Fault detection method suitable for permanent magnet motor sewage pump

Abstract

The invention discloses a fault detection method suitable for a permanent magnet motor sewage pump, and particularly relates to the technical field of motor equipment operation fault detection.The fault detection method comprises the following steps that stage division is conducted on the operation process of the permanent magnet motor sewage pump, and staged dynamic baselines corresponding to all operation stages are established; extracting transient abrupt change of the operating parameters under the dynamic baseline constraint and forming a transient abnormal event sequence; the method comprises the following steps: judging the damage degree grade of a transient abnormal event, continuously accumulating risk integrals, outputting early warning information when a risk reaches a trigger condition, and generating an operation disposal instruction; according to the invention, by constructing a staged dynamic baseline and extracting a transient abnormal event, accurate identification of abnormal behaviors in different operation stages is realized; and in combination with a continuous accumulation mechanism of the damage degree grade and the risk integral, when the risk reaches a triggering condition, an early warning is output, and an operation disposal instruction is generated, so that the accuracy, timeliness and operation safety of fault detection of the permanent magnet motor sewage pump are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of motor equipment operation fault detection, more specifically, the present application relates to a fault detection method suitable for permanent magnet motor sewage pump. BACKGROUND

[0002] Permanent magnet motor sewage pump has been widely used in municipal drainage, sewage treatment and industrial wastewater transportation due to its compact structure, high efficiency and low energy consumption. In actual operation, permanent magnet motor sewage pump is often in a complex environment with high humidity, high corrosion, high impurity content and frequent working condition changes. During operation, it inevitably goes through different stages such as starting, steady running, disturbance and shutdown. The change rule and allowable fluctuation range of operating parameters in each operating stage are obviously different.

[0003] In the prior art, the fault detection method of permanent magnet motor sewage pump is mostly based on fixed threshold or empirical model. The current, voltage, speed, temperature and other operating parameters are periodically sampled, and the sampling data are averaged, filtered or smoothed, and then compared with the preset threshold to determine the equipment operating state. This method has certain effect when the operating state is stable and the abnormal duration is long, but its judgment logic usually assumes that the abnormality has a long duration or obvious amplitude characteristics, which is difficult to adapt to the transient abnormal characteristics of sewage pump under complex working conditions.

[0004] In actual operation, permanent magnet motor sewage pump is prone to abnormal operating state with extremely short duration but strong impact due to instantaneous entry of foreign matter in sewage, short-term stall, sudden change of load and other reasons. Such abnormalities often appear and recover quickly in a very short time, and their amplitude may not meet the fixed threshold setting condition, or they may be weakened or even completely covered by filtering and smoothing algorithms during periodic sampling and averaging processing, resulting in that the existing detection method misjudges the actual existing destructive impact as normal operating state. Moreover, although the duration of such abnormality is short, its instantaneous impact may have caused potential irreversible damage to the rotor, magnet steel or mechanical structure of the permanent magnet motor sewage pump, such as rotor deformation, local demagnetization of magnet steel, fatigue accumulation of shaft or impeller structure, etc. Such damage is difficult to identify directly at the initial stage by traditional threshold judgment method, but it will gradually enlarge in the subsequent operation process and eventually may cause sudden and disastrous failure, seriously affecting the safety of equipment operation and system reliability.

[0005] The existing fault detection method mostly uses single parameter out-of-range as the basis for judgment, lacks comprehensive consideration of the cumulative effect of abnormal risk with running time, and is difficult to reflect the real process of gradual accumulation of potential fault hidden danger of permanent magnet motor sewage pump in a period of running time, and the early warning is lagging. Therefore, the present application proposes a fault detection method suitable for permanent magnet motor sewage pump to solve the above problems. SUMMARY

[0006] To achieve the above object, the present application provides the following technical solutions. A fault detection method suitable for a permanent magnet motor sewage pump, comprising the following steps: The operation process of the permanent magnet motor sewage pump is divided into stages, and the operation parameters of the permanent magnet motor sewage pump are collected in each operation stage. According to the change relationship of the operation parameters in each operation stage, a dynamic response curve and a change slope boundary corresponding to each operation stage are established to form a staged dynamic baseline for subsequent fault detection; Under the constraint of the staged dynamic baseline, the instantaneous mutation of the operation parameters in the sampling process is subjected to peak value retention and time marking processing, so that abnormal changes with a duration shorter than the conventional sampling period are completely retained to form a transient abnormal event sequence containing mutation amplitude, mutation duration and mutation rising characteristics; Based on the formed transient abnormal event sequence, the corresponding mutation energy index and multi-parameter coupling offset index are calculated for each transient abnormal event, and the calculation results are associated with the staged dynamic baseline for analysis to determine whether the transient abnormal event presents a preset different level of impact characteristics, and accordingly determine the damage level of the transient abnormal event to the permanent magnet motor sewage pump; According to the damage level, the risk of the transient abnormal event is quantified, the corresponding risk score is generated for each transient abnormal event, and the risk score is continuously accumulated during the operation process. When the risk score reaches the corresponding trigger condition within a preset operation time range, the warning information that the permanent magnet motor sewage pump has potential fault hidden danger is output; After outputting the warning information, the running disposal instruction corresponding to the risk score and the damage level is generated.

[0007] The operation process includes a starting stage, a steady state operation stage, a disturbance stage and a shutdown stage, and the operation parameters at least include current parameters, voltage parameters, speed parameters and temperature parameters.

[0008] In a preferred embodiment, the staged dynamic baseline is generated by the following steps: The time series of the continuously collected operation parameters in each operation stage is constructed, and the change direction and change amplitude of the operation parameters are calculated by the numerical difference of adjacent sampling points to determine the time series change trend of the operation parameters in the operation stage; Based on the determination of the change trend, the change amount per unit time is calculated based on the numerical difference of the operation parameters with time to obtain the change slope of the operation parameters in the operation stage, and a slope curve reflecting the change speed of the operation state is formed accordingly; The change slopes corresponding to different operation parameters at the same sampling time points are time-aligned, the consistency of change direction and the difference of change amplitude of each operation parameter in the same time window are compared, and the synchronization degree and deviation degree of the operation parameters in the time dimension are obtained as the mutual correlation between the operation parameters; Based on the change trend, the change slope and the mutual correlation, a dynamic response curve of the corresponding operation stage is generated, and the upper limit boundary and the lower limit boundary of the change slope are determined based on the dynamic response curve, forming a staged dynamic baseline for subsequent fault detection.

[0009] In a preferred embodiment, the synchronization degree and deviation degree between the operation parameters are calculated as follows: In the same time window, the change amount between adjacent sampling points of each operation parameter is calculated, and the change direction is represented by the positive and negative properties of the change amount; When any two operation parameters have the same change direction at the same sampling time, it is recorded as a direction synchronization event, and the ratio of the number of direction synchronization events to the total number of sampling points in the time window is determined as the synchronization degree value; The absolute value of the numerical difference between the change amounts of any two operation parameters at the same sampling time is taken, and then normalized, and the numerical difference at each sampling time in the time window is accumulated to obtain the deviation degree value.

[0010] In a preferred embodiment, forming the transient abnormal event sequence includes: Under the constraint of the staged dynamic baseline, a continuous sampling time window is set for the operation parameter, and the maximum value of the operation parameter and the time point at which the maximum value occurs in each time window are recorded as the peak value retention result and the time marker result; Based on the peak value retention result, the absolute value of the difference between the mutation reference value corresponding to the peak value and the mutation peak value is determined as the mutation amplitude, and the time span of the operation parameter from the mutation reference value entering the mutation peak value interval and returning to the mutation reference value interval is determined as the mutation duration; The unit time change amount of the operation parameter from the mutation reference value to the mutation peak value in the mutation duration is calculated as the mutation rising feature; The mutation amplitude, mutation duration and mutation rising feature are bound and sorted by time marker to form a transient abnormal event sequence.

