A method and system for estimating remaining time in multi-channel water sampling and detection

By constructing a heterogeneous event graph and a real-time state sequence, the remaining time of multi-channel sampling detection is dynamically estimated, solving the problem of remaining time estimation that cannot comprehensively consider multiple factors in existing technologies, and achieving accurate time prediction and anomaly handling.

CN123117416APending Publication Date: 2026-11-17HUANENG LANZHOU XIGU THERMAL POWER CO LTD +1
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202611082097.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-21
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies have not yet developed a method for estimating the remaining time of the entire multi-channel sampling and detection process that can integrate sampling status, fluid stability status, actuator feedback, instrument resource occupancy, data validity, communication confirmation, and shared resource queuing relationships, and transform abnormal states into time-related quantities.

Method used

Construct a heterogeneous event graph for multi-channel sampling and detection, collect real-time status time series of nodes, perform reliable completion judgment, calculate the time cost of anomalies, introduce shared resource queuing constraints, integrate the remaining time of the entire process, identify key impact chains, and output the expected remaining time and reliable interval.

Benefits of technology

Dynamically calculate the remaining time, reduce the inconsistency between fixed countdown and progress, avoid misjudgment, and be suitable for the whole process estimation of multi-waterway sampling and detection. Identify key influencing factors and output the estimated remaining time after anomaly correction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123117416A_ABST
    Figure CN123117416A_ABST
Patent Text Reader

Abstract

This invention discloses a method and system for estimating the remaining time of multi-channel sampling and detection. The method includes: S1, constructing a heterogeneous event graph of the entire process of multi-channel sampling and detection; S2, collecting and encoding the real-time status time series of each node; S3, performing a reliable completion determination; S4, calculating the abnormal time cost; S5, estimating the remaining time of a single node node by node; S6, introducing shared resource queuing constraints; S7, fusing the remaining time of the entire process and identifying key influence chains; S8, outputting the remaining time result probabilistically and updating the historical sample database. This invention uses a heterogeneous event graph to uniformly represent the sampling, flow and pressure stabilization, flow storage or buffering, channel switching, sample delivery, instrument stabilization, detection, data determination, and result output processes. It can dynamically calculate the remaining time based on nodes that have not yet been reliably completed, reducing the problem of inconsistency between fixed countdowns and actual progress.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of online water quality monitoring and industrial process state estimation technology, specifically relating to a method and system for estimating the remaining time of the entire process in a multi-channel online sampling and detection device. Background Technology

[0002] Online water quality monitoring devices typically include a sampling unit, a flow and pressure stabilization unit, a flow storage or buffer unit, a water path switching unit, online monitoring instruments, a controller, a human-machine interface (HMI), and a communication module. After sampling, flow and pressure stabilization, flow storage, or pretreatment, the water sample is fed into the monitoring instruments via valves, pumps, and switching mechanisms to detect parameters such as conductivity, pH, iron content, and particulate matter, and the data is then uploaded to the HMI, a host computer, or a DCS system.

[0003] Existing technical solution 1: Multi-parameter online water quality monitoring system. CN114965922A can realize the switching, sequential injection, and waste discharge of multiple water samples, standard solutions, reagents, and pure water through the sample injection flow path, pretreatment, detection module, multi-way valve, peristaltic pump, etc. However, its focus is on flow path integration and detection implementation, and it has not established a mechanism for estimating the remaining time of the entire process from sampling to the output of effective detection results.

[0004] Existing technical solution 2: Online monitoring data anomaly identification method. CN112287988A can identify anomalies, mark missing data, analyze fluctuations, and classify anomaly causes in monitoring data. However, it is usually used for alarm, rejection, or marking of anomalies, and does not convert data spikes, disconnections, flow fluctuations, instrument instability, or communication anomalies into waiting time, recovery time, or retest time.

[0005] Existing technical solution three: a general method for predicting remaining time in a process. CN111178623B mainly relies on business process logs, event timestamps, key activities, and historical process characteristics for prediction, but it does not establish a unified time impact model for fluid state, pump and valve feedback, instrument occupancy, data reliability, and communication confirmation in online sampling and detection.

[0006] Existing technical solution four: Graph structure learning and time series prediction methods. Although graph structure learning and time series prediction methods can express node relationships or state change trends, when used alone, they do not establish a full-process remaining time estimation scheme for process dependencies, fluid connections, shared resource occupation, anomaly recovery, data validity determination, and communication confirmation in multi-channel online sampling and detection.

