A time-series correlation-based steady voltage pump starting event classification method and system

Abstract

The present application relates to the technical field of fire engineering and Internet of Things data analysis, and particularly relates to a time-series correlation-based stable pressure pump starting event classification method and system, which comprises the following steps: in response to a stable pressure pump starting trigger signal, determining a stable pressure pump starting event to be classified and collecting corresponding event correlation parameters; extracting event discrimination features from the event correlation parameters and inputting the event discrimination features into a preset classification discrimination model; matching the event discrimination features according to classification discrimination conditions corresponding to each event category in the preset classification discrimination model to obtain at least one candidate event category; and determining a final event category of the stable pressure pump starting event to be classified according to a classification priority in the preset classification discrimination model. The present application can realize automatic classification of stable pressure pump starting events without additional sensors, distinguish different event types, improve the recognition accuracy of fire Internet of Things platforms for pipe network abnormalities, equipment failures and actual water use events, and reduce false positives and false negatives.

Description

Technical Field

[0001] This invention relates to the fields of fire protection engineering and Internet of Things data analysis technology, and in particular to a method and system for classifying pressure-stabilizing pump start-up events based on time-series correlation. This method is used to automatically classify pressure-stabilizing pump start-up events in fire water supply systems by combining the pressure time-series characteristics before and after the start-up of the pressure-stabilizing pump, the pump's daily start-up and shutdown statistics, and the time-series correlation of the water flow indicator. Background Technology

[0002] In fire-fighting water supply systems, pressure-stabilizing pumps, also known as pressure-maintaining pumps, are typically used to automatically activate and replenish pressure when the pressure in the fire-fighting water supply network drops, maintaining the network within the specified operating pressure range. Therefore, the activation of pressure-stabilizing pumps reflects pressure fluctuations in the fire-fighting water supply network and is a crucial signal for assessing the network's operational status, equipment status, and any abnormal water usage. With the development of fire-fighting IoT technology, existing IoT platforms can now collect data on the activation time, shutdown time, number of activations, runtime, and network pressure of pressure-stabilizing pumps, and provide anomaly alerts based on the number of activations or activation frequency.

[0003] However, existing IoT platforms typically only count the number or frequency of voltage regulator pump start-up events, without further identifying the cause of each voltage regulator pump start-up event. The existing technology has at least the following shortcomings: (1) There may be situations caused by real fire extinguishing water or fire drill water in the start-up of the pressure stabilizing pump. Such events usually have a high priority for handling. However, the existing platform does not perform time-series correlation analysis between the start-up time of the pressure stabilizing pump and the action record of the water flow indicator in the same fire water supply system. As a result, the suspected real water use signal is easily submerged by a large number of ordinary pressure replenishment events and is difficult to identify in time. (2) The low-pressure start-up of the pressure stabilizing pump may include two different situations: serious failure and occasional pressure disturbance. Among them, serious failure usually manifests as the pressure being lower than the lower threshold multiple times before the start-up, while occasional low-pressure start-up may only be caused by short-term disturbance. The existing method is difficult to distinguish between the two based on the pressure before the start-up and the cumulative number of low-pressure start-ups on the day, resulting in inaccurate judgment of the urgency of handling. (3) Planned operations such as maintenance water discharge test and functional test may also cause the pressure stabilizing pump to start. If the existing platform cannot identify the start-up time, the number of times it occurs on the day and the pressure recovery status, it is easy to misjudge the maintenance operation as an abnormal event, thereby interfering with the operation statistics and equipment health assessment. Summary of the Invention

[0004] To address the problems in existing fire protection IoT platforms that only record the number of times pressure-stabilizing pumps are started, their start frequency, or pressure alarms, without combining pressure timing, water flow indicator action timing, and daily start-stop statistics to classify the causes of start-ups, leading to difficulties in identifying suspected real water usage events, distinguishing the degree of low-pressure failure, and making maintenance tests prone to misjudgment, this application provides a time-series correlation-based method and system for classifying pressure-stabilizing pump start-up events. This method achieves automatic classification of pressure-stabilizing pump start-up events without the need for additional sensors, providing data support for pipeline network status diagnosis and maintenance decisions. The objective of this invention can be achieved through the following technical solutions: This application provides a method for classifying pressure-stabilizing pump start-up events based on time-series correlation, including: Step S1: Respond to the pressure stabilizing pump start-up trigger signal, determine the pressure stabilizing pump start-up event to be classified, and collect the corresponding event association parameters. The event association parameters include pressure stabilizing pump start-up and shutdown parameters, pipeline pressure timing parameters, and water flow indicator action parameters. Step S2: Extract event discrimination features from event association parameters and input the event discrimination features into a preset classification discrimination model. The preset classification discrimination model will match according to the classification discrimination conditions corresponding to each event category to obtain at least one candidate event category. Step S3: Based on the classification priority in the preset classification and discrimination model, determine the final event category of the pressure stabilizing pump start-up event to be classified from at least one candidate event category. The final event category is one of the following: suspected real water use event, frequent start-stop event, severe low pressure start-up event of pressure stabilizing pump, occasional low pressure start-up event of pressure stabilizing pump, suspected maintenance test event, one-time pressure adjustment event, and unclassified event.

[0005] Further, step S1 includes, The platform responds to the pressure stabilizing pump start-up trigger signal, which is a trigger signal generated when the platform detects that the pressure stabilizing pump has switched from a stopped state to a running state. Based on the start-up trigger signal of the pressure stabilizing pump, the start-up time of the pressure stabilizing pump is recorded, and a single operation process of the pressure stabilizing pump starting from the start-up time is identified as the pressure stabilizing pump start-up event to be classified. Based on the start-up time of the pressure-stabilizing pump and the system identifier to which the pressure-stabilizing pump belongs in the start-up events to be classified, the corresponding data collection range is determined and the corresponding event-related parameters are collected.

[0006] Furthermore, the start-up and shutdown parameters of the pressure stabilizing pump include the pressure stabilizing pump identifier, the system identifier to which the pressure stabilizing pump belongs, the start time, stop time, running time, number of start-ups and shutdowns per day, and average running time; the pipeline pressure timing parameters include the pressure sampling time, pressure sampling value, pressure before start-up, pressure recovery status after stop, and single pressure change; the water flow indicator action parameters include the water flow indicator identifier, the system identifier to which the water flow indicator belongs, action time, action status, and action duration.