[0011] The mutation reference value is the operation parameter value of the operation parameter under the constraint of the staged dynamic baseline when the operation parameter is in a stable change state before the occurrence of the transient abnormality, and the operation parameter value is obtained by averaging or taking the median value of the operation parameter in a preset time window before the occurrence of the peak value.

[0012] In a preferred embodiment, determining the level of damage includes the following steps: For each transient abnormal event in the transient abnormal event sequence, the mutation intensity value is calculated based on the numerical product of the mutation amplitude and the mutation duration, which serves as a mutation energy index characterizing the transient impact intensity of a single operating parameter. Within the time window corresponding to the duration of the mutation, the synchronization degree and deviation degree values ​​obtained by combining all operating parameters in pairs for current, speed, voltage and temperature parameters are summarized to generate a multi-parameter coupling offset index to characterize the degree of deviation of the overall coordination relationship of multiple operating parameters. The mutation energy index and the multi-parameter coupling offset index are compared with the preset impact characteristic reference range of the corresponding operating stage in the phased dynamic baseline. Based on the degree of difference between the mutation energy index and the multi-parameter coupling offset index in the numerical dimension, the preset impact characteristic level that is closest to the current transient abnormal event is determined. Based on the one-to-one mapping relationship between the preset impact characteristic level and the damage level, the damage level corresponding to the current transient abnormal event is determined.

[0013] In a preferred embodiment, the logic for generating and accumulating risk points is as follows: The transient anomaly event is assigned a corresponding basic integral based on the degree of damage. The over-limit amplitudes of the mutation energy index and the multi-parameter coupling offset index are then added to the basic integral as corrections to obtain the risk integral of the transient anomaly event. During operation, a sliding time window is set according to a preset running time. Each risk component within the sliding time window is attenuated and then accumulated to obtain a cumulative risk value used to characterize potential fault hazards. The cumulative risk value is compared with the trigger threshold corresponding to the level of damage. When the cumulative risk value reaches the trigger threshold, an early warning message is generated, and the trigger time and trigger stage are recorded.

[0014] In a preferred embodiment, the cumulative accumulation of each risk integral within the sliding time window after attenuation processing refers to: Within the sliding time window, each risk score is assigned a time weight corresponding to its occurrence time according to the order in which transient abnormal events occur. The closer the time is to the current moment, the greater the time weight of the risk score. Each risk integral is multiplied by its corresponding time weight to obtain the decayed risk integral. The cumulative risk value is obtained by summing all the decayed risk integrals within the sliding time window.

[0015] The technical effects and advantages of this invention are as follows: This invention divides the operation process of a permanent magnet motor sewage pump into stages and establishes dynamic response curves and slope boundaries that match the changes in operating parameters at different stages, forming a staged dynamic baseline. This allows fault detection to move away from relying on a unified, static judgment standard and instead perform differentiated comparative analysis based on the inherent characteristics of different stages such as startup and operational fluctuations. This effectively distinguishes between normal stage fluctuations and abnormal operating states, avoiding misjudgments and omissions caused by changes in operating conditions, and significantly improving the accuracy and stability of fault detection for permanent magnet motor sewage pumps under complex operating conditions.

[0016] Under the constraint of a phased dynamic baseline, this invention performs peak preservation and time stamping processing on instantaneous mutations in operating parameters, forming a transient abnormal event sequence that includes mutation amplitude, mutation duration, and mutation rise characteristics. This allows short-term abnormal changes with durations shorter than the conventional sampling period to be fully captured and utilized, avoiding the weakening or loss of transient impact information caused by traditional periodic sampling and averaging methods. As a result, it can promptly identify potential anomalies caused by instantaneous load changes, short-term impacts, etc., providing a more comprehensive and realistic basis for subsequent damage level determination.

[0017] This invention determines the degree of damage of transient abnormal events and continuously accumulates risk points based on this. When the risk points reach the trigger condition, it outputs early warning information and generates corresponding operation and disposal instructions. This realizes the transformation from single abnormality detection to risk accumulation assessment and operation and disposal linkage. This makes the operation status assessment of permanent magnet motor sewage pumps no longer limited to a single moment or a single event, but can comprehensively reflect the cumulative impact of abnormal behavior over a period of time. Thus, it can provide early warning and intervention before potential faults evolve into serious failures, thereby improving the safety, reliability and maintainability of equipment during operation. Attached Figure Description

[0018] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings; Figure 1 This is a schematic diagram of a fault detection method for a permanent magnet motor sewage pump according to the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0020] Reference Figure 1 The following examples were obtained: Example 1: A fault detection method for permanent magnet motor sewage pumps, comprising the following steps: The operation process of the permanent magnet motor sewage pump is divided into stages, and operating parameters are collected in each stage. This allows for the differentiation of parameter change characteristics under different operating conditions, such as start-up, steady state, disturbance, and shutdown, avoiding confusion between normal fluctuations and abnormal states due to the use of a uniform judgment standard. Based on this, dynamic response curves and slope boundaries are established according to the relationship between the operating parameters in each stage, providing a clear reference for the normal range and rate of change of operating parameters. This forms a staged dynamic baseline, providing a reliable comparative basis for the subsequent identification and judgment of transient anomalies.

[0021] Under the constraint of a phased dynamic baseline, peak preservation and time stamping of transient mutations in operating parameters can avoid the masking of short-term anomalies by conventional sampling and averaging, thus fully preserving destructive anomalies with durations shorter than the conventional sampling period. By extracting mutation amplitude, mutation duration, and mutation rise characteristics, transient anomalies that were originally scattered on the time axis are transformed into a structured sequence of transient anomaly events, providing basic data for subsequent quantitative analysis and destructive assessment of anomalies.

[0022] Based on the formed transient abnormal event sequence, the mutation energy index and multi-parameter coupling offset index are calculated for each transient abnormal event. The abnormal event can be characterized from two aspects: the transient impact intensity of a single operating parameter and the synergistic change offset between multiple operating parameters. The calculation results are correlated with the staged dynamic baseline to compare the performance of the abnormal event with the normal characteristics of the corresponding operating stage, thereby distinguishing between general fluctuations and impactful abnormal states, and determining the degree of damage caused by the transient abnormal event to the permanent magnet motor sewage pump.

[0023] Transient anomalies are quantified based on their degree of damage, and a risk score is generated for each transient anomaly. This transforms the destructive impact of a single anomaly into an accumulative risk. During operation, the risk score is continuously accumulated, allowing multiple low-frequency but destructive anomalies to be considered comprehensively. When the risk score reaches the trigger condition within a preset operating time range, a timely warning of potential fault hazards is output, thus avoiding missed detections caused by relying solely on single out-of-limit judgments.

[0024] After outputting early warning information, corresponding operation and disposal instructions are generated based on risk scores and damage levels. This allows fault detection results to directly serve operational decisions. By adopting differentiated disposal methods for different risk levels, the transformation from fault detection to risk management is realized. This allows for intervention in permanent magnet motor sewage pumps before potential faults develop into serious failures, thereby improving the safety and reliability of equipment operation.

[0025] The operation process is divided based on the changes in operating parameters over time, clearly defining the operation process into four stages: startup, steady-state operation, disturbance, and shutdown. The startup stage is defined as follows: when the permanent magnet motor sewage pump transitions from standstill to operation, and the speed parameter increases from 0 within 0.5 seconds and continues to rise, while the current parameter experiences a startup impact and then declines into a predictable downward trend. The steady-state operation stage is defined as follows: when the speed parameter fluctuates within ±1.5% of the rated speed for 10 consecutive seconds, the voltage parameter remains within ±5% of the rated voltage, and the temperature parameter exhibits a slow, monotonous change. When a sudden load change causes the current parameter to deviate from the reference value by more than 15% instantaneously, accompanied by a significant drop or rise in the speed parameter within 1 second, or a short-term dip in the voltage parameter followed by recovery within 2 seconds, the system is considered to be in the disturbance stage. When a shutdown action is received, the speed parameter continuously decreases and drops to below 10% of the rated speed within 3 seconds, and the current parameter returns to the no-load level, the system is considered to be in the shutdown stage. This division method, which uses "speed change - current change - voltage stability - temperature change" as a joint constraint, makes the normal fluctuation boundaries of different stages distinguishable and reproducible.