[0007] Therefore, existing technologies have not yet developed a method for estimating the remaining time of the entire multi-channel sampling and detection process that can integrate sampling status, fluid stability status, actuator feedback, instrument resource occupancy, data validity, communication confirmation, and shared resource queuing relationships, and transform abnormal states into time-related quantities. Summary of the Invention

[0008] This invention provides a method and system for estimating the remaining time of multi-channel sampling and detection, which solves the problem that existing multi-channel online sampling and detection devices mainly rely on fixed delays or single-state judgments and are difficult to dynamically reflect the on-site operating status and task progress.

[0009] To achieve the above objectives, this invention provides a method for estimating the remaining time of multi-channel sampling and detection, which is used by a multi-channel online sampling and detection device to calculate the remaining time of the entire process from the current moment to the output of a valid detection result, including the following steps: S1. Construct a heterogeneous event graph for the entire process of multi-channel sampling and testing; abstract the online sampling and testing device into water channel nodes, pump and valve execution nodes, flow stabilization and pressure stabilization nodes, buffer and storage nodes, water channel switching nodes, testing instrument nodes, data judgment nodes, communication output nodes, and task nodes; establish process dependency edges, fluid connection edges, resource occupation edges, state influence edges, and communication confirmation edges between each node to uniformly represent the entire process of sampling, flow stabilization and pressure stabilization, buffer and storage, water channel switching, sample delivery, instrument testing, data judgment, and result output; S2. Collect and encode the real-time status time series of each node; collect flow rate, pressure, liquid level, pump and valve commands and feedback, flow and pressure stability status, instrument working status, detection data quality status and communication interaction status according to the preset sampling period, and store the continuous status sequence within the time window to represent the trend of working condition changes. S3. Execute a reliable completion judgment; based on the consistency of pump and valve feedback, the stability of flow and pressure conditions, the amount of buffer storage, the stability of instruments, the continuity and validity of detection data, and the communication upload confirmation flag, determine whether each node has been reliably completed. Reliably completed nodes will no longer participate in the remaining time calculation. S4. Calculate the time cost of anomalies; For various anomalies such as unstable flow, abnormal pressure, insufficient liquid level, pump and valve malfunction, unstable instruments, missing / peak / over-range / invalid data, and communication interruption / delay, the time cost of anomalies is obtained by combining the duration of the anomaly, the historical recovery time of similar anomalies, and the compensation time for restabilization / re-sampling / re-detection, and then added to the remaining time of the corresponding node; When the anomaly exceeds the estimable range, the remaining time is indeterminate and the corresponding anomaly node is output. S5. Estimate the remaining time of a single node on a node-by-node basis; for nodes that have not started, determine the basic remaining time based on the historical time consumption of the same waterway, the same instrument, and the same working conditions; for nodes that are currently executing, adjust the remaining time in real time by combining the running time, the trend of status changes, and the degree of satisfaction of the reliable completion conditions; for abnormally blocked nodes, add the abnormal time cost to the basic remaining time; when there are enough historical samples, use a graph time series prediction model to input the node relationship and the real-time status sequence to output the node's remaining time consumption and uncertainty range. S6. Introduce shared resource queuing constraints; For scenarios where multiple waterways share detection instruments, pumps, valves, switching mechanisms, and communication interfaces, calculate the resource waiting time based on the resource occupancy status, the expected release time of the occupied task, and the task queue order. The expected start time of a node is determined by the completion time of the preceding node, the resource release time, the anomaly recovery time, and the scheduling queue order. S7. Integrate the remaining time of the entire process and identify key impact chains; based on the heterogeneous event graph path, superimpose the time consumption of each node, sequentially accumulate serial nodes, take the longest completion time of parallel branches, superimpose the waiting time of shared resources, and superimpose the abnormal time cost of abnormal nodes; take the longest time consumption of the entire path as the estimated remaining time of the entire process, and simultaneously identify the key impact path with the longest consumption as the key impact chain. S8. Probabilistically output the remaining time result and update the historical sample library; output the estimated remaining time, conservative remaining time, confidence interval, current execution step, key impact chain, abnormal blocked node, and shared resource waiting node; after a single detection task is completed, record the actual start and stop time, time consumed, waiting time, anomaly type, resource occupation status, data validity, and communication confirmation time of each node, and update the three types of time distribution and prediction models of node time consumption, anomaly recovery, and resource waiting.

