Water pump intelligent monitoring control system and application method thereof

Abstract

The invention provides a water pump intelligent monitoring control system and an application method thereof. In the method, a control system firstly extracts a mechanical noise component corresponding to a real-time rotating speed from initial baseline data by utilizing the relevance between the rotating speed and mechanical noise, and then constructs a baseline noise template to obtain mechanical noise interference during normal operation of a water pump. And the control system subtracts the baseline noise template from the original acoustic vibration signal to obtain a residual signal containing impurity impact information. And then the control system carries out continuous wavelet transform on the residual signal, and a generated time-frequency energy distribution diagram can identify the time node and frequency characteristics of the impact event so as to realize positioning of the impact event. And finally, the control system calculates the impact event rate and the average impact energy value in a preset statistical time window, and quantifies the progressive accumulation process of blockage. According to the method, the technical problem that early blockage cannot be identified by fixed threshold alarm is relieved, and the timeliness of fault prediction is improved.

Description

Technical Field

[0001] This application relates to the field of mechanical equipment testing technology, and in particular to an intelligent monitoring and control system for water pumps and its application methods. Background Technology

[0002] As a core piece of equipment for fluid transport, water pumps are widely used in industrial production, municipal water supply, agricultural irrigation, and other fields. During long-term operation, water pumps are prone to blockage due to impurities, particles, or foreign objects in the medium, leading to decreased equipment efficiency, increased energy consumption, and even equipment damage and production interruption.

[0003] In related technologies, pump failure prediction mainly employs regular maintenance and threshold alarms. Specifically, by installing monitoring devices such as pressure sensors and flow sensors on the pump, the pump's operating parameters are collected in real time. When a parameter exceeds a preset fixed threshold, an alarm signal is triggered, alerting the operator that the pump may be malfunctioning.

[0004] However, pump blockage is a gradual, cumulative process. In the early stages of blockage, changes in operating parameters are often minor and slow. Fixed threshold alarm mechanisms rely on parameters reaching preset critical values ​​to trigger an alarm. During the slow development of blockage, parameter changes remain below the threshold for an extended period, failing to trigger an alarm until the blockage accumulates to a certain extent, causing a sudden parameter change. For equipment like deep well pumps, which have long maintenance cycles and complex preparation processes, the time required from receiving an alarm signal to completing maintenance preparations and cleaning often exceeds the time needed. By this time, the pump may have already overloaded due to severe blockage, missing the optimal maintenance window. Summary of the Invention

[0005] This application provides an intelligent monitoring and control system for water pumps and its application method, which alleviates the technical problem of delayed early warning of water pump blockage and improves the timeliness of fault prediction.

[0006] In a first aspect, this application provides an intelligent monitoring and control method for a water pump, comprising: acquiring acoustic vibration signals and real-time speed signals during the operation of the water pump to be monitored, wherein the acoustic vibration signals represent time-series data of sound and vibration generated by the water pump to be monitored; extracting mechanical noise components corresponding to the real-time speed signals from the initial baseline data of the water pump to be monitored, and constructing a baseline noise template based on the mechanical noise components, wherein the initial baseline data is the initial acoustic vibration data when the water pump is not blocked; subtracting the baseline noise template from the acoustic vibration signals to obtain residual signals; performing continuous wavelet transform on the residual signals to generate a time-frequency energy distribution map, wherein the time-frequency energy distribution map presents the distribution of signal energy in the time and frequency dimensions; determining, in the time-frequency energy distribution map, that an impact event has occurred in the region where the signal energy value exceeds the energy threshold and the duration is less than the duration threshold; calculating the impact event rate based on the total number of impact events within a preset statistical time window, and calculating the average impact energy value based on the sum of the energy values ​​of all impact events; determining the blockage risk level of the water pump to be monitored based on the impact event rate and the average impact energy value, and generating a corresponding early warning signal.

[0007] By adopting the above technical solution, the control system first utilizes the correlation between rotational speed and mechanical noise to extract the mechanical noise component corresponding to the real-time rotational speed from the initial baseline data, and then constructs a baseline noise template to obtain the mechanical noise interference during normal pump operation. The control system then subtracts the baseline noise template from the original acoustic vibration signal to obtain a residual signal mainly containing impurity impact information. Subsequently, the control system performs continuous wavelet transform on the residual signal, and the generated time-frequency energy distribution map can identify the time nodes and frequency characteristics of impact events, thus achieving the location of impact events. Finally, within a preset statistical time window, the control system calculates the impact event rate and average impact energy value to quantify the gradual accumulation process of blockage. This method alleviates the technical problem that fixed threshold alarms cannot identify early blockages and improves the timeliness of fault prediction.

[0008] In conjunction with some embodiments of the first aspect, in some embodiments, mechanical noise components corresponding to real-time speed signals are extracted from the initial baseline data of the pump to be monitored, and a baseline noise template is constructed based on the mechanical noise components. Specifically, this includes: finding the two speed signals with the smallest absolute difference from the real-time speed signal in the speed-noise characteristic baseline database, the speed-noise characteristic baseline database being established based on the initial baseline data of the pump to be monitored; performing linear interpolation on the mechanical noise components corresponding to the two speed signals with the smallest absolute difference from the real-time speed signal to obtain the real-time mechanical noise components corresponding to the real-time speed signal; and converting the real-time mechanical noise components into time-domain signals to obtain the baseline noise template.

[0009] By adopting the above technical solution, the control system generates mechanical noise components adapted to the real-time speed based on a speed-noise characteristic baseline database and using a linear interpolation method. This method can accurately match the dynamic changes in pump speed during operation, alleviate the residual normal noise caused by speed fluctuations, and improve the separation purity of abnormal impact signals.

[0010] In conjunction with some embodiments of the first aspect, in some embodiments, the establishment of the speed-noise characteristic baseline database specifically includes: acquiring the acoustic vibration signal and speed signal of the water pump under test in its initial state, operating at variable speed within the speed range, and generating initial baseline data based on the acoustic vibration signal and speed signal, wherein the initial state is an unblocked state; dividing the speed range into multiple continuous speed intervals; for each speed interval, extracting the current acoustic vibration signal and the corresponding speed signal within the current speed interval, and performing order analysis on the current acoustic vibration signal to calculate the average order spectrum within the current speed interval, obtaining the mechanical noise component corresponding to the current speed interval, wherein the mechanical noise component represents the energy distribution of the mechanical vibration noise corresponding to the current speed interval at each order; determining the reference speed of each speed interval based on the mechanical noise component of each speed interval; establishing a mapping relationship based on each speed interval and the corresponding mechanical noise component, and constructing the speed-noise characteristic baseline database based on the mapping relationship.

[0011] By adopting the above technical solution, the control system first collects variable-speed operation data of the water pump in its initial state, covering the entire speed range. Then, the speed is divided into continuous intervals, and order analysis is performed on each interval to extract stable mechanical noise components. Finally, a mapping relationship between speed intervals and mechanical noise components is established. This method ensures the baseline data covers the entire speed operating condition and simplifies the subsequent real-time matching search logic through interval division, alleviating the contradiction between baseline data adaptability and search efficiency.

[0012] In conjunction with some embodiments of the first aspect, in some embodiments, in the time-frequency energy distribution map, determining that an impact event has occurred in a region where the signal energy value exceeds an energy threshold and the duration is less than a duration threshold specifically includes: in the time-frequency energy distribution map, marking pixels where the signal energy value exceeds the energy threshold and the duration is less than the duration threshold as candidate regions, and merging pixels of spatially adjacent candidate regions into energy patches; calculating morphological feature parameters for each energy patch, including time duration, frequency bandwidth, energy concentration, and vertical morphological index, where energy concentration is the ratio of peak energy to average energy within the energy patch, and the vertical morphological index is determined by calculating the aspect ratio of the minimum bounding rectangle of the energy patch; scoring each energy patch based on the morphological feature parameters, and identifying energy patches with scores greater than a score threshold as impact events.

[0013] By adopting the above technical solution, the control system first preliminarily screens candidate regions based on energy and duration thresholds, then merges adjacent pixels to form energy patches, and finally scores and filters them using multi-dimensional morphological feature parameters (duration of time, energy concentration, etc.). This method utilizes the morphological differences between impact signals and interference signals to effectively mitigate interference from environmental noise and electrical noise, reduce the probability of false alarms and missed alarms, and improve the accuracy of impact event recognition.

