Abnormality diagnosis method and device for drainage equipment of water pump house

Abstract

The invention belongs to the technical field of drainage equipment fault diagnosis, and particularly relates to a water pump room drainage equipment abnormity diagnosis method and device. The method comprises the following steps of collecting multi-source time sequence data of a water pump in a complete operation period, analyzing linear correlation between water pump flow and pipeline pressure based on a Pearson's correlation coefficient, diagnosing water leakage of a pipeline valve, and performing clustering analysis on the water pump flow data by adopting a K-means clustering algorithm, the water pump packing water leakage time period is recognized, and the actual drainage capacity is calculated; the device comprises an attribute monitoring component, a server, an alarm platform and a visual platform. According to the method, the subjectivity of manual inspection and the dependence of deep learning on samples and computing power are overcome, automatic, accurate and self-adaptive diagnosis of the abnormality of the drainage equipment is realized, and the diagnosis efficiency and reliability are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of drainage equipment fault diagnosis technology, specifically relating to a method and device for diagnosing abnormalities in pump room drainage equipment. Background Technology

[0002] In mining production, the normal operation of drainage equipment in pump stations is directly related to production safety and stable production capacity. Currently, in actual production, the judgment of abnormalities in drainage equipment mainly relies on the daily inspections of on-site operators, who judge whether the equipment is abnormal based on their personal experience. This method has obvious limitations: on the one hand, manual judgment is highly subjective, and different personnel have different levels of experience and judgment standards, which can easily lead to misjudgments or omissions; on the other hand, manual inspections are inefficient, cannot achieve real-time monitoring, and inspections are conducted during the intervals between drainage equipment operations, making it difficult to capture instantaneous abnormalities.

[0003] Further analysis of the data characteristics generated during the operation of drainage equipment in mine production pump stations reveals that the flow time-series data often exhibits a dynamic pattern of multiple step-like increases or decreases, such as... Figure 1 As shown, in this type of flow time-series data, the timing, frequency, and duration of data fluctuations are all uncertain, making it difficult for traditional thresholding methods or simple statistical models to effectively identify such non-stationary and irregular fluctuations. Although deep learning neural network models are widely used in current mainstream equipment anomaly diagnosis methods, capable of training diagnostic models based on operational data and possessing a certain degree of versatility, they rely on massive labeled samples and large amounts of computing resources, resulting in high costs and long training cycles in practical applications. Furthermore, for multi-attribute coupled systems such as drainage equipment in pumping stations, historical information and correlation characteristics between different attributes are crucial for accurately identifying anomalies, but deep learning neural network model prediction methods cannot fully mine and utilize this key information, thus limiting the accuracy and reliability of the diagnosis.

[0004] Chinese patent CN115163513A discloses a fault diagnosis method and device for a subway submersible sewage pump. The method includes: real-time acquisition of the pump's operating data; the operating data includes: current data, outflow rate data, and water level data in the collection tank; obtaining a feature vector of the operating data; and inputting the feature vector into a preset submersible sewage pump fault diagnosis expert system for fault diagnosis. The expert system is a module pre-established based on expert judgment rules for various faults, capable of judging faults based on input data. This patent achieves self-diagnosis of submersible sewage pump faults, reducing manpower consumption. However, this patent is essentially based on reasoning using preset fault judgment rules. For different types of pump equipment, the preset fault judgment rules need to be redefined, and these rules need to be adjusted with the service life, resulting in insufficient flexibility and adaptability of the diagnostic method.

[0005] Furthermore, in existing studies applying clustering methods to anomaly diagnosis, most methods are still limited to anomaly detection and usually require pre-specifying the number of clusters, failing to effectively cover the detection of continuous, irregular, step-like changes in time series. Meanwhile, correlation coefficient analysis in anomaly diagnosis is mostly used to determine the general correlation between attributes, and its application in determining the installation location of associated equipment has not been reported. Summary of the Invention

[0006] The purpose of this invention is to provide a method for diagnosing abnormalities in pump room drainage equipment, which can solve the problems of low efficiency and strong subjectivity in existing technologies that rely on manual judgment, as well as the high requirements for samples and computing power and poor adaptability of deep learning. This invention reduces subjective errors and manpower consumption in manual inspections, enhances the adaptability and flexibility of abnormality diagnosis for different equipment types, and reduces management costs. This invention also provides a corresponding diagnostic device.

