract pressure, flow rate, strain, and synchronously acquired motor power signals at the same time. Direction of change sign , , , The direction sign of the pressure signal is set as the reference vector. A step-by-step XOR comparison is performed, and the number of signals whose direction sign matches the reference is counted. Calculate the ratio of its value to the total number of signals (i.e., 4). For example, if the pressure increases by 1 at a certain moment, and the flow rate and strain increase by 1 but the power decreases by 1, then the consistency quantity is 3, the synchronization ratio index at that moment is calculated to be 0.75, and a synchronization judgment threshold parameter is set. The threshold is 0.8. This threshold is set based on the average synchronization rate statistics of 1000 start-stop cycles under factory health conditions. The calculation results are iterated through all time periods, and all... The continuous time intervals are marked as asynchronous operation segments. These segments represent the discretization and degradation of the electromechanical system response, forming time series data that describes the degree of coordinated degradation of the electromechanical-hydraulic multidimensional system, and obtaining the synchronization ratio index.

[0034] S403: Based on the synchronicity ratio index, identify the period of continuous decline of the index, calculate the rate of change within the decline segment and draw a time trend sequence, compare the direction of change of each segment within the continuous operating cycle, screen the interval of widening signal response difference, extract the decay trend of synchronicity change within the cycle, establish a time curve reflecting the decline relationship of equipment operating status, and generate the equipment residual life change curve. The rate of change of the synchronicity index in the continuously decreasing segment is calculated as follows: The difference between the synchronicity index values ​​in adjacent time intervals is divided by the interval length to obtain the synchronicity decay rate. Based on the change of decay rate over time, a continuous time trend curve is plotted to establish a time curve reflecting the degradation relationship of equipment operating status, so as to generate the equipment residual life change curve. Based on the synchronicity ratio index, the time period of continuous decline in the index is identified. A sliding window algorithm is applied to scan the synchronicity ratio sequence, with a window length of 50 sampling points, and the linear regression slope of the index values ​​within the window is calculated. Filter out Furthermore, intervals lasting longer than 0.5 seconds are defined as signal response difference amplification intervals. For each defined descent interval, the synchronization index value at its starting moment is calculated. index value at the end time The difference, divided by the duration of the interval. The synchronous decay rate of this interval is obtained. For example, if a certain interval lasts for 2 seconds and the index decreases from 0.75 to 0.65, then the decay rate is... According to the progression of the operating cycle, the decay rate values ​​of each decreasing interval are connected into a monotonically increasing cumulative damage curve. The exponential smoothing method is used to filter out random fluctuation noise, and a trend equation reflecting the nonlinear decline of the equipment's health status over the service time is fitted. This trend sequence is used as the target predictor variable (Label) of the Long Short-Term Memory Network (LSTM) to train the Remaining Useful Life (RUL) prediction model, quantifying the remaining time span from the current state of the equipment to complete failure, and obtaining the equipment's residual life change curve.