[0014] In conjunction with some embodiments of the first aspect, in some embodiments, the blockage risk level of the water pump to be monitored is determined based on the impact event rate and average impact energy, and a corresponding early warning signal is generated. Specifically, this includes: extracting the energy distribution of each impact event in the time-frequency energy distribution map, and calculating the spectral centroid and peak energy of the energy distribution, where the spectral centroid represents the weighted average center of the impact energy on the frequency axis, and the peak energy represents the maximum energy value of the impact event; calculating the average spectral centroid and average impact energy of all impact events within a preset statistical time window based on the spectral centroid and peak energy of the energy distribution of each impact event; determining the blockage type based on the comparison results of the average spectral centroid and the preset centroid threshold, and the comparison results of the average impact energy and the preset energy threshold, whereby the blockage type includes hard materials, soft materials, and mixed materials; and determining the blockage risk level of the water pump to be monitored based on the blockage type, impact event rate, and average impact energy value, and generating a corresponding early warning signal.

[0015] By adopting the above technical solution, the control system incorporates spectral centroid parameters, combining peak energy and impact event rate to achieve a dual assessment of "blockage type" and "blockage frequency." The spectral centroid can accurately distinguish between hard, soft, and mixed-type blockages based on frequency distribution characteristics, clearly defining the damage risk level of the blockage. This is further combined with the impact event rate to quantify the blockage accumulation rate. This method makes the classification of blockage risk levels more closely aligned with actual hazard characteristics, improving the precision of fault prediction.

[0016] In conjunction with some embodiments of the first aspect, in some embodiments, the blockage risk level of the monitored water pump is determined based on the type of blockage and the impact event rate, and a corresponding early warning signal is generated. Specifically, this includes: calculating the blockage accumulation index of the monitored water pump based on the impact event rate and the average impact energy value; obtaining the real-time operating parameters of the monitored water pump and calculating the actual performance parameters of the monitored water pump based on the real-time operating parameters, including inlet pressure, outlet pressure, flow rate, and motor power; calculating the performance deviation index between the actual performance parameters and the baseline performance parameters of the monitored water pump at the same speed; and determining the final blockage risk level of the monitored water pump based on the blockage accumulation index, the performance deviation index, and the type of blockage, and generating a corresponding early warning signal.

[0017] By adopting the above technical solution, the control system quantifies the accumulation of blockage at the signal level using a blockage accumulation index, verifies the actual operational differences at the performance level using a performance deviation index, and then clarifies the degree of harm by combining the type of blockage. This method alleviates misjudgments such as "high impact rate but no performance degradation" or "performance degradation but not caused by blockage," and improves the reliability of early warning signals.

[0018] In conjunction with some embodiments of the first aspect, in some embodiments, the final clogging risk level of the pump under monitoring is determined based on the clogging accumulation index, performance deviation index, and clogging type. Specifically, this includes: performing moving average processing on the time series data of the clogging accumulation index and the time series data of the performance deviation index to obtain smoothed time series data; performing linear regression fitting on the smoothed time series data within a preset fitting time window to obtain a first fitted line for the clogging accumulation index and a second fitted line for the performance deviation index; extracting the slope of the first fitted line as the first future change slope of the clogging accumulation index, and extracting the slope of the second fitted line as the second future change slope of the performance deviation index; when the first future change slope or the second future change slope exceeds a slope threshold, determining the final clogging risk level of the pump under monitoring based on the percentage by which the first future change slope or the second future change slope exceeds the slope threshold.

[0019] By adopting the above technical solution, the control system first performs moving average noise reduction on the time series of the congestion accumulation index and performance deviation index, and then extracts the future change slope through linear regression fitting. The slope magnitude and the percentage exceeding the threshold are used to quantify the rate of congestion deterioration. This method predicts the congestion development trend in advance, allowing sufficient time for maintenance preparation, alleviating the technical problem of delayed early warning in gradual congestion control, and reducing the risk of equipment damage.

[0020] In a second aspect, this application provides a control system including one or more processors and a memory; the memory is coupled to the one or more processors and is used to store computer program code including computer instructions, which the one or more processors call to cause the control system to perform the methods described in the first aspect and any possible implementation thereof.

[0021] Thirdly, this application provides a computer-readable storage medium including instructions that, when executed on a control system, cause the control system to perform the method described in the first aspect and any possible implementation thereof.

[0022] Fourthly, this application provides a computer program product that, when run on a control system, causes the control system to perform the method described in the first aspect and any possible implementation thereof.

[0023] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: 1. By adopting the technical means of acquiring acoustic vibration signals and real-time speed signals by the control system, constructing a baseline noise template adapted to the real-time speed to separate abnormal impact signals, and identifying impact events and quantifying the impact event rate and average impact energy through continuous wavelet transform, the technical problem of delayed early warning of water pump blockage is effectively alleviated and the timeliness of fault prediction is improved.

[0024] 2. By employing the technical means of extracting the spectral centroid and peak energy of impact events from the control system to distinguish the type of blockage, and combining the impact event rate to construct a dual evaluation system of "blockage type" and "blockage frequency", the technical problem of fault warnings being able to only indicate abnormalities but not distinguish the characteristics of blockages and lacking specificity is effectively alleviated. This improves the precision of fault prediction and makes the warning signal more in line with the actual hazard characteristics of the blockage.

[0025] 3. By adopting a three-dimensional evaluation system that includes a blockage accumulation index, a performance deviation index, and a blockage type, the technical problem that a single indicator evaluation cannot fully reflect the actual impact of blockages is effectively alleviated, and the reliability of the early warning signal is improved. Attached Figure Description

[0026] Figure 1 This is a flowchart illustrating a pump failure prediction method in an embodiment of this application. Figure 2 This is another flowchart illustrating the pump failure prediction method in the embodiments of this application; Figure 3 This is a schematic diagram of the hardware structure of the control system in an embodiment of this application. Detailed Implementation

[0027] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification and appended claims of this application, the singular expressions “a,” “an,” “the,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to any or all possible combinations including one or more of the listed items.

[0028] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature, and in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0029] For ease of understanding, the method provided in this implementation is described in process below. Please refer to [link / reference]. Figure 1 This is a flowchart illustrating a pump failure prediction method in an embodiment of this application.

[0030] 101. Acquire acoustic vibration signals and real-time speed signals during the operation of the water pump to be monitored. The acoustic vibration signals represent the time series data of the sound and vibration generated by the water pump to be monitored.

[0031] The water pump to be monitored refers to the water pump equipment that needs to be predicted for failure; the acoustic vibration signal represents the sound and mechanical vibration generated by the water pump during operation, and is recorded in the form of time series data; the real-time speed signal refers to the rotational speed data of the water pump motor or rotor during operation, usually expressed in revolutions per minute, and is used to characterize the working status and load of the water pump; time series data refers to a series of values ​​collected in chronological order.

[0032] Specifically, the control system collects three types of signals through a sensor network connected to the pump under monitoring: first, acoustic vibration signals, acquired by accelerometers, acoustic sensors, or vibration sensors installed on the pump casing or pipeline; second, real-time speed signals, acquired by speed sensors, Hall effect sensors, or high-precision photoelectric encoders installed on the pump motor shaft, representing the instantaneous rotational speed of the pump rotor; and third, phase reference signals (key phase signals), typically acquired by key phase sensors or high-precision photoelectric encoders (e.g., one pulse per revolution). This signal provides a reference pulse synchronized with the rotor's angular position, marking the starting point of each rotor revolution. The control system controls these sensors to ensure high-precision synchronous acquisition of the acoustic vibration signals, real-time speed signals, and phase reference signals, or acquisition with a unified high-precision timestamp. The acoustic vibration signals collected by the control system include mechanical noise from normal pump operation, fluid flow noise, and abnormal signals generated by potential foreign object impacts or friction. The control system digitizes these raw signals to form time-series data.

[0033] 102. Extract the mechanical noise component corresponding to the real-time speed signal from the initial baseline data of the pump to be monitored, and construct a baseline noise template based on the mechanical noise component. The initial baseline data is the initial acoustic vibration data when the pump is not blocked.

[0034] Initial baseline data refers to the dataset of acoustic vibration signals collected by the water pump under normal, unblocked conditions; mechanical noise component represents the acoustic vibration signals caused by factors such as the movement of mechanical parts and fluid flow during normal operation of the water pump; baseline noise template refers to a reference template constructed based on the initial baseline data, used to characterize the normal acoustic vibration characteristics of the water pump at different speeds.