[0007] Further, in step S2, the event discrimination features are extracted from the event association parameters. Extract the first, second, third, fourth, and fifth discriminant features from the event association parameters, where, The first discriminant features include the start time of the pressure stabilizing pump, the system identifier of the pressure stabilizing pump, the system identifier of the water flow indicator, and the action time. These features are used to determine whether the start-up event of the pressure stabilizing pump to be classified has a temporal correlation with the action of the water flow indicator in the same fire water supply system, corresponding to a suspected real water use event. The second discriminant features include the number of start-stop cycles and the average runtime, which are used to determine whether the pressure stabilizing pump start-up event to be classified corresponds to a frequent start-stop event. The third discriminant features include the pressure before startup and the cumulative number of low-pressure startups on the day, which are used to determine whether the startup event of the pressure stabilizing pump to be classified corresponds to a severe low-pressure startup event of the pressure stabilizing pump or an occasional low-pressure startup event of the pressure stabilizing pump. The fourth discriminant feature includes the start time of the pressure stabilizing pump, the number of starts on the same day, and the pressure recovery status after shutdown, which are used to determine whether the pressure stabilizing pump start-up event to be classified corresponds to a suspected maintenance test event; The fifth discriminant feature includes the single pressure change and the pressure stability state after startup, which are used to determine whether the startup event of the pressure-stabilizing pump to be classified corresponds to a one-time pressure regulation event.

[0008] Further, in step S2, the event discrimination features are input into a preset classification discrimination model, which then matches the events according to the classification discrimination conditions corresponding to each event category to obtain at least one candidate event category, including: The first discrimination feature is matched with the discrimination conditions for suspected real water use. If there is a water flow indicator action record in the preset time window corresponding to the start time of the pressure stabilizing pump that belongs to the same fire water supply system as the start event of the pressure stabilizing pump to be classified, then it corresponds to the suspected real water use event. The second discriminant feature is matched with the frequent start-stop classification criteria. If the number of start-stops of the pressure stabilizing pump on a given day is greater than the preset frequent start-stop number threshold and the average running time is less than or equal to the preset short running time threshold, then it is a frequent start-stop event. The third discriminant feature is matched with the low-pressure start classification criteria. If the pressure before start is lower than the preset low-pressure threshold, it is either a severe low-pressure start event of the pressure stabilizing pump or an occasional low-pressure start event of the pressure stabilizing pump. The fourth discrimination feature is matched with the discrimination conditions for suspected maintenance test. If the pressure stabilizing pump start-up event to be classified occurs in a preset daytime period, the same pressure stabilizing pump starts only once in the same statistical day, and the pressure returns to the preset working pressure range after stopping, then it is a suspected maintenance test event. The fifth discrimination feature is matched with the one-time pressure regulation classification discrimination condition. When the single pressure change is greater than the preset multiple of the normal pump circulation pressure amplitude, and the pipeline pressure after the start-up event of the pressure-stabilizing pump to be classified is stable at the adjusted pressure level, it is a one-time pressure regulation event.

[0009] Furthermore, the third discriminant feature is matched with the low-pressure start classification criteria. If the pressure before start-up is lower than the preset low-pressure threshold, it is classified as a severe low-pressure start-up event of the pressure-stabilizing pump or an occasional low-pressure start-up event of the pressure-stabilizing pump. This also includes... When the pressure before startup is lower than the preset low pressure threshold, and the number of low pressure starts within the same statistical day is greater than the preset severe start number threshold, it is determined as a severe low pressure start event of the pressure stabilizing pump. When the pressure before startup is lower than the preset low-pressure threshold, and the number of low-pressure starts within the same statistical day is lower than the preset severe start number threshold, it is determined as an occasional low-pressure start event of the pressure stabilizing pump.

[0010] Further, in step S2, the event discrimination features are input into a preset classification discrimination model, which then matches the events according to the classification discrimination conditions corresponding to each event category to obtain at least one candidate event category. This also includes... When the pressure-stabilizing pump start-up event to be classified does not match any of the candidate event categories such as suspected real water usage event, frequent start-stop event, severe low-pressure start-up event of pressure-stabilizing pump, occasional low-pressure start-up event of pressure-stabilizing pump, suspected maintenance test event, and one-time pressure adjustment event, the pressure-stabilizing pump start-up event to be classified is determined as an unclassified event, and a review prompt message is generated to prompt manual review.

[0011] Furthermore, in step S3, the classification priority is from high to low as follows: suspected real water use event, frequent start-stop event, severe low pressure start event of pressure stabilizing pump, occasional low pressure start event of pressure stabilizing pump, suspected maintenance test event, and one-time pressure adjustment event.

[0012] Furthermore, it also includes step S4, which generates corresponding handling priority information based on the determined final event category, and outputs the final event category and handling priority information to the platform. When the final event category is suspected actual water usage event, the highest priority notification message is generated; when the final event category is frequent start-stop event or severe low-pressure start event of pressure stabilizing pump, a high priority alarm message is generated; when the final event category is occasional low-pressure start event of pressure stabilizing pump, a medium priority monitoring message is generated; when the final event category is suspected maintenance test event or one-time pressure adjustment event, normal record information is generated; when the final event category is unclassified event, a manual review prompt message is generated.

[0013] Furthermore, when the final event category is a suspected genuine water usage incident, the highest priority notification information is generated, including: When the final event category is suspected real water use event, the real water use event association verification is triggered; Acquire the start signal of the pressure stabilizing pump, the action signal of the water flow indicator, and the signal of the decrease in pipeline pressure, and determine whether the three signals are all within the same preset time window; When all three signals are within the same preset time window and their system identifiers match, the verification result of the real water use event is output, and the confidence level of the verification result of the real water use event is raised to high confidence. The preset time window is a time window formed by extending a preset time period forward and backward based on the occurrence time of any signal. When the verification result of a real water use event is high confidence, a high-priority alarm notification is generated; when only the condition of a suspected real water use event is met, a second-lower priority notification is generated.

[0014] Based on the same inventive concept, this application also provides a time-series correlation-based classification system for pressure-regulating pump start-up events, employing the time-series correlation-based classification method for pressure-regulating pump start-up events as described above, including: The response module is used to respond to the start-up trigger signal of the pressure stabilizing pump, determine the start-up event of the pressure stabilizing pump to be classified, and collect the corresponding event-related parameters, including the start-up and stop parameters of the pressure stabilizing pump, the pipeline pressure timing parameters, and the action parameters of the water flow indicator. The matching module is used to extract event discrimination features from event association parameters and input the event discrimination features into a preset classification discrimination model. The preset classification discrimination model matches according to the classification discrimination conditions corresponding to each event category to obtain at least one candidate event category. The output module is used to determine the final event category of the pressure-stabilizing pump start-up event to be classified from at least one candidate event category according to the classification priority in the preset classification and discrimination model. The final event category is one of the following: suspected real water use event, frequent start-stop event, severe low pressure start-up event of pressure-stabilizing pump, occasional low pressure start-up event of pressure-stabilizing pump, suspected maintenance test event, one-time pressure regulation event, and unclassified event.