[0010] In one embodiment of the present invention, optionally, each node in the heterogeneous event graph is defined as follows: Water path nodes represent the flow paths of each water sample and the real-time sampling status; pump and valve execution nodes represent the status of pump and valve control commands, execution feedback, and action completion; flow and pressure stabilization nodes represent the flow rate, pressure values, and fluid stability judgment status of the pipeline; buffer and storage nodes represent the buffer liquid level, water sample storage volume, and storage compliance status; water path switching nodes represent the on / off status of multi-channel water sample switching valves; detection instrument nodes represent the instrument idle, occupied, preheating and stabilizing, detection, and fault shutdown status; data judgment nodes represent the status of missing detection data, spikes, disconnections, over-range, and valid / invalid status; communication output nodes represent the status of detection result upload, communication delay, link interruption, and upper-level confirmation; task nodes represent the current detection task and the queuing status of multi-water path tasks.

[0011] In one embodiment of the present invention, optionally, the process dependency edge represents the sequential execution logic of each detection step; the fluid connection edge represents the flow path relationship of water sample between water circuit, heat exchanger, buffer, and instrument; the resource occupation edge represents the occupation and queuing constraints of shared pumps, valves, detection instruments, and communication interfaces in multiple water circuits; the state influence edge represents the prolongation effect of abnormal and non-compliant states on the time consumption of subsequent nodes; and the communication confirmation edge represents the complete timing constraints of the detection data being uploaded to the host computer and receiving the confirmation signal.

[0012] In one embodiment of the present invention, optionally, the graph time series prediction model takes the node association weights of the heterogeneous event graph, the real-time state time series of the nodes, and historical working condition samples as inputs, and outputs the remaining basic time consumption and the time consumption confidence interval of each node. The node association weights are trained by the degree of influence of different nodes on the total detection time in historical data.

[0013] In one embodiment of the present invention, optionally, in step S3, the reliable completion determination rule includes: the pump valve control command is consistent with the field feedback signal, the pipeline flow and pressure are continuously stable for a preset time, the buffer water sample volume meets the minimum sampling threshold, the instrument preheating stability time reaches the standard, multiple sets of continuous detection data are without distortion and within the effective range, and the host computer returns a data upload confirmation signal. When all of these conditions are met, the corresponding node is determined to be reliable completed.

[0014] In one embodiment of the present invention, optionally, in step S6, the shared resource queuing constraint calculation logic is as follows: if the resources required by the target waterway are in an idle state and there is no additional waiting time; if the resources are occupied by other waterways, the resource release time is determined based on the remaining detection time of the occupied task, and the resource release time is used as the earliest start time of the target waterway node, which is added to the remaining time of the node.

[0015] This invention also provides a multi-channel sampling and detection remaining time estimation system, which is used to implement the above-mentioned multi-channel sampling and detection remaining time estimation method, including: The status acquisition module is used to acquire real-time status data such as flow rate, pressure, liquid level, pump and valve feedback, instrument operating conditions, detection data, and communication interaction within the multi-channel sampling and testing device, and generate a status time series. The heterogeneous event graph construction module is used to construct a full-process heterogeneous event graph containing multiple types of nodes and five types of associated edges, and to establish a logical and physical relationship model between various detection links, devices, and tasks. The trusted completion determination module has built-in trusted completion determination rules to determine whether each node meets the completion conditions and filter out completed nodes. The abnormal time cost calculation module identifies various equipment, fluid, data, and communication anomalies, quantifies the time loss caused by the anomalies, and outputs the abnormal time cost. The node remaining time estimation module includes a basic time calculation unit, an anomaly overlay unit, and a graph time series prediction unit, which respectively output the uncorrected and corrected remaining time and uncertainty range of each node. The shared resource queuing calculation module identifies shared equipment resources across multiple waterways, calculates resource queuing waiting time, and corrects node startup times. The remaining time fusion module for the entire process traverses all execution paths in the heterogeneous event graph, merges the time consumed by serial, parallel, waiting, and abnormal processes, and solves the total estimated remaining time. The critical impact chain identification module traverses all detection paths, locates the most time-consuming critical path, and outputs the anomalies, waiting, and device nodes on the path. The probabilistic result output module outputs the estimated remaining time, conservative remaining time, confidence interval, key impact chain, abnormal blocking nodes, and shared waiting nodes. The historical sample update module collects full-process time-series data after a single detection and updates the time consumption distribution, anomaly recovery distribution, resource waiting distribution, and graph time-series prediction model parameters.

[0016] In one embodiment of the present invention, optionally, the system communicates with the controller, human-machine interface, and DCS host system of the multi-channel online sampling and detection device, and pushes the probabilistic remaining time and key influence chain to the human-machine interface for display in real time, while uploading it to the host system for intelligent scheduling of multi-channel detection tasks.