[0035] Specifically, the control system first retrieves the initial baseline data of the pump to be monitored from the database. This data consists of acoustic vibration data collected after the pump has been installed, commissioned, and is operating normally without blockages. The control system analyzes the relationship between the initial baseline data and the rotational speed to identify the mechanical noise characteristics related to the rotational speed. Since the mechanical noise of the pump exhibits a certain functional relationship with the rotational speed, the control system extracts the speed-dependent noise components from the initial baseline data using methods such as spectrum analysis and correlation analysis. These components include bearing noise, impeller rotation noise, and motor noise. Based on the extracted mechanical noise components (including amplitude spectrum and corresponding phase information), the control system constructs a parameterized baseline noise template. This template can generate the acoustic vibration characteristics (including specific amplitude and phase information) that the pump should exhibit during normal operation at the corresponding rotational speed, based on the input real-time rotational speed signal and its instantaneous phase information.

[0036] 103. Subtract the baseline noise template from the acoustic vibration signal to obtain the residual signal.

[0037] The residual signal refers to the difference between the actual acquired acoustic vibration signal and the baseline noise template, representing the deviation between the current operating state and the normal state of the water pump.

[0038] Specifically, the control system first generates a baseline noise template (essentially a time-domain waveform reflecting the normal operating state of the pump) from the baseline noise template based on the currently acquired real-time speed signal and the corresponding instantaneous phase information. The generated baseline noise template has strict time-domain synchronization with the real-time acquired acoustic vibration signal. Then, the control system performs point-to-point subtraction between the real-time acquired acoustic vibration signal and the baseline noise template generated after high-precision phase calibration. That is, for the sampled value at each time point, the template prediction value is subtracted from the actual acquired value to obtain the residual signal. This effectively eliminates the background noise generated by the normal operation of the pump, making abnormal signals (such as vibrations caused by foreign object impacts or friction) stand out in the residual signal.

[0039] 104. Perform continuous wavelet transform on the residual signal to generate a time-frequency energy distribution map, which shows the distribution of signal energy in the time and frequency dimensions.

[0040] Continuous wavelet transform is a time-frequency analysis method used to decompose a signal into wavelet coefficients at different time and frequency scales, which can simultaneously provide the local characteristics of the signal in the time and frequency domains; the time-frequency energy distribution map represents a two-dimensional image of how the signal energy is distributed on the time and frequency axes, and usually uses color depth or brightness to represent the energy magnitude; signal energy refers to the intensity of the signal at a specific time point and a specific frequency, reflecting the intensity of the vibration.

[0041] Specifically, the control system applies a continuous wavelet transform algorithm to the residual signal, selecting a suitable wavelet basis function (such as the Morlet wavelet or the Mexicanhat wavelet) to obtain the impact-type transient signal. The control system performs multi-scale decomposition of the residual signal by adjusting the wavelet scaling parameters, calculating wavelet coefficients at different time points and frequencies. The squares or absolute values ​​of these wavelet coefficients represent the energy of the signal at the corresponding time-frequency point. The control system maps the calculated energy values ​​onto a two-dimensional plane, forming a time-frequency energy distribution map, where the horizontal axis represents time, the vertical axis represents frequency, and the color intensity (or brightness) represents the energy magnitude.

[0042] 105. In the time-frequency energy distribution map, it is determined that an impact event occurred in the region where the signal energy value exceeds the energy threshold and the duration is less than the duration threshold.

[0043] The energy threshold is a reference value used to determine whether the signal energy is abnormal. It is usually set based on the statistical characteristics under normal operating conditions. The duration threshold represents the maximum time limit for the abnormal signal energy to continue, and is used to distinguish between short-term impact events and continuous abnormalities. An impact event refers to a short-term high-energy vibration phenomenon caused by the collision of foreign objects inside the water pump with components such as the impeller and pump wall.

[0044] Specifically, the control system first determines an appropriate energy threshold based on the statistical characteristics of energy distribution under normal pump operation. This threshold is typically set as the mean of normal background energy plus several times the standard deviation to ensure effective differentiation between normal fluctuations and abnormal impacts. Simultaneously, the control system sets a duration threshold to distinguish between brief impact events and persistent abnormal vibrations (such as friction or resonance). The control system scans and analyzes the time-frequency energy distribution map to find regions that meet two conditions: first, the energy value within the region exceeds the preset energy threshold, indicating the presence of abnormally high energy; second, the duration of the high-energy state is less than the preset duration threshold, consistent with the brief and instantaneous characteristics of an impact event. When both conditions are met simultaneously, the control system marks the region as an impact event.

[0045] 106. Within a preset statistical time window, calculate the impact event rate based on the total number of impact events, and calculate the average impact energy value based on the sum of the energy values ​​of all impact events.

[0046] The preset statistical time window refers to a fixed period of time for statistical analysis of impact events, which can be a few minutes, a few hours, or a few days, set according to the pump's operating characteristics and monitoring needs. The impact event rate refers to the frequency of impact events occurring within the preset statistical time window, reflecting the frequency of foreign object impacts. The energy value of impact events refers to the sum of the energy of all impact events within the statistical window; the average impact energy value represents the average energy intensity of a single impact event, reflecting the severity of the impact.

[0047] Specifically, the control system first sets an appropriate statistical time window. This window's length needs to include a representative sample of events and ensure timely reflection of changes in the pump's status. Within the preset statistical time window, the control system counts all identified impact events and divides this number by the time window length (usually converted to hours or days) to obtain the impact event rate. Simultaneously, the control system sums the peak energy or energy integral value of all impact events within the preset statistical time window to obtain the total energy value. This total energy value is then divided by the number of impact events to calculate the average impact energy value, which reflects the average intensity of the impact. These two indicators together constitute key parameters for assessing the risk of pump blockage: an increased impact frequency indicates a potential increase in the number of foreign objects inside the pump, while an increase in impact energy indicates a potential increase in the size of the foreign objects or a worsening of the blockage.

[0048] 107. Based on the impact event rate and average impact energy value, determine the blockage risk level of the water pump to be monitored and generate the corresponding early warning signal.

[0049] The blockage risk level refers to the classification of the current severity of blockage of a water pump based on the impact event rate and average impact energy value. It is usually divided into multiple levels (such as normal, slight, moderate, severe, etc.). The warning signal is a warning message generated based on the blockage risk level.

[0050] Specifically, the control system first extracts energy distribution data for each identified impact event from the time-frequency energy distribution map. This data reflects the distribution of impact energy at different frequencies. For each impact event, the control system calculates two key characteristic parameters: first, the spectral centroid, obtained by weighted averaging of energy values ​​at various points on the frequency axis, representing the concentrated location of impact energy in the frequency dimension. Impacts from foreign objects of different materials will produce energy distributions with different frequency characteristics, resulting in differences in the spectral centroid; second, the peak energy, which is the maximum energy value of the impact event on the time-frequency plane, reflecting the severity of the impact. Within a preset statistical time window, the control system averages the spectral centroid and peak energy of all impact events to obtain the average spectral centroid and average impact energy. Then, the control system compares the calculated average spectral centroid with a preset centroid threshold, and simultaneously compares the average impact energy with a preset energy threshold. Based on these two comparison results, the control system determines the clogging risk level of the monitored water pump: a high-spectral centroid typically indicates hard materials (such as sand, gravel, metal fragments, etc.), whose high-frequency vibrations from impact can cause severe wear even with a low impact rate, and is therefore classified as a severe level; a low-spectral centroid typically indicates soft materials (such as fibers, rubber, etc.), whose low-frequency vibrations from impact are therefore classified as a slight level; and materials with both centroid and energy values ​​in the middle range indicate mixed materials, and are classified as a moderate level. Based on the determined clogging risk level, the control system generates an early warning message containing the type of clogging material, risk level, current indicator values, and recommended treatment measures, and sends the warning message to maintenance personnel.

[0051] In some embodiments, the determination of congestion risk level and the generation of early warning signals can be achieved in various ways: Optionally, the control system can employ a threshold grading method. First, multiple threshold levels (such as normal, attention, warning, danger, etc.) are set for the impact event rate and average impact energy value. Then, based on the respective levels of the two indicators, the final congestion risk level is determined through preset combination rules (such as taking the higher level, weighted average, etc.). Next, an early warning message containing detailed information is generated based on the risk level, and finally, it is sent to relevant personnel through multiple channels and recorded in the system log. It is understood that other methods can also be used to determine the congestion risk level and generate early warning signals, such as fuzzy logic reasoning, Bayesian networks, etc., which are not limited here.