[0026] Operating parameters of the permanent magnet motor sewage pump are collected at each stage of operation. These parameters include at least current, voltage, speed, and temperature parameters, and can be further supplemented with vibration, power, power factor, insulation resistance, leakage current, bearing temperature, stator winding temperature, flow rate, pressure, and liquid level parameters, depending on the actual situation. This is to cover multi-dimensional anomalies from the electromagnetic, mechanical, hydraulic, and insulation sides. For example, with a sampling period of 1 millisecond, the current parameter can be recorded as a peak current of 42 amperes at startup, a steady-state average of 18 amperes, and a peak current during disturbance. The voltage parameter can be recorded as 32 amperes, steady-state 380 volts, momentarily dropping to 342 volts during disturbance and recovering in 1.2 seconds; the rotational speed parameter can be recorded as 0→2850 rpm at startup, steady-state 2870±30 rpm, dropping to 2600 rpm during disturbance and recovering; the temperature parameter can be recorded as the winding temperature slowly rising from 46 degrees Celsius to 58 degrees Celsius; simultaneously, the vibration parameter can be recorded as 1.8 mm / s steady-state, momentarily rising to 6.5 mm / s during disturbance; and the flow rate parameter can be recorded as 45 cubic meters per hour steady-state, dropping to 30 cubic meters per hour during disturbance, providing sufficient samples for subsequent characteristic boundary establishment.

[0027] A phased dynamic baseline is generated through the following steps: Within each operational phase, a time series of continuously acquired operational parameters is constructed. The direction and magnitude of parameter changes are calculated by differentiating values ​​at adjacent sampling points to determine the time series trend of the operational parameters within that operational phase. Specifically, this can be implemented by continuously acquiring current, voltage, speed, and temperature parameters during the startup, steady-state operation, disturbance, and shutdown phases, and sorting them by timestamp to form a time series. For example, during the steady-state operation phase, continuous acquisition for 10 seconds with a sampling period of 1 millisecond yields a time series. 000 sets of data; taking adjacent sampling points at a certain moment as an example, the current parameter changes from 18.2 Amperes to 18.6 Amperes, the voltage parameter changes from 379 V to 381 V, the rotation speed parameter changes from 2865 rpm to 2872 rpm, and the temperature parameter changes from 52.3 degrees Celsius to 52.31 degrees Celsius. By differentiating the values ​​of adjacent sampling points, the direction and magnitude of change are obtained, thus obtaining a reproducible time series change trend expression such as "gradual increase, slow decrease, approximately stable, sudden surge followed by fall" during this operation phase, providing input consistency for the subsequent slope curve generation.

[0028] Based on the determined trend of change, the change in operating parameters per unit time is calculated based on the numerical difference results of the changes over time, and the slope of the change in operating parameters during this operating stage is obtained. Based on this, a slope curve reflecting the rate of change of operating state is formed. Specifically, it can be implemented as follows: Based on the adjacent differences obtained in the previous sub-step, the difference value is divided by the sampling interval to obtain the change in unit time and form a change slope sequence. For example, the current parameter difference is 0.4 amperes and the sampling interval is 0.001 seconds, so the change slope is 400 amperes per second; the speed parameter difference of 7 revolutions per minute corresponds to a change slope of 7000 revolutions per minute per second; in the disturbance stage, the voltage parameter may experience a momentary dip, and the difference of -15 volts corresponds to a change slope of -15000 volts per second. The temperature parameter usually has a small slope, such as 0.01 degrees Celsius per second. The slopes at each moment are connected by timestamps to form a slope curve, so that the rate of change characteristics of different operating stages can be compared at the numerical level. For example, in the steady-state operation stage, the slope curve stays close to the zero line for a long time, and in the disturbance stage, the slope curve shows a peak and then quickly returns.

[0029] By aligning the slopes of different operating parameters at the same sampling time point, and comparing the consistency of the direction of change and the difference in the magnitude of change of each operating parameter within the same time window, the degree of synchronization and deviation between operating parameters in the time dimension is obtained, which serves as the correlation between operating parameters. Specifically, this can be implemented as follows: based on multiple slope curves, they are aligned one by one according to the same sampling time point. Within the time window, the consistency of the direction of change and the difference in the magnitude of change of the slopes of current parameter, speed parameter, voltage parameter, and temperature parameter are observed synchronously. For example, if, during a disturbance phase, the slope of the current parameter is continuously positive while the slope of the speed parameter is continuously negative, and the slope of the voltage parameter is initially negative and then positive, it indicates that there is a significant synchronization mismatch and magnitude difference between the operating parameters. During a steady-state operation phase, within a 1-second window, the slopes of the current parameter and the speed parameter are mostly in the same direction with small magnitude differences, indicating that the correlation is stable.

[0030] Based on the changing trends, slopes, and interrelationships, dynamic response curves are generated for the corresponding operating stages. Upper and lower limits of the slope are then determined based on these dynamic response curves, forming a phased dynamic baseline for subsequent fault detection. Specifically, the changing trends are used to determine the main curve shape, the slopes are used to determine the curve's speed characteristics, and the interrelationships are used to determine the consistency constraints of multi-parameter coupling. This generates dynamic response curves for each operating stage. For example, in the steady-state operating stage, a "current parameter - speed parameter" response curve is generated, with sample points falling within the range of 18±2 amperes and 2870±30 revolutions per minute and exhibiting a narrow band distribution; in the disturbance stage, a "voltage parameter dip - current parameter" response curve is generated. The "overshoot" response curve, with sample points covering the combined trajectory of voltage parameter dips from 380V to 340V and current parameter overshoots from 18A to 32A, is used to statistically analyze the slope distribution of the corresponding operating stage based on each dynamic response curve. The upper and lower bounds of common stable distributions are taken as the upper and lower limits of the slope change. For example, the upper limit of the slope change of current parameter in the steady-state operating stage can be limited to 1200A / s, and the lower limit of the slope change can be limited to -1200A / s. In the disturbance stage, wider boundaries are allowed to cover short-term spikes. This yields a staged dynamic baseline for subsequent fault detection, ensuring that subsequent comparisons and judgments of transient anomalies always fall within the reproducible reference range of the corresponding operating stage.

[0031] The synchronization and deviation between operating parameters are calculated as follows: Within the same time window, the change between adjacent sampling points of each operating parameter is calculated, and the positive or negative attribute of the change represents the direction of change. Specifically, this can be implemented as follows: Under the constraint of a phased dynamic baseline, the same time window is selected, for example, a 200-millisecond time window with a sampling period of 1 millisecond to obtain 200 sampling points. Differential sequences of adjacent sampling points are constructed for current, voltage, speed, and temperature parameters respectively. Taking the current parameter as an example, if the k-th sampling point is 18.2 amperes and the (k+1)-th sampling point is 18.6 amperes, the change is +0.4 amperes, and the positive or negative attribute represents the direction of change as positive. Taking voltage parameters as an example, if the k-th sampling point is 380 volts and the (k+1)-th sampling point is 378 volts, the change is -2 volts, indicating a negative direction of change. Similarly, for speed parameters, if the k-th sampling point is 2865 revolutions per minute and the (k+1)-th sampling point is 2852 revolutions per minute, the change is -13 revolutions per minute, indicating a negative direction of change. And for temperature parameters, if the k-th sampling point is 52.30 degrees Celsius and the (k+1)-th sampling point is 52.31 degrees Celsius, the change is +0.01 degrees Celsius, indicating a positive direction of change. This method of constructing the change and direction point-by-point according to time windows ensures that the direction consistency judgment among operating parameters has the same time reference.