[0017] In one embodiment of the present invention, optionally, the method is applicable to two types of operating conditions: single-channel independent detection and multi-channel time-sharing instrument / pump valve sharing. Under the multi-channel time-sharing operating condition, the waiting time of shared resources is superimposed sequentially according to the task queue order, the remaining time of the entire process of each water sample detection task is calculated independently, and the key influence chain corresponding to each channel is identified.

[0018] In one embodiment of the present invention, optionally, when a serious fault anomaly is determined to exist that cannot be quantified in terms of recovery time, the system blocks the output of the remaining time value, and only pushes the location of the abnormal node, the type of the anomaly, and the fault prompt to the human-machine interface, suspends the iterative calculation of the remaining time, and restarts the full-process estimation logic after the fault is eliminated and the state returns to normal.

[0019] The present invention has the following beneficial technical effects: Compared with fixed delay or sequential control methods, this invention uses a heterogeneous event diagram to uniformly represent the sampling, flow and voltage stabilization, current storage or buffering, water circuit switching, sampling, instrument stabilization, detection, data judgment and result output processes. It can dynamically calculate the remaining time based on nodes that have not yet been reliably completed, reducing the problem of inconsistency between fixed countdowns and actual progress.

[0020] Compared with single-state judgment methods, the present invention avoids mistaking the issuance of instructions, the receipt of commands by valves, or the existence of instrument readings as the completion of steps by reliable completion judgment. This can reduce the misjudgment of progress caused by the actuator not being in place, the fluid not being stable, the instrument data not being valid, or the result not being confirmed and uploaded.

[0021] Compared with general prediction methods based solely on process logs, this invention uses flow rate, pressure, liquid level, pump and valve feedback, instrument occupancy, instrument stability, data validity, and communication confirmation as node status features, making it more suitable for estimating the entire process from sampling to valid result output in multi-channel sampling and detection.

[0022] Compared with simple anomaly identification methods, this invention converts unstable flow, abnormal pressure, insufficient liquid level, abnormal pump and valve operation, unstable instruments, missing or invalid data, and communication anomalies into anomaly time costs and includes them in the remaining time, and can output the estimated remaining time after anomaly correction.

[0023] When multiple water channels share instruments, pumps, valves, switching mechanisms, or communication interfaces, this invention includes the resource queuing waiting time in the estimated start time or remaining time of the node, and identifies the key impact chain, thereby explaining which water sample, which instrument, which data judgment link, which anomaly recovery or communication confirmation link mainly affects the current remaining time.

[0024] This invention outputs the estimated remaining time, conservative remaining time, confidence interval, current execution step, key impact chain, abnormal blocking nodes, and shared resource waiting nodes; at the same time, it gradually corrects the remaining time estimate under different waterways, instruments, operating conditions, and anomaly types by updating historical samples after the task is completed. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a schematic diagram of the overall structure of a multi-channel online sampling and detection device; Figure 2 The flowchart shows the method for estimating the remaining time of the entire process. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] The technical problem to be solved by this invention is: to dynamically correct the remaining time by comprehensively utilizing water circuit status, flow and pressure stability, pump and valve feedback, instrument occupancy, data reliability, and communication status during the processes of sampling, flow and pressure stabilization, flow storage or buffering, sample delivery, instrument testing, data validity determination, and result output; to identify key water circuits, key instruments, and key waiting links that affect the completion time when multiple water circuits share instruments, pumps and valves, or when time-sharing testing is used; and to quantify the impact of abnormalities on waiting, retesting, and result output time under abnormal conditions.

[0029] This invention aims to provide time status information for human-machine interface display, task scheduling, and upper-level system linkage by using heterogeneous event graphs, real-time node status time series, reliable completion determination, abnormal time cost, and shared resource queuing estimation to output the estimated remaining time, conservative remaining time, reliable interval, and key impact chain.

[0030] This invention does not use a fixed countdown or rely solely on historical average detection time. Instead, it models the multi-channel sampling and detection process as a heterogeneous event graph that includes process flow, equipment status, detection resources, data quality, and communication confirmation relationships. It also dynamically estimates the remaining completion time of the current detection task or detection queue by combining the real-time status time series of nodes, reliable completion judgment, abnormal time cost, and shared resource queuing constraints.

[0031] First embodiment: Figure 1 This is a schematic diagram of the overall structure of a multi-channel online sampling and testing device, used to illustrate the connection relationships of the sampling unit, the flow and pressure stabilization unit, the flow storage or buffering unit, the water channel switching unit, the online testing instrument, the controller, the human-machine interface, and the communication module. Figure 2 This is a flowchart of the whole process remaining time estimation method, which is used to illustrate the process of state acquisition, heterogeneous event graph construction, real-time state time series encoding of nodes, reliable completion determination, abnormal time cost calculation, node remaining time estimation, shared resource queuing, whole process remaining time fusion, key impact chain identification, probabilistic remaining time output and historical sample update.

