Sewage pump intelligent maintenance method based on fault prediction

Abstract

The invention discloses an intelligent sewage pump maintenance method based on fault prediction, and relates to the technical field of sewage pump management, and the method comprises the following steps: 1, collecting a state jump signal appearing in the operation process of a sewage pump control system; marking each type of state jump signals as state jump events, and establishing a state jump event sequence according to an actual occurrence time sequence; 2, for each state jump event identified in the step 1, setting a response window with a fixed time length, collecting a transient response process of a plurality of operation parameters in the window, extracting behavior characteristic indexes reflecting the stability and the recovery capability of the sewage pump system, and forming a structured response vector set; according to the method, the response atlas model with the state jump event as the core is constructed, behavior characteristic indexes in various running states are bound to the atlas nodes in a structured mode, and the system stability performance and evolution path characteristics are fused while the running event time sequence relation is reserved.

Description

Technical Field

[0001] This invention relates to the field of sewage pump management technology, specifically to an intelligent maintenance method for sewage pumps based on fault prediction. Background Technology

[0002] In wastewater treatment systems, wastewater pumps are one of the key power equipment, and their operating status has a significant impact on the overall hydraulic transport and wastewater treatment efficiency. During operation, wastewater pumps often experience frequent start-stop, flow fluctuations, and increased vibration due to factors such as changes in water quality, mechanical wear, and electrical fluctuations. These changes reflect, to some extent, the changing trends in equipment status and operational stability. Accurately identifying these changes and conducting subsequent analysis is the foundation for achieving efficient operation and maintenance management. Existing wastewater pump monitoring methods generally employ distributed sensor systems to collect parameters such as current, voltage, flow rate, pressure, and temperature in real time, and identify and warn of abnormal states through threshold judgment and trend analysis. These methods are highly adaptable in engineering practice, capable of responding promptly to some sudden anomalies, and are particularly suitable for rapid single-point detection of key monitoring indicators. However, when faced with multi-variable interconnected anomalies and systemic degradation trends in complex operating environments, traditional methods still have certain limitations in terms of the continuity of event identification, the ability to model response behavior, and the foresight of fault warnings. With the development of intelligent sensing and graph modeling technologies, graph-based behavioral modeling and cross-cycle path analysis have gradually become effective means to improve predictive capabilities. By using state transition events as graph nodes and behavioral features as node attributes to construct response graphs, not only can the temporal relationships between events be preserved, but the stability change patterns of the system can also be extracted from the overall structural evolution, providing a data foundation and modeling support for realizing path-feature-oriented intelligent maintenance. System degradation identification methods based on graph evolution and response offset are becoming a research hotspot in predictive maintenance of equipment. Therefore, this application proposes an intelligent maintenance method for sewage pumps based on fault prediction. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent maintenance method for sewage pumps based on fault prediction, so as to solve the problems mentioned in the background art.

[0004] This invention can be achieved through the following technical solution: an intelligent maintenance method for sewage pumps based on fault prediction, comprising the following steps: Step 1: Collect the status change signals that occur during the operation of the sewage pump control system; Each type of state transition signal is marked as a state transition event, and a sequence of state transition events is established according to the actual time order of their occurrence. Step 2: For each state transition event identified in Step 1, set a response window of fixed duration, collect the transient response process of multiple operating parameters within the window, extract behavioral feature indicators that reflect the stability and recovery capability of the sewage pump system, and form a set of structured response vectors. Step 3: Use the state transition events obtained in Step 1 as graph nodes to construct graph structure connections; bind the behavioral feature indicators extracted in Step 2 to the corresponding graph nodes to form a response graph model, and evaluate the behavioral feature indicators of the graph nodes in multiple dimensions. Nodes that exceed the anomaly judgment threshold are marked as abnormal nodes. Step 4: Statistically analyze the historical operation records of various state transition events in the response graph model, calculate the recovery failure rate, and calculate the minimum risk path cost from the current node to the historical failure node by combining the number of abnormal nodes and the degree of feature degradation. Step 5: Based on the preset degradation judgment conditions, determine whether the system has a trend of stability degradation; Step 6: If stability degradation is determined, a maintenance strategy is generated by combining the offset direction of the map path features and the historical evolution pattern.

[0005] A further technical improvement of the present invention is that: in step two, morphological clustering is performed based on the response curves of similar state transition events within multiple historical periods to form multiple response behavior morphological clusters, and a unique identifier is assigned to each response morphological cluster. After each transition event occurs, multiple operating parameter change curves within its response window are collected to determine the response pattern cluster to which it belongs and to calculate its offset from the cluster center. The type identifier of the transition event, the cluster identifier, the behavior index group, and the intra-cluster offset value are combined to form a structured response vector, which is then used as a feature of the subsequent graph nodes.

[0006] A further technical improvement of the present invention is that the step of obtaining the index value includes: In step two, for operating parameters that have an overshoot peak within the response window and form a monotonic decline segment, the overshoot peak in its response curve is identified as the starting point, and the time point when it first enters the steady-state tolerance range is determined as the ending point. The time point corresponding to the overshoot peak and the time point when it first enters the steady-state tolerance range are defined as the "stable decline region". Based on the response curves corresponding to the stable fallback regions in similar state transition events within historical cycles, a "standard fallback behavior envelope" is constructed. The overlap ratio and envelope penetration rate between the response curves corresponding to the current stable fallback region and the envelope are calculated as "stable trend consistency indicators". Meanwhile, the trend reversal point and response inflection point in the response curve of the operating parameters are extracted, and the time length and magnitude change value between the first trend reversal and the entry into steady state are calculated to obtain the "response inertia offset index". The stable trend consistency index and the response inertia offset index are written as structured indicators into the behavioral feature index set of the corresponding state transition event nodes in the response graph model.

[0007] A further technical improvement of the present invention is that, in step four, the evaluation process for the degree of feature degradation includes: Z1. Construction of Historical Stable Indicator Bands: For each type of state transition event node, the system statistically analyzes its set of behavioral characteristic indicators in historical normal cycles, and then uses the historical average of each behavioral characteristic indicator in the set as a basis. with standard deviation Construct the stability range of each behavioral characteristic indicator. ; Z2, Current Period Indicator Deviation Calculation: Behavioral characteristic indicators of target nodes in the current operating cycle Calculate its deviation from the historical mean. ; Z3. Degradation Level Judgment Rules: If behavioral characteristic indicators If it falls outside the stability range, it is marked as an "abnormal indicator"; If deviation >Preset deviation threshold ( Based on standard deviation If you configure it, it will be marked as "severely degraded"; For each node, count the number of abnormal indicators and their average deviation to determine the characteristic degradation value of that node. .