[0052] The pump fault prediction method employed in this application involves a control system that collects acoustic vibration signals and real-time speed signals from the monitored pump during operation. It extracts the mechanical noise component corresponding to the real-time speed from the initial baseline data and constructs a baseline noise template. The acoustic vibration signal is then subtracted from the baseline noise template to obtain a residual signal containing only impurity impact information. Subsequently, the control system performs continuous wavelet transform on the residual signal to generate a time-frequency energy distribution map. Impact events with signal energy exceeding a threshold and duration meeting requirements are identified from this map. The impact event rate and average impact energy value are then calculated, and combined to determine the blockage risk level and generate an early warning signal, thus achieving early prediction of pump blockage faults. This method effectively alleviates the technical problem of fixed threshold alarms failing to identify early, progressive blockages and improves the timeliness of fault prediction.

[0053] Based on the above, the following is a more detailed description of the process provided in this implementation. Please refer to [link / reference]. Figure 2 This is another flowchart illustrating the pump failure prediction method in this application.

[0054] 201. Collect acoustic vibration signals and real-time speed signals of the water pump under monitoring during its operation. The acoustic vibration signals represent the time-series data of the sound and vibration generated by the water pump under monitoring. (This step has been explained in step 101.) 202. In the speed-noise characteristic baseline database, find the two speed signals with the smallest absolute value of the difference between them and the real-time speed signal, and their corresponding mechanical noise components. The speed-noise characteristic baseline database is established based on the initial baseline data of the pump to be monitored.

[0055] The speed-noise characteristic baseline database represents a set of data that stores the mechanical noise characteristics of the water pump under monitoring at different speeds.

[0056] Specifically, the control system first reads the real-time speed signal value, then searches the speed-noise characteristic baseline database, which stores the pump's speed signals and corresponding mechanical noise components at various speed ranges. The control system compares the absolute values ​​of the differences between the real-time speed value and each speed value in the database to determine the two speed signals with the smallest absolute differences from the real-time speed signal value. These two speed signals are located at the upper and lower boundaries of the real-time speed signal, respectively. Then, the control system extracts the mechanical noise component data corresponding to these two closest speed signals. This mechanical noise component data contains the energy distribution characteristics of the pump's mechanical vibration noise at each order under these two specific speeds.

[0057] The establishment of the speed-noise characteristic baseline database specifically includes steps 2021-2025: 2021. Acquire the acoustic vibration signal and speed signal of the water pump under monitoring in its initial state and during variable speed operation within the speed range, and generate initial baseline data based on the acoustic vibration signal and speed signal. The initial state is a non-blocked state.

[0058] The initial state refers to the normal working state of the water pump after installation, commissioning, or maintenance, at which time there are no blockages inside the water pump; the speed range indicates the range of speeds the water pump may operate under normal working conditions, from the lowest operating speed to the highest operating speed; the speed signal indicates the rotational speed data of the water pump during operation; the initial baseline data refers to the reference dataset of the acoustic vibration characteristics and speed relationship of the water pump under normal, unblocked conditions.

[0059] Specifically, the control system controls the monitored water pump to perform a variable speed operation test according to a preset speed change scheme in an initial state where it is confirmed to be completely unblocked. This test covers the entire operating speed range of the water pump, gradually increasing from the lowest operating speed to the highest operating speed, and then gradually decreasing back to the lowest operating speed, forming a complete speed change cycle. During this process, the control system synchronously collects the acoustic vibration signal, speed signal, and phase reference signal (e.g., key phase signal or encoder signal) of the water pump through a sensor network installed on the water pump. The control system performs time synchronization pairing of the collected acoustic vibration signal, the corresponding speed signal, and the phase reference signal to form an initial baseline dataset containing the speed-acoustic vibration correspondence (including the characteristics of vibration amplitude changing with speed and its phase information). This data represents the acoustic vibration characteristics of the water pump under different speed conditions in a blockage-free normal state.

[0060] 2022. Divide the speed range into multiple continuous speed intervals.

[0061] A speed range refers to dividing the entire speed range into multiple sub-ranges, each containing a certain range of speed values.

[0062] Specifically, the control system first determines the complete operating speed range of the pump to be monitored, from the lowest to the highest operating speed. Then, based on the pump's operating characteristics and data processing requirements, the control system divides the entire speed range into multiple consecutive speed intervals. This division can use an equal-width method, where each interval has the same speed span; or a non-equal-width method, setting narrower intervals near the pump's critical operating speeds to obtain a more refined characteristic description. The control system ensures that these speed intervals are interconnected, seamlessly covering the entire speed range and avoiding any gaps in speed measurement.

[0063] 2023. For each speed range, extract the current acoustic vibration signal and the corresponding speed signal within the current speed range, perform order analysis on the current acoustic vibration signal, calculate the average order spectrum within the current speed range, and obtain the mechanical noise component corresponding to the current speed range. The mechanical noise component represents the energy distribution of the mechanical vibration noise corresponding to the current speed range at each order.

[0064] The current speed range refers to a specific range among multiple speed ranges; the current acoustic vibration signal refers to the acoustic vibration data collected within the current speed range; order analysis is an analysis method that converts time-domain vibration signals into speed-dependent frequency components, used to identify vibration components proportional to the speed; the order spectrum represents the energy distribution of the vibration signal at each order, where order refers to the ratio of vibration frequency to speed; the average order spectrum is the average result of all order spectra within the current speed range.

[0065] Specifically, the control system first filters all data points from the initial baseline data whose rotational speed values ​​fall within the current rotational speed range, including the corresponding acoustic vibration signals and rotational speed signals. Then, the control system performs order analysis on the filtered acoustic vibration signals. In order analysis, the control system first converts the time-domain acoustic vibration signals into frequency-domain signals, and then, combined with synchronously acquired phase reference signals, converts the frequency axis into an order axis based on the corresponding rotational speed signal, thereby obtaining the amplitude and phase information of each order. The order represents the ratio of vibration frequency to rotational speed. This conversion ensures that vibration components proportional to rotational speed (such as vibrations caused by impeller rotation, bearing movement, etc.) can correspond to the same order at different rotational speeds, facilitating the identification of mechanical characteristics. The control system calculates the order spectrum (containing the amplitude and phase information of each order) of all data points within the current rotational speed range, and averages these order spectra to obtain the average order spectrum of the rotational speed range (which includes the average amplitude and relative phase of each order at a specific rotational speed). The average order spectrum is the mechanical noise component corresponding to this rotational speed range.

[0066] 2024. Based on the mechanical noise components of each speed range, determine the reference speed for each speed range.

[0067] The reference speed is a specific speed value that can represent the characteristics of the current speed range.

[0068] Specifically, the control system analyzes all speed data and corresponding mechanical noise components within the current speed range, and determines the speed value that best represents the characteristics of that range through statistical methods. The control system can choose the center value of the current speed range as the reference speed; it can also calculate the average speed of all sampled points within the range as the reference speed; or it can select the speed point with the richest data and the highest signal-to-noise ratio within the range as the reference speed. The purpose of determining the reference speed is to establish a clear index for the mechanical noise components of each speed range.

[0069] 2025. Establish a mapping relationship based on each speed range and the corresponding mechanical noise component, and construct a speed-noise characteristic baseline database based on the mapping relationship.

[0070] The mapping relationship represents the correspondence between the speed range (or reference speed) and the corresponding mechanical noise component.

[0071] Specifically, the control system creates a data record for each speed range, containing the upper and lower boundary values ​​of the speed range, the reference speed value, and the corresponding mechanical noise component data (i.e., the average order spectrum). The control system arranges these records in ascending or descending order of speed values, forming an ordered data structure that facilitates rapid location using efficient algorithms such as binary search. The control system can also create an index for the database to further improve query efficiency. The final constructed speed-noise characteristic baseline database enables rapid mapping from any speed value to the corresponding mechanical noise characteristic.

[0072] 203. Perform linear interpolation on the mechanical noise components corresponding to the two speed signals with the smallest absolute difference from the real-time speed signal to obtain the real-time mechanical noise components corresponding to the real-time speed signal.

[0073] Linear interpolation is a mathematical method for estimating intermediate point values ​​based on the linear relationship between known points; real-time mechanical noise components represent mechanical noise characteristics that precisely correspond to the current real-time rotational speed signal.