[0035] Please see Figure 6 The steps for obtaining equipment life assessment results are as follows: S501: Based on the time segment in the equipment residual life change curve where the life change rate decreases or an inflection point occurs, extract the corresponding time segment data, calculate the life value change rate within the segment and identify the rate decrease interval, establish the life step value sequence for each segment, and determine the affiliation of each life change segment in the operating cycle based on the time axis position of the fire pump unit start-up and shutdown and the operating cycle signal of the pressure stabilizing equipment, thereby obtaining the life stage positioning interval. Based on the time intervals in the equipment residual life change curve where the rate of life change decreases or inflection points occur, data from that curve is retrieved. The rate of change calculation window is set to 10 operating cycles, and a second-order difference algorithm is used to calculate the acceleration of the life value within each window. If three consecutive windows (i.e., accelerated decay), then this segment is marked as a rapid decay period. A change from a negative value to a positive value or close to zero is marked as an inflection point. For example, if an acceleration of -0.08 is detected continuously within the 200th to 250th cycle, this is extracted as a target time segment. For each extracted segment, the difference between the predicted lifetime value at its start and end times is calculated and divided by the time span of the segment to obtain the average decay rate. If the absolute value of this rate is greater than a set benchmark value (e.g., 0.1% per hour), it is determined to be a period of significant rate decline. A lifetime ladder feature vector containing the start and end points of this period and the decay rate is constructed. At the same time, the start and stop logs of the fire pump unit and the pressure stabilizing pump in the SCADA system are loaded. The pressure maintenance cycle records are unified to the Unix standard time format. A timeline matching algorithm is executed to check whether the center time point of each lifespan interval falls within the time range of a pump start-up event (status code 1) or a pressure stabilization operation event (status code 2). For example, if it is confirmed that the accelerated decline in the 230th cycle occurred during a high-load start-up process lasting 45 minutes, the lifespan decline segment is bound to the corresponding operating condition ID to generate a lifespan segment index table containing operating condition context information. This table serves as the basis for subsequent fault attribution analysis and the lifespan stage location interval is obtained.

[0036] S502: Based on the life stage positioning interval, calculate the average rate of decline and duration of life values ​​in each operating stage, arrange the declining stage and the stable stage in chronological order, compare the ratio of decline magnitude to dwell time, extract the difference interval between the steady state stage and the decay stage, establish a persistent distribution sequence of life change trend, and obtain the stage trend ratio sequence. Based on the positioning intervals of the lifespan stages, the average rate of decline and duration of the lifespan value for each operational stage are calculated. For each positioning interval, the corresponding lifespan value sequence is read. Linear regression was used to fit the descending slope within this interval. Simultaneously record the duration of this interval. Set the steady-state determination threshold as ,like If the slope is gentler, it is marked as a steady-state segment; otherwise, it is marked as a decaying segment. All marked segments are reordered according to their chronological order to form alternating "steady-state-decaying" sequences. For each pair of adjacent "decaying-steady-state segments," the magnitude of the decline in the decaying segment is calculated. Duration of the steady-state phase ratio For example, if the lifetime value decreases by 5% during a certain decline phase, and the subsequent steady-state phase lasts for 20 hours, then... This ratio reflects the device's ability to recover after experiencing a damaging impact or the rate of disease deterioration. It is calculated by traversing the entire sequence. Value, if a certain consecutive three stages If the value shows an increasing trend (0.25->0.35->0.5), it indicates that the device's self-recovery capability is weakening. These difference features and their corresponding time-series positions are encoded into vectors to construct a distribution sequence describing the dynamic characteristics of lifespan evolution, thus obtaining the stage trend ratio sequence.

[0037] S503: Based on the phase trend ratio sequence, identify the trend reversal position and continuous decline section in the life curve, calculate the time interval between the turning points of each section and arrange them in sequence to form a life change time chain covering the entire operating cycle. Merge the sequences with the same trend decline direction in the time chain into a continuous regular line, establish the equipment life decline pattern map, and obtain the equipment life assessment result. Based on the phase trend ratio sequence, identify the trend reversal points and continuous decline segments in the life curve, and perform a first derivative scan on the trend ratio sequence to find... The index position where the value undergoes a sign change (from positive to negative or vice versa) is defined as the trend reversal point, for example, at the 500th running hour. When a mutation changes from a gradual decrease to a sharp increase, this point is marked as a potential failure mutation point, and the sequence is simultaneously searched. Subsequences with values ​​continuously greater than 0.5 and a duration exceeding 5 stages are identified as irreversible continuous decay segments, and the time difference between two adjacent reversal points or the starting point of the decay segment is calculated. Construct a time interval sequence, if The exponential shortening over time (e.g., 100h->50h->10h) verifies the cumulative effect of the fault. All identified monotonic decay segments are pieced together along the time axis, and transient recovery fluctuations are removed to fit a smooth exponential decay curve. The radius of curvature, asymptote position, and current tangent slope of this curve together constitute a topological map of the equipment's health status. After identifying the lifespan change characteristics, a lifespan prediction model is established using deep learning algorithms based on the lifespan decay patterns of each operating stage and multi-cycle signal data. Pressure, flow rate, power, strain, and energy flux characteristic sequences are used as training data. The system trains on input samples, uses the network structure to adaptively learn the temporal dependencies between features, and optimizes the model parameters to achieve lifespan trend fitting after multi-feature fusion. It then performs inference calculations on real-time signals in subsequent operation phases, outputting the equipment lifespan evolution trend and potential failure probability. This enables intelligent prediction and lifespan forecasting of the operational health status of the fire protection installation system. The system inputs the map parameters and prediction model output results into a pre-trained deep belief network (DBN), which outputs the current lifespan stage of the equipment (such as early wear, stable operation, and wear and tear failure period) and the corresponding failure probability distribution, thus obtaining the equipment lifespan assessment results.