[0032] like Figure 2 As shown, this invention provides a method for estimating the remaining time of multi-channel sampling and detection, which is used by a multi-channel online sampling and detection device to calculate the remaining time of the entire process from the current moment to the output of a valid detection result, including the following steps: S1. Construct a heterogeneous event graph for the entire process of multi-channel sampling and testing; abstract the online sampling and testing device into water channel nodes, pump and valve execution nodes, flow stabilization and pressure stabilization nodes, buffer and storage nodes, water channel switching nodes, testing instrument nodes, data judgment nodes, communication output nodes, and task nodes; establish process dependency edges, fluid connection edges, resource occupation edges, state influence edges, and communication confirmation edges between each node to uniformly represent the entire process of sampling, flow stabilization and pressure stabilization, buffer and storage, water channel switching, sample delivery, instrument testing, data judgment, and result output; S2. Collect and encode the real-time status time series of each node; collect flow rate, pressure, liquid level, pump and valve commands and feedback, flow and pressure stability status, instrument working status, detection data quality status and communication interaction status according to the preset sampling period, and store the continuous status sequence within the time window to represent the trend of working condition changes. S3. Execute a reliable completion judgment; based on the consistency of pump and valve feedback, the stability of flow and pressure conditions, the amount of buffer storage, the stability of instruments, the continuity and validity of detection data, and the communication upload confirmation flag, determine whether each node has been reliably completed. Reliably completed nodes will no longer participate in the remaining time calculation. S4. Calculate the time cost of anomalies; For various anomalies such as unstable flow, abnormal pressure, insufficient liquid level, pump and valve malfunction, unstable instruments, missing / peak / over-range / invalid data, and communication interruption / delay, the time cost of anomalies is obtained by combining the duration of the anomaly, the historical recovery time of similar anomalies, and the compensation time for restabilization / re-sampling / re-detection, and then added to the remaining time of the corresponding node; When the anomaly exceeds the estimable range, the remaining time is indeterminate and the corresponding anomaly node is output. S5. Estimate the remaining time of a single node on a node-by-node basis; for nodes that have not started, determine the basic remaining time based on the historical time consumption of the same waterway, the same instrument, and the same working conditions; for nodes that are currently executing, adjust the remaining time in real time by combining the running time, the trend of status changes, and the degree of satisfaction of the reliable completion conditions; for abnormally blocked nodes, add the abnormal time cost to the basic remaining time; when there are enough historical samples, use a graph time series prediction model to input the node relationship and the real-time status sequence to output the node's remaining time consumption and uncertainty range. S6. Introduce shared resource queuing constraints; For scenarios where multiple waterways share detection instruments, pumps, valves, switching mechanisms, and communication interfaces, calculate the resource waiting time based on the resource occupancy status, the expected release time of the occupied task, and the task queue order. The expected start time of a node is determined by the completion time of the preceding node, the resource release time, the anomaly recovery time, and the scheduling queue order. S7. Integrate the remaining time of the entire process and identify key impact chains; based on the heterogeneous event graph path, superimpose the time consumption of each node, sequentially accumulate serial nodes, take the longest completion time of parallel branches, superimpose the waiting time of shared resources, and superimpose the abnormal time cost of abnormal nodes; take the longest time consumption of the entire path as the estimated remaining time of the entire process, and simultaneously identify the key impact path with the longest consumption as the key impact chain. S8. Probabilistically output the remaining time result and update the historical sample library; output the estimated remaining time, conservative remaining time, confidence interval, current execution step, key impact chain, abnormal blocked node, and shared resource waiting node; after a single detection task is completed, record the actual start and stop time, time consumed, waiting time, anomaly type, resource occupation status, data validity, and communication confirmation time of each node, and update the three types of time distribution and prediction models of node time consumption, anomaly recovery, and resource waiting.

[0033] In one embodiment of the present invention, optionally, each node in the heterogeneous event graph is defined as follows: Water path nodes represent the flow paths of each water sample and the real-time sampling status; pump and valve execution nodes represent the status of pump and valve control commands, execution feedback, and action completion; flow and pressure stabilization nodes represent the flow rate, pressure values, and fluid stability judgment status of the pipeline; buffer and storage nodes represent the buffer liquid level, water sample storage volume, and storage compliance status; water path switching nodes represent the on / off status of multi-channel water sample switching valves; detection instrument nodes represent the instrument idle, occupied, preheating and stabilizing, detection, and fault shutdown status; data judgment nodes represent the status of missing detection data, spikes, disconnections, over-range, and valid / invalid status; communication output nodes represent the status of detection result upload, communication delay, link interruption, and upper-level confirmation; task nodes represent the current detection task and the queuing status of multi-water path tasks.