[0074] Specifically, the control system first acquires the two speed values ​​with the smallest absolute difference from the real-time speed signal, denoted as speed 1 and speed 2 (assuming speed 1 < real-time speed < speed 2), and their corresponding mechanical noise component data, denoted as noise component 1 and noise component 2. Then, the control system calculates the relative position of the real-time speed between these two speed points, i.e., the interpolation weighting coefficient: weight = (real-time speed - speed 1) / (speed 2 - speed 1). Finally, the control system performs a weighted average of noise component 1 and noise component 2: real-time mechanical noise component = noise component 1 × (1 - weight) + noise component 2 × weight. The real-time mechanical noise component includes amplitude and phase information of each order, where both amplitude and phase are interpolated according to weights.

[0075] 204. Convert the real-time mechanical noise components into time-domain signals to obtain the baseline noise template.

[0076] A time-domain signal is a signal waveform that changes over time, representing the amplitude change of vibration or sound on the time axis.

[0077] Specifically, the control system first determines the representation of the real-time mechanical noise component (order spectrum or frequency spectrum), and then applies an inverse Fourier transform algorithm to convert it into a time-domain signal. If the real-time mechanical noise component is represented in order spectrum form, the control system needs to first convert the order axis to the frequency axis based on the current real-time rotational speed to obtain the frequency spectrum representation. Then, the control system applies an inverse Fourier transform to convert the frequency spectrum into a time-domain signal. Specifically, the process of converting the real-time mechanical noise component into a time-domain signal by the control system employs order tracking reconstruction technology. First, the control system performs time integration on the acquired real-time rotational speed signal to calculate the instantaneous rotation angle (i.e., instantaneous phase) of the rotor at each moment. Then, the control system uses the interpolated real-time mechanical noise component (which already contains the amplitude A of each order corresponding to the real-time rotational speed) to calculate the instantaneous rotation angle (i.e., instantaneous phase) of the rotor at each moment. k and initial phase θ k Information), combined with the instantaneous phase ϕ(t), is used to reconstruct the time-domain waveform using the cosine wave synthesis formula. The formula is: Template(t) = ∑[A k ·cos(k·ϕ(t)+θ k )], where A k Let ϕ(t) be the k-th order amplitude obtained by interpolation, and θ be the instantaneous phase obtained by real-time integration. k This is the initial phase for this order (the initial phase has been established in the speed-noise characteristic baseline database and acquired along with the mechanical noise component). The resulting time-domain signal is the baseline noise template, which represents the acoustic vibration characteristics that the pump should have under normal operation at the current real-time speed.

[0078] 205. Subtract the baseline noise template from the acoustic vibration signal to obtain the residual signal. (This step has been explained in section 103) Before subtracting the baseline noise template from the acoustic vibration signal, the control system performs a phase alignment operation. Specifically, the control system determines the zero-phase moment of the acoustic vibration signal based on the acquired phase reference signal (such as the key phase pulse). Simultaneously, when generating the baseline noise template, the initial phase of the template is adjusted using the instantaneous phase obtained from the real-time rotational speed integration, ensuring strict synchronization with the rotation period of the acoustic vibration signal. Alternatively, the control system can calculate the cross-correlation function between the acoustic vibration signal and the baseline noise template, and perform a time-domain shift of the baseline noise template based on the time delay corresponding to the cross-correlation peak, thus achieving phase alignment.

[0079] Since the baseline noise template is generated based on the real-time rotational speed signal and its instantaneous phase reconstruction, it maintains phase consistency with its original acoustic vibration signal on the time axis (i.e., peak to peak, trough to trough). Therefore, to address potential errors during actual acquisition and reconstruction, the control system first calculates the time difference between the real-time acoustic vibration signal and the baseline noise template using a cross-correlation function. Based on this time difference, the baseline noise template is time-shifted to achieve high-precision phase alignment in the time domain. After alignment, the control system effectively cancels out the periodic mechanical background noise (i.e., synchronization component) generated by the shaft rotation through point-to-point subtraction, while retaining the non-periodic transient impact component (i.e., asynchronous component) generated by impurity impacts.

[0080] 206. Perform continuous wavelet transform on the residual signal to generate a time-frequency energy distribution map, which shows the distribution of signal energy in the time and frequency dimensions. (This step has been explained in section 104.) It should be noted that, in this embodiment of the application, in order to use image morphology algorithms to extract features more accurately from non-steady-state impact signals, the time-frequency energy distribution map constructed by the control system is regarded as a two-dimensional digital image matrix at the data processing level.

[0081] Specifically, the control system constructs a two-dimensional array from the squares of the wavelet coefficient moduli output by the continuous wavelet transform, where the horizontal axis index corresponds to the sampling time point, the vertical axis index corresponds to the scale (frequency), and the element values ​​in the array correspond to the signal energy. In this two-dimensional data space, each discrete time-frequency coordinate point is defined as a "pixel" as described in this application, and its corresponding energy value is used as the "grayscale value" or "brightness value" of that pixel.

[0082] Therefore, the "regions," "patches," and "morphological feature parameters" mentioned in subsequent steps are essentially the results of the control system calling computer vision or image processing algorithms (such as connected component analysis algorithms) to perform calculations on the two-dimensional digital image matrix. This processing method can capture two-dimensional topological features (such as vertical stripe distributions) that cannot be identified by traditional one-dimensional signal threshold determination, thereby more accurately distinguishing real impact events from random noise spots.

[0083] 207. In the time-frequency energy distribution map, it is determined that an impact event occurred in the region where the signal energy value exceeds the energy threshold and the duration is less than the duration threshold.

[0084] Step 207 specifically includes steps 2071 to 2073.

[0085] 2071. In the time-frequency energy distribution map, pixels whose signal energy value exceeds the energy threshold and whose duration is less than the duration threshold are marked as candidate regions, and pixels of spatially adjacent candidate regions are merged into energy patches.

[0086] Candidate regions refer to the set of pixels that meet the preliminary screening criteria; energy patches represent spatially continuous high-energy regions, which may represent an impact event.

[0087] Specifically, the control system sets two key threshold parameters: an energy threshold and a duration threshold. The control system scans each pixel in the time-frequency energy distribution map, initially marking pixels with energy values ​​exceeding the energy threshold as candidate points. Then, the control system checks the continuity of the candidate points in the time dimension; if the duration of a high-energy state at a certain frequency exceeds the duration threshold, these points are excluded because impact events typically manifest as brief transient signals. For candidate points that pass the initial screening, the control system further analyzes their spatial connectivity in the time-frequency plane, merging spatially adjacent (connected in the time or frequency direction) candidate points into a complete energy patch. This energy patch represents a possible impact event, and its morphological characteristics need further analysis for confirmation.

[0088] 2072. Calculate the morphological characteristic parameters for each energy patch. The morphological characteristic parameters include time duration, frequency bandwidth, energy concentration and vertical morphological index. The energy concentration is the ratio of the peak energy to the average energy within the energy patch. The vertical morphological index is determined by calculating the aspect ratio of the minimum bounding rectangle of the energy patch.

[0089] Morphological parameters represent the geometric and statistical properties of energy patches in the time-frequency plane; temporal duration refers to the span of energy patches on the time axis, indicating the duration of the event; frequency bandwidth represents the span of energy patches on the frequency axis, reflecting the frequency distribution range of the signal; energy concentration measures the degree of concentration of energy distribution, with high values ​​indicating sharp energy distribution; vertical morphology index reflects the shape characteristics of energy patches, with high values ​​indicating a slender shape extending in the frequency direction, which is a typical feature of impact events.

[0090] Specifically, the control system first calculates the temporal duration of each energy patch, i.e., the maximum span of the patch on the time axis; impact events typically have a short duration. Next, the control system calculates the frequency bandwidth of the energy patch, i.e., the coverage area of ​​the patch on the frequency axis; impact events usually generate energy over a wide frequency range. Then, the control system calculates the energy concentration, i.e., the ratio of the maximum energy value to the average energy value within the energy patch; impact events typically have a high energy concentration, manifested as a sharp energy distribution. Finally, the control system calculates the vertical morphology index by finding the minimum bounding rectangle of the energy patch and calculating its aspect ratio (the ratio of frequency span to time span). Impact events typically exhibit a vertical stripe shape on the time-frequency plane (i.e., the frequency span is greater than the time span), thus possessing a high vertical morphology index. These morphological characteristic parameters collectively constitute a multidimensional feature space that distinguishes impact events from other interference signals.