[0032] When any two operating parameters change in the same direction at the same sampling moment, it is recorded as a direction synchronization event. The ratio of the number of direction synchronization events to the total number of sampling points within the time window is determined as the synchronization degree value. Specifically, based on the change direction sequence obtained in the previous step, a comparison is performed on a sampling-by-sampling-moment basis for any two operating parameters. For example, selecting the current parameter and the speed parameter as a group, the number of times their change directions are the same in 200 sampling moments is counted; if the current parameter changes in the positive direction and the speed parameter also changes in the positive direction in 120 sampling moments, or the current... If both the parameter change direction and the speed parameter change direction are negative, this is recorded as 120 directional synchronization events. If the total number of sampling points within the time window is 200, then the synchronization degree value is 120 / 200 = 0.60. For example, if there are 150 directional synchronization events between the voltage parameter and the speed parameter in 200 sampling moments, then the synchronization degree value is 150 / 200 = 0.75. By converting the number of directional synchronization events into a proportional value, the directional consistency between different operating stages and different parameter pairs can be directly quantified and compared, providing a parallel reference for the subsequent calculation of the deviation degree value.

[0033] For any two operating parameters, the absolute value of the difference between their changes at the same sampling time is taken, and then normalized. The difference between the values ​​at each sampling time is then accumulated within the time window to obtain the deviation value. Specifically, this can be implemented as follows: Based on any two selected operating parameters, for example, the current parameter and the speed parameter are grouped together. The changes of the two parameters are taken at each sampling time, and the absolute value of the difference is calculated. For example, if the change of the current parameter is +0.4 amperes and the change of the speed parameter is -13 revolutions per minute at the k-th sampling time, the absolute value of the difference is normalized according to the preset normalization benchmark to obtain 0.08. At the (k+1)th sampling time, the change in current parameter is +0.1 amperes and the change in rotational speed parameter is -5 revolutions per minute, which is normalized to 0.03. Within a 200-millisecond time window, the difference between the normalized values ​​at 200 sampling times is accumulated item by item. For example, if the accumulated difference is 6.4, then the deviation value is 6.4. Then, the voltage parameter and rotational speed parameter are grouped together. If the accumulated difference is 4.9, then the deviation value is 4.9. The deviation value is formed by accumulating the normalized differences, so that the cumulative deviation of the amplitude difference over time can reflect the strength of the offset of the coupling relationship between the operating parameters, and form a complementary characterization with the synchronization value.

[0034] By embedding synchronization and deviation values ​​into the phased dynamic baseline generation and update process, parameter pairs with low synchronization and high deviation values ​​are marked as enhanced coupling offset regions during the disturbance phase, while parameter pairs with high synchronization and low deviation values ​​are marked as stable coupling regions during the steady-state operation phase. For example, during the steady-state operation phase, the synchronization value of current and speed parameters can be stable above 0.80 and the deviation value below 3.0, while during the disturbance phase, the synchronization value may drop to 0.50 and the deviation value may rise to above 6.0. Through this method of backfilling quantification results in units of time windows, the interrelationships between operating parameters can be reproducibly characterized in different operating phases, providing a consistent data foundation and judgment basis for the subsequent calculation of transient abnormal event sequences, mutation energy indicators, and multi-parameter coupling offset indicators.

[0035] The formation of a transient abnormal event sequence includes the following steps: Under the constraint of a phased dynamic baseline, a continuous sampling time window is set for the operating parameters, and the maximum value of the operating parameter and the time point when the maximum value occurs are recorded within each time window as peak hold results and time stamp results. Specifically, this can be implemented as follows: Select any operating phase from the startup phase, steady-state operation phase, disturbance phase, and shutdown phase, while keeping the phased dynamic baseline unchanged. Set continuous sampling time windows for current parameters, voltage parameters, speed parameters, and temperature parameters. For example, if the sampling period is 1 millisecond and the continuous sampling time window is 20 milliseconds, then each time window contains 20 sampling points. Within each time window, search for the maximum value of the operating parameter and record the time point when the maximum value occurs. For example, if the current parameter has sampled values ​​of 18.2, 18.4, 18.1, 21.6, and 19.0 amperes within a certain time window, and reaches a maximum value of 21.6 amperes at the fourth sampling point, corresponding to a time point of 3 milliseconds after the start of the window, then the maximum value of 21.6 amperes is taken as the peak hold result, and the time point of 3 milliseconds is taken as the time stamp result. The voltage parameter has a maximum value of 382 volts within the same time window, and the time point is 5 milliseconds. The speed parameter has a maximum value of 2895 revolutions per minute, and the time point is 2 milliseconds. The temperature parameter has a maximum value of 52.35 degrees Celsius, and the time point is 19 milliseconds. Through this peak hold and time stamping based on time windows, abnormal peak values ​​with durations shorter than the normal sampling period are not diluted by averaging.

[0036] Based on the peak hold results, the absolute value of the difference between the mutation baseline value corresponding to the peak and the mutation peak value is determined as the mutation amplitude, and the time span from the mutation baseline value to the mutation peak interval and back to the mutation baseline value interval is taken as the mutation duration. Specifically, the mutation baseline value is determined by a preset time window before the peak hold result. The mutation baseline value is the value of the operating parameters when the operating parameters are in a stable change state under the staged dynamic baseline constraint before the transient anomaly occurs. This value of the operating parameters is obtained by averaging or taking the median value of the operating parameters within the preset time window before the peak occurs. For example, the average value of the current parameter within 10 milliseconds before the peak occurs is used as the reference value for the sudden change. If the average value is 18.0 amps and the peak value is 21.6 amps, then the absolute value of the difference, 3.6 amps, is used as the amplitude of the sudden change. At the same time, the duration of the sudden change is determined by the boundary between the current parameter entering the peak value range and returning to the reference value range. For example, if the current parameter rises from 18.0 amps to 19.2 amps 2 milliseconds before the peak occurs and falls back to 18.1 amps 6 milliseconds after the peak, then the time span from the start of entering the peak value range to the end of returning to the reference value range is 8 milliseconds, which is used as the duration of the sudden change. Similarly, if the reference value of the speed parameter is 2870 rpm and the peak value is 2895 rpm, then the amplitude of the sudden change is 25 rpm. If the time span between entering and returning is 6 milliseconds, then the duration of the sudden change is 6 milliseconds. By uniformly using the absolute value of the difference and the time span, it is ensured that the amplitude of the sudden change and the duration of the sudden change can be consistently expressed across different operating parameters.

[0037] Calculate the change in operating parameters per unit time from the reference value to the peak value within the duration of the mutation, as the mutation rise characteristic. Specifically, based on the obtained mutation amplitude and duration, determine the rise time from the reference value to the peak value. For example, if the current parameter rises from 18.0 amps to 21.6 amps in 2 milliseconds, the change per unit time is 3.6 amps. Dividing this by 0.002 seconds yields 1800 amps per second, which is used as the mutation rise characteristic. If the voltage parameter's reference value is... If the voltage is 380 volts, the peak voltage is 382 volts, and the rise time is 1 millisecond, then the rise time characteristic of the sudden change is 2000 volts per second. Temperature parameters usually have a slower rise time. For example, if the baseline value of the sudden change is 52.30 degrees Celsius, the peak value of the sudden change is 52.35 degrees Celsius, and the rise time is 20 milliseconds, then the rise time characteristic of the sudden change is 2.5 degrees Celsius per second. Through this change in unit time based on the rise time, the rise time characteristic of the sudden change can characterize the difference between rapid impact type and slow drift type, and provide a comparable velocity quantity for the subsequent discrimination of transient abnormal event sequences.