[0007] To achieve the above objectives, the technical solution of the present invention is as follows: A method for diagnosing abnormalities in pump room drainage equipment includes the following steps: S1. Collect time-series data of the water pump from start to stop within a complete operating cycle. The time-series data includes the water pump operating status, water pump flow rate, drainage pipeline pressure, and pipeline valve position status. S2. Based on time series data, Pearson correlation coefficient is used to analyze the linear correlation between pump flow and drainage pipeline pressure to determine whether there is leakage in pipeline valves. S3. Based on time-series data, the K-means clustering algorithm is used to perform cluster analysis on the flow data during pump operation, identify the specific time of leakage of the pump packing and calculate the actual drainage capacity of the pump.

[0008] Furthermore, in step S1, the pump operating status, pump flow rate, drainage pipeline pressure, and pipeline valve position status all come from different transmission protocols. The pump operating status, pump flow rate, drainage pipeline pressure, and pipeline valve position status are uniformly converted into a two-dimensional sequence of timestamps and attribute values, and linear interpolation is used to fill in the missing values.

[0009] Furthermore, in step S1, the water pump includes a running pump and a standby pump, the pipeline valve includes a pressure valve and a check valve, and the position of the pipeline valve includes the pressure valve being installed above the check valve and the pressure valve being installed below the check valve.

[0010] Furthermore, in step S2, the formula for calculating the Pearson correlation coefficient is:

[0011] Where: r xy Let x be the Pearson correlation coefficient. iLet y be the i-th observation of the pump flow rate. i Let i be the i-th observation value of the drainage pipeline pressure. This represents the average flow rate of the water pump. is the average pressure in the drainage pipeline, and n is the number of samples.

[0012] Furthermore, in step S2, when the pressure valve is installed above the check valve, the linear correlation coefficient r between the operating pump drainage pipeline pressure and the standby pump flow rate is calculated using the Pearson correlation coefficient. xy1 When r xy1 When the pressure is less than -0.8, it is determined that the check valve or pressure valve on the operating pump's drainage pipeline is leaking. When the pressure valve is installed below the check valve, the linear correlation coefficient r between the operating pump flow rate and the operating pump's drainage pipeline pressure is calculated using the Pearson correlation coefficient. xy2 When r xy2 If the value is greater than 0.8, it is determined that the check valve or pressure valve on the drainage pipeline of the operating pump is leaking.

[0013] Furthermore, in step S3, the K-means clustering algorithm is used to calculate the flow time series data of the pump during its stable operating period. The stable operating period of the pump is the data period excluding the initial 10 minutes of pump start-up and the 10 minutes before pump shutdown.

[0014] Furthermore, in step S3, the K-means clustering algorithm calculation formula is as follows:

[0015] Where: WCSS(K) is the sum of the squares within all clusters when the number of clusters is K; K is the number of clusters, j is the j-th cluster; x is each data point in the cluster; u j It is the center of the j-th cluster; by calculating the sum of squares within the clusters WCSS corresponding to different cluster numbers K, we find the x-coordinate value K1 where the rate of decrease of WCSS slows down significantly, and take it as the optimal cluster number K1; The flow time series data is divided into K1 clusters using a clustering algorithm. The cluster with the most data points contains the flow data when the pump is running stably, and the rest contain the flow data when the pump is running abnormally. The timestamp of the flow data when the pump is running abnormally is the leakage period. The center value of the flow data when the pump is running stably is the actual drainage capacity of the pump.

[0016] Where c represents the actual drainage capacity of the water pump. The category number is the cluster containing the most points.