[0008] A further technical improvement of the present invention is as follows: after the response graph model is constructed, the state transition event paths with completely consistent structures in multiple operating cycles are aggregated to construct a cross-cycle state transition event path cluster, and the behavioral characteristic index set of the state transition event nodes in the cross-cycle state transition event path cluster is subjected to time series analysis in different cycles to determine whether there is a trend of the behavioral characteristic index value deviating from its historical stable statistical range in the path cluster period by period. If the deviation widens in multiple consecutive cycles, it is determined that the path cluster of cross-cycle state transition events has a deterioration trend, and this trend is incorporated into the graph model as an enhanced criterion for the risk of system stability degradation.

[0009] A further technical improvement of the present invention is that the method for obtaining the minimum risk path cost value includes: S1. Based on the response graph model constructed in the current cycle, select the nodes in the graph that are currently active or in operation. It also retrieves the set of nodes that have been marked as "known fault nodes" in all historical cycles. ; S2. For each edge in the response graph model (from node) point to Define its path weights. The risk level of its termination node is calculated as follows: ; In the formula, For nodes The anomaly flag value is set to 1 if the node is identified as an anomaly, and 0 otherwise. For nodes The characteristic degradation level value; For nodes The Boolean value indicating whether the cross-cycle state transition event path cluster has a deterioration trend is set to 1 if it does, otherwise it is set to 0. These are the risk weight coefficients for the corresponding items; S3, Minimum Path Cost Calculation: Based on step S2, the edge Path weight With the current node Starting from the point, use the shortest path search algorithm in the weighted graph to search for all paths leading to the historical failure node. Find the path with the minimum total value, denoted as: In the formula, Indicates starting from the current node To the historical fault node A feasible path; Minimize the total value of the path The risk value of the current operating state is recorded and written into the response graph model for subsequent stability degradation assessment and intelligent maintenance strategy generation.

[0010] A further technical improvement of the present invention is that: in multiple historical operating cycles, state transition event paths with consistent structure are extracted, and their corresponding periodic behavior index matrices are constructed into a set of path behavior trajectories to form a standard path pool; In the current operating cycle, the periodic behavior index matrix corresponding to the path to be evaluated is extracted and matched with the path behavior trajectory in the standard path pool to calculate the degree of trajectory matching deviation. If the current path's behavior trajectory shows a continuous tendency to deviate from the central region of the main trajectory cluster of the standard path pool in multiple consecutive periods, and the direction of deviation is consistent, then the path is determined to have a risk of self-evolutionary feature shift. The trajectory deviation trend score of this path is used as a new indicator dimension and written into the corresponding state transition event path structure in the response graph model. It is also used as an additional weighting factor in the path risk cost calculation formula to enhance the ability to identify the evolutionary degradation trend of the system in advance.

[0011] Compared with the prior art, the present invention has the following beneficial effects: This invention constructs a response graph model centered on state transition events, binding behavioral characteristic indicators of various operating states to graph nodes in a structured manner. While preserving the temporal relationship of operating events, it integrates system stability performance and evolution path characteristics, thereby achieving the ability to identify global degradation trends driven by multi-dimensional data. Compared with traditional monitoring methods based on single points or fixed dimensions, this graph modeling method is more systematic and scalable, and can more comprehensively depict the equipment state evolution process. Furthermore, the "stable fallback region" identification and "response behavior clustering" mechanism proposed in this invention can extract representative dynamic behavior templates across multiple operating cycles, and quickly assign and calculate the offset of new events based on historical templates, thereby constructing a structured response vector driven by a dual factor of "cluster offset-behavioral index," which improves the system's accuracy in identifying abnormal response behaviors. In addition, through the cycle-by-cycle trend analysis of behavioral indicators at the path scale, the tracking of cross-cycle degradation trends is realized, which helps to identify state paths with hidden degradation risks in advance. On the other hand, this invention introduces "minimum risk path cost" as a fault prediction indicator, and combines it with multi-factor modeling such as path structure, abnormal node status, indicator degradation degree, and trajectory deviation trend to effectively construct a quantitative risk mapping between the system's operating status and historical fault status. Through this indicator, a continuous quantitative assessment of the system's current health status can be achieved, and personalized intelligent maintenance strategies can be pushed accordingly, including inspection suggestions, operation strategy adjustment suggestions, and control linkage execution schemes. This enables the transformation from passive response to proactive prediction in intelligent maintenance, improving the safety, reliability, and maintenance economy of equipment operation. Attached Figure Description

[0012] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0013] Figure 1This is a schematic diagram of the method logic of the present invention. Detailed Implementation

[0014] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0015] Example 1: Please refer to Figure 1 As shown, this invention provides an intelligent maintenance method for sewage pumps based on fault prediction, comprising the following steps: Step 1: Trigger Event Recognition Collect state transition signals that occur during the operation of the sewage pump control system. State transition signals include, but are not limited to, significant changes in operating conditions that occur during operation, such as pump start-up, pump shutdown, sudden changes in water level, sudden increases or decreases in flow rate, and valve opening and closing. Specifically, significant changes in operating conditions refer to sudden jumps in the operating state of a sewage pump control system caused by certain key operations or external conditions during operation. These typically manifest as: Key parameters change drastically in a short period of time, for example: The rate at which monitored parameters such as current, voltage, speed, flow rate, liquid level, and pump chamber pressure rise or fall exceeds the preset first speed threshold. The rate of temperature rise exceeds the corresponding second speed threshold, or the temperature cannot return to a steady state within a set time.

[0016] The operating mode changes abruptly, for example: Suddenly restarted from a stopped state; Forced to shut down or restart during continuous operation; The control logic triggers an emergency bypass or switches the pump unit.

[0017] The system stability is disturbed, for example: The response time after a certain startup exceeds the preset delay threshold (e.g., from the usual 5 seconds to more than 8 seconds). The steady-state process, which should have been established quickly, exhibited phenomena such as overshoot, oscillation, or delayed recovery.

[0018] These changes in operating conditions are not characterized by continuity over time, but rather by discrete, abrupt events, which can be identified and captured through state monitoring of the control system, equipment log recording, or abnormal event detection models.