[0038] Fire protection installation engineering fault prediction and life assessment system, the system includes: The signal coupling analysis module acquires pressure, flow and power signals when the pump unit starts and stops, aligns and compares the responses according to the time sequence, identifies time difference and hysteresis characteristics to construct load records, analyzes the continuity and fluctuation segments of load signals in continuous start-stop cycles and forms a chain to obtain the load hysteresis change sequence. The weld degradation identification module, based on the load fluctuation segment in the load hysteresis change sequence, modulates the strain and current signals at the weld of the anti-epidemic pipeline, identifies the response difference time and analyzes the direction, reflects the material degradation trend according to the frequency of the repeating segment, filters continuous segments, and generates the pipeline strain decay change sequence. The energy difference correlation extraction module locates the operating stage of the pressure stabilizing equipment based on the strain decay time segment in the pipeline strain decay change sequence, retrieves inlet and outlet power and heat flow data, compares the signal trends and identifies segments with inconsistent change directions, and extracts energy difference segments from the correlation offset segments to obtain the energy flux dissipation offset sequence. The multi-parameter synchronization module compares the monitoring component signals based on the time segment of energy shift in the energy flux dissipation offset sequence, constructs the relationship between load response and energy change, calculates the proportion of multi-signal unidirectional fluctuation as a synchronization index, analyzes the continuous downward trend, and generates the equipment residual life change curve. The lifespan trend analysis module extracts lifespan change characteristics based on the time intervals in the equipment's residual lifespan change curve where the lifespan change rate significantly decreases or inflection points appear. It then combines the start-up and shutdown information of the fire pump unit and the operating cycle information of the pressure stabilizing equipment to divide the lifespan change trend, tracks the decline and stagnation segments in each stage and identifies the turning points. By correlating these in time sequence, it forms a decay pattern map and obtains the equipment lifespan assessment results.

[0039] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for fault prediction and life assessment of fire protection installation engineering, characterized in that, Includes the following steps: S1: Acquire pressure, flow and power signals when the pump unit starts and stops, align and compare the response according to the time sequence, identify time difference and hysteresis characteristics to construct load records, analyze the continuity and fluctuation segments of load signals in the continuous start and stop cycle and form a chain to obtain the load hysteresis change sequence. S2: Based on the load fluctuation segment in the load hysteresis change sequence, adjust the strain and current signals at the anti-epidemic pipeline weld, identify the response difference time and analyze the direction, reflect the material degradation trend according to the frequency of the repeating segment, screen the continuous segment, and generate the pipeline strain decay change sequence. S3: Based on the strain decay time segment in the pipeline strain decay change sequence, locate the operating stage of the pressure stabilizing equipment, retrieve inlet and outlet power and heat flow data, compare the signal trends and identify segments with inconsistent change directions, correlate offset segments to extract energy difference segments, and obtain the energy flux dissipation offset sequence. S4: Based on the time interval of energy shift in the energy flux dissipation shift sequence, compare the monitoring component signals to determine the trend, construct the relationship between load response and energy change, calculate the proportion of multi-signal unidirectional fluctuation as a synchronization index, analyze the continuous downward trend, and generate the equipment residual life change curve.