[0017] Furthermore, the elbow rule is used to determine the optimal number of clusters K1. The preset number of clusters is K. The sum of squares within each cluster under each preset K is calculated (WCSS(K)). By plotting the relationship curve between K and WCSS(K), the point where the rate of decline of the curve suddenly slows down is found; the K1 corresponding to this point is the optimal number of clusters. After dividing the flow data into K1 clusters, the cluster with the most data points represents normal flow, and the remaining clusters represent abnormal flow. The time period corresponding to the abnormal flow cluster is the leakage period, and the centroid of the normal flow cluster is the actual drainage capacity of the pump. For the flow data during abnormal pump operation, the data is sorted in descending order according to the number of cluster points. The time series corresponding to the flow data during abnormal pump operation after descending order is obtained, which can reveal the start and end times and duration of the abnormal pump operation, thus providing the diagnostic time for pump leakage.

[0018] A diagnostic device for abnormalities in pump room drainage equipment includes an attribute monitoring component, a server, an alarm platform, and a visualization platform. The equipment attribute monitoring component records the pump operating status, pump flow rate, drainage pipeline pressure, and pipeline valve position status every minute. The server receives the data collected by the attribute monitoring component, calculates the Pearson correlation coefficient, and performs a K-means clustering algorithm to generate real-time alarm information, which is then sent to the alarm platform. The alarm platform receives the real-time alarm information from the pump room drainage equipment and pushes it to maintenance personnel in the form of instant messages. Maintenance personnel process the information online in real time and simultaneously report the situation back to the server. The visualization platform summarizes the alarm information and generates alarm information statistical bar charts, a list of historical diagnostic reports, and abnormal segment data curves.

[0019] Furthermore, data exchange occurs between the server and the alarm platform, and between the server and the visualization platform, via the network.

[0020] The beneficial effects of this invention are as follows: This invention combines Pearson correlation coefficient analysis with K-means clustering algorithm to construct a diagnostic scheme for anomalies in drainage equipment in pumping stations. This method does not rely on massive labeled samples or high-performance computing hardware, effectively overcoming the limitations of high deployment costs and long training cycles of deep learning models in real-world industrial scenarios. It is more suitable for resource-constrained mining pumping station environments and has strong engineering applicability.

[0021] The K-means clustering algorithm determines the optimal number of clusters through the elbow rule, and the Pearson correlation coefficient performs multi-attribute coupling analysis through data correlation. It breaks away from the dependence on the preset fault rule base, is not limited to specific equipment models or service years, and can flexibly adapt to the abnormal diagnosis needs of different types of water pump equipment. It has strong adaptive diagnosis capabilities and good versatility.

[0022] Cluster analysis can accurately identify the specific time period of water pump packing leakage, enabling time-series localization of abnormal events; by combining valve installation location information for correlation discrimination, the fault location of valve leakage in a specific pipeline can be accurately located; based on the clustering results, the centroid of stable operating clusters can be extracted, and the actual drainage capacity of the water pump can be calculated, providing reliable data support for equipment performance evaluation and preventive maintenance.

[0023] This invention changes the traditional operation and maintenance model that relies on manual inspection and experience judgment by automatically collecting data and using algorithms for analysis. It overcomes the drawbacks of manual judgment, such as strong subjectivity, low efficiency and inability to monitor in real time. It realizes objective and real-time diagnosis of equipment anomalies, reduces manpower consumption and the risk of misjudgment and omission, and improves the overall safety management level of pump room drainage equipment. Attached Figure Description

[0024] Figure 1 This is a schematic diagram showing the flow rate and time of the drainage equipment in the pump room in the background technology; Figure 2 This is a flowchart of the method for diagnosing abnormalities in drainage equipment in this invention; Figure 3 This is a flowchart of the method for clustering analysis of water pump flow rate and time series data in this invention; Figure 4 This is a logic flowchart for diagnosing pipeline valve leakage in this invention; Figure 5 This is a data illustration of determining the optimal number of clusters based on the elbow rule in this invention; Figure 6 This is a schematic diagram of the drainage equipment abnormality diagnosis device in this invention; In the picture: 1. Attribute monitoring component; 2. Server; 3. Alarm platform; 4. Visualization platform. Detailed Implementation

[0025] The present invention will now be described and illustrated in detail with reference to the embodiments.