[0091] These morphological parameters were chosen because impact events, as typical transient shocks, exhibit significant morphological differences in time-frequency energy distribution maps compared to persistent background noise or random interference. Impact events are typically characterized by short duration, broad frequency distribution, and concentrated energy bursts within a very short time (e.g., high energy concentration), and appear as bands perpendicular to the time axis on the time-frequency map (e.g., a high vertical morphological index can be reflected by the aspect ratio of the minimum bounding rectangle). By comprehensively evaluating these characteristics, real impact events can be distinguished from background noise, collectively forming a multidimensional feature space that differentiates impact events from other interference signals.

[0092] 2073. Based on morphological feature parameters, each energy patch is scored, and energy patches with scores greater than the score threshold are identified as impact events.

[0093] The score refers to the degree to which an energy patch conforms to the characteristics of an impact event, calculated comprehensively based on morphological feature parameters; the score threshold represents the minimum score requirement for determining an impact event.

[0094] Specifically, the control system first assigns a weight coefficient to each morphological feature parameter, reflecting the importance of that feature in identifying impact events. For example, duration may have a high negative weight (shorter duration, more likely an impact); frequency bandwidth may have a medium positive weight (wider bandwidth, more likely an impact); energy concentration may have a high positive weight (more concentrated energy, more likely an impact); and vertical morphological indicators may have the highest positive weight (more vertical shape, more likely an impact). The control system multiplies each feature parameter value by its corresponding weight, then sums them or uses other mathematical combinations to calculate a comprehensive score for the energy patch. The control system compares this score with a preset score threshold. If the score exceeds the threshold, the energy patch is identified as an impact event; otherwise, it is considered interference or noise and excluded. This multi-feature comprehensive scoring method effectively reduces false alarms and false negatives, improving the accuracy of impact event identification.

[0095] 208. Within a preset statistical time window, calculate the impact event rate based on the total number of impact events, and calculate the average impact energy value based on the sum of the energy values ​​of all impact events. (This step has been explained in 106.) 209. Extract the energy distribution of each impact event in the time-frequency energy distribution map, and calculate the spectral centroid and peak energy of the energy distribution. The spectral centroid represents the weighted average center of the impact energy on the frequency axis, and the peak energy represents the maximum energy value of the impact event.

[0096] The spectral centroid is the weighted average of frequencies, with the weights being the energy values ​​at each frequency point, reflecting the concentration trend of energy in the frequency dimension; the peak energy is the maximum energy value that occurs in an impact event, representing the impact intensity.

[0097] Specifically, the control system first extracts the energy patch region corresponding to each impact event from the time-frequency energy distribution map, obtaining the energy values ​​of all time-frequency points within that energy patch region. Then, the control system calculates the spectral centroid of the impact event, i.e., a weighted average frequency is calculated for all frequency points within the energy patch, weighted by their energy values: Spectral centroid = Σ(frequency i × energy i) / Σ(energy i), where the summation covers all points within the energy patch. The spectral centroid reflects the concentrated location of impact energy in the frequency dimension. Impacts from blockages of different materials typically produce significantly different spectral centroids; for example, impacts from hard materials usually produce a higher spectral centroid, while impacts from soft materials produce a lower spectral centroid. Simultaneously, the control system calculates the peak energy of the impact event, i.e., the maximum energy value among all points within the energy patch. The peak energy reflects the intensity of the impact and is related to factors such as the mass and hardness of the blockage and the impact velocity. These two parameters together constitute the key indicators characterizing the impact event.

[0098] 210. Based on the spectral centroid and peak energy of the energy distribution of each impact event, calculate the average spectral centroid and average impact energy of all impact events within a preset statistical time window.

[0099] The average spectral centroid refers to the arithmetic mean of the spectral centroids of all impact events within a preset statistical time window; the average impact energy refers to the arithmetic mean of the peak energies of all impact events within a preset statistical time window.

[0100] Specifically, within a preset statistical time window, the control system collects all identified impact events along with their spectral centroids and peak energy values. Then, the control system calculates the average spectral centroid of these impact events, which is the arithmetic mean of the spectral centroids of all events: Average spectral centroid = Σ(spectral centroid of event i) / total number of events. Similarly, the control system calculates the average impact energy, which is the arithmetic mean of the peak energies of all events: Average impact energy = Σ(peak energy of event i) / total number of events. These two statistical indicators reflect the average material properties of the blockage and the average impact intensity within the current time window, respectively.

[0101] 211. Based on the comparison results of the average spectral centroid and the preset centroid threshold, and the comparison results of the average impact energy and the preset energy threshold, determine the type of blockage. The types of blockage include hard materials, soft materials, and mixed materials.

[0102] The preset centroid threshold is a frequency reference value for distinguishing different types of blockages; the preset energy threshold is an energy reference value for distinguishing different intensities of impacts; hard materials refer to materials with high hardness such as metals and stones; soft materials refer to soft materials such as fibers and plastic films; and mixed materials refer to composite materials that combine the characteristics of both hard and soft materials.

[0103] Specifically, the control system first compares the calculated average spectral centroid with a preset centroid threshold. Typically, the acoustic vibrations produced by impacts from hard materials (such as metals or stones) have higher frequency components, resulting in a higher spectral centroid; while the acoustic vibrations produced by impacts from soft materials (such as fibers or plastic films) are mainly concentrated at lower frequencies, resulting in a lower spectral centroid. The control system can set one or more centroid thresholds, such as a low centroid threshold and a high centroid threshold, to distinguish between different materials. Simultaneously, the control system compares the average impact energy with a preset energy threshold; hard materials typically produce higher-energy impacts. Based on the comparison results of these two dimensions, the control system can classify the blockage as: hard materials (high spectral centroid, high energy), soft materials (low spectral centroid, low energy), or mixed materials (medium spectral centroid or cases where the spectral centroid and energy are mismatched).

[0104] 212. Calculate the blockage accumulation index of the water pump to be monitored based on the impact event rate and the average impact energy value.

[0105] The impact event rate represents the number of impact events occurring per unit time; the average impact energy value represents the average energy intensity of a single impact event, reflecting the severity of the impact; the blockage accumulation index is a comprehensive indicator that reflects the severity and development trend of pump blockage.

[0106] Specifically, the control system first calculates the impact event rate within the current statistical time window, which is the total number of impact events divided by the time window length, yielding the impact frequency within the current statistical time window. The impact event rate reflects the frequency of blockage movement inside the pump and typically increases with the severity of blockage. Then, the control system combines the impact event rate with the average impact energy value, calculating the blockage accumulation index using a mathematical function (such as a weighted product): Blockage Accumulation Index = f(impact event rate, average impact energy value). The blockage accumulation index considers both the frequency and intensity of impacts, providing a more comprehensive reflection of the pump's blockage status.

[0107] 213. Obtain the real-time operating parameters of the pump to be monitored, and calculate the actual performance parameters of the pump to be monitored based on the real-time operating parameters. The real-time operating parameters include inlet pressure, outlet pressure, flow rate and motor power.

[0108] Real-time operating parameters refer to physical quantities that describe the current operating status of the water pump; inlet pressure and outlet pressure represent the fluid pressure at the pump inlet and outlet, respectively; flow rate represents the volume of fluid passing through the pump per unit time; motor power represents the electrical energy consumed by the motor driving the pump; actual performance parameters refer to indicators that characterize the pump performance, such as pump efficiency and head, calculated based on the operating parameters.

[0109] Specifically, the control system first collects real-time operating parameters such as inlet pressure (P1), outlet pressure (P2), flow rate, and motor power from various sensors connected to the pump monitoring system. These parameters are typically measured and transmitted to the control system by devices such as pressure sensors, flow meters, and power analyzers. Then, based on these raw operating parameters, the control system calculates the actual performance parameters of the pump. The main performance parameter includes the pump head (H), calculated using the formula H = (P2 - P1) / (ρg) + Δz + v 2 / (2g), where ρ is the fluid density, g is the gravitational acceleration, Δz is the inlet / outlet height difference, and v is the flow velocity; pump shaft power (P), which is the motor power multiplied by the motor efficiency; pump efficiency (η), calculated by the formula η=ρgQH / P, where Q is the flow rate. These performance parameters comprehensively reflect the actual working state and efficiency of the pump.

[0110] 214. Calculate the performance deviation index between the actual performance parameters and the baseline performance parameters of the pump under monitoring at the same speed.

[0111] Baseline performance parameters represent the standard performance indicators of a water pump under normal, unclogging conditions and at the same speed. The performance deviation index is a dimensionless index that measures the degree of difference between actual performance and baseline performance, reflecting the degree of performance degradation of the water pump.