2. The method for fault prediction and life assessment of fire protection installation engineering according to claim 1, characterized in that: The load hysteresis change sequence includes pressure and flow response time difference, power response hysteresis coefficient, and load combination signal fluctuation interval identifier. The pipeline strain decay change sequence specifically includes strain current response difference amplitude, difference change monotonicity duration, and weld material degradation frequency characteristic value. The energy flux dissipation offset sequence specifically refers to the input and output power deviation of the voltage regulator, the correspondence between heat flow change and power difference, and the energy transmission asynchronous period marker. The equipment residual life change curve includes the consistency ratio of multiple signal fluctuation directions, the attenuation rate of operating synchronization index, and the signal response mode difference curve.

3. The method for fault prediction and life assessment of fire protection installation engineering according to claim 1, characterized in that: The steps for obtaining the load hysteresis change sequence are as follows: S101: Acquire continuous signal data from pressure sensors, flow sensors, and motor power measurement devices during the start-up and shutdown phases of the fire pump unit; arrange each signal according to timestamps; compare the pressure, flow, and power values ​​at the same time point; identify the response time of flow before and after pressure signal changes; calculate the delay interval of motor power relative to pressure changes; and generate the start-up and shutdown response time difference interval. S102: Based on the start-stop response time difference interval, call the time series data of pressure, flow and power signals, align them according to time sequence, calculate the corresponding offset values ​​of the three types of signals at adjacent time points, connect the offset values ​​in time sequence to form a continuous combination sequence, reflect the correlation between signals, and obtain the load coupling offset sequence. S103: Based on the load coupling offset sequence, calculate the amplitude change rate of pressure, flow and power signals within a continuous time period, identify time segments with consistent change rate directions, integrate adjacent segments according to time sequence, construct a trend chain reflecting signal hysteresis characteristics, and obtain the load hysteresis change sequence.

4. The method for fault prediction and life assessment of fire protection installation engineering according to claim 1, characterized in that: The steps for obtaining the pipeline strain decay change sequence are as follows: S201: Based on the load fluctuation segment in the load hysteresis change sequence, retrieve the strain signal and current density signal data of the fire pipeline weld at the corresponding time period, align the two sets of signals according to the time axis, calculate the difference between the strain value and the current density value at each time point, extract the time position where the difference exceeds the set deviation threshold, and generate the strain-current difference interval. S202: Call the strain current difference interval, identify the direction and duration of the difference change within the continuous start-stop cycle, calculate the duration of the direction value in different time periods, determine the time segments in which the direction remains consistent, merge adjacent segments with the same direction in chronological order, and obtain a sequence of consistent direction duration. S203: Based on the consistent direction duration sequence, count the number of times the same direction segment appears in each start-stop cycle, calculate the recurrence frequency and arrange the time periods with continuously increasing frequency, integrate the time periods into a continuous trend line, establish a time series reflecting the strain decay trend of the weld, and obtain the pipeline strain decay change sequence.

5. The method for fault prediction and life assessment of fire protection installation engineering according to claim 1, characterized in that: The steps for obtaining the energy flux dissipation offset sequence are as follows: S301: Based on the time segments in the pipeline strain decay change sequence where the strain decay rate increases or the fluctuations are concentrated, retrieve the operating signal data of the pressure stabilizing equipment within the same time period, compare the start and end times of each segment with the time marker of the equipment's operating cycle, calculate the time difference between the segment start point and the cycle stage marker, determine the correspondence between strain fluctuations and operating stages, and obtain the corresponding interval of the operating stage. S302: Based on the corresponding interval of the operation stage, call the inlet and outlet power signals and heat flow change data of the voltage regulator, calculate the difference curve between the input power value and the output power value, compare the change value of the heat flow change amplitude with the power difference curve, extract the time period where the input and output power change directions are inconsistent, establish the time sorted offset segment sequence, and generate the energy offset segment interval sequence. S303: Based on the energy offset segment interval sequence, calculate the time interval between adjacent offset segments, filter sequences with continuous intervals and maintained difference directions, merge them into a continuous energy difference segment set in chronological order, accumulate the time lag between energy output and input within the set, establish an energy difference chain for continuous operation, and obtain the energy flux dissipation offset sequence.