[0038] Binding the amplitude, duration, and rise characteristics of the sudden change in voltage (SPD) and sorting them by time stamps to form a sequence of transient abnormal events can be implemented as follows: using the time stamp results within each time window as the sorting key, the amplitude, duration, and rise characteristics of the sudden change in voltage obtained for the same operating parameter within consecutive time windows are bound as a transient abnormal event, and arranged in order of time stamp results from earliest to latest to form a sequence of transient abnormal events; for example, the current parameter forms transient abnormal events A, B, and C in three consecutive time windows, where event A has a time stamp of 3 milliseconds, a sudden change amplitude of 3.6 amperes, a sudden change duration of 8 milliseconds, and a sudden change rise characteristic of 1800 amperes. If event B is time-stamped at 26 milliseconds, with a sudden change amplitude of 2.1 amperes, a sudden change duration of 5 milliseconds, and a sudden change rise characteristic of 1400 amperes per second, and event C is time-stamped at 49 milliseconds, with a sudden change amplitude of 4.8 amperes, a sudden change duration of 10 milliseconds, and a sudden change rise characteristic of 1600 amperes per second, then the transient abnormal event sequence A→B→C can be obtained by sorting according to the time stamp. Similarly, transient abnormal event sequences can be formed for voltage parameters, speed parameters, and temperature parameters, so that transient abnormalities of each operating parameter on the same time axis can be completely recorded and traced, thereby meeting the purpose of "completely preserving abnormal changes with a duration shorter than the normal sampling period".

[0039] Determining the level of damage includes the following steps: For each transient abnormal event in the transient abnormal event sequence, calculate the mutation intensity value based on the numerical product of the mutation amplitude and the mutation duration. This value serves as a mutation energy index characterizing the transient impact intensity of a single operating parameter. Specifically, this can be implemented as follows: In the transient abnormal event sequence formed under staged dynamic baseline constraints, read the mutation amplitude and mutation duration corresponding to each transient abnormal event one by one, and directly perform numerical multiplication to obtain the mutation intensity value; for example, for a transient abnormal event of the current parameter, the mutation amplitude is 3.6 amperes and the mutation duration is 0.008 seconds, then the mutation intensity value is 0.0288 ampere-seconds and is used as the mutation energy index. For example, for a transient abnormal event in the speed parameter, if the amplitude of the sudden change is 25 revolutions per minute and the duration of the sudden change is 0.006 seconds, then the value of the sudden change intensity is 0.15 revolutions per minute per second, which is used as the sudden change energy index; if the amplitude of the voltage parameter sudden change is 40 volts and the duration of the sudden change is 0.010 seconds, then the value of the sudden change intensity is 0.40 volts per second; if the amplitude of the temperature parameter sudden change is 0.05 degrees Celsius and the duration of the sudden change is 0.020 seconds, then the value of the sudden change intensity is 0.001 degrees Celsius per second. By uniformly converting the amplitude and duration of the sudden change into the value of the sudden change intensity, the transient impact intensity of different operating parameters can be used in subsequent comparative analysis in the same form of "amplitude × time".

[0040] Within the time window corresponding to the duration of the mutation, the synchronization and deviation values ​​obtained from pairwise combinations of all operating parameters (current, speed, voltage, and temperature) are summarized to generate a multi-parameter coupling offset index characterizing the overall shift in the cooperative relationship of various operating parameters. Specifically, this can be implemented as follows: Taking the same transient abnormal event processed in the previous sub-step as the center, a time window corresponding to the duration of its mutation is selected. For example, if the mutation duration is 8 milliseconds, a time window covering this 8-millisecond period is selected. Within this time window, the synchronization degree is read for each of the six pairwise combinations: current and speed parameters, current and voltage parameters, current and temperature parameters, speed and voltage parameters, speed and temperature parameters, and voltage and temperature parameters. The numerical values ​​and deviation values ​​are collected and summarized to obtain a multi-parameter coupling offset index. For example, within this time window, the synchronization degree value of the current parameter and the speed parameter is 0.60 and the deviation value is 6.4, the synchronization degree value of the current parameter and the voltage parameter is 0.72 and the deviation value is 4.9, and the synchronization degree value of the speed parameter and the voltage parameter is 0.55 and the deviation value is 7.1. The corresponding values ​​are also obtained for other combinations. The synchronization degree value and deviation value of each combination are combined into a summary result according to a preset summarization method as the multi-parameter coupling offset index. This allows the multi-parameter coupling offset index to reflect the overall coordination relationship deviation of multiple operating parameters during the same transient abnormal event, and to form a parallel input with the mutation energy index at the same event granularity.

[0041] The mutation energy index and multi-parameter coupling offset index are compared with the preset impact feature reference range of the corresponding operational stage in the staged dynamic baseline. Based on the degree of difference between the mutation energy index and the multi-parameter coupling offset index in the numerical dimension, the preset impact feature level closest to the current transient anomaly is determined. Specifically, this can be implemented as follows: First, according to the operational stage corresponding to the occurrence time of the transient anomaly, the preset impact feature reference range of that operational stage is extracted from the staged dynamic baseline, and the mutation energy index and multi-parameter coupling offset index are simultaneously mapped to this reference range for numerical comparison. For example, in the disturbance stage, the preset impact feature reference range can include multiple impact features of different levels, and each different level of impact feature corresponds to a set of mutation energy index ranges. Given a set of multi-parameter coupled offset index ranges, if the mutation energy index of the current transient anomaly is 0.0288 ampere-seconds and the aggregated result of the multi-parameter coupled offset index falls near the center of a certain level of impact characteristic range, then the minimum numerical difference between the mutation energy index and the multi-parameter coupled offset index and the center of that level of impact characteristic range is used as the criterion to determine the preset impact characteristic level closest to the current transient anomaly. For example, in the steady-state operation phase, if the mutation energy index is much higher than the low-level impact characteristic range of the steady-state operation phase and the multi-parameter coupled offset index simultaneously falls into a higher-level impact characteristic range, then a higher-level preset impact characteristic level is determined, thereby achieving a judgment path that matches different levels of impact characteristics according to the degree of numerical difference within the same operation phase.

[0042] Based on the one-to-one mapping relationship between preset impact characteristic levels and damage severity levels, the damage severity level corresponding to the current transient abnormal event is determined. Specifically, after obtaining the preset impact characteristic level, the pre-established one-to-one mapping relationship between the preset impact characteristic level and the damage severity level is directly retrieved and the damage severity level is output, so that the damage severity level has stable interpretability and reproducibility. For example, the preset impact characteristic levels are divided into Level 1, Level 2, Level 3, and Level 4, and each corresponds one-to-one with the damage severity level. If the current transient abnormal event matches the Level 2 preset impact characteristic level, then the Level 2 damage severity level is output and used as the input for subsequent risk score generation. If it matches the Level 4 preset impact characteristic level, then the Level 4 damage severity level is output and assigned a higher base score in the risk quantification process. This allows the mutation energy index, multi-parameter coupling offset index, preset impact characteristic level, and damage severity level to form a continuous closed loop at the same transient abnormal event granularity, meeting the requirements for the classification and discrimination of transient abnormal damage of permanent magnet motor sewage pumps.

[0043] The logic for generating and accumulating risk points is as follows: Based on the level of damage, a corresponding base point is assigned to the transient abnormal event, and the over-limit amplitudes of the mutation energy index and the multi-parameter coupling offset index are successively added to the base point as corrections to obtain the risk point of the transient abnormal event. Specifically, after determining the level of damage corresponding to each transient abnormal event, a correspondence between the level of damage and the base point is established in advance. For example, a level 1 level of damage corresponds to a base point of 10, a level 2 level of damage corresponds to a base point of 30, a level 3 level of damage corresponds to a base point of 60, and a level 4 level of damage corresponds to a base point of 100.