[0019] Each type of state transition signal is marked as a state transition event, and a sequence of state transition events is established according to the actual time order of occurrence, which serves as an event index for subsequent data organization.

[0020] Step 2: Event Response Data Collection and Behavioral Feature Extraction For each state transition event identified in step one, a response window of fixed duration is set, and the transient response process of multiple operating parameters within the response window is collected; Operating parameters include, but are not limited to, current, flow rate, pressure, temperature, and vibration; Based on this transient response process, behavioral characteristic indicators reflecting the stability and resilience of the sewage pump system are extracted. These behavioral characteristic indicators include at least the following: Steady-state settling time: refers to the length of time required for the system operating parameters to gradually transition from an initial disturbance state to a stable operating state after a state transition event (such as pump start-up, valve action, etc.) occurs; Its acquisition method is based on the dynamic change curve of the target parameters (such as flow rate, current, temperature, etc.) during the response process; Set a tolerance range for stable operation (e.g., ±5% of steady-state value); The length of time from the time the event is triggered for the measured parameter to first enter the tolerance zone and remain there for a set time (e.g., 2 seconds).

[0021] Maximum overshoot amplitude: This is the maximum deviation of the target operating parameter from its steady-state value after a state transition event occurs; it is obtained by acquiring stable values ​​(such as steady-state flow rate and steady-state current). Find the maximum or minimum values ​​reached by the parameters during the response process; Maximum overshoot = |Extremum value - Steady-state value|.

[0022] Response recovery rate: refers to the rate of change per unit time of a system parameter falling back from its peak (or trough) after a jump to the steady-state tolerance range; it is obtained by obtaining the time period from the maximum deviation point to the stable range in the response curve; Calculate the average rate of decrease / recovery of parameter amplitude per unit time; Recovery rate = (maximum deviation - steady-state threshold) / recovery time.

[0023] Residual amplitude of response fluctuation: This is the maximum residual fluctuation amplitude of the parameters that still appears within a set time window after the system response has reached a steady state; it is obtained within a few seconds (e.g., 5 seconds) after the system enters the steady-state threshold range. The maximum amplitude of the parameter fluctuation during this period is recorded. That is, half the difference between the maximum and minimum values, or the absolute difference.

[0024] Each state transition event and its corresponding behavioral characteristic indicators are combined to form an event response feature group, which forms a set of structured response vectors. Step 3: Response Map Construction The state transition events obtained in step one are used as graph nodes, and the graph structure connection relationship is constructed according to the actual execution sequence of the events. The behavioral feature indicators extracted in step two are bound to the corresponding graph nodes to form a response graph model containing event nodes, event connection edges, and node feature attributes, which is used to reflect the state evolution path and dynamic stability performance of the sewage pump under different operating cycles. The behavioral characteristic indicators of each state transition event node in the response graph model are evaluated from multiple dimensions and their indicator values ​​are obtained. If the indicator value exceeds the preset anomaly judgment threshold, the node is marked as an abnormal node. The steps to obtain the indicator value include: In step two, for operating parameters that have an overshoot peak within the response window and form a monotonic decline segment, the overshoot peak in its response curve is identified as the starting point, and the time point when it first enters the steady-state tolerance range is determined as the ending point. The time point corresponding to the overshoot peak and the time point when it first enters the steady-state tolerance range are defined as the "stable decline region". Based on the response curves corresponding to the stable fallback regions in similar state transition events within historical cycles, a "standard fallback behavior envelope" is constructed. The overlap ratio and envelope penetration rate between the response curves corresponding to the current stable fallback region and the envelope are calculated as "stable trend consistency indicators". Meanwhile, the trend reversal point and response inflection point in the response curve of the operating parameters are extracted, and the time length and magnitude change value between the first trend reversal and the entry into steady state are calculated to obtain the "response inertia offset index". The stable trend consistency index and the response inertia offset index are written as structured indicators into the behavioral feature index set of the corresponding state transition event nodes in the response graph model.

[0025] Specifically, in step two, for operating parameters (such as current, flow rate, pressure, etc.) that have overshoot peaks and form monotonic decline segments within the response window, the "stable decline region" in their response process is extracted. The stable decline region refers to the time period in the response curve from the time point corresponding to the overshoot peak to the time point when the operating parameter first enters and remains within the steady-state tolerance range. It is used to characterize the dynamic behavior of the system transitioning from an abnormal response state to a steady state. Based on the response curves corresponding to the stable fallback regions of similar state transition events within multiple historical periods, a "standard fallback behavior envelope" is constructed. This envelope represents the typical fallback behavior range of the system under normal stable conditions in history. The system's response curve to the stable fallback region of the current state transition event is compared with the standard fallback behavior envelope, and their intersection overlap ratio and envelope penetration rate are calculated. This is used as an indicator to assess whether the current behavior deviates from the stable trend, and is defined as the "stable trend consistency index". Meanwhile, the system performs derivative analysis on the response curve formed by the time series of response parameters, extracts the trend reversal point and response inflection point, and counts the time length and magnitude change from the first trend reversal point to the final entry into the steady state range. It then compares this with the mean of historical normal samples to obtain an index used to measure the change in system response inertia, which is defined as the "response inertia offset index". The stability trend consistency index and the response inertia offset index constitute a structured response vector, which is used for multi-dimensional evaluation in this embodiment. As part of the set of behavioral feature indicators of the corresponding node in the response graph model for the current state transition event, it is used for subsequent graph modeling, abnormal node identification and stability degradation evaluation.

[0026] Step 4: Recovery Failure Rate Modeling and Path Risk Assessment The historical running records of each type of state transition event in the response graph model are statistically analyzed, and the recovery failure rate of the event type in the past running cycle is calculated. The recovery failure rate refers to the proportion of the event response characteristics that fail to recover to the stable threshold within a preset time window. By combining the number of marked abnormal nodes in the current graph path and their characteristic degradation degree, the minimum risk path cost from the current state node to the historical known fault node is calculated as a quantitative indicator of the current system stability degradation degree. In step four, the assessment process for the degree of feature degradation includes: Z1. Construction of Historical Stable Indicator Bands: For each type of state transition event node, the system statistically analyzes its set of behavioral characteristic indicators in historical normal cycles, and then uses the historical average of each behavioral characteristic indicator in the set as a basis. with standard deviation Construct the stability range of each behavioral characteristic indicator. ; Z2, Current Period Indicator Deviation Calculation: Behavioral characteristic indicators of target nodes in the current operating cycle Calculate its deviation from the historical mean. ; Z3. Degradation Level Judgment Rules: If behavioral characteristic indicators If it falls outside the stability range, it is marked as an "abnormal indicator"; If deviation >Preset deviation threshold ( Based on standard deviation If you configure it, it will be marked as "severely degraded"; For each node, count the number of abnormal indicators and their average deviation to determine the characteristic degradation value of that node. .