[0015] Compared with the prior art, the present invention has at least one of the following technical advantages: This invention, through joint analysis of pressure-stabilizing pump start-up and shutdown information, pipeline pressure timing information, and water flow indicator action information, can automatically classify pressure-stabilizing pump start-up events without adding additional sensors. It distinguishes between different event types such as suspected actual water use, frequent start-up and shutdown, low-pressure start-up, maintenance testing, one-time pressure adjustment, and unclassified events. This improves the accuracy of the fire protection IoT platform in identifying pipeline anomalies, equipment failures, and actual water use events, reduces false alarms and missed alarms, and provides more refined data support for alarm classification, maintenance scheduling, and fire protection system health assessment. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below: Figure 1 This is a decision tree diagram of the time-series correlation-based classification method for pressure-stabilizing pump start-up events in an embodiment of the present invention. Figure 2 This is a structural diagram of the pressure-stabilizing pump start-up event classification system based on time-series correlation in an embodiment of the present invention. Detailed Implementation

[0017] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. It should be noted that the specific embodiments described herein are merely some embodiments of the present application, not all embodiments, and are only used to explain the present application and are not intended to limit the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present application.

[0018] First Embodiment In existing fire-fighting water supply systems, pressure-stabilizing pumps are typically used to maintain stable pipeline pressure. When the pipeline pressure drops below a set range, the pressure-stabilizing pump automatically starts to replenish the pressure. With the application of fire-fighting IoT platforms, operational information such as the start / stop status of pressure-stabilizing pumps, pipeline pressure, and the action of water flow indicators can be collected and uploaded in real time. However, most existing platforms only statistically analyze the number of pressure-stabilizing pump starts, start frequency, or low-pressure alarms, without further analyzing the specific causes of each pressure-stabilizing pump start event. In practical applications, pressure-stabilizing pump starts may be caused by various situations, such as actual water usage, pipeline leakage, low-pressure failure of the pressure-stabilizing pump, maintenance water discharge testing, and system pressure regulation. Different situations have different priorities and management significance. If alarms are only based on the number of starts or pressure thresholds, it is easy to cause problems such as the flooding of actual water usage events, confusion of fault levels, and misjudgment of maintenance operations.

[0019] During the operation monitoring of fire protection IoT and the operation and maintenance management of fire water supply systems, the applicant discovered that pressure-stabilizing pump start-up events are not isolated signals. They exhibit clear temporal correlations with changes in pipeline pressure before and after start-up, start-stop patterns within the same statistical day, and the actions of water flow indicators within the same fire water supply system. Based on this understanding, a temporal correlation-based classification method for pressure-stabilizing pump start-up events is proposed. This method collects event correlation parameters after the pressure-stabilizing pump is triggered, extracts event discrimination features related to start-stop, pressure, and water flow actions, and uses a preset classification discrimination model to match candidate event categories. Finally, the event category is determined according to classification priority. This enables automatic identification of suspected real water usage events, frequent start-stop events, severe low-pressure start-up events of pressure-stabilizing pumps, occasional low-pressure start-up events of pressure-stabilizing pumps, suspected maintenance test events, one-time pressure adjustment events, and unclassified events. This provides data support for fire water supply network status diagnosis, alarm classification, and maintenance decisions. The specific implementation method is as follows: This application provides a method for classifying pressure-stabilizing pump start-up events based on time-series correlation, including: Step S1: Respond to the pressure stabilizing pump start-up trigger signal, determine the pressure stabilizing pump start-up event to be classified, and collect the corresponding event association parameters. The event association parameters include pressure stabilizing pump start-up and shutdown parameters, pipeline pressure timing parameters, and water flow indicator action parameters. Step S2: Extract event discrimination features from event association parameters and input the event discrimination features into a preset classification discrimination model. The preset classification discrimination model will match according to the classification discrimination conditions corresponding to each event category to obtain at least one candidate event category. Step S3: Based on the classification priority in the preset classification and discrimination model, determine the final event category of the pressure stabilizing pump start-up event to be classified from at least one candidate event category. The final event category is one of the following: suspected real water use event, frequent start-stop event, severe low pressure start-up event of pressure stabilizing pump, occasional low pressure start-up event of pressure stabilizing pump, suspected maintenance test event, one-time pressure adjustment event, and unclassified event.

[0020] Further, step S1 includes, The platform responds to the pressure stabilizing pump start-up trigger signal, which is a trigger signal generated when the platform detects that the pressure stabilizing pump has switched from a stopped state to a running state. Based on the start-up trigger signal of the pressure stabilizing pump, the start-up time of the pressure stabilizing pump is recorded, and a single operation process of the pressure stabilizing pump starting from the start-up time is identified as the pressure stabilizing pump start-up event to be classified. Based on the start-up time of the pressure-stabilizing pump and the system identifier to which the pressure-stabilizing pump belongs in the start-up events to be classified, the corresponding data collection range is determined and the corresponding event-related parameters are collected.

[0021] Specifically, the fire protection IoT platform continuously receives operating status data uploaded by the pressure-stabilizing pump control cabinet, pressure sensors, and water flow indicators. When the operating status of a pressure-stabilizing pump changes from stopped to running, the platform determines that the pump has started and generates a pressure-stabilizing pump start-up trigger signal. For example, if pressure-stabilizing pump P1 in a building's fire hydrant system switches from stopped to running at 14:32:00, the platform records 14:32:00 as the start-up time of the pump and identifies the entire operating process from that start-up time until the pump stops again as a pressure-stabilizing pump start-up event to be classified. Subsequently, based on the system identifier of pressure-stabilizing pump P1, such as "Building 1 Fire Hydrant System," the system determines the fire water supply system scope corresponding to this event and, using the start-up time as a time reference, retrieves the pressure-stabilizing pump start / stop parameters, pipeline pressure timing parameters, and water flow indicator action parameters within a preset time range before and after the start-up time. The parameters for starting and stopping the pressure-stabilizing pump include the start time, stop time, running time, number of starts and stops per day, and average running time. The pipeline pressure time-series parameters include pressure sampling values ​​before and after start-up, pressure before start-up, pressure recovery status after stop-up, and single pressure change. The parameters for the water flow indicator include the start time, status, and duration of action of the water flow indicator belonging to the same fire-fighting water supply system as the pressure-stabilizing pump. Through these methods, a pressure-stabilizing pump start-up event can be correlated with the pressure changes before and after it, the start-up and stop statistics for the same day, and the action of water flow indicators within the same system, providing a data foundation for event discrimination feature extraction and classification.