[0044] After assigning a corresponding basic integral to the transient anomaly event based on the degree of damage, for the same transient anomaly event, the mutation energy index calculated by different operating parameters is compared with the preset impact characteristic reference range of the corresponding operating stage in the staged dynamic baseline. The over-limit magnitude of each mutation energy index relative to the upper limit of the corresponding reference range is calculated. When the mutation energy index corresponding to a certain operating parameter does not exceed the upper limit of the reference range, its over-limit magnitude is recorded as zero. When it exceeds the upper limit of the reference range, the excess part is taken as the over-limit magnitude corresponding to that operating parameter.

[0045] After obtaining the excess range of the mutation energy index corresponding to each operating parameter, the excess range of the mutation energy index of all operating parameters is converted into mutation energy correction amount according to the preset quantization ratio, and the mutation energy correction amount is accumulated to form the first type of correction amount used to characterize the cumulative effect of single-parameter transient shock.

[0046] For the multi-parameter coupled offset index formed by the pairwise combination of operating parameters in the same transient abnormal event, the coupled offset index corresponding to each operating parameter is compared with the preset impact characteristic reference range of the corresponding operating stage in the staged dynamic baseline. The over-limit magnitude of the coupled offset index corresponding to each operating parameter pair relative to the upper limit of the corresponding reference range is calculated. When the coupled offset index corresponding to a certain operating parameter pair does not exceed the upper limit of the reference range, its over-limit magnitude is recorded as zero. When it exceeds the upper limit of the reference range, the excess part is taken as the over-limit magnitude of the operating parameter pair.

[0047] After obtaining the over-limit amplitude of the coupling offset index corresponding to each pair of operating parameters, the over-limit amplitudes of the coupling offset index of all pairs of operating parameters are converted into coupling offset correction quantities according to a preset quantization ratio. These correction quantities are then accumulated to form a second type of correction quantity characterizing the cumulative effect of multi-parameter coordinated offset. The basic integral corresponding to the damage severity level is added to the first type of correction quantity formed by summing the over-limit amplitudes of the sudden energy index of each operating parameter, and to the second type of correction quantity formed by summing the over-limit amplitudes of the multi-parameter coupled offset index of each operating parameter. This yields the risk integral corresponding to the transient anomaly event, enabling the risk integral to simultaneously reflect the cumulative effect of differences in damage severity levels, the transient impact intensity of a single operating parameter, and the comprehensive impact of the coordinated offset degree of multiple operating parameters.

[0048] During operation, a sliding time window is set according to a preset runtime. Risk scores within the sliding time window are attenuated and then accumulated to obtain a cumulative risk value characterizing potential faults. Specifically, this can be implemented as follows: A preset runtime is determined based on the operating scenario of the permanent magnet motor sewage pump, and a sliding time window is set. For example, the preset runtime is 30 minutes, updated in 1-minute increments, ensuring each update covers all transient abnormal event risk scores occurring within the last 30 minutes. Within this sliding time window, the risk score of each transient abnormal event is included in the same accumulation process, forming a cumulative risk value characterizing potential faults. For example, if 8 transient abnormal events are recorded in the last 30 minutes, with risk scores of 18, 22, 60, 15, 35, 40, 12, and 28 respectively, the accumulated risk value after attenuation is 210. By continuously updating the cumulative risk value, low-frequency but high-risk transient abnormal events can form a traceable risk accumulation effect over time.

[0049] The cumulative risk value is compared with the trigger threshold corresponding to the damage level, and an early warning message is generated when the cumulative risk value reaches the trigger threshold. Simultaneously, the trigger time and trigger stage are recorded. Specifically, this can be implemented by pre-establishing trigger thresholds corresponding to different damage levels. These thresholds are consistent with a preset runtime and are used to constrain the sensitivity of the early warning. For example, a trigger threshold of 120 points corresponds to a level 1 damage level, 180 points to a level 2 damage level, 240 points to a level 3 damage level, and 300 points to a level 4 damage level. When the cumulative risk value is 210 points, if the highest damage level in the current window is level 2... For each level of damage, a threshold of 180 points corresponding to the second-level damage level is selected for comparison. If the cumulative risk value reaches the threshold, an early warning message is generated, and the trigger time (e.g., 14:32 on December 17, 2025) and the trigger stage (e.g., the disturbance stage) are recorded simultaneously. If the cumulative risk value is 150 points and the highest damage level in the current window is the second-level damage level, then the cumulative risk value has not reached the threshold and no early warning message is generated. Instead, the cumulative risk value is updated and the evaluation continues. By outputting the warning message in groups of "cumulative risk value - trigger threshold - trigger time - trigger stage", the early warning message has clear semantics regarding time and operational stage.

[0050] After generating early warning information, the warning information is associated with the trigger time and trigger stage, saved, and output. This information is used to support the decision-making input for generating corresponding operational and handling instructions based on risk scores and damage levels. This allows the results of risk quantification to be directly linked to the operational and handling process. For example, when the early warning information is triggered during the disturbance stage and the cumulative risk value is 210 points, and the window contains a transient abnormal event with a risk score of 60 points, priority can be given to reducing the operating load or adjusting the operating rhythm in the subsequent operational and handling instruction generation. When the early warning information is triggered during the steady-state operation stage and the cumulative risk value continues to rise, priority can be given to arranging equipment inspection and maintenance in the subsequent operational and handling instruction generation. By outputting the early warning information along with the trigger time and trigger stage, the potential fault hazards of the permanent magnet motor sewage pump have a closed-loop basis that is locatable, traceable, and manageable.

[0051] The trigger threshold is set based on distinguishing the cumulative risk level of the permanent magnet motor sewage pump during the preset operating time range from the risk characteristics of known destructive transient abnormal events. By statistically analyzing the long-term distribution of risk points in each operating stage under the constraints of the phased dynamic baseline, the upper limit of the risk points when no substantial damage occurs is determined. Combined with the risk amplification effect corresponding to different levels of damage, a critical value that can be triggered before the risk points deviate significantly from the normal accumulation range but before irreversible damage occurs is selected as the trigger threshold. This allows the trigger threshold to effectively suppress false alarms caused by normal fluctuations and low-risk anomalies, and to provide early warning of potential fault hazards that accumulate continuously.

[0052] The process of accumulating risk integrals after attenuation within a sliding time window refers to assigning a time weight to each risk integral based on the chronological order of occurrence of transient anomalies within the sliding time window. Risk integrals closer to the current time have a higher time weight. This can be implemented by determining a preset running time and forming a sliding time window accordingly (e.g., a 30-minute window updated in 1-minute increments), arranging transient anomalies within the sliding time window in chronological order, and marking the time interval between their occurrence and the current time. For example, if the current time is 14:32 on December 17, 2025, and the sliding time window contains 1... The four transient anomalies that occurred at 4:31, 14:28, 14:20, and 14:05 correspond to risk scores of 40, 60, 22, and 18, respectively. A time weight is assigned to each risk score: a time weight of 1.00 for a 1-minute time interval, 0.85 for a 4-minute time interval, 0.55 for a 12-minute time interval, and 0.20 for a 27-minute time interval. This method of assigning time weights in descending order of time interval ensures that more recently occurring transient anomalies contribute more to the cumulative risk value, while the contribution of earlier events is naturally suppressed, consistent with the engineering principle that potential fault hazards decay over time.

[0053] The attenuated risk score is obtained by multiplying each risk score by its corresponding time weight. Specifically, after obtaining the time weight for each risk score, the values ​​are multiplied one by one to obtain the attenuated risk score. For example, a risk score of 40 points corresponds to a time weight of 1.00, so the attenuated risk score is 40 points; a risk score of 60 points corresponds to a time weight of 0.85, so the attenuated risk score is 51 points; a risk score of 22 points corresponds to a time weight of 0.55, so the attenuated risk score is 12.1 points; a risk score of 18 points corresponds to a time weight of 0.20, so the attenuated risk score is 3.6 points. Furthermore, when the sliding time window contains more transient abnormal events, the attenuated risk scores are still obtained item by item using the unified method of "risk score × time weight," ensuring consistent reproducibility of the attenuation process and providing an input sequence with the same dimensions for subsequent summarization within the sliding time window.