[0027] Methods for obtaining the minimum risk path cost include: S1. Based on the response graph model constructed in the current cycle, select the nodes in the graph that are currently active or in operation. It also retrieves the set of nodes that have been marked as "known fault nodes" in all historical cycles. ; S2. For each edge in the response graph model (from node) point to Define its path weights. The risk level of its termination node is calculated as follows: ; In the formula, For nodes The anomaly flag value is set to 1 if the node is identified as an anomaly, and 0 otherwise. For nodes The characteristic degradation level value; These are the risk weight coefficients for the corresponding items; S3, Minimum Path Cost Calculation: Based on step S2, the edge Path weight With the current node Starting from the point, use the shortest path search algorithm in the weighted graph to search for all paths leading to the historical failure node. Find the path with the minimum total value, denoted as: In the formula, Indicates starting from the current node To the historical fault node A feasible path; Minimize the total value of the path The risk value of the current operating state is recorded and written into the response graph model for subsequent stability degradation assessment and intelligent maintenance strategy generation.

[0028] Step 5: Determine operational stability degradation. Determine whether the current operating status of the sewage pump exhibits a trend of stability degradation based on at least one of the following criteria: The response characteristic indicators of a certain type of state transition event continuously deviate from its historical cluster center in multiple consecutive operating cycles; The failure rate of recovery for a certain type of event shows a continuous upward trend within the sliding time window and exceeds the preset statistical threshold. The minimum risk path cost from the current node to the historical failure node decreases, and there are multiple consecutive weak recovery nodes in the path. If any of the judgment conditions are met, the system is considered to be in a state of stability degradation. Step Six: Intelligent Maintenance Strategy Generation and Linkage Control Execution: After determining that the system has stability degradation, based on the feature offset direction in the current map path and the evolution pattern of historical degradation paths, a maintenance recommendation strategy containing the following content is generated: Recommended key inspection items; Recommended maintenance execution time window; Recommendations for adjusting the operating strategy of control parameters; Recommendations for limiting the operating load; Furthermore, by linking with the control system, it can push maintenance suggestions, issue early warning signals, or generate maintenance work orders, thereby enabling early intervention and refined intelligent maintenance for potential faults.

[0029] Example 2: A method for intelligent maintenance of sewage pumps based on fault prediction, comprising the following steps: Step 1: Trigger Event Recognition Collect status change signals that occur during the operation of the sewage pump control system; Each type of state transition signal is marked as a state transition event, and a sequence of state transition events is established according to the actual time order of occurrence, which serves as an event index for subsequent data organization.

[0030] Step 2: Event Response Data Collection and Behavioral Feature Extraction For each state transition event identified in step one, a response window of fixed duration is set, and the transient response process of multiple operating parameters within the response window is collected; Operating parameters include, but are not limited to, current, flow rate, pressure, temperature, and vibration; Based on this transient response process, behavioral characteristic indicators reflecting the stability and recovery capability of the sewage pump system are extracted. These behavioral characteristic indicators include at least steady-state establishment time, maximum overshoot amplitude, response recovery rate, and residual amplitude of response fluctuation. Each state transition event and its corresponding behavioral characteristic indicators are combined to form an event response feature group, which forms a set of structured response vectors. Compared to Example 1, Example 2 performs morphological clustering based on the response curves of similar state transition events within historical multi-cycle periods to form multiple response behavior morphological clusters, and assigns a unique identifier to each response morphological cluster. After each transition event occurs, multiple operating parameter change curves within its response window are collected to determine the response pattern cluster to which it belongs and to calculate its offset from the cluster center. The type identifier of the transition event, the cluster identifier, the behavior index group, and the intra-cluster offset value are combined to form a structured response vector, which is then used as a feature of the subsequent graph nodes.

[0031] Specifically, including: Firstly, based on historical operating data over multiple periods, the response parameter change curves for the same type of state transition events (such as "start-up event", "shutdown event", "load surge event", etc.) are analyzed. The response parameters can include multiple monitoring dimensions such as pump current, voltage, flow rate, liquid level, speed, and vibration intensity. For each state transition event, a fixed-length response time window (e.g., 10 seconds) is set starting from its occurrence. The time series of each response parameter within this window is extracted as the original set of response curves. For these sets of response curves: First, perform normalization processing (such as Z-score or Min-Max standardization). Then, the differences between different response curves are calculated using DTW (Dynamic Time Warping) distance, Euclidean distance, or similarity measures of Fourier transform coefficients. Further, algorithms such as K-Means, DBSCAN, or hierarchical clustering are used to cluster response curves with similar dynamic behavior patterns into a cluster, resulting in multiple response behavior pattern clusters (each cluster represents a typical response behavior pattern). Each response behavior morphology cluster is assigned a unique cluster identifier, and the curve with the smallest average distance from all other curves in the cluster is selected as the morphology center template of the cluster for subsequent matching and judgment.

[0032] Secondly, after a new state transition event occurs, the system immediately collects the change curves of each monitored parameter within its response window, forming a response sample for the current event. Then, the following steps are executed: The current response sample is processed using the same normalization method as the historical curve; Calculate the similarity (e.g., DTW distance) between this sample and the central templates of all historical behavior clusters. The sample is assigned to the behavioral pattern cluster with the smallest distance, and the corresponding cluster identifier is obtained. Simultaneously, the offset value between the response sample and its cluster center template is recorded. The offset value can be defined as: the DTW distance value, or the Euclidean distance of key response features (such as the square root of the sum of squares of the differences in behavioral indicators such as steady-state settling time, maximum overshoot amplitude, recovery rate, residual fluctuation amplitude, etc.), or a threshold is set to determine whether the response is an "abnormal response behavior" or "drift behavior".