[0026] Example 1 like Figure 1-6 As shown, a method for diagnosing abnormalities in pump room drainage equipment includes the following steps: S1. Collect time-series data of the water pump from start to stop within a complete operating cycle. The time-series data includes the water pump operating status, water pump flow rate, drainage pipeline pressure, and pipeline valve position status. S2. Based on time series data, Pearson correlation coefficient is used to analyze the linear correlation between pump flow and drainage pipeline pressure to determine whether there is leakage in pipeline valves. S3. Based on time-series data, the K-means clustering algorithm is used to perform cluster analysis on the flow data during pump operation, identify the specific time of leakage of the pump packing and calculate the actual drainage capacity of the pump.

[0027] In step S1, the pump operating status, pump flow rate, drainage pipeline pressure, and pipeline valve position status all come from different transmission protocols. The pump operating status, pump flow rate, drainage pipeline pressure, and pipeline valve position status are uniformly converted into a two-dimensional sequence of timestamps and attribute values, and linear interpolation is used to fill in the missing values.

[0028] In step S1, the water pump includes a running pump and a standby pump, the pipeline valves include a pressure valve and a check valve, and the position of the pipeline valves includes the pressure valve being installed above the check valve and the pressure valve being installed below the check valve.

[0029] In step S2, the formula for calculating the Pearson correlation coefficient is:

[0030] Where: r xy Let x be the Pearson correlation coefficient. i Let y be the i-th observation of the pump flow rate. i Let i be the i-th observation value of the drainage pipeline pressure. This represents the average flow rate of the water pump. is the average pressure in the drainage pipeline, and n is the number of samples.

[0031] In step S2, when the pressure valve is installed above the check valve, the linear correlation coefficient r between the operating pump drainage pipeline pressure and the standby pump flow rate is calculated using the Pearson correlation coefficient. xy1 When r xy1 When the pressure is less than -0.8, it is determined that the check valve or pressure valve on the operating pump's drainage pipeline is leaking. When the pressure valve is installed below the check valve, the linear correlation coefficient r between the operating pump flow rate and the operating pump's drainage pipeline pressure is calculated using the Pearson correlation coefficient. xy2 When r xy2 If the value is greater than 0.8, it is determined that the check valve or pressure valve on the drainage pipeline of the operating pump is leaking.

[0032] In step S3, the K-means clustering algorithm is used to calculate the flow time series data of the pump during the stable operating period. The stable operating period of the pump is the data period excluding the first 10 minutes after starting the pump and the first 10 minutes before shutting down the pump.

[0033] In step S3, the K-means clustering algorithm calculation formula is as follows:

[0034] Where: WCSS(K) is the sum of the squares within all clusters when the number of clusters is K; K is the number of clusters, j is the j-th cluster; x is each data point in the cluster; u j It is the center of the j-th cluster; by calculating the sum of squares within the clusters WCSS corresponding to different cluster numbers K, we find the x-coordinate value K1 where the rate of decrease of WCSS slows down significantly, and take it as the optimal cluster number K1; The flow time series data is divided into K1 clusters using a clustering algorithm. The cluster with the most data points contains the flow data when the pump is running stably, and the rest contain the flow data when the pump is running abnormally. The timestamp of the flow data when the pump is running abnormally is the leakage period. The center value of the flow data when the pump is running stably is the actual drainage capacity of the pump.

[0035] Where c represents the actual drainage capacity of the water pump. The category number is the cluster containing the most points.

[0036] In step S3, the elbow rule is used to determine the optimal number of clusters K1. The preset number of clusters is K. The sum of squares within each cluster under each preset K is calculated as WCSS(K). By plotting the relationship curve between K and WCSS(K), the point where the rate of decline of the curve suddenly slows down is found. The K1 corresponding to this point is the optimal number of clusters. After dividing the flow data into K1 clusters, the cluster with the most data points represents normal flow, and the remaining clusters represent abnormal flow. The time period corresponding to the abnormal flow cluster is the leakage period, and the centroid of the normal flow cluster is the actual drainage capacity of the water pump. For the flow data of the water pump during abnormal operation, it is sorted in descending order according to the number of cluster points contained. The time series corresponding to the flow data of each water pump during abnormal operation after descending order is obtained. The start and end times and duration of the abnormal operation of the water pump can be obtained, and thus the diagnosis time of water pump leakage can be given.