[0034] In one embodiment of the present invention, optionally, the process dependency edge represents the sequential execution logic of each detection step; the fluid connection edge represents the flow path relationship of water sample between water circuit, heat exchanger, buffer, and instrument; the resource occupation edge represents the occupation and queuing constraints of shared pumps, valves, detection instruments, and communication interfaces in multiple water circuits; the state influence edge represents the prolongation effect of abnormal and non-compliant states on the time consumption of subsequent nodes; and the communication confirmation edge represents the complete timing constraints of the detection data being uploaded to the host computer and receiving the confirmation signal.

[0035] In one embodiment of the present invention, optionally, the graph time series prediction model takes the node association weights of the heterogeneous event graph, the real-time state time series of the nodes, and historical working condition samples as inputs, and outputs the remaining basic time consumption and the time consumption confidence interval of each node. The node association weights are trained by the degree of influence of different nodes on the total detection time in historical data.

[0036] In one embodiment of the present invention, optionally, in step S3, the reliable completion determination rule includes: the pump valve control command is consistent with the field feedback signal, the pipeline flow and pressure are continuously stable for a preset time, the buffer water sample volume meets the minimum sampling threshold, the instrument preheating stability time reaches the standard, multiple sets of continuous detection data are without distortion and within the effective range, and the host computer returns a data upload confirmation signal. When all of these conditions are met, the corresponding node is determined to be reliable completed.

[0037] In one embodiment of the present invention, optionally, in step S6, the shared resource queuing constraint calculation logic is as follows: if the resources required by the target waterway are in an idle state and there is no additional waiting time; if the resources are occupied by other waterways, the resource release time is determined based on the remaining detection time of the occupied task, and the resource release time is used as the earliest start time of the target waterway node, which is added to the remaining time of the node.

[0038] like Figure 1 As shown, the present invention also provides a multi-channel sampling and detection remaining time estimation system, which is used to implement the above-mentioned multi-channel sampling and detection remaining time estimation method, including: The status acquisition module is used to acquire real-time status data such as flow rate, pressure, liquid level, pump and valve feedback, instrument operating conditions, detection data, and communication interaction within the multi-channel sampling and testing device, and generate a status time series. The heterogeneous event graph construction module is used to construct a full-process heterogeneous event graph containing multiple types of nodes and five types of associated edges, and to establish a logical and physical relationship model between various detection links, devices, and tasks. The trusted completion determination module has built-in trusted completion determination rules to determine whether each node meets the completion conditions and filter out completed nodes. The abnormal time cost calculation module identifies various equipment, fluid, data, and communication anomalies, quantifies the time loss caused by the anomalies, and outputs the abnormal time cost. The node remaining time estimation module includes a basic time calculation unit, an anomaly overlay unit, and a graph time series prediction unit, which respectively output the uncorrected and corrected remaining time and uncertainty range of each node. The shared resource queuing calculation module identifies shared equipment resources across multiple waterways, calculates resource queuing waiting time, and corrects node startup times. The remaining time fusion module for the entire process traverses all execution paths in the heterogeneous event graph, merges the time consumed by serial, parallel, waiting, and abnormal processes, and solves the total estimated remaining time. The critical impact chain identification module traverses all detection paths, locates the most time-consuming critical path, and outputs the anomalies, waiting, and device nodes on the path. The probabilistic result output module outputs the estimated remaining time, conservative remaining time, confidence interval, key impact chain, abnormal blocking nodes, and shared waiting nodes. The historical sample update module collects full-process time-series data after a single detection and updates the time consumption distribution, anomaly recovery distribution, resource waiting distribution, and graph time-series prediction model parameters.

[0039] In one embodiment of the present invention, optionally, the system communicates with the controller, human-machine interface, and DCS host system of the multi-channel online sampling and detection device, and pushes the probabilistic remaining time and key influence chain to the human-machine interface for display in real time, while uploading it to the host system for intelligent scheduling of multi-channel detection tasks.

[0040] In one embodiment of the present invention, optionally, the method is applicable to two types of operating conditions: single-channel independent detection and multi-channel time-sharing instrument / pump valve sharing. Under the multi-channel time-sharing operating condition, the waiting time of shared resources is superimposed sequentially according to the task queue order, the remaining time of the entire process of each water sample detection task is calculated independently, and the key influence chain corresponding to each channel is identified.