[0112] Specifically, the control system first queries or interpolates baseline performance parameters at the same speed based on the real-time pump speed from the baseline performance database, including baseline head and baseline efficiency. This baseline data may come from performance curves provided by the pump manufacturer or from calibrated measured performance data from the initial installation phase. Then, the control system calculates the deviation between the actual performance parameters and the baseline performance parameters, standardizing these deviations into performance deviation indices. For example, the head deviation index can be calculated as: Head Deviation Index = (Baseline Head - Actual Head) / Baseline Head × 100%; the efficiency deviation index can be calculated as: Efficiency Deviation Index = (Baseline Efficiency - Actual Efficiency) / Baseline Efficiency × 100%. These performance deviation indices reflect the degree of pump performance degradation.

[0113] 215. Based on the blockage accumulation index, performance deviation index, and blockage type, determine the final blockage risk level of the pump to be monitored and generate a corresponding early warning signal.

[0114] The blockage risk level is a graded assessment of the severity of water pump blockage, usually divided into multiple levels (such as low risk, medium risk, high risk, and extremely high risk); the warning signal is a warning message that includes the risk level and recommended measures.

[0115] Specifically, the control system smooths and performs linear regression fitting on the time-series data of the blockage accumulation index and performance deviation index to extract the slope reflecting future trends. When the slope exceeds a threshold, the final blockage risk level of the monitored pump is determined based on the excess ratio. After determining the risk level, the control system generates a corresponding early warning signal, including information such as the risk level, blockage type, performance degradation degree, and recommended measures, and sends it to relevant technical personnel so that maintenance measures can be taken in a timely manner.

[0116] Specifically, determining the final blockage risk level of the water pump to be monitored includes steps 2151 to 2154.

[0117] 2151. The time series data of the congestion accumulation index and the performance deviation index are processed by moving average to obtain smoothed time series data.

[0118] Moving average processing is a time series smoothing technique; smoothed time series data retains the long-term trend of the original data and reduces random fluctuations and noise.

[0119] Specifically, the control system first collects historical data on the congestion accumulation index and performance deviation index over a period of time (such as the past few days or weeks), forming two time series. This raw data may contain short-term random fluctuations caused by factors such as measurement errors and temporary operating condition variations, which are not conducive to accurately judging long-term trends. Therefore, the control system applies a moving average algorithm to smooth these two time series. The formula for calculating the moving average is: Smoothed value (t) = Σ[original value (ti)] / n, where i ranges from 0 to n-1, and n is the size of the moving average window. The choice of the moving average window size needs to balance the smoothing effect and response speed, and is typically chosen between 4 and 24 hours, depending on the data acquisition frequency and application requirements. Through moving average processing, short-term random fluctuations are filtered out, while long-term trends are preserved.

[0120] 2152. Within the preset fitting time window, perform linear regression fitting on the smoothed time series data to obtain the first fitting line of the blockage accumulation index and the second fitting line of the performance deviation index.

[0121] The preset fitting time window refers to the time range used for trend analysis, usually selecting data from a recent period; linear regression fitting is a statistical method used to find the straight line that best represents the trend of data change; the fitted line is determined by two parameters: slope and intercept, where the slope reflects the rate of change of the indicator.

[0122] Specifically, the control system first determines an appropriate fitting time window, such as data from the most recent 24 hours, 48 ​​hours, or 7 days. The length of the fitting time window needs to balance short-term responsiveness and long-term stability; too short a window may be affected by local fluctuations, while too long a window may lag behind actual changes. Within this preset time window, the control system applies linear regression algorithms to the smoothed congestion accumulation index and performance deviation index time series data, respectively. The goal of linear regression is to find a straight line of the form y = ax + b that minimizes the sum of the squared vertical distances from data points to the line, where y is the index value, x is time, a is the slope, and b is the intercept. Using the least squares method, the control system calculates the optimal slope and intercept parameters, obtaining two fitted lines: a first fitted line for the congestion accumulation index and a second fitted line for the performance deviation index. These two fitted lines represent the changing trends of the two key indicators within the current time window.

[0123] 2153. Extract the slope of the first fitted line as the first future change slope of the blockage accumulation index, and extract the slope of the second fitted line as the second future change slope of the performance deviation index.

[0124] The first future slope represents the trend of the congestion accumulation index; a positive value indicates that the congestion situation is worsening, while a negative value indicates that the congestion situation is improving. The second future slope represents the trend of the performance deviation index; a positive value indicates that the performance degradation is intensifying, while a negative value indicates that the performance is recovering.

[0125] Specifically, the control system extracts slope parameters from the two fitted lines obtained in the previous step. For the first fitted line y=a1x+b1 of the blockage accumulation index, the control system extracts slope a1 as the first future change slope; for the second fitted line y=a2x+b2 of the performance deviation index, the control system extracts slope a2 as the second future change slope. The first future change slope represents the change in the blockage accumulation index within the preset fitting time window, reflecting the rate of increase or decrease in blockage impact activity; the second future change slope represents the change in the performance deviation index within the preset fitting time window, reflecting the rate of acceleration or deceleration of pump performance degradation. A positive slope indicates that the index value is increasing, i.e., the blockage condition is worsening; a negative slope indicates that the index value is decreasing, i.e., the blockage condition is improving; a slope close to zero indicates that the blockage condition is relatively stable.

[0126] 2154. When the first future change slope or the second future change slope exceeds the slope threshold, the final blockage risk level of the pump to be monitored is determined based on the percentage by which the first future change slope or the second future change slope exceeds the slope threshold.

[0127] The slope threshold is a reference value for judging whether a trend change is significant; the percentage exceeding the slope threshold represents the relative difference between the actual slope and the threshold; the final congestion risk level is a risk rating determined by comprehensively considering the current state and future trends.

[0128] Specifically, the control system first compares the extracted first and second future slopes with preset slope thresholds. These slope thresholds are typically safety boundary values ​​determined based on historical data analysis and expert experience, representing the upper limit of the acceptable rate of change. When either slope exceeds its corresponding threshold, it indicates that the congestion is worsening at an abnormal rate, requiring an increase in the risk level. The control system calculates the percentage by which the slope exceeds the threshold: (actual slope - slope threshold) / slope threshold × 100%. This percentage reflects the degree of trend abnormality. Then, the control system determines the final congestion risk level based on the percentage exceeding the threshold. For example, exceeding 0-20% is considered low risk, 20%-50% is medium risk, 50%-100% is high risk, and above 100% is extremely high risk. This trend-based risk assessment method considers not only the current congestion status but also the dynamic trend of congestion development, enabling early warning of potential severe congestion risks and providing decision support for preventative maintenance.

[0129] The pump failure prediction method adopted in this application involves a control system that constructs a speed-adapted baseline noise template to obtain a high signal-to-noise ratio residual signal, and then uses time-frequency morphological features to identify impact events. The control system then distinguishes between soft and hard blockages using spectral centroid features, and analyzes the slope of the changes in the blockage accumulation index and performance deviation index using linear regression fitting. This achieves a comprehensive assessment of the "qualitative" (material type), "quantitative" (accumulation degree), and "potential" (development trend) of pump blockage risk, alleviating the technical problem that a single indicator cannot distinguish the harmful characteristics of blockages (such as distinguishing between soft fibers and hard stones) and is easily affected by speed fluctuations, leading to misjudgments. Furthermore, by monitoring the future slope of the indicator changes, it achieves an improvement from passive threshold alarms to proactive trend prediction, alleviating the problem of delayed early warning of progressive blockage failures and providing maintenance personnel with sufficient preparation time for repairs.

[0130] The methods provided in the above embodiments can be executed by a control system. The control system in the embodiments of this invention is described below from a hardware processing perspective; please refer to [link to relevant documentation]. Figure 3 This is a schematic diagram of the physical device structure of the control system in an embodiment of this application.

[0131] It should be noted that, Figure 3 The structure of the control system shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of the present invention.

[0132] like Figure 3 As shown, the control system includes a Central Processing Unit (CPU) 301, which can perform various appropriate actions and processes based on programs stored in Read-Only Memory (ROM) 302 or programs loaded from storage section 308 into Random Access Memory (RAM) 303, such as executing the methods described in the above embodiments. The RAM 303 also stores various programs and data required for system operation. The CPU 301, ROM 302, and RAM 303 are interconnected via a bus 304. An Input / Output (I / O) interface 305 is also connected to the bus 304.