[0054] The cumulative risk value is obtained by summing all attenuated risk scores within the sliding time window. Specifically, after obtaining the attenuated risk score for each item within the sliding time window, the cumulative risk value is obtained by summing the scores in chronological or arbitrary order. For example, summing the attenuated risk scores of 40, 51, 12.1, and 3.6 points yields a cumulative risk value of 106.7 points. Another example is a transient anomaly event occurring at 14:32 within the sliding time window, with a risk score of 35 points and a time weight of 1.00. The attenuated risk score is then 35 points, and the cumulative risk value is updated to 141.7 points. Simultaneously, when the sliding time window updates and removes transient anomalies occurring earlier than 30 minutes, the risk score and time weight corresponding to that event are no longer included in the summation. The cumulative risk value naturally decreases with the window update, ensuring that the cumulative risk value always represents the "recent risk accumulation level within the preset runtime range" and maintains consistency in time scale with subsequent trigger thresholds.

[0055] The cumulative risk value is used as the core output of risk quantification. It is compared with the trigger threshold corresponding to the level of damage and used to generate early warning information. At the same time, the trigger time and trigger stage are recorded, so that the time weight, the decayed risk score and the cumulative risk value form a continuous chain from transient abnormal events to early warning information. For example, when the cumulative risk value rises from 106.7 points to 141.7 points and the trigger threshold for the corresponding level of damage is 120 points, the cumulative risk value reaches the trigger threshold and an early warning information is generated. The trigger time is recorded as 14:32 on December 17, 2025 and the trigger stage is recorded as the disturbance stage. This allows the decay process, which states that "the closer the time is to the current moment, the greater the time weight of the risk score", to play a direct role in the early warning judgment. This avoids false alarms caused by the accumulation of low-risk events in the long term and ensures that high-risk events in the near future have sufficient sensitivity to early warning information.

[0056] In designs where the risk integral, closer to the current moment, corresponds to a larger time weight, the expression of time weight can draw on the common ideas in existing technologies for modeling the degree of time decay and the influence of historical data. Common and mature expressions mainly include the following categories: The first type is the linear mapping representation. This method uses the start point of the sliding time window and the current time as two endpoints, mapping the time interval to a weight value that changes linearly between a preset maximum and minimum weight. For example, when the transient abnormal event occurs at the same time as the current time, the corresponding time weight is at its maximum value; when the transient abnormal event occurs at the start point of the sliding time window, the corresponding time weight is at its minimum value. Transient abnormal events in between receive a time weight that decreases linearly according to the time interval ratio. This method is intuitive to calculate, has few parameters, and is suitable for operational scenarios where the impact of risk is expected to decay uniformly over time, facilitating engineering implementation and debugging.

[0057] The second type is the piecewise linear expression method. This method divides the sliding time window into multiple time intervals, using different linear change rates in different time intervals. For example, in the first few minutes close to the current moment, the time weight remains at a high level and decreases slowly, while in subsequent time intervals far from the current moment, the time weight decreases rapidly until it approaches zero. This method can highlight the impact of recent transient anomalies while accelerating the reduction of the impact of earlier events, making it suitable for operating environments where the impact of transient shocks on equipment has a significant "short-term sensitivity".

[0058] The third type is the exponential decay representation. This method calculates the time weight using an exponential function based on the time interval between the occurrence of the transient event and the current moment, causing the time weight to decay rapidly and non-linearly as the time interval increases. For example, the weight decreases slowly when the time interval is short, but rapidly approaches zero as the time interval increases. This method is widely used in reliability assessment and aging effect modeling, and is suitable for scenarios where it is desirable to highlight the most recent risks and quickly ignore historical low-impact events.

[0059] The fourth type is the tiered weighting method. This method does not continuously change the time weights, but instead divides the sliding time window into several time periods and assigns a fixed weight to each time period. For example, the risk score within the most recent 5 minutes is given the highest weight, the score within 5 to 15 minutes is given a medium weight, and the score within 15 to 30 minutes is given a lower weight. This method has a clear structure and strong interpretability, and is suitable for application scenarios that require clear risk level intervals and are easy for human understanding and operational decision-making.

[0060] In practical implementation, the application of any of the above time weighting methods can be selected based on the operating characteristics of the permanent magnet motor sewage pump, the time sensitivity of transient abnormalities to equipment damage, and the early warning response requirements. This allows for flexible control and engineering adaptation of risk integral decay behavior while ensuring that the basic principle of "the closer the time is to the current moment, the greater the time weight of the risk integral" remains unchanged.

[0061] After issuing the early warning information, corresponding operational instructions are generated based on the risk score and the level of damage. In practice, the operational instructions are generated according to preset standards. For example: Operational handling instructions for medium-risk, controllable disturbance scenarios: When the cumulative risk value corresponding to the output warning information is in the medium range, and the highest level of damage corresponding to the transient abnormal event in the current window is level two damage, the generated operational handling instruction is to limit the operating load and strengthen monitoring. Specifically, without stopping the operation of the permanent magnet motor sewage pump, the operating speed is reduced or the instantaneous start frequency is limited, so that the variation amplitude of current parameters and speed parameters falls back to the allowable range of the phased dynamic baseline. At the same time, the acquisition frequency of operating parameters is increased to continuously track the changing trend of risk integral. This operational handling instruction is applicable to scenarios that are judged to be short-term disturbances but have not yet caused significant damage to the mechanical structure and permanent magnets, and can achieve risk mitigation without affecting normal drainage function.

[0062] Operational handling instructions for high-risk, potential structural damage scenarios: When the cumulative risk value corresponding to the output warning information significantly exceeds the trigger threshold, and the highest damage level corresponding to the transient abnormal event is level three, the generated operational handling instruction is to execute controlled shutdown and arrange inspection. Specifically, after completing the minimum necessary operating cycle in the current operating phase, the permanent magnet motor sewage pump is controlled to enter the shutdown phase, so that the speed parameters and current parameters decrease smoothly according to the shutdown slope boundary specified by the phased dynamic baseline to avoid secondary impact. At the same time, the sudden energy index, multi-parameter coupling offset index, and risk integral changes during the time period of triggering the warning are recorded as important basis for subsequent inspection of permanent magnet demagnetization, rotor impact deformation, or impeller jamming. This operational handling instruction is applicable to preventive intervention scenarios where obvious impact characteristics have appeared, but catastrophic failure has not yet occurred.

[0063] Operational handling instructions for extremely high-risk scenarios that may lead to catastrophic failures: When the cumulative risk value corresponding to the output warning information continues to rise and far exceeds the highest level trigger threshold, and the damage level corresponding to the transient abnormal event is level four, the generated operational handling instruction is to immediately execute a safe shutdown and prohibit restarting; specifically, the operation of the permanent magnet motor sewage pump is directly interrupted, causing it to enter the emergency shutdown process, while locking the current operating parameter status, prohibiting it from re-entering the startup phase before maintenance confirmation is completed; this operational handling instruction is mainly used to identify situations with strong impacts, high coupling offsets, and multiple rapid accumulations of risk integrals, in order to prevent irreversible consequences such as severe demagnetization of the permanent magnet, rotor breakage, or pump body structural damage.

[0064] Operational handling instructions for low-risk, early-stage hazard warning scenarios: When the cumulative risk value corresponding to the output warning information just reaches the minimum trigger threshold, and the highest level of damage corresponding to the transient abnormal event is Level 1 damage, the generated operational handling instruction is to maintain the current operating status and prompt maintenance attention; specifically, the operating parameters of the permanent magnet motor sewage pump are not actively adjusted, but the corresponding time period and operating stage are marked in the operation record, and in subsequent operations, priority is given to whether similar transient abnormal events recur in the same stage; this operational handling instruction is used to track and observe early, low-intensity but recurring abnormal patterns without affecting normal operation.