[0041] In one embodiment of the present invention, optionally, when a serious fault anomaly is determined to exist that cannot be quantified in terms of recovery time, the system blocks the output of the remaining time value, and only pushes the location of the abnormal node, the type of the anomaly, and the fault prompt to the human-machine interface, suspends the iterative calculation of the remaining time, and restarts the full-process estimation logic after the fault is eliminated and the state returns to normal.

[0042] Second embodiment: Single-channel full-process estimation. The system collects flow rate, pressure, level, pump and valve feedback, instrument status, detection values, data validity, and communication confirmation status of a single water sample. It constructs a heterogeneous event graph including nodes for sampling, flow and pressure stabilization, flow storage or buffering, sample delivery, instrument detection, data judgment, and communication output. The system sequentially determines whether each node has been reliably completed. For incomplete nodes, it estimates the remaining time based on status trends, historical time consumption, and abnormal time costs, and merges these estimates according to the process sequence to obtain the estimated remaining time from the current moment to the output of a valid result.

[0043] Third embodiment: Multi-channel shared instrument estimation. Of the three water samples, the first channel is currently using a monitoring instrument, the second channel has completed flow storage and is awaiting instrumentation, and the third channel is still in the sampling or flow / pressure stabilization phase. The system estimates the instrument release time based on the monitoring status of the first channel, overlays the instrument waiting time on the second channel, and first estimates the remaining time for sampling, flow / pressure stabilization, and flow storage for the third channel. Then, combining the instrument release time and queue sequence, it calculates the expected start and end times and identifies the critical impact chains of the current queue.

[0044] Fourth embodiment: Abnormal operating condition correction. When flow fluctuations occur during the steady flow and pressure stabilization phase of a certain water sample, the system collects the flow status time series and determines that the node does not meet the stability requirements, marking it as an abnormal blockage node. The system calculates the abnormal time cost based on the duration of the abnormality, the historical recovery time of similar abnormalities, and the re-stabilization compensation time, and adds it to the remaining time of the relevant nodes. If the abnormality continues beyond the estimable range or there is severe data loss, the system outputs the remaining time as uncertain and the corresponding abnormal node.

[0045] Fifth embodiment: Results output and sample updates. Based on the remaining time of nodes, shared resource waiting time, and the cost of abnormal time, the system outputs the estimated remaining time, conservative remaining time, confidence interval, current execution step, key impact chain, abnormal blocked nodes, shared resource waiting nodes, and update time. After the detection task is completed, the system records the actual start time, completion time, time consumed, waiting time, abnormal type, abnormal duration, resource occupancy status, data validity status, and result confirmation time for each node. This information is used to update the historical sample database and subsequent estimation models.

[0046] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of one embodiment, and the modules or processes shown in the drawings are not necessarily essential for implementing the present invention.

[0047] Those skilled in the art will understand that the modules in the apparatus of the embodiments can be distributed in the apparatus of the embodiments as described in the embodiments, or they can be located in one or more devices different from this embodiment with corresponding changes. The modules of the above embodiments can be combined into one module, or they can be further divided into multiple sub-modules.