[0033] Third, combine the following elements to form a structured response vector, which will be used as the feature attribute input for event nodes in the response graph: Event type identifier (e.g., "startup event", "valve jump event", etc., Event_Type_ID); Cluster affiliation identifier; A set of behavioral characteristic indicators, such as: steady-state establishment time (the time from event triggering to key parameters entering a stable range), maximum overshoot amplitude (the maximum deviation from the steady-state target), recovery rate (the speed at which parameters return to the steady-state range, such as slope), and residual fluctuation amplitude (the average / peak value of small oscillations after steady-state). Intra-cluster offset (quantified value of the difference from the center of the behavioral morphology to which it belongs); Third, after each new state transition event occurs, the running parameter curves in its response window are collected, matched according to the preset behavior pattern template, the response pattern cluster to which it belongs is determined, and the degree of offset from the cluster center is calculated to obtain the drift distance or deviation metric. Fourth, the type identifier, cluster identifier, behavioral feature index set, and intra-cluster offset value of the aforementioned state transition events are integrated into a structured response vector, which serves as the feature attribute of subsequent response map nodes.

[0034] Step 3: Response Map Construction The state transition events obtained in step one are used as graph nodes, and the graph structure connection relationship is constructed according to the actual execution sequence of the events. The behavioral feature indicators extracted in step two are bound to the corresponding graph nodes to form a response graph model containing event nodes, event connection edges, and node feature attributes, which is used to reflect the state evolution path and dynamic stability performance of the sewage pump under different operating cycles. The behavioral characteristic indicators of each state transition event node in the response graph model are evaluated from multiple dimensions and their indicator values ​​are obtained. If the indicator value exceeds the preset anomaly judgment threshold, the node is marked as an abnormal node. Step 4: Recovery Failure Rate Modeling and Path Risk Assessment The historical running records of each type of state transition event in the response graph model are statistically analyzed, and the recovery failure rate of the event type in the past running cycle is calculated. The recovery failure rate refers to the proportion of the event response characteristics that fail to recover to the stable threshold within a preset time window. By combining the number of marked abnormal nodes in the current graph path and their characteristic degradation degree, the minimum risk path cost from the current state node to the historical known fault node is calculated as a quantitative indicator of the current system stability degradation degree. Step 5: Determine operational stability degradation. Determine whether the current operating status of the sewage pump exhibits a trend of stability degradation based on at least one of the following criteria: The response characteristic indicators of a certain type of state transition event continuously deviate from its historical cluster center in multiple consecutive operating cycles; The failure rate of recovery for a certain type of event shows a continuous upward trend within the sliding time window and exceeds the preset statistical threshold. The minimum risk path cost from the current node to the historical failure node decreases, and there are multiple consecutive weak recovery nodes in the path. If any of the judgment conditions are met, the system is considered to be in a state of stability degradation. Step Six: Intelligent Maintenance Strategy Generation and Linkage Control Execution: After determining that the system has stability degradation, based on the feature offset direction in the current map path and the evolution pattern of historical degradation paths, a maintenance recommendation strategy containing the following content is generated: Recommended key inspection items; Recommended maintenance execution time window; Recommendations for adjusting the operating strategy of control parameters; Recommendations for limiting the operating load; Furthermore, by linking with the control system, it can push maintenance suggestions, issue early warning signals, or generate maintenance work orders, thereby enabling early intervention and refined intelligent maintenance for potential faults.

[0035] Example 3: A method for intelligent maintenance of sewage pumps based on fault prediction, comprising the following steps: Step 1: Trigger Event Recognition Collect status change signals that occur during the operation of the sewage pump control system; Each type of state transition signal is marked as a state transition event, and a sequence of state transition events is established according to the actual time order of occurrence, which serves as an event index for subsequent data organization.

[0036] Step 2: Event Response Data Collection and Behavioral Feature Extraction For each state transition event identified in step one, a response window of fixed duration is set, and the transient response process of multiple operating parameters within the response window is collected; Operating parameters include, but are not limited to, current, flow rate, pressure, temperature, and vibration; Based on this transient response process, behavioral characteristic indicators reflecting the stability and recovery capability of the sewage pump system are extracted. These behavioral characteristic indicators include at least steady-state establishment time, maximum overshoot amplitude, response recovery rate, and residual amplitude of response fluctuation. Each state transition event and its corresponding behavioral characteristic indicators are combined to form an event response feature group, which forms a set of structured response vectors. Step 3: Response Map Construction The state transition events obtained in step one are used as graph nodes, and the graph structure connection relationship is constructed according to the actual execution sequence of the events. The behavioral feature indicators extracted in step two are bound to the corresponding graph nodes to form a response graph model containing event nodes, event connection edges, and node feature attributes, which is used to reflect the state evolution path and dynamic stability performance of the sewage pump under different operating cycles. The behavioral characteristic indicators of each state transition event node in the response graph model are evaluated from multiple dimensions and their indicator values ​​are obtained. If the indicator value exceeds the preset anomaly judgment threshold, the node is marked as an abnormal node. The steps to obtain the indicator value include: In step two, for operating parameters that have an overshoot peak within the response window and form a monotonic decline segment, the overshoot peak in its response curve is identified as the starting point, and the time point when it first enters the steady-state tolerance range is determined as the ending point. The time point corresponding to the overshoot peak and the time point when it first enters the steady-state tolerance range are defined as the "stable decline region". Based on the response curves corresponding to the stable fallback regions in similar state transition events within historical cycles, a "standard fallback behavior envelope" is constructed. The overlap ratio and envelope penetration rate between the response curves corresponding to the current stable fallback region and the envelope are calculated as "stable trend consistency indicators". Meanwhile, the trend reversal point and response inflection point in the response curve of the operating parameters are extracted, and the time length and magnitude change value between the first trend reversal and the entry into steady state are calculated to obtain the "response inertia offset index". The stable trend consistency index and the response inertia offset index are written as structured indicators into the behavioral feature index set of the corresponding state transition event nodes in the response graph model.