6. The method for fault prediction and life assessment of fire protection installation engineering according to claim 1, characterized in that: The steps for obtaining the equipment residual life variation curve are as follows: S401: Based on the energy offset time segment in the energy flux dissipation offset sequence, determine its position in the overall operating time axis, retrieve the pressure signal, flow signal and strain signal recorded by the fire network monitoring component and the pressure stabilizing pump control component during the same time period, synchronize and align them with the energy offset segment according to the timestamp, calculate the time offset and trend change rate of each signal, establish the correspondence between signal and energy change on the time axis, and obtain the energy load corresponding time chain; S402: Based on the time chain corresponding to the energy load, select the continuous data segment of the pressure signal and flow signal, calculate the ratio of the change amplitude between the pressure value and the flow value at each time point, construct the load change curve, compare the fluctuation direction of the strain signal and the power signal in the same time segment, count the number of time points with the same fluctuation direction and the proportion of the total count, establish a numerical index reflecting the consistency of operation, and obtain the synchronization ratio index. S403: Based on the synchronicity ratio index, identify the time period during which the index continues to decline, calculate the rate of change within the declining segment and draw a time trend sequence, compare the direction of change of each segment within the continuous operating cycle, filter the interval where the signal response difference expands, extract the attenuation trend of synchronicity changes within the cycle, establish a time curve reflecting the decline relationship of equipment operating status, and generate the equipment residual life change curve.

7. The method for fault prediction and life assessment of fire protection installation engineering according to claim 1, characterized in that: The method further includes: S5: Based on the time intervals in the equipment residual life change curve where the life change rate decreases significantly or an inflection point appears, extract life change characteristics, combine the fire pump unit start-up and shutdown and the operating cycle information of the pressure stabilizing equipment to divide the life change trend, track the decline and stagnation segments of each stage and identify the inflection points, form a decay law map according to the time sequence, and obtain the equipment life assessment result. The specific equipment life assessment results include life trend inflection points, full-cycle decline pattern graphs, and operation phase failure probability mapping tables.

8. The method for fault prediction and life assessment of fire protection installation engineering according to claim 7, characterized in that: The steps for obtaining the equipment life assessment results are as follows: S501: Based on the time segment in the residual life change curve of the equipment where the life change rate decreases or an inflection point occurs, extract the corresponding time segment data, calculate the life value change rate within the segment and identify the rate decrease interval, establish the life step value sequence of each segment, and determine the affiliation of each life change segment in the operating cycle according to the time axis position of the fire pump unit start-up and shutdown and the operating cycle signal of the pressure stabilizing equipment, thereby obtaining the life stage positioning interval. S502: Based on the life stage positioning interval, calculate the average rate of decline and duration of life value in each operating stage, arrange the declining stage and the stable stage in chronological order, compare the ratio of decline magnitude to dwell time, extract the difference interval between the steady state segment and the decay segment, establish a persistent distribution sequence of life change trend, and obtain the stage trend ratio sequence. S503: Based on the aforementioned stage trend ratio sequence, identify the trend reversal positions and continuous decline segments in the lifespan curve, calculate the time interval between the turning points of each segment and arrange them sequentially to form a lifespan change time chain covering the entire operating cycle. Merge sequences with consistent trend decline directions in the time chain into continuous regular lines, establish an equipment lifespan decline pattern map, and obtain the equipment lifespan assessment results.