[0133] The following components are connected to I / O interface 305: input section 306 including audio input devices, push-button switches, etc.; output section 307 including a liquid crystal display (LCD) and audio output devices, indicator lights, etc.; storage section 308 including a hard disk, etc.; and communication section 309 including a network interface card such as a LAN (Local Area Network) card, modem, etc. Communication section 309 performs communication processing via a network such as the Internet. Drive 310 is also connected to I / O interface 305 as needed. Removable media 311, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 310 as needed so that computer programs read from them can be installed into storage section 308 as needed.

[0134] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing computer programs for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 309, and / or installed from removable medium 311. When the computer program is executed by central processing unit (CPU) 301, it performs the various functions defined in the present invention.

[0135] It should be noted that specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0136] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. Each block in a flowchart or block diagram may represent a module, program segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those shown in the drawings.

[0137] Specifically, the control system of this embodiment includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, it implements the water pump failure prediction method provided in the above embodiment.

[0138] In another aspect, the present invention also provides a computer-readable storage medium, which may be included in the control system described in the above embodiments; or it may exist independently and not be assembled into the control system. The storage medium carries one or more computer programs that, when executed by a processor of the control system, cause the control system to implement the pump failure prediction method provided in the above embodiments.

[0139] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0140] As used in the above embodiments, depending on the context, the term "when..." can be interpreted as meaning "if...", "after...", "in response to determining...", or "in response to detecting...". Similarly, depending on the context, the phrase "when determining..." or "if (the stated condition or event) is interpreted as meaning "if determining...", "in response to determining...", "when (the stated condition or event) is detected", or "in response to detecting (the stated condition or event)".

[0141] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive), etc.

[0142] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A water pump intelligent monitoring control method, characterized in that, include: Acoustic vibration signals and real-time speed signals of the water pump under monitoring are collected during its operation. The acoustic vibration signals represent the time series data of the sound and vibration generated by the water pump under monitoring. From the initial baseline data of the water pump to be monitored, the mechanical noise component corresponding to the real-time speed signal is extracted, and a baseline noise template is constructed based on the mechanical noise component. The initial baseline data is the initial acoustic vibration data when the water pump is not blocked. The baseline noise template is subtracted from the acoustic vibration signal to obtain the residual signal; Perform continuous wavelet transform on the residual signal to generate a time-frequency energy distribution map, which shows the distribution of signal energy in the time and frequency dimensions. In the time-frequency energy distribution map, it was determined that an impact event occurred in a region where the signal energy value exceeded the energy threshold and the duration was less than the duration threshold; Within a preset statistical time window, the impact event rate is calculated based on the total number of impact events, and the average impact energy value is calculated based on the sum of the energy values ​​of all impact events. Based on the impact event rate and the average impact energy value, the blockage risk level of the water pump to be monitored is determined, and a corresponding early warning signal is generated.

2. The method according to claim 1, characterized in that, From the initial baseline data of the pump under monitoring, the mechanical noise component corresponding to the real-time speed signal is extracted, and a baseline noise template is constructed based on the mechanical noise component, specifically including: In the speed-noise characteristic baseline database, find the two speed signals with the smallest absolute value of the difference with the real-time speed signal and their corresponding mechanical noise components. The speed-noise characteristic baseline database is established based on the initial baseline data of the water pump to be monitored. Linear interpolation is performed on the mechanical noise components corresponding to the two speed signals with the smallest absolute difference from the real-time speed signal to obtain the real-time mechanical noise component corresponding to the real-time speed signal. The real-time mechanical noise component is converted into a time-domain signal to obtain a baseline noise template.

3. The method according to claim 2, characterized in that, The establishment of the speed-noise characteristic baseline database specifically includes: Acquire the acoustic vibration signal and speed signal of the water pump under monitoring in its initial state, when it is running at variable speed within the speed range, and generate initial baseline data based on the acoustic vibration signal and the speed signal, wherein the initial state is a non-blocked state; The speed range is divided into multiple consecutive speed intervals; For each speed range, the current acoustic vibration signal and the corresponding speed signal within the current speed range are extracted, and the order analysis of the current acoustic vibration signal is performed to calculate the average order spectrum within the current speed range, thereby obtaining the mechanical noise component corresponding to the current speed range. The mechanical noise component represents the energy distribution of the mechanical vibration noise corresponding to the current speed range at each order. Based on the mechanical noise components of each speed range, the reference speed of each speed range is determined; A mapping relationship is established based on each speed range and its corresponding mechanical noise component, and a speed-noise characteristic baseline database is constructed based on the mapping relationship.

4. The method according to claim 1, characterized in that, In the time-frequency energy distribution map, it is determined that an impact event occurred within a region where the signal energy value exceeds the energy threshold and the duration is less than the duration threshold, specifically including: In the time-frequency energy distribution map, pixels with signal energy values ​​exceeding the energy threshold and durations less than the duration threshold are marked as candidate regions, and pixels in spatially adjacent candidate regions are merged into energy patches. For each energy patch, morphological characteristic parameters are calculated, including time duration, frequency bandwidth, energy concentration, and vertical morphological index. The energy concentration is the ratio of peak energy to average energy within the energy patch, and the vertical morphological index is determined by calculating the aspect ratio of the minimum bounding rectangle of the energy patch. Based on the morphological feature parameters, each energy patch is scored, and energy patches with scores greater than a threshold are identified as impact events.

5. The method according to claim 1, characterized in that, Based on the impact event rate and the average impact energy value, the blockage risk level of the monitored water pump is determined, and a corresponding early warning signal is generated, specifically including: Extract the energy distribution of each impact event from the time-frequency energy distribution map, and calculate the spectral centroid and peak energy of the energy distribution. The spectral centroid represents the weighted average center of the impact energy on the frequency axis, and the peak energy represents the maximum energy value of the impact event. Based on the spectral centroid and peak energy of the energy distribution of each impact event, calculate the average spectral centroid and average impact energy of all impact events within the preset statistical time window. Based on the comparison results of the average spectral centroid and the preset centroid threshold, and the comparison results of the average impact energy and the preset energy threshold, the blockage type is determined, and the blockage type includes hard material, soft material, and mixed material. Based on the type of blockage, the impact event rate, and the average impact energy value, the blockage risk level of the water pump to be monitored is determined, and a corresponding early warning signal is generated.

6. The method according to claim 5, characterized in that, Based on the type of blockage, the impact event rate, and the average impact energy value, the blockage risk level of the monitored water pump is determined, and a corresponding early warning signal is generated, specifically including: Based on the impact event rate and the average impact energy value, the blockage accumulation index of the water pump to be monitored is calculated; The real-time operating parameters of the pump to be monitored are obtained, and the actual performance parameters of the pump to be monitored are calculated based on the real-time operating parameters. The real-time operating parameters include inlet pressure, outlet pressure, flow rate and motor power. Calculate the performance deviation index between the actual performance parameters and the baseline performance parameters of the pump under monitoring at the same speed; Based on the blockage accumulation index, the performance deviation index, and the blockage type, the final blockage risk level of the water pump to be monitored is determined, and a corresponding early warning signal is generated.

7. The method according to claim 6, characterized in that, The determination of the final blockage risk level of the monitored water pump based on the blockage accumulation index, the performance deviation index, and the blockage type specifically includes: The time series data of the congestion accumulation index and the time series data of the performance deviation index are respectively processed by moving average to obtain smoothed time series data; Within a preset fitting time window, linear regression fitting is performed on the smoothed time series data to obtain the first fitting line of the congestion accumulation index and the second fitting line of the performance deviation index. The slope of the first fitted line is extracted as the first future change slope of the blockage accumulation index, and the slope of the second fitted line is extracted as the second future change slope of the performance deviation index. When the first future change slope or the second future change slope exceeds the slope threshold, the final blockage risk level of the water pump to be monitored is determined based on the percentage by which the first future change slope or the second future change slope exceeds the slope threshold.

8. A control system, characterized in that, Includes one or more processors and memory; The memory is coupled to the one or more processors, the memory being used to store computer program code, the computer program code including computer instructions, the one or more processors invoking the computer instructions to cause the control system to perform the method as described in any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that, When the instruction is executed on the control system, it causes the control system to perform the method as described in any one of claims 1-7.

10. A computer program product, characterized in that, When the computer program product is run on the control system, the control system performs the method as described in any one of claims 1-7.

Images