[0022] Furthermore, the start-up and shutdown parameters of the pressure stabilizing pump include the pressure stabilizing pump identifier, the system identifier to which the pressure stabilizing pump belongs, the start time, stop time, running time, number of start-ups and shutdowns per day, and average running time; the pipeline pressure timing parameters include the pressure sampling time, pressure sampling value, pressure before start-up, pressure recovery status after stop, and single pressure change; the water flow indicator action parameters include the water flow indicator identifier, the system identifier to which the water flow indicator belongs, action time, action status, and action duration.

[0023] Further, in step S2, the event discrimination features are extracted from the event association parameters. Extract the first, second, third, fourth, and fifth discriminant features from the event association parameters, where, The first discriminant features include the start time of the pressure stabilizing pump, the system identifier of the pressure stabilizing pump, the system identifier of the water flow indicator, and the action time. These features are used to determine whether the start-up event of the pressure stabilizing pump to be classified has a temporal correlation with the action of the water flow indicator in the same fire water supply system, corresponding to a suspected real water use event. The second discriminant features include the number of start-stop cycles and the average runtime, which are used to determine whether the pressure stabilizing pump start-up event to be classified corresponds to a frequent start-stop event. The third discriminant features include the pressure before startup and the cumulative number of low-pressure startups on the day, which are used to determine whether the startup event of the pressure stabilizing pump to be classified corresponds to a severe low-pressure startup event of the pressure stabilizing pump or an occasional low-pressure startup event of the pressure stabilizing pump. The fourth discriminant feature includes the start time of the pressure stabilizing pump, the number of starts on the same day, and the pressure recovery status after shutdown, which are used to determine whether the pressure stabilizing pump start-up event to be classified corresponds to a suspected maintenance test event; The fifth discriminant feature includes the single pressure change and the pressure stability state after startup, which are used to determine whether the startup event of the pressure-stabilizing pump to be classified corresponds to a one-time pressure regulation event.

[0024] Further, in step S2, the event discrimination features are input into a preset classification discrimination model, which then matches the events according to the classification discrimination conditions corresponding to each event category to obtain at least one candidate event category, including: The first discrimination feature is matched with the discrimination conditions for suspected real water use. If there is a water flow indicator action record in the preset time window corresponding to the start time of the pressure stabilizing pump that belongs to the same fire water supply system as the start event of the pressure stabilizing pump to be classified, then it corresponds to the suspected real water use event. Specifically, the preset time window is a time window formed by extending forward by a first preset time period and backward by a second preset time period, based on the start time of the pressure-stabilizing pump. The water flow indicator action record of the same fire water supply system refers to an action record where the system identifier of the water flow indicator is consistent with the system identifier of the pressure-stabilizing pump in the pressure-stabilizing pump start-up event to be classified, and the action time of the water flow indicator falls within the preset time window. The temporal correlation between the water flow indicator action record and the start time of the pressure-stabilizing pump is a necessary condition for generating the candidate event category corresponding to the suspected real water use event. Both the first and second preset time periods are 10 minutes, and the preset time window is a 20-minute detection window formed from 10 minutes before to 10 minutes after the start time of the pressure-stabilizing pump.

[0025] When identifying suspected real water usage events, the pre-defined classification model first reads the start time of the pressure-stabilizing pump, the system identifier of the pressure-stabilizing pump, the action time of the water flow indicator, and the system identifier of the water flow indicator from the first discrimination feature, and establishes a pre-defined time window based on the start time of the pressure-stabilizing pump. For example, the pre-defined time window can be a time range from 10 minutes before to 10 minutes after the start time of the pressure-stabilizing pump. Subsequently, the pre-defined classification model searches for water flow indicator action records within the pre-defined time window and determines whether the searched water flow indicator action records belong to the same fire water supply system as the pressure-stabilizing pump start event to be classified. If there are water flow indicator action records of the same fire water supply system within the pre-defined time window, it indicates that the pressure-stabilizing pump start event is temporally related to the actual water flow action in the fire pipeline network, indicating that the fire water supply system may experience actual fire extinguishing water use, fire drill water use, or large-scale water release. Therefore, the pressure-stabilizing pump start event is generated as a candidate event category corresponding to suspected real water usage events. Because such events involve actual water discharge risks, their handling priority is higher than that of events such as frequent start-stop, low-pressure start-up, maintenance testing, and pressure regulation operations. After determining the final event category, the system can generate the highest priority notification information and push the suspected real water use event to the relevant person in charge so as to verify the on-site situation in a timely manner.

[0026] It should be noted that the suspected real water use events described in this application mainly refer to scenarios where, under non-fire conditions, a slow drop in pipeline pressure is caused by small-flow water use, minor leaks, or planned water release, triggering the pressure-stabilizing pump to replenish pressure. It also refers to the initial drop in pipeline pressure at the start of actual water use, where the pressure-stabilizing pump responds before the main fire pump. In these scenarios, the pressure-stabilizing pump is involved in the timing correlation and joint verification. However, in actual fire conditions with large-scale water use, the pump activated by the water flow indicator is usually the main fire pump or sprinkler pump, while the pressure-stabilizing pump enters a dormant state. These are different pump types under different conditions and should not be confused. Therefore, the results of the suspected real water use event identification in this application are used to trigger on-site verification and early warning, rather than being directly equivalent to a confirmed fire water supply status.

[0027] Furthermore, after generating candidate event categories corresponding to suspected real water use events, the process also includes a real water use event association verification step, which includes: Three signals are acquired: a pressure stabilizing pump start signal, a pipeline pressure drop signal, and a water flow indicator activation signal. The pressure stabilizing pump start signal indicates the time of the pressure stabilizing pump start-up event and its associated system identifier; the pipeline pressure drop signal indicates the time of the pipeline pressure drop event and its associated system identifier; and the water flow indicator activation signal indicates the time of the water flow indicator activation and its associated system identifier. Using the occurrence time of any one of the three signals as a reference, a preset time window is formed by extending forward and backward by preset time periods. For example, if the preset time period is 10 minutes, then the time window is 20 minutes long, extending forward and backward by 10 minutes from the reference signal. Then, it is determined whether all three signals are within the preset time window, and whether the system identifiers of the three signals match the same fire water supply system. When the pressure stabilizing pump start signal, the pipeline pressure drop signal, and the water flow indicator action signal are all within the same preset time window, and the three signals belong to the same fire water supply system, the actual water use event correlation verification result is output and treated as a high-confidence event. If only two signals meet the time window and matching conditions, a low-confidence candidate event category is generated to assist in further judgment. Through the above processing, the preset classification and discrimination model can first generate candidate event categories for suspected real water use events based on the two-way time-series correlation between the pressure stabilizing pump start signal and the water flow indicator action signal; then, based on the three-way time-series correlation between the pressure stabilizing pump start signal, the pipeline pressure drop signal and the water flow indicator action signal, the candidate events are further verified, thereby improving the reliability and confidence of real water use event identification, and providing accurate basis for event handling and management.