[0037] A diagnostic device for abnormalities in pump room drainage equipment includes an attribute monitoring component 1, a server 2, an alarm platform 3, and a visualization platform 4. The equipment attribute monitoring component 1 records the pump operating status, pump flow rate, drainage pipeline pressure, and pipeline valve position status every minute. The server 2 receives the data collected by the attribute monitoring component 1, calculates the Pearson correlation coefficient, and performs a K-means clustering algorithm to generate real-time alarm information, which is then sent to the alarm platform 3. The alarm platform 3 receives the real-time alarm information from the pump room drainage equipment and pushes it to maintenance personnel in the form of instant messages. The maintenance personnel process the information online in real time and simultaneously report the situation back to the server 2. The visualization platform 4 summarizes the alarm information and generates alarm information statistical bar charts, historical diagnostic report lists, and abnormal segment data curves.

[0038] Server 2 and alarm platform 3, as well as server 2 and visualization platform 4, exchange data via network.

[0039] Example 2 This embodiment takes the drainage equipment of a mine pumping station as the implementation scenario. The pumping station is equipped with a running pump and a standby pump. A check valve and a pressure valve are installed on the drainage pipeline, and the pressure valve is located below the check valve.

[0040] The diagnostic method includes the following steps: S1. Collect time-series data of the water pump from start to stop within a complete operating cycle. The time-series data includes the water pump operating status, water pump flow rate, drainage pipeline pressure, and pipeline valve position status. Since the data comes from sensors and control systems using different protocols, all attribute data were first converted into a two-dimensional sequence of "timestamp-attribute value". Missing values ​​were filled using linear interpolation. To eliminate interference from surges and water hammer effects during pump start-up and shutdown, data from the initial 10 minutes of pump start-up and the 10 minutes before pump shutdown were removed, retaining only data from the stable operating period for subsequent analysis.

[0041] S2. Based on the preprocessed time-series data, the Pearson correlation coefficient was used to analyze the linear correlation between the operating pump flow rate and the pressure in its drainage pipeline. The formula for calculating the Pearson correlation coefficient is as follows:

[0042] Where: r xy Let x be the Pearson correlation coefficient. i Let y be the i-th observation of the pump flow rate. i Let i be the i-th observation value of the drainage pipeline pressure. This represents the average flow rate of the water pump. is the average pressure in the drainage pipeline, and n is the sample size; In this embodiment, the pressure valve is installed below the check valve. When leakage occurs in the external discharge pipeline valve, the leak point will dominate the operating load characteristics of the pump, causing the pump flow rate and the discharge pipeline pressure to show a significant synchronous change trend. Therefore, if the calculated Pearson correlation coefficient r xy A value greater than 0.8 indicates a strong positive correlation between flow rate and pressure, which suggests that there is a leak in the check valve or pressure valve on the external discharge pipeline corresponding to the operating pump.

[0043] S3. The K-means clustering algorithm is used to analyze the flow time series data of the pump during the stable operation period to identify the specific time of leakage of packing and to calculate the actual drainage capacity of the pump.

[0044] Determining the optimal number of clusters K1: Using the elbow rule, the preset range of K values ​​is 1, 2, 3, ... K1, K1+1...; Calculate the sum of squares (WCSS) within each cluster corresponding to each K value and plot the K-WCSS relationship curve. Observation shows that when K=K1, the rate of descent of the curve slows down significantly, that is, K1 is the "inflection point" where the rate of descent slows down significantly, therefore, the optimal number of clusters is determined to be K1.

[0045] K-means clustering is performed: the flow time series data is clustered with K=K1, dividing the data into K1 clusters; the number of data points in each cluster is counted, and the cluster with the most data points is identified as the flow data of the pump in "stable operation" state. The value of its cluster center, i.e., the average value, is calculated as the actual drainage capacity of the pump. The remaining clusters are identified as "abnormal operation" state, corresponding to the period when the pump packing leaks. By extracting the timestamps corresponding to the abnormal clusters, the start time, end time, and duration of the leakage event can be accurately located.