[0037] Specifically, in step two, for operating parameters (such as current, flow rate, pressure, etc.) that have overshoot peaks and form monotonic decline segments within the response window, the "stable decline region" in their response process is extracted. The stable decline region refers to the time period in the response curve from the time point corresponding to the overshoot peak to the time point when the operating parameter first enters and remains within the steady-state tolerance range. It is used to characterize the dynamic behavior of the system transitioning from an abnormal response state to a steady state. Based on the response curves corresponding to the stable fallback regions of similar state transition events within multiple historical periods, a "standard fallback behavior envelope" is constructed. This envelope represents the typical fallback behavior range of the system under normal stable conditions in history. The system's response curve to the stable fallback region of the current state transition event is compared with the standard fallback behavior envelope, and their intersection overlap ratio and envelope penetration rate are calculated. This is used as an indicator to assess whether the current behavior deviates from the stable trend, and is defined as the "stable trend consistency index". Meanwhile, the system performs derivative analysis on the response curve formed by the time series of response parameters, extracts the trend reversal point and response inflection point, and counts the time length and magnitude change from the first trend reversal point to the final entry into the steady state range. It then compares this with the mean of historical normal samples to obtain an index used to measure the change in system response inertia, which is defined as the "response inertia offset index". The stability trend consistency index and the response inertia offset index constitute a structured response vector, which serves as part of the set of behavioral characteristic indicators of the corresponding node in the response graph model for the current state transition event. This vector is used for subsequent graph modeling, abnormal node identification, and stability degradation assessment.

[0038] Step 4: Recovery Failure Rate Modeling and Path Risk Assessment The historical running records of each type of state transition event in the response graph model are statistically analyzed, and the recovery failure rate of the event type in the past running cycle is calculated. The recovery failure rate refers to the proportion of the event response characteristics that fail to recover to the stable threshold within a preset time window. By combining the number of marked abnormal nodes in the current graph path and their characteristic degradation degree, the minimum risk path cost from the current state node to the historical known fault node is calculated as a quantitative indicator of the current system stability degradation degree. The assessment process for the degree of feature degradation includes: Z1. Construction of Historical Stable Indicator Bands: For each type of state transition event node, the system statistically analyzes its set of behavioral characteristic indicators in historical normal cycles, and then uses the historical average of each behavioral characteristic indicator in the set as a basis. with standard deviation Construct the stability range of each behavioral characteristic indicator. ; Z2, Current Period Indicator Deviation Calculation: Behavioral characteristic indicators of target nodes in the current operating cycle Calculate its deviation from the historical mean. ; Z3. Degradation Level Judgment Rules: If behavioral characteristic indicators If it falls outside the stability range, it is marked as an "abnormal indicator"; If deviation >Preset deviation threshold ( Based on standard deviation If you configure it, it will be marked as "severely degraded"; For each node, count the number of abnormal indicators and their average deviation to determine the characteristic degradation value of that node. .

[0039] After the response graph model is constructed, state transition event paths with completely consistent structures in multiple operating cycles are aggregated to construct a cross-cycle state transition event path cluster. Then, a time series analysis is performed on the set of behavioral characteristic indicators of the state transition event nodes in the cross-cycle state transition event path cluster in different cycles to determine whether there is a trend of the behavioral characteristic indicator values ​​deviating from their historical stable statistical range in the path cluster cycle by cycle. If the deviation widens in multiple consecutive cycles, it is determined that the path cluster of cross-cycle state transition events has a deterioration trend, and this trend is incorporated into the graph model as an enhanced criterion for the risk of system stability degradation.

[0040] Specifically, "a state transition event path with completely consistent structure" means that the types of state transition event nodes, the order of event connections, and the graph structure topology contained in each path are consistent. For example, "startup event structured response vector → structured response vector load surge event structured response vector → structured response vector vibration fluctuation event". If this path repeats in multiple cycles, it is defined as the same cross-cycle state transition event path cluster. For each state transition event node in the path cluster, the system statistically analyzes its corresponding set of behavioral characteristic indicators in different operating cycles, including steady-state establishment time, maximum overshoot amplitude, recovery rate, and residual fluctuation amplitude. All indicators are organized according to event node number and period number, forming a two-dimensional time series indicator matrix: = The k-th index value of the j-th node in the i-th period; where: i∈[1,n], is the cycle number; j∈[1,m], is the node number of the state transition event in the path; k∈[1,k], where k is the type number of the behavioral characteristic index.

[0041] Specifically, the two-dimensional time series index matrix can also be understood as a periodic behavior index matrix that constitutes the state transition event path in cross-cycle operation, which is used for subsequent path trajectory matching and offset analysis. Performing trend analysis on the above indicator matrix includes the following steps: Constructing a historical stable statistical interval: Determining the mean of each behavioral characteristic indicator over a historical period T. with standard deviation Calculate the stability interval: ; Calculate the deviation of behavioral characteristic indicators: Within the current sliding window, calculate the deviation of the k-th indicator value from the center value in the i-th period. ; Conditions for identifying degradation trends: If any of the following conditions are met, the path cluster is determined to have a degradation trend: Deviation of a certain behavioral characteristic indicator over L consecutive periods Continuously increasing; At least P% of the behavioral characteristics indicators exceeded their stable range for L consecutive periods; If the slope of the linear fit of the indicator deviation sequence exceeds the set slope threshold, it indicates that the behavioral characteristic indicator is deteriorating rapidly.

[0042] Once a path cluster that meets any of the above degradation conditions is identified, the system marks the path cluster of the cross-cycle state transition event as a degradation path cluster and writes its marking information into the response graph model, which is used as the basis for identifying stability degradation risks in the subsequent fault prediction and intelligent maintenance strategy generation steps.

[0043] Methods for obtaining the minimum risk path cost include: S1. Based on the response graph model constructed in the current cycle, select the nodes in the graph that are currently active or in operation. It also retrieves the set of nodes that have been marked as "known fault nodes" in all historical cycles. ; S2. For each edge in the response graph model (from node) point to Define its path weights. The risk level of its termination node is calculated as follows: ; In the formula, For nodes The anomaly flag value is set to 1 if the node is identified as an anomaly, and 0 otherwise. For nodes The characteristic degradation level value; For nodes The Boolean value indicating whether the cross-cycle state transition event path cluster has a deterioration trend is set to 1 if it does, otherwise it is set to 0. These are the risk weight coefficients for the corresponding items; S3, Minimum Path Cost Calculation: Based on step S2, the edge Path weight With the current node Starting from the point, use the shortest path search algorithm in the weighted graph to search for all paths leading to the historical failure node. Find the path with the minimum total value, denoted as: In the formula, Indicates starting from the current node To the historical fault node A feasible path; Minimize the total value of the path The risk value of the current operating state is recorded and written into the response graph model for subsequent stability degradation assessment and intelligent maintenance strategy generation.