[0065] It should be noted that the operation and disposal instructions are not single actions, but are generated differently based on the combination of risk score level and damage level. Furthermore, the operation and disposal instructions are generated according to corresponding standards, which can refer to the safety operation and management regulations and instruction manual of permanent magnet motor sewage pumps, so that the operation and management of permanent magnet motor sewage pumps can achieve a balance between safety, continuity and maintenance costs.

[0066] The above algorithms or formulas are all dimensionless and numerical calculations, and the results are obtained by software simulation based on a large amount of collected data to obtain the most recent real-world results. The preset parameters are set by those skilled in the art according to the actual situation.

[0067] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0068] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0069] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0070] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A fault detection method for permanent magnet motor sewage pumps, characterized in that, Includes the following steps: The operation process of the permanent magnet motor sewage pump is divided into stages, and the operating parameters of the permanent magnet motor sewage pump are collected in each stage. Based on the relationship between the changes in the operating parameters in each stage, dynamic response curves and change slope boundaries corresponding to each stage are established to form a staged dynamic baseline for subsequent fault detection. Under the staged dynamic baseline constraint, the instantaneous mutations of the operating parameters during the sampling process are processed by peak preservation and time stamping, so that abnormal changes with a duration shorter than the normal sampling period are fully preserved, forming a transient abnormal event sequence containing mutation amplitude, mutation duration and mutation rise characteristics. Based on the formed transient abnormal event sequence, the corresponding mutation energy index and multi-parameter coupling offset index are calculated for each transient abnormal event. The calculation results are then correlated with the staged dynamic baseline to determine whether the transient abnormal events exhibit different preset levels of impact characteristics, and the degree of damage caused by the transient abnormal events to the permanent magnet motor sewage pump is determined accordingly. The transient abnormal events are quantified according to the degree of damage. A corresponding risk score is generated for each transient abnormal event. The risk score is continuously accumulated during operation. When the risk score reaches the corresponding trigger condition within the preset running time range, an early warning message is output that there is a potential fault in the permanent magnet motor sewage pump. After issuing the early warning information, corresponding operational and handling instructions are generated based on the risk score and the level of damage.

2. The fault detection method for a permanent magnet motor sewage pump according to claim 1, characterized in that, The operation process includes the startup phase, steady-state operation phase, disturbance phase, and shutdown phase. The operating parameters include at least current parameters, voltage parameters, speed parameters, and temperature parameters.

3. The fault detection method for a permanent magnet motor sewage pump according to claim 2, characterized in that, A phased dynamic baseline is generated through the following steps: In each operating phase, a time series of continuously collected operating parameters is constructed. The direction and magnitude of change of the operating parameters are calculated by the numerical difference between adjacent sampling points to determine the time series change trend of the operating parameters in that operating phase. Based on the determined trend of change, the change in the operating parameters per unit time is calculated based on the numerical difference results of the change of operating parameters over time, and the slope of the change of operating parameters in this operating stage is obtained, and a slope curve reflecting the rate of change of operating status is formed accordingly. The slopes of change of different operating parameters at the same sampling time point are time-aligned. The consistency of the direction of change and the difference in the magnitude of change of each operating parameter within the same time window are compared to obtain the degree of synchronization and deviation between operating parameters in the time dimension, which is used as the correlation between operating parameters. Based on the changing trend, the changing slope and the interrelationship, dynamic response curves for the corresponding operating stages are generated. Based on the dynamic response curves, the upper limit boundary and the lower limit boundary of the changing slope are determined to form a staged dynamic baseline for subsequent fault detection.

4. The fault detection method for a permanent magnet motor sewage pump according to claim 3, characterized in that, The degree of synchronization and deviation between operating parameters is calculated as follows: Within the same time window, the change between adjacent sampling points of each running parameter is calculated, and the positive or negative attribute of the change is used to characterize the direction of change; When any two operating parameters change in the same direction at the same sampling time, it is recorded as a direction synchronization event. The ratio of the number of direction synchronization events to the total number of sampling points in the time window is determined as the synchronization degree value. For any two operating parameters, take the absolute value of the difference between their changes at the same sampling time, normalize it, and then sum the differences between the values ​​at each sampling time within the time window to obtain the deviation value.

5. A fault detection method for a permanent magnet motor sewage pump according to claim 4, characterized in that, The formation of a transient abnormal event sequence includes the following steps: Under the staged dynamic baseline constraint, a continuous sampling time window is set for the operating parameters, and the maximum value of the operating parameters and the time point when the maximum value occurs are recorded in each time window as peak hold results and time stamp results; Based on the peak hold results, the absolute value of the difference between the mutation baseline value corresponding to the peak value and the mutation peak value is determined as the mutation amplitude, and the time span from the mutation baseline value to the mutation peak value interval and back to the mutation baseline value interval is taken as the mutation duration. The change in operating parameters from the mutation baseline to the mutation peak within the mutation duration is calculated as the mutation rise characteristic. The mutation amplitude, mutation duration, and mutation rise characteristics are bound together and sorted by time markers to form a transient abnormal event sequence.

6. A fault detection method for a permanent magnet motor sewage pump according to claim 5, characterized in that, The mutation baseline value is the value of the operating parameters when the operating parameters are in a stable state of change under the staged dynamic baseline constraint before the transient anomaly occurs. This value of the operating parameters is obtained by averaging or taking the median value of the operating parameters within a preset time window before the peak occurs.

7. A fault detection method for a permanent magnet motor sewage pump according to claim 5, characterized in that, Determining the level of damage includes the following steps: For each transient abnormal event in the transient abnormal event sequence, the mutation intensity value is calculated based on the numerical product of the mutation amplitude and the mutation duration, which serves as a mutation energy index characterizing the transient impact intensity of a single operating parameter. Within the time window corresponding to the duration of the mutation, the synchronization degree and deviation degree values ​​obtained by combining all operating parameters in pairs for current, speed, voltage and temperature parameters are summarized to generate a multi-parameter coupling offset index to characterize the degree of deviation of the overall coordination relationship of multiple operating parameters. The mutation energy index and the multi-parameter coupling offset index are compared with the preset impact characteristic reference range of the corresponding operating stage in the phased dynamic baseline. Based on the degree of difference between the mutation energy index and the multi-parameter coupling offset index in the numerical dimension, the preset impact characteristic level that is closest to the current transient abnormal event is determined. Based on the one-to-one mapping relationship between the preset impact characteristic level and the damage level, the damage level corresponding to the current transient abnormal event is determined.

8. A fault detection method for a permanent magnet motor sewage pump according to claim 7, characterized in that, The logic for generating and accumulating risk points is as follows: The transient anomaly event is assigned a corresponding basic integral based on the degree of damage. The over-limit amplitudes of the mutation energy index and the multi-parameter coupling offset index are then added to the basic integral as corrections to obtain the risk integral of the transient anomaly event. During operation, a sliding time window is set according to a preset running time. Each risk component within the sliding time window is attenuated and then accumulated to obtain a cumulative risk value used to characterize potential fault hazards. The cumulative risk value is compared with the trigger threshold corresponding to the level of damage. When the cumulative risk value reaches the trigger threshold, an early warning message is generated, and the trigger time and trigger stage are recorded.

9. A fault detection method for a permanent magnet motor sewage pump according to claim 8, characterized in that, The cumulative effect after attenuation processing of each risk integral within the sliding time window refers to: Within the sliding time window, each risk score is assigned a time weight corresponding to its occurrence time according to the order in which transient abnormal events occur. Each risk integral is multiplied by its corresponding time weight to obtain the decayed risk integral. The cumulative risk value is obtained by summing all the decayed risk integrals within the sliding time window.

Images