[0046] The K-means clustering algorithm calculation formula is:

[0047] Where: WCSS(K) is the sum of the squares within all clusters when the number of clusters is K; K is the number of clusters, j is the j-th cluster; x is each data point in the cluster; u j It is the center of the j-th cluster; by calculating the sum of squares within the clusters WCSS corresponding to different cluster numbers K, we find the x-coordinate value K1 where the rate of decrease of WCSS slows down significantly, and take it as the optimal cluster number K1; The flow time series data is divided into K1 clusters using a clustering algorithm. The cluster with the most data points contains the flow data when the pump is running stably, and the rest contain the flow data when the pump is running abnormally. The timestamp of the flow data when the pump is running abnormally is the leakage period. The center value of the flow data when the pump is running stably is the actual drainage capacity of the pump.

[0048] Where c represents the actual drainage capacity of the water pump. The category number is the cluster containing the most points.

[0049] The operation process of the diagnostic device in this embodiment is as follows: The attribute monitoring component 1, deployed in the pump room, automatically records the pump operating status, pump flow rate, drainage pipeline pressure, and pipeline valve position status data once per minute. Server 2 receives monitoring data and executes the diagnostic algorithms S1 to S3 above. After the diagnosis is completed, it generates an alarm message containing information such as the anomaly type, location, and time. Alarm platform 3 receives alarm information sent by server 2 and pushes it to the mobile terminal of on-site equipment maintenance personnel in the form of instant message. On-site equipment maintenance personnel can fill in the handling measures online and feed them back to server 2 to form a closed loop of handling. The visualization platform 4 obtains data from server 2, dynamically generates and displays alarm information statistical bar charts, historical diagnostic report lists, and traffic pressure data curves for abnormal periods.

[0050] The remaining steps are the same as in Example 1.

[0051] Example 3 This embodiment takes the drainage equipment of a mine pumping station as the implementation scenario. The pumping station is equipped with a running pump and a standby pump. A check valve and a pressure valve are installed on the drainage pipeline, and the pressure valve is located above the check valve.

[0052] The diagnostic method includes the following steps: S1. Collect time-series data of the water pump from start to stop within a complete operating cycle. The time-series data includes the water pump operating status, water pump flow rate, drainage pipeline pressure, and pipeline valve position status. Since the data comes from sensors and control systems using different protocols, all attribute data were first converted into a two-dimensional sequence of "timestamp-attribute value". Missing values ​​were filled using linear interpolation. To eliminate interference from surges and water hammer effects during pump start-up and shutdown, data from the initial 10 minutes of pump start-up and the 10 minutes before pump shutdown were removed, retaining only data from the stable operating period for subsequent analysis.

[0053] S2. Based on the preprocessed time-series data, the Pearson correlation coefficient was used to analyze the linear correlation between the operating pump flow rate and the pressure in its drainage pipeline. The formula for calculating the Pearson correlation coefficient is as follows:

[0054] Where: r xy Let x be the Pearson correlation coefficient. i Let y be the i-th observation of the pump flow rate. i Let i be the i-th observation value of the drainage pipeline pressure. This represents the average flow rate of the water pump. is the average pressure in the drainage pipeline, and n is the sample size; In this embodiment, because the pressure valve is installed above the check valve, leakage pressure from the valves in the operating pump pipeline will impact the standby pump pipeline, causing abnormal backflow in the standby pump. This abnormality manifests as a negative correlation between the pressure in the operating pump's drainage pipeline and the flow rate of the standby pump. The linear correlation coefficient r between the pressure in the operating pump's drainage pipeline and the flow rate of the standby pump is calculated. xy If rxy If 1 < -0.8, it indicates that the pressure of the operating pump is strongly negatively correlated with the flow rate of the standby pump, which means that there is a leak in the check valve or pressure valve on the drainage pipeline of the operating pump.

[0055] S3. The K-means clustering algorithm is used to analyze the flow time series data of the pump during the stable operation period to identify the specific time of leakage of packing and to calculate the actual drainage capacity of the pump.

[0056] Determining the optimal number of clusters K1: Using the elbow rule, the preset range of K values ​​is 1, 2, 3, ... K1, K1+1...; Calculate the sum of squares (WCSS) within each cluster corresponding to each K value and plot the K-WCSS relationship curve. Observation shows that when K=K1, the rate of descent of the curve slows down significantly, that is, K1 is the "inflection point" where the rate of descent slows down significantly, therefore, the optimal number of clusters is determined to be K1.