[0028] The second discriminant feature is matched with the frequent start-stop classification criteria. If the number of start-stops of the pressure stabilizing pump on a given day is greater than the preset frequent start-stop number threshold and the average running time is less than or equal to the preset short running time threshold, then it is a frequent start-stop event. Specifically, the preset threshold for the number of frequent start-stops is 10 times, and the preset threshold for short running time is 30 seconds. When the number of start-stops of the same pressure stabilizing pump in the same statistical day is greater than 10 times and the average running time is less than or equal to 30 seconds, the second discrimination feature is determined to match the frequent start-stop event, and the candidate event category corresponding to the frequent start-stop event is generated.

[0029] When identifying frequent start-stop events, the preset classification model reads the number of start-stops and the average runtime from the second discrimination feature and matches them with the frequent start-stop classification criteria. The frequent start-stop classification criteria can be set as follows: the number of start-stops of the same pressure-stabilizing pump within the same statistical day is greater than a preset frequent start-stop threshold, and the average runtime is less than or equal to a preset short runtime threshold. For example, the preset frequent start-stop threshold can be 10 times, and the preset short runtime threshold can be 30 seconds. When a pressure-stabilizing pump P1 in a building's fire hydrant system starts a total of 12 times within the same day, with runtimes of 22 seconds, 25 seconds, 18 seconds, 24 seconds, 27 seconds, 20 seconds, 26 seconds, 23 seconds, 21 seconds, 28 seconds, 19 seconds, and 24 seconds respectively, and the calculated average runtime is approximately 23 seconds, satisfying the condition that the number of start-stops exceeds 10 times and the average runtime does not exceed 30 seconds, the preset classification model generates a candidate event category corresponding to the frequent start-stop event for this pressure-stabilizing pump start event. Therefore, when the final event category is determined to be a frequent start-stop event, the system can generate high-priority alarm information to prompt maintenance personnel to promptly investigate pipeline leaks, valve closure status, or pressure stabilizing pump control parameter settings, and arrange corresponding repairs.

[0030] The third discriminant feature is matched with the low-pressure start-up classification criteria. If the pressure before start-up is lower than the preset low-pressure threshold, it is classified as a severe low-pressure start-up event or an occasional low-pressure start-up event of the pressure-stabilizing pump. When the pressure before startup is lower than the preset low pressure threshold, and the number of low pressure starts within the same statistical day is greater than the preset severe start number threshold, it is determined as a severe low pressure start event of the pressure stabilizing pump. When the pressure before startup is lower than the preset low-pressure threshold, and the number of low-pressure starts within the same statistical day is lower than the preset severe start number threshold, it is determined as an occasional low-pressure start event of the pressure stabilizing pump.

[0031] Specifically, the preset low-pressure threshold is 0.15 MPa, and the preset severe frequency threshold is 2 times. When the preset classification and discrimination model determines that the pressure before startup is lower than 0.15 MPa, it matches the pressure stabilizing pump startup event to be classified as the corresponding candidate event category. When the same pressure stabilizing pump has more than 2 low-pressure startups in the same statistical day, it is determined as a severe low-pressure startup event of the pressure stabilizing pump. When the same pressure stabilizing pump has 1 to 2 low-pressure startups in the same statistical day, it is determined as an occasional low-pressure startup event of the pressure stabilizing pump.

[0032] When identifying low-pressure start-up events, the preset classification model reads the pre-start pressure and the cumulative number of low-pressure starts within the same statistical day from the third discrimination feature, and matches them with the low-pressure start-up classification criteria. The pre-start pressure can be the most recent effective pressure sample value before the start-up time of the pressure stabilizing pump, and the preset low-pressure threshold can be set to 0.15 MPa. When the pre-start pressure is lower than 0.15 MPa, it indicates that the pipeline pressure before the pressure stabilizing pump starts is significantly lower than the normal operating pressure range. The preset classification model first identifies this start-up event as a severe low-pressure start-up event or an occasional low-pressure start-up event of the pressure stabilizing pump; subsequently, it further counts the cumulative number of times the pre-start pressure of the same pressure stabilizing pump is lower than 0.15 MPa within the same statistical day. If the cumulative number of low-pressure starts exceeds the preset severity threshold, for example, more than 3 times in a single day, it indicates that the low-pressure start is not an occasional disturbance, but a continuous or recurring low-pressure anomaly. This may be caused by malfunction of the pressure-stabilizing pump itself, incomplete opening of the inlet valve, insufficient water supply pressure, or failure of the pipeline pressure source. Therefore, it is identified as a severe low-pressure start event of the pressure-stabilizing pump, and a high-priority alarm message is generated to prompt maintenance personnel to urgently check the working status of the pressure-stabilizing pump and the water supply pressure. If the cumulative number of low-pressure starts in the same statistical day is 1 to 2, it indicates that the low-pressure start may be caused by short-term water usage, instantaneous pressure fluctuations, or occasional disturbances. Its severity is lower than that of a severe low-pressure start event of the pressure-stabilizing pump. Therefore, it is identified as an occasional low-pressure start event of the pressure-stabilizing pump, and a medium-priority monitoring message is generated to continuously track its subsequent frequency. When the number of subsequent low-pressure starts continues to increase and exceeds the preset severity threshold, it can be further escalated to a severe low-pressure start event of the pressure-stabilizing pump.