[0044] In this embodiment, according to The size determines the warning level, for example: <1.0: Low risk; 1.0≤ <2.5: Medium risk; ≥2.5: High risk, triggering maintenance strategy generation.

[0045] Step 5: Determine operational stability degradation. Determine whether the current operating status of the sewage pump exhibits a trend of stability degradation based on at least one of the following criteria: The response characteristic indicators of a certain type of state transition event continuously deviate from its historical cluster center in multiple consecutive operating cycles; The failure rate of recovery for a certain type of event shows a continuous upward trend within the sliding time window and exceeds the preset statistical threshold. The minimum risk path cost from the current node to the historical failure node decreases, and there are multiple consecutive weak recovery nodes in the path. If any of the judgment conditions are met, the system is considered to be in a state of stability degradation. Step Six: Intelligent Maintenance Strategy Generation and Linkage Control Execution: After determining that the system has stability degradation, based on the feature offset direction in the current map path and the evolution pattern of historical degradation paths, a maintenance recommendation strategy containing the following content is generated: Recommended key inspection items; Recommended maintenance execution time window; Recommendations for adjusting the operating strategy of control parameters; Recommendations for limiting the operating load; Furthermore, by linking with the control system, it can push maintenance suggestions, issue early warning signals, or generate maintenance work orders, thereby enabling early intervention and refined intelligent maintenance for potential faults.

[0046] Example 4: A method for intelligent maintenance of sewage pumps based on fault prediction, comprising the following steps: Step 1: Trigger Event Recognition Collect status change signals that occur during the operation of the sewage pump control system; Each type of state transition signal is marked as a state transition event, and a sequence of state transition events is established according to the actual time order of occurrence, which serves as an event index for subsequent data organization.

[0047] Step 2: Event Response Data Collection and Behavioral Feature Extraction For each state transition event identified in step one, a response window of fixed duration is set, and the transient response process of multiple operating parameters within the response window is collected; Operating parameters include, but are not limited to, current, flow rate, pressure, temperature, and vibration; Based on this transient response process, behavioral characteristic indicators reflecting the stability and recovery capability of the sewage pump system are extracted. These behavioral characteristic indicators include at least steady-state establishment time, maximum overshoot amplitude, response recovery rate, and residual amplitude of response fluctuation. Each state transition event and its corresponding behavioral characteristic indicators are combined to form an event response feature group, which forms a set of structured response vectors. Step 3: Response Map Construction The state transition events obtained in step one are used as graph nodes, and the graph structure connection relationship is constructed according to the actual execution sequence of the events. The behavioral feature indicators extracted in step two are bound to the corresponding graph nodes to form a response graph model containing event nodes, event connection edges, and node feature attributes, which is used to reflect the state evolution path and dynamic stability performance of the sewage pump under different operating cycles. The behavioral characteristic indicators of each state transition event node in the response graph model are evaluated from multiple dimensions and their indicator values ​​are obtained. If the indicator value exceeds the preset anomaly judgment threshold, the node is marked as an abnormal node. Step 4: Recovery Failure Rate Modeling and Path Risk Assessment The historical running records of each type of state transition event in the response graph model are statistically analyzed, and the recovery failure rate of the event type in the past running cycle is calculated. The recovery failure rate refers to the proportion of the event response characteristics that fail to recover to the stable threshold within a preset time window. By combining the number of marked abnormal nodes in the current graph path and their characteristic degradation degree, the minimum risk path cost from the current state node to the historical known fault node is calculated as a quantitative indicator of the current system stability degradation degree. Compared to Example 3, Example 4 extracts state transition event paths with consistent structure from multiple historical operation cycles, and constructs a set of path behavior trajectories from their corresponding periodic behavior index matrices to form a standard path pool. In the current operating cycle, extract the periodic behavior index matrix corresponding to the path to be evaluated, match it with the path behavior trajectory in the standard path pool, and calculate the degree of trajectory matching deviation. If the current path's behavior trajectory shows a continuous tendency to deviate from the central region of the main trajectory cluster of the standard path pool in multiple consecutive periods, and the direction of deviation is consistent, then the path is determined to have a risk of self-evolutionary feature shift. The trajectory deviation trend score of this path is used as a new indicator dimension and written into the corresponding state transition event path structure in the response graph model. It is also used as an additional weighting factor in the path risk cost calculation formula to enhance the ability to identify the evolutionary degradation trend of the system in advance.

[0048] Step 5: Determine operational stability degradation. Determine whether the current operating status of the sewage pump exhibits a trend of stability degradation based on at least one of the following criteria: The response characteristic indicators of a certain type of state transition event continuously deviate from its historical cluster center in multiple consecutive operating cycles; The failure rate of recovery for a certain type of event shows a continuous upward trend within the sliding time window and exceeds the preset statistical threshold. The minimum risk path cost from the current node to the historical failure node decreases, and there are multiple consecutive weak recovery nodes in the path. If any of the judgment conditions are met, the system is considered to be in a state of stability degradation. Step Six: Intelligent Maintenance Strategy Generation and Linkage Control Execution: After determining that the system has stability degradation, based on the feature offset direction in the current map path and the evolution pattern of historical degradation paths, a maintenance recommendation strategy containing the following content is generated: Recommended key inspection items; Recommended maintenance execution time window; Recommendations for adjusting the operating strategy of control parameters; Recommendations for limiting the operating load; Furthermore, by linking with the control system, it can push maintenance suggestions, issue early warning signals, or generate maintenance work orders, thereby enabling early intervention and refined intelligent maintenance for potential faults.

[0049] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0050] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for intelligent maintenance of sewage pumps based on fault prediction, characterized in that, include: Step 1: Collect the status change signals that occur during the operation of the sewage pump control system; Each type of state transition signal is marked as a state transition event, and a sequence of state transition events is established according to the actual time order of their occurrence. Step 2: For each state transition event identified in Step 1, set a response window of fixed duration, collect the transient response process of multiple operating parameters within the window, extract behavioral feature indicators that reflect the stability and recovery capability of the sewage pump system, and form a set of structured response vectors. Step 3: Use the state transition events obtained in Step 1 as graph nodes to construct the graph structure connection relationships; The behavioral feature indicators extracted in step two are bound to the corresponding graph nodes to form a response graph model. The behavioral feature indicators of the graph nodes are evaluated in multiple dimensions, and nodes that exceed the anomaly judgment threshold are marked as abnormal nodes. Step 4: Statistically analyze the historical operation records of various state transition events in the response graph model, calculate the recovery failure rate, and calculate the minimum risk path cost from the current node to the historical failure node by combining the number of abnormal nodes and the degree of feature degradation. Step 5: Based on the preset degradation judgment conditions, determine whether the system has a trend of stability degradation; Step 6: If stability degradation is determined, a maintenance strategy is generated by combining the offset direction of the map path features and the historical evolution pattern.