[0057] K-means clustering is performed: the flow time series data is clustered with K=K1, dividing the data into K1 clusters; the number of data points in each cluster is counted, and the cluster with the most data points is identified as the flow data of the pump in "stable operation" state. The value of its cluster center, i.e., the average value, is calculated as the actual drainage capacity of the pump. The remaining clusters are identified as "abnormal operation" state, corresponding to the period when the pump packing leaks. By extracting the timestamps corresponding to the abnormal clusters, the start time, end time, and duration of the leakage event can be accurately located.

[0058] The K-means clustering algorithm calculation formula is:

[0059] Where: WCSS(K) is the sum of the squares within all clusters when the number of clusters is K; K is the number of clusters, j is the j-th cluster; x is each data point in the cluster; u j It is the center of the j-th cluster; by calculating the sum of squares within the clusters WCSS corresponding to different cluster numbers K, we find the x-coordinate value K1 where the rate of decrease of WCSS slows down significantly, and take it as the optimal cluster number K1; The flow time series data is divided into K1 clusters using a clustering algorithm. The cluster with the most data points contains the flow data when the pump is running stably, and the rest contain the flow data when the pump is running abnormally. The timestamp of the flow data when the pump is running abnormally is the leakage period. The center value of the flow data when the pump is running stably is the actual drainage capacity of the pump.

[0060] Where c represents the actual drainage capacity of the water pump. The category number is the cluster containing the most points.

[0061] The operation process of the diagnostic device in this embodiment is as follows: The attribute monitoring component 1, deployed in the pump room, automatically records the pump operating status, pump flow rate, drainage pipeline pressure, and pipeline valve position status data once per minute. Server 2 receives monitoring data and executes the diagnostic algorithms S1 to S3 above. After the diagnosis is completed, it generates an alarm message containing information such as the anomaly type, location, and time. Alarm platform 3 receives alarm information sent by server 2 and pushes it to the mobile terminal of on-site equipment maintenance personnel in the form of instant message. On-site equipment maintenance personnel can fill in the handling measures online and feed them back to server 2 to form a closed loop of handling. The visualization platform 4 obtains data from server 2, dynamically generates and displays alarm information statistical bar charts, historical diagnostic report lists, and traffic pressure data curves for abnormal periods.

[0062] The remaining steps are the same as in Example 1.

Claims

1. A water pump house drainage equipment abnormality diagnosis method characterized by comprising: The method comprises the following steps: S1, collecting time sequence data of the water pump from start to stop in a complete running cycle, the time sequence data comprising water pump running state, water pump flow, drain pipeline pressure and pipeline valve position state; S2, based on the time sequence data, using Pearson correlation coefficient to analyze the linear correlation between the water pump flow and the drain pipeline pressure, and judging whether there is pipeline valve leakage; S3, based on the time sequence data, using K-means clustering algorithm to cluster analyze the flow data when the water pump is running, identifying the specific time of water pump packing leakage and calculating the actual water pump drainage capacity.

2. The water pump house drainage equipment abnormality diagnosis method according to claim 1, characterized by, In step S1, the water pump running state, the water pump flow, the drain pipeline pressure and the pipeline valve position state are all from different transmission protocols, and the water pump running state, the water pump flow, the drain pipeline pressure and the pipeline valve position state are uniformly converted into a two-dimensional sequence of timestamp and attribute value, and the linear interpolation method is used to fill in the null value.

3. The water pump house drainage equipment abnormality diagnosis method according to claim 1, characterized by, In step S1, the water pump comprises a running pump and a standby pump, the pipeline valve comprises a pressure valve and a check valve, and the pipeline valve position state comprises the pressure valve installed above the check valve and the pressure valve installed below the check valve.

4. The water pump house drainage equipment abnormality diagnosis method according to claim 3, characterized by, In step S2, the Pearson correlation coefficient calculation formula is: where: r xy is the Pearson correlation coefficient, x i is the ith observation of the water pump flow rate, y i is the ith observation of the drain line pressure, is the average of the water pump flow rate, is the average of the drain line pressure, and n is the number of samples.