[0033] The fourth discrimination feature is matched with the discrimination conditions for suspected maintenance test. If the pressure stabilizing pump start-up event to be classified occurs in a preset daytime period, the same pressure stabilizing pump starts only once in the same statistical day, and the pressure returns to the preset working pressure range after stopping, then it is a suspected maintenance test event. Specifically, when identifying suspected maintenance test events, the preset classification and discrimination model reads the start time of the pressure-stabilizing pump, the number of starts within the same statistical day, and the pressure recovery status after stopping from the fourth discrimination feature, and matches them with the suspected maintenance test classification and discrimination conditions. The preset daytime period can be set to 06:00 to 22:00. When the pressure-stabilizing pump start event to be classified occurs within this time period, and the same pressure-stabilizing pump starts only once within the same statistical day, and the pipeline pressure after the pressure-stabilizing pump stops can recover to the preset working pressure range without a sustained low pressure state, the preset classification and discrimination model identifies the start event as a suspected maintenance test event. This type of event usually corresponds to pressure-stabilizing pump function tests, water discharge tests, or pressure recovery tests performed by maintenance personnel. It is a pump start caused by planned operations and should not be directly treated as equipment failure or pipeline abnormality. Therefore, when the final event category is determined to be a suspected maintenance test event, the system can save it as a normal record and verify it with the daily maintenance plan, maintenance work order or manual maintenance record to distinguish between normal maintenance operations and unregistered abnormal start events.

[0034] The fifth discrimination feature is matched with the one-time pressure regulation classification discrimination condition. When the single pressure change is greater than the preset multiple of the normal pump circulation pressure amplitude, and the pipeline pressure after the start-up event of the pressure-stabilizing pump to be classified is stable at the adjusted pressure level, it is a one-time pressure regulation event.

[0035] Specifically, when identifying a one-time pressure regulation event, the preset classification model reads the single pressure change and the pressure stabilization state after startup from the fifth discrimination feature and matches them with the one-time pressure regulation classification criteria. The single pressure change can be calculated from the pressure sampling values ​​before and after the pressure-stabilizing pump startup, and the normal pump circulation pressure amplitude can be determined from the historical pressure change range of the fire water supply system during normal pressure replenishment. When the single pressure change caused by a pressure-stabilizing pump startup is greater than a preset multiple of the normal pump circulation pressure amplitude (e.g., more than 3 times the normal pump circulation pressure amplitude), and after the startup event, the pipeline pressure does not return to the original pressure fluctuation range but remains stable near the new pressure level, the preset classification model identifies the startup event as a one-time pressure regulation event. Such events typically indicate that the system has undergone intentional pressure adjustment, such as pipeline water filling and pressurization, pressure recovery after maintenance, adjustment of the pressure-stabilizing pump pressure setpoint, or system pressure reset, and should not be directly identified as a fault event. Therefore, when the final event category is determined to be a one-time pressure adjustment event, the system can save it as a normal record and generate a pressure adjustment verification prompt to remind maintenance personnel to verify whether the pressure adjustment is an authorized operation and whether it is consistent with the corresponding maintenance record, debugging record or pressure setting change record.

[0036] Further, in step S2, the event discrimination features are input into a preset classification discrimination model, which then matches the events according to the classification discrimination conditions corresponding to each event category to obtain at least one candidate event category. This also includes... When the pressure-stabilizing pump start-up event to be classified does not match any of the candidate event categories such as suspected real water usage event, frequent start-stop event, severe low-pressure start-up event of pressure-stabilizing pump, occasional low-pressure start-up event of pressure-stabilizing pump, suspected maintenance test event, and one-time pressure adjustment event, the pressure-stabilizing pump start-up event to be classified is determined as an unclassified event, and a review prompt message is generated to prompt manual review.

[0037] Furthermore, in step S3, the classification priority is from high to low as follows: suspected real water use event, frequent start-stop event, severe low pressure start event of pressure stabilizing pump, occasional low pressure start event of pressure stabilizing pump, suspected maintenance test event, and one-time pressure adjustment event.

[0038] Specifically, the preset classification and discrimination model can adopt, for example... Figure 1The decision tree structure is shown. This decision tree uses the pressure-stabilizing pump start-up event to be classified as an input node, and uses the first to fifth discriminant features as branching criteria to classify a single pressure-stabilizing pump start-up event into its corresponding event category step by step. First, the decision tree uses the first discriminant feature to determine whether there are any records of water flow indicator actions in the same fire water supply system within a preset time window before and after the pressure-stabilizing pump start-up time. If so, it indicates that the pressure-stabilizing pump start-up event and the actual water discharge behavior are correlated in time and system scope, and a suspected real water use event is output. If not, the second discriminant feature is used to determine whether the number of start-ups and shutdowns and the average running time on that day meet the frequent start-up / shutdown condition. If so, a frequent start-up / shutdown event is output. If not, the third discriminant feature is used to determine whether the pressure before start-up is lower than a preset low-pressure threshold, and combined with the cumulative number of low-pressure starts on that day, it is further classified into a severe low-pressure start-up event or an occasional low-pressure start-up event of the pressure-stabilizing pump. If the pressure before startup is not lower than the preset low-pressure threshold, the fourth discrimination feature determines whether the startup event occurred during a preset daytime period, whether it is a single startup of the day, and whether the pressure after shutdown returns to the preset working pressure range. If these conditions are met, a suspected maintenance test event is output. If these conditions are not met, the fifth discrimination feature determines whether the single pressure change is greater than a preset multiple of the normal pump circulation pressure amplitude, and whether the pressure after startup is stable at the adjusted pressure level. If these conditions are met, a one-time pressure adjustment event is output. If none of the above branch conditions are met, an unclassified event is output, and manual review is prompted. Through the above decision tree structure, the preset classification discrimination model can incorporate the timing association of water flow indicator actions, daily start-stop statistics of pressure stabilizing pumps, pressure before startup, pressure recovery status after shutdown, and single pressure change into the same discrimination process. When multiple discrimination conditions may be met simultaneously, the final event category is output according to the priority order from suspected real water use events to one-time pressure adjustment events.

[0039] It should be noted that the specific values ​​given in the embodiments of this specification, such as the duration of the preset timing window and preset time window, the preset threshold for the number of frequent start-stop cycles, the preset threshold for short running time, the preset low pressure threshold, the preset threshold for the number of severe cycles, the preset multiple of the normal pump circulation pressure amplitude, and the preset daytime time period, are all merely exemplary values ​​for ease of understanding and do not constitute a limitation on the scope of protection of this invention. Those skilled in the art can adjust the above values ​​according to the engineering parameters, operating rules, and management needs of the specific fire water supply system, and the corresponding solutions should all fall within the scope of protection of this application.