9. The method for fault prediction and life assessment of fire protection installation engineering according to claim 6, characterized in that: The process of calculating the time offset and trend change rate of each signal is as follows: The difference between the start time of each energy offset time segment in the energy flux dissipation offset sequence and the timestamp of the pressure signal is obtained. The obtained difference is used as the time offset reference value. The offset interval of the flow signal and strain signal relative to the pressure signal is calculated. The numerical changes of each signal within the offset interval are continuously differentiated to obtain the ratio of the change amplitude of adjacent time moments. The obtained ratio is defined as the trend change rate. The process of establishing the correspondence between signal and energy changes on the time axis is as follows: Within the period of continuous and stable trend change rate, the operating time corresponding to the energy offset time interval is determined. Based on the time offset reference value, the pressure signal, flow signal and strain signal are arranged on a unified time axis and sequentially associated according to the energy fluctuation direction. The consistency determination result of the trend change rate of each signal within the continuous time interval is recorded to form a correspondence table containing time offset, trend change rate and energy offset direction. The process of calculating the proportion in a consistent direction is as follows: In the load change curve, the change direction marks of the pressure signal, flow signal, strain signal and power signal are extracted respectively. The consistency of the four types of signal direction marks at the same time point is compared. The ratio of the number of consistent marks to the total number of marks is calculated. The ratio is used as the operation synchronization index value. When the index value is lower than the set synchronization threshold, it is marked as a continuous decline in the synchronization index. The rate of change of the continuously declining segment of the synchronicity index is calculated as follows: The difference between the synchronicity index values ​​in adjacent time intervals is divided by the interval length to obtain the synchronicity decay rate. A continuous time trend curve is plotted according to the change order of the decay rate over time to establish a time curve reflecting the degradation relationship of the equipment's operating status, thereby generating the equipment's residual life change curve.

10. A fire protection installation engineering fault prediction and life assessment system, characterized in that the system is used to execute the fire protection installation engineering fault prediction and life assessment method according to any one of claims 1-9, the system comprising: The signal coupling analysis module acquires pressure, flow and power signals when the pump unit starts and stops, aligns and compares the responses according to the time sequence, identifies time difference and hysteresis characteristics to construct load records, analyzes the continuity and fluctuation segments of load signals in continuous start-stop cycles and forms a chain to obtain the load hysteresis change sequence. The weld degradation identification module, based on the load fluctuation segment in the load hysteresis change sequence, adjusts the strain and current signals at the weld of the anti-epidemic pipeline, identifies the response difference time and analyzes the direction, reflects the material degradation trend according to the frequency of the repeating segment, filters continuous segments, and generates the pipeline strain decay change sequence. The energy difference correlation extraction module locates the operating stage of the pressure stabilizing equipment based on the strain decay time segment in the pipeline strain decay change sequence, retrieves inlet and outlet power and heat flow data, compares the signal trend and identifies segments with inconsistent change directions, correlates the offset segments to extract energy difference segments, and obtains the energy flux dissipation offset sequence. The multi-parameter synchronization module compares the monitoring component signals to trend based on the time segment of energy shift in the energy flux dissipation offset sequence, constructs the relationship between load response and energy change, calculates the proportion of multi-signal unidirectional fluctuation as a synchronization index, analyzes the continuous downward trend, and generates the equipment residual life change curve. The lifespan trend analysis module extracts lifespan change characteristics based on the time segments in the equipment's residual lifespan change curve where the lifespan change rate significantly decreases or inflection points appear. It then combines the start-up and shutdown information of the fire pump unit and the operating cycle information of the pressure stabilizing equipment to divide the lifespan change trend, tracks the decline and stagnation segments in each stage and identifies the turning points. By correlating these segments in time sequence, it forms a decay pattern map and obtains the equipment lifespan assessment results.

Images