5. The water pump house drainage equipment abnormality diagnosis method according to claim 4, characterized by, Step S2, when the pressure valve is installed above the check valve, the Pearson correlation coefficient is used to calculate the linear correlation coefficient r between the running pump drain line pressure and the standby pump flow xy1 , when r xy1 <-0.8, it is determined that the check valve or pressure valve on the running pump drain line leaks; when the pressure valve is installed below the check valve, the Pearson correlation coefficient is used to calculate the linear correlation coefficient r between the running pump flow and the running pump drain line pressure xy2 , when r xy2 >0.8, it is determined that the check valve or pressure valve on the running pump drain line leaks.

6. The water pump house drainage equipment abnormality diagnosis method according to claim 3, characterized by In step S3, the K-means clustering algorithm is used to calculate the flow time sequence data of the running pump in the stable running period, and the stable running period of the running pump is the data period excluding the initial 10 minutes of starting the pump and the last 10 minutes of stopping the pump.

7. The water pump house drainage equipment abnormality diagnosis method according to claim 3, characterized by, In step S3, the K-means clustering algorithm calculation formula is: Where: WCSS(K) is the sum of the squares within all clusters when the number of clusters is K; K is the number of clusters, j is the j-th cluster; x is each data point in the cluster; u j It is the center of the j-th cluster; by calculating the sum of squares within the clusters WCSS corresponding to different cluster numbers K, we find the x-coordinate value K1 where the rate of decrease of WCSS slows down significantly, and take it as the optimal cluster number K1; The flow time sequence data is divided into K1 clusters by the clustering algorithm, wherein the cluster containing the most data points is the flow data of the stable running of the water pump, and the rest is the flow data of the abnormal running of the water pump; the time stamp of the flow data of the abnormal running of the water pump is the leakage period; the center value of the flow data of the stable running of the water pump is the actual drainage capacity of the water pump; Wherein c is the actual water pump discharge capacity, is the class number containing the cluster with the most points.

8. The water pump house drainage equipment abnormality diagnosis method according to claim 7, characterized by, In step S3, the elbow rule is used to determine the optimal cluster number K1, the preset cluster number is K, the within-cluster sum of squares WCSS(K) of all clusters under each preset K is calculated, the relationship curve of K and WCSS(K) is drawn, the point where the curve suddenly slows down is found, and the K1 corresponding to the point is the optimal cluster number; after dividing the flow data into K1 clusters, the cluster with the most data points represents normal flow, and the rest represents abnormal flow; the time period corresponding to the abnormal flow cluster is the leakage period, and the centroid of the normal flow cluster is the actual drainage capacity of the water pump; for the flow data of the abnormal running of the water pump, the number of cluster points contained is arranged in descending order, the time sequence corresponding to each flow data of the abnormal running of the water pump in descending order is obtained, the start and end time and the duration of the abnormal running of the water pump are obtained, and then the diagnosis time of the water pump leakage is given.

9. A diagnosis device based on the water pump house drainage equipment abnormality diagnosis method according to any one of claims 1 to 8, characterized by It comprises attribute monitoring component (1), server (2), alarm platform (3) and visualization platform (4), the equipment attribute monitoring component (1) records water pump running state, water pump flow, drain pipeline pressure and pipeline valve position state once a minute; the server (2) receives the data collected by the attribute monitoring component (1), calculates the Pearson correlation coefficient and executes the K-means clustering algorithm, generates real-time alarm information, and sends the real-time alarm information to the alarm platform (3); the alarm platform (3) receives the real-time alarm information of the water pump house drainage equipment, and pushes it to the operation and maintenance personnel in the form of instant message, the operation and maintenance personnel fill in and report the processing according to the instant information online, and feed back the situation to the server (2); the visualization platform (4) summarizes the alarm information, generates alarm information statistical column chart, historical diagnosis report list and abnormal segment data curve.

10. The diagnosis device of the water pump house drainage equipment abnormality diagnosis method according to claim 9, characterized by Data interaction is carried out between the server (2) and the alarm platform (3), and between the server (2) and the visualization platform (4) through the network.

Images