[0040] In summary, this application, by jointly analyzing pressure-stabilizing pump start-up events with the timing of water flow indicator actions, pipeline pressure timing, and daily start-up and shutdown statistics, enables refined identification and classification of pressure-stabilizing pump start-up events without the need for additional sensors. On one hand, by establishing a temporal correlation between water flow indicator actions and pressure-stabilizing pump start-up times—for example, by detecting water flow indicator action records within a preset time window before and after pump start-up—suspected real water usage events can be automatically flagged, improving the fire protection IoT platform's early detection capability for real firefighting water usage or large-scale water usage events. On the other hand, by distinguishing between severe low-pressure start-up events and occasional low-pressure start-up events using features such as the number of low-pressure starts and pre-start pressure, it facilitates pressure-stabilizing pump fault classification, making maintenance priority scheduling more accurate and preventing occasional events from triggering unnecessary emergency responses. Furthermore, this application can also identify suspected maintenance test events and one-time pressure adjustment events, cross-checking them with manual maintenance records and pressure adjustment records to reduce the misjudgment of normal maintenance operations as abnormal events. Therefore, the classification results can not only be used for alarm classification and handling prompts on the fire protection IoT platform, but also serve as a refined input for the health assessment and comprehensive scoring of building fire protection systems. In particular, the counts of events such as low-pressure start-up and frequent start-stop of pressure-stabilizing pumps can be directly used for subsequent pipeline operation status diagnosis and maintenance decisions.

[0041] Furthermore, it also includes step S4, which generates corresponding handling priority information based on the determined final event category, and outputs the final event category and handling priority information to the platform. When the final event category is suspected actual water usage event, the highest priority notification message is generated; when the final event category is frequent start-stop event or severe low-pressure start event of pressure stabilizing pump, a high priority alarm message is generated; when the final event category is occasional low-pressure start event of pressure stabilizing pump, a medium priority monitoring message is generated; when the final event category is suspected maintenance test event or one-time pressure adjustment event, normal record information is generated; when the final event category is unclassified event, a manual review prompt message is generated.

[0042] Second Embodiment Based on the same inventive concept, such as Figure 2 As shown, this application also provides a time-series correlation-based classification system for pressure-regulating pump start-up events, employing the time-series correlation-based classification method for pressure-regulating pump start-up events as described above, including: The response module is used to respond to the start-up trigger signal of the pressure stabilizing pump, determine the start-up event of the pressure stabilizing pump to be classified, and collect the corresponding event-related parameters, including the start-up and stop parameters of the pressure stabilizing pump, the pipeline pressure timing parameters, and the action parameters of the water flow indicator. The matching module is used to extract event discrimination features from event association parameters and input the event discrimination features into a preset classification discrimination model. The preset classification discrimination model matches according to the classification discrimination conditions corresponding to each event category to obtain at least one candidate event category. The output module is used to determine the final event category of the pressure-stabilizing pump start-up event to be classified from at least one candidate event category according to the classification priority in the preset classification and discrimination model. The final event category is one of the following: suspected real water use event, frequent start-stop event, severe low pressure start-up event of pressure-stabilizing pump, occasional low pressure start-up event of pressure-stabilizing pump, suspected maintenance test event, one-time pressure regulation event, and unclassified event.

[0043] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the foregoing claims.

[0044] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

[0045] It should be understood that "multiple" as used in this article refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0046] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0047] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for classifying pressure-stabilizing pump start-up events based on time-series correlation, characterized in that, include: Step S1: Respond to the pressure stabilizing pump start-up trigger signal, determine the pressure stabilizing pump start-up event to be classified, and collect the corresponding event association parameters. The event association parameters include pressure stabilizing pump start-up and stop parameters, pipeline pressure timing parameters, and water flow indicator action parameters. Step S2: Extract event discrimination features from the event association parameters, and input the event discrimination features into a preset classification discrimination model. The preset classification discrimination model matches the events according to the classification discrimination conditions corresponding to each event category to obtain at least one candidate event category. Step S3: Based on the classification priority in the preset classification and discrimination model, determine the final event category of the pressure-stabilizing pump start-up event to be classified from at least one of the candidate event categories, wherein the final event category is one of the following: suspected real water use event, frequent start-stop event, severe low-pressure start-up event of pressure-stabilizing pump, occasional low-pressure start-up event of pressure-stabilizing pump, suspected maintenance test event, one-time pressure adjustment event, and unclassified event.

2. The method for classifying pressure-stabilizing pump start-up events based on time-series correlation according to claim 1, characterized in that, Step S1 includes, In response to the pressure stabilizing pump start trigger signal, which is a trigger signal generated by the platform when it detects that the pressure stabilizing pump has switched from a stopped state to a running state; Based on the pressure stabilizing pump start-up trigger signal, the start-up time of the pressure stabilizing pump is recorded, and a single operation of the pressure stabilizing pump starting from the start-up time is identified as the pressure stabilizing pump start-up event to be classified. Based on the start-up time of the pressure stabilizing pump and the system identifier to which the pressure stabilizing pump belongs in the pressure stabilizing pump start-up event to be classified, the corresponding data collection range is determined and the corresponding event-related parameters are collected.

3. The method for classifying pressure-stabilizing pump start-up events based on time-series correlation according to claim 1, characterized in that, The pressure stabilizing pump start / stop parameters include the pressure stabilizing pump identifier, the system identifier to which the pressure stabilizing pump belongs, the start time, stop time, running time, number of start / stop cycles per day, and average running time; the pipeline pressure time sequence parameters include the pressure sampling time, pressure sampling value, pressure before start, pressure recovery status after stop, and single pressure change; the water flow indicator action parameters include the water flow indicator identifier, the system identifier to which the water flow indicator belongs, action time, action status, and action duration.

4. The method for classifying pressure-stabilizing pump start-up events based on time-series correlation according to claim 3, characterized in that, In step S2, the event discrimination features are extracted from the event association parameters. Extract the first discriminant feature, the second discriminant feature, the third discriminant feature, the fourth discriminant feature, and the fifth discriminant feature from the event association parameters, wherein, The first discrimination feature includes the start time of the pressure stabilizing pump, the system identifier of the pressure stabilizing pump, the system identifier of the water flow indicator, and the action time, which are used to determine whether the start event of the pressure stabilizing pump to be classified has a temporal correlation with the action of the water flow indicator in the same fire water supply system, corresponding to the suspected real water use event; The second discrimination feature includes the number of start-stop times on the same day and the average running time, which is used to determine whether the pressure stabilizing pump start-up event to be classified corresponds to the frequent start-stop event; The third discrimination feature includes the pre-start pressure and the cumulative number of low-pressure starts on the day, which is used to determine whether the pressure stabilizing pump start-up event to be classified corresponds to the severe low-pressure start-up event of the pressure stabilizing pump or the occasional low-pressure start-up event of the pressure stabilizing pump. The fourth discrimination feature includes the start time of the pressure stabilizing pump, the number of starts on the same day, and the pressure recovery status after stopping, which is used to determine whether the pressure stabilizing pump start event to be classified corresponds to the suspected maintenance test event; The fifth discriminant feature includes the single pressure change and the pressure stabilization state after startup, used to determine whether the pressure stabilizing pump startup event to be classified corresponds to the one-time pressure regulation event.