2. The intelligent maintenance method for sewage pumps based on fault prediction according to claim 1, characterized in that, The degradation criteria in step five include: The response characteristic indicators of a certain type of state transition event continuously deviate from its historical cluster center in multiple consecutive operating cycles; The failure rate of recovery for a certain type of event shows a continuous upward trend within the sliding time window and exceeds the preset statistical threshold. The minimum risk path cost from the current node to the historical failure node decreases, and there are multiple consecutive weak recovery nodes in the path. If any of the judgment conditions are met, the sewage pump control system is considered to be in a state of stability degradation.

3. The intelligent maintenance method for sewage pumps based on fault prediction according to claim 1, characterized in that, In step two, morphological clustering is performed based on the response curves of similar state transition events within historical multi-cycle periods to form multiple response behavior morphological clusters, and a unique identifier is assigned to each response morphological cluster. After each transition event occurs, multiple operating parameter change curves within its response window are collected to determine the response pattern cluster to which it belongs and to calculate its offset from the cluster center. The type identifier of the transition event, the cluster identifier, the behavior index group, and the intra-cluster offset value are combined to form a structured response vector, which is then used as a feature of the subsequent map nodes.

4. The intelligent maintenance method for sewage pumps based on fault prediction according to claim 1, characterized in that, The steps to obtain the indicator value include: In step two, for operating parameters that have an overshoot peak within the response window and form a monotonic decline segment, the overshoot peak in its response curve is identified as the starting point, and the time point when it first enters the steady-state tolerance range is determined as the ending point. The time point corresponding to the overshoot peak and the time point when it first enters the steady-state tolerance range are defined as the "stable decline region". Based on the response curves corresponding to the stable fallback regions in similar state transition events within historical cycles, a "standard fallback behavior envelope" is constructed. The overlap ratio and envelope penetration rate between the response curve corresponding to the current stable fallback region and the envelope are calculated as "stable trend consistency indicators". Meanwhile, the trend reversal point and response inflection point in the response curve of the operating parameters are extracted, and the time length and magnitude change value between the initial trend reversal and the entry into steady state are calculated to obtain the "response inertia offset index". The stable trend consistency index and the response inertia offset index are written as structured indicators into the behavioral feature index set of the corresponding state transition event nodes in the response graph model.

5. The intelligent maintenance method for sewage pumps based on fault prediction according to claim 1, characterized in that, In step four, the assessment process for the degree of feature degradation includes: Z1. Construction of Historical Stable Indicator Bands: For each type of state transition event node, the system statistically analyzes its set of behavioral characteristic indicators in historical normal cycles, and then uses the historical average of each behavioral characteristic indicator in the set as a basis. with standard deviation Construct the stability range of each behavioral characteristic indicator. ; Z2, Current Period Indicator Deviation Calculation: Behavioral characteristic indicators of target nodes in the current operating cycle Calculate its deviation from the historical mean. ; Z3. Degradation Level Judgment Rules: If behavioral characteristic indicators If it falls outside the stability range, it is marked as an "abnormal indicator"; If deviation >Preset deviation threshold If so, it will be marked as "severely degraded"; For each node, count the number of abnormal indicators and their average deviation to determine the characteristic degradation value of that node. .

6. The intelligent maintenance method for sewage pumps based on fault prediction according to claim 1, characterized in that, After the response graph model is constructed, state transition event paths with completely consistent structures in multiple operating cycles are aggregated to construct a cross-cycle state transition event path cluster. Then, a time series analysis is performed on the set of behavioral characteristic indicators of the state transition event nodes in the cross-cycle state transition event path cluster in different cycles to determine whether there is a trend of the behavioral characteristic indicator values ​​deviating from their historical stable statistical range in the path cluster cycle by cycle. If the deviation widens in multiple consecutive cycles, it is determined that the path cluster of cross-cycle state transition events has a deterioration trend, and this trend is incorporated into the graph model as an enhanced criterion for the risk of system stability degradation.

7. The intelligent maintenance method for sewage pumps based on fault prediction according to claim 6, characterized in that, The method for obtaining the minimum risk path cost value includes: S1. Based on the response graph model constructed in the current cycle, select the nodes in the graph that are currently active or in operation. And retrieve the set of nodes that have been marked as "known fault nodes" in all historical cycles { ; S2. For each edge in the response graph model Define its path weight The risk level of its termination node is calculated as follows: ； In the formula, For nodes The anomaly flag value is set to 1 if the node is identified as an anomaly, and 0 otherwise. For nodes The characteristic degradation level value; For nodes The Boolean value indicating whether the cross-cycle state transition event path cluster has a deterioration trend is set to 1 if it does, otherwise it is set to 0. These are the risk weight coefficients for the corresponding items; S3. Based on step S2, the edge Path weight With the current node Starting from the point, use the shortest path search algorithm in the weighted graph to search for all paths leading to the historical failure node. Find the path with the minimum total value, denoted as: In the formula, Indicates starting from the current node To the historical fault node A feasible path; Minimize the total value of the path The risk value of the current operating state is used as a basis for writing into the response graph model.

8. The intelligent maintenance method for sewage pumps based on fault prediction according to claim 7, characterized in that, In multiple historical operating cycles, state transition event paths with consistent structures are extracted, and their corresponding periodic behavior index matrices are constructed into a set of path behavior trajectories to form a standard path pool. In the current operating cycle, the periodic behavior index matrix corresponding to the path to be evaluated is extracted and matched with the path behavior trajectory in the standard path pool to calculate the degree of trajectory matching deviation. If the current path's behavior trajectory shows a continuous tendency to deviate from the central region of the main trajectory cluster of the standard path pool in multiple consecutive periods, and the direction of deviation is consistent, then the path is determined to have a risk of self-evolutionary feature shift. The trajectory deviation trend score of this path is used as a new indicator dimension and written into the corresponding state transition event path structure in the response graph model. It is also used as an additional weighting factor in the path risk cost calculation formula to enhance the ability to identify the evolutionary degradation trend of the system in advance.

Images