5. The method for classifying pressure-stabilizing pump start-up events based on time-series correlation according to claim 4, characterized in that, In step S2, the event discrimination features are input into a preset classification discrimination model, which then matches the events according to the classification discrimination conditions corresponding to each event category to obtain at least one candidate event category. include, The first discrimination feature is matched with the suspected real water use classification discrimination condition. If there is a water flow indicator action record in the preset time window corresponding to the start time of the pressure stabilizing pump that belongs to the same fire water supply system as the start event of the pressure stabilizing pump to be classified, then it corresponds to the suspected real water use event. The second discrimination feature is matched with the frequent start-stop classification discrimination condition. If the number of daily start-stops of the pressure stabilizing pump is greater than the preset frequent start-stop number threshold, and the average running time is less than or equal to the preset short running time threshold, then it is the frequent start-stop event. The third discrimination feature is matched with the low-pressure start classification discrimination condition. When the pressure before start is lower than the preset low-pressure threshold, it is either a severe low-pressure start event of the pressure stabilizing pump or an occasional low-pressure start event of the pressure stabilizing pump. The fourth discrimination feature is matched with the classification criteria for suspected maintenance test. If the pressure stabilizing pump start-up event to be classified occurs during a preset daytime period, the same pressure stabilizing pump starts only once in the same statistical day, and the pressure returns to the preset working pressure range after stopping, then it is the suspected maintenance test event. The fifth discrimination feature is matched with the one-time pressure regulation classification discrimination condition. When the single pressure change is greater than a preset multiple of the normal pump circulation pressure amplitude, and the pipeline pressure after the start-up event of the pressure-stabilizing pump to be classified is stable at the adjusted pressure level, it is the one-time pressure regulation event.

6. The method for classifying pressure-stabilizing pump start-up events based on time-series correlation according to claim 5, characterized in that, The third discriminant feature is matched with the low-pressure start-up classification criteria. If the pressure before start-up is lower than a preset low-pressure threshold, it is considered a severe low-pressure start-up event of the pressure-stabilizing pump or an occasional low-pressure start-up event of the pressure-stabilizing pump. This also includes... When the pressure before startup is lower than the preset low pressure threshold, and the number of low pressure starts within the same statistical day is greater than the preset severe start number threshold, it is determined as a severe low pressure start event of the pressure stabilizing pump. When the pressure before startup is lower than the preset low-pressure threshold, and the number of low-pressure starts within the same statistical day is lower than the preset severe start count threshold, it is determined as an occasional low-pressure start event of the pressure stabilizing pump.

7. The method for classifying pressure-stabilizing pump start-up events based on time-series correlation according to claim 6, characterized in that, In step S2, the event discrimination features are input into a preset classification discrimination model, which then matches the data according to the classification discrimination conditions corresponding to each event category to obtain at least one candidate event category. This also includes... When the pressure-stabilizing pump start-up event to be classified does not match any of the candidate event categories of the suspected real water usage event, the frequent start-stop event, the pressure-stabilizing pump severe low-pressure start-up event, the pressure-stabilizing pump occasional low-pressure start-up event, the suspected maintenance test event, and the one-time pressure adjustment event, the pressure-stabilizing pump start-up event to be classified is determined as the unclassified event, and a review prompt message is generated to prompt manual review.

8. The method for classifying pressure-stabilizing pump start-up events based on time-series correlation according to claim 7, characterized in that, In step S3, the classification priority is from high to low as follows: suspected real water use event, frequent start-stop event, severe low pressure start event of pressure stabilizing pump, occasional low pressure start event of pressure stabilizing pump, suspected maintenance test event, and one-time pressure adjustment event.

9. The method for classifying pressure-stabilizing pump start-up events based on time-series correlation according to claim 8, characterized in that, The method also includes step S4, which generates corresponding handling priority information based on the determined final event category, and outputs the final event category and the handling priority information to the platform. When the final event category is the suspected real water usage event, a highest priority notification message is generated; when the final event category is the frequent start-stop event or the severe low-pressure start event of the pressure stabilizing pump, a high-priority alarm message is generated; when the final event category is the occasional low-pressure start event of the pressure stabilizing pump, a medium-priority monitoring message is generated; when the final event category is the suspected maintenance test event or the one-time pressure adjustment event, normal record information is generated; when the final event category is the unclassified event, a manual review prompt message is generated.

10. The method for classifying pressure-stabilizing pump start-up events based on time-series correlation according to claim 8, characterized in that, In step S4, when the final event category is the suspected real water usage event, a highest priority notification message is generated. include, When the final event category is the suspected real water use event, the real water use event association verification is triggered; Acquire the start signal of the pressure stabilizing pump, the action signal of the water flow indicator, and the signal of the decrease in pipeline pressure, and determine whether the three signals are all within the same preset time window; When all three signals are within the same preset time window and their system identifiers match, a real water usage event verification result is output, and the confidence level of the real water usage event verification result is raised to high confidence. The preset time window is a time window formed by extending a preset time period forward and backward based on the occurrence time of any signal. When the verification result of the real water use event is high confidence, a high-priority alarm notification is generated; when only the conditions of the suspected real water use event are met, a second-lower priority notification is generated.

11. A time-series correlation-based classification system for pressure-stabilized pump start-up events, employing the time-series correlation-based classification method for pressure-stabilized pump start-up events as described in any one of claims 1 to 10, characterized in that, include, The response module is used to respond to the start-up trigger signal of the pressure stabilizing pump, determine the start-up event of the pressure stabilizing pump to be classified, and collect the corresponding event association parameters. The event association parameters include the start-up and stop parameters of the pressure stabilizing pump, the pipeline pressure timing parameters, and the action parameters of the water flow indicator. The matching module is used to extract event discrimination features from the event association parameters and input the event discrimination features into a preset classification discrimination model. The preset classification discrimination model matches according to the classification discrimination conditions corresponding to each event category to obtain at least one candidate event category. The output module is used to determine the final event category of the pressure-stabilizing pump start-up event to be classified from at least one of the candidate event categories according to the classification priority in the preset classification and discrimination model. The final event category is one of the following: suspected real water use event, frequent start-stop event, severe low pressure start-up event of pressure-stabilizing pump, occasional low pressure start-up event of pressure-stabilizing pump, suspected maintenance test event, one-time pressure adjustment event, and unclassified event.

Images