[0048] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for estimating the remaining time of multi-channel sampling and detection, characterized in that, The multi-channel online sampling and detection device calculates the remaining time from the current moment until the output of a valid detection result, including the following steps: S1. Construct a heterogeneous event graph for the entire process of multi-channel sampling and testing; abstract the online sampling and testing device into water channel nodes, pump and valve execution nodes, flow stabilization and pressure stabilization nodes, buffer and storage nodes, water channel switching nodes, testing instrument nodes, data judgment nodes, communication output nodes, and task nodes; establish process dependency edges, fluid connection edges, resource occupation edges, state influence edges, and communication confirmation edges between each node to uniformly represent the entire process of sampling, flow stabilization and pressure stabilization, buffer and storage, water channel switching, sample delivery, instrument testing, data judgment, and result output; S2. Collect and encode the real-time status time series of each node; collect flow rate, pressure, liquid level, pump and valve commands and feedback, flow and pressure stability status, instrument working status, detection data quality status and communication interaction status according to the preset sampling period, and store the continuous status sequence within the time window to represent the trend of working condition changes. S3. Execute a reliable completion judgment; based on the consistency of pump and valve feedback, the stability of flow and pressure conditions, the amount of buffer storage, the stability of instruments, the continuity and validity of detection data, and the communication upload confirmation flag, determine whether each node has been reliably completed. Reliably completed nodes will no longer participate in the remaining time calculation. S4. Calculate the time cost of anomalies; For various anomalies such as unstable flow, abnormal pressure, insufficient liquid level, pump and valve malfunction, unstable instruments, missing / peak / over-range / invalid data, and communication interruption / delay, the time cost of anomalies is obtained by combining the duration of the anomaly, the historical recovery time of similar anomalies, and the compensation time for restabilization / re-sampling / re-detection, and then added to the remaining time of the corresponding node; When the anomaly exceeds the estimable range, the remaining time is indeterminate and the corresponding anomaly node is output. S5. Estimate the remaining time of a single node on a node-by-node basis; for nodes that have not started, determine the basic remaining time based on the historical time consumption of the same waterway, the same instrument, and the same working conditions; for nodes that are currently executing, adjust the remaining time in real time by combining the running time, the trend of status changes, and the degree of satisfaction of the reliable completion conditions; for abnormally blocked nodes, add the abnormal time cost to the basic remaining time; when there are enough historical samples, use a graph time series prediction model to input the node relationship and the real-time status sequence to output the node's remaining time consumption and uncertainty range. S6. Introduce shared resource queuing constraints; For scenarios where multiple waterways share detection instruments, pumps, valves, switching mechanisms, and communication interfaces, calculate the resource waiting time based on the resource occupancy status, the expected release time of the occupied task, and the task queue order. The expected start time of a node is determined by the completion time of the preceding node, the resource release time, the anomaly recovery time, and the scheduling queue order. S7. Integrate the remaining time of the entire process and identify key impact chains; based on the heterogeneous event graph path, superimpose the time consumption of each node, sequentially accumulate serial nodes, take the longest completion time of parallel branches, superimpose the waiting time of shared resources, and superimpose the abnormal time cost of abnormal nodes; take the longest time consumption of the entire path as the estimated remaining time of the entire process, and simultaneously identify the key impact path with the longest consumption as the key impact chain. S8. Probabilistically output the remaining time result and update the historical sample library; output the estimated remaining time, conservative remaining time, confidence interval, current execution step, key impact chain, abnormal blocked node, and shared resource waiting node; after a single detection task is completed, record the actual start and stop time, time consumed, waiting time, anomaly type, resource occupation status, data validity, and communication confirmation time of each node, and update the three types of time distribution and prediction models of node time consumption, anomaly recovery, and resource waiting.

2. The method for estimating the remaining time of multi-channel sampling and detection according to claim 1, characterized in that, The nodes in the heterogeneous event graph are defined as follows: Waterway nodes represent the flow paths of each water sample and the real-time sampling status; pump and valve execution nodes represent the control commands, execution feedback, and action status of pumps and valves; flow and pressure stabilization nodes represent the flow rate, pressure values, and fluid stability judgment status of the pipeline; buffer and storage nodes represent the buffer liquid level, water sample storage volume, and storage compliance status. The water path switching node represents the on / off status of multiple water sample switching valves; The instrument node represents the instrument's idle, occupied, preheating and stable, detection, and fault shutdown status. The data judgment node represents the status of missing, spike, disconnection, over-range, and valid / invalid data; the communication output node represents the status of uploaded detection results, communication delay, link interruption, and upper-level confirmation; the task node represents the status of the current detection task and the queue status of multi-channel tasks.

3. The method for estimating the remaining time of multi-channel sampling and detection according to claim 1, characterized in that, The process dependency edge represents the sequential execution logic of each detection step; the fluid connection edge represents the flow path relationship of the water sample between the water circuit, heat exchanger, buffer, and instrument. Resource occupancy edges represent the occupancy and queuing constraints of shared pumps, valves, detection instruments, and communication interfaces in multiple waterways; The state influences the extension of time consumption for subsequent nodes due to abnormal or unqualified edge representations; the communication confirms the complete timing constraint of uploading edge representation detection data to the host computer and receiving an acknowledgment signal.

4. The method for estimating the remaining time of multi-channel sampling and detection according to claim 1, characterized in that, The graph time series prediction model takes the node association weights of the heterogeneous event graph, the real-time state time series of nodes, and historical working condition samples as inputs, and outputs the remaining basic time consumption and the time consumption confidence interval of each node. The node association weights are trained by the degree of influence of different nodes on the total detection time in historical data.

5. The method for estimating the remaining time of multi-channel sampling and detection according to claim 1, characterized in that, In step S3, the reliable completion determination rules include: the pump valve control command is consistent with the field feedback signal, the pipeline flow and pressure are continuously stable for a preset time, the buffer water sample volume meets the minimum sampling threshold, the instrument preheating is stable for a specified time, multiple sets of continuous test data are without distortion and within the effective range, and the host computer returns a data upload confirmation signal. When all of these conditions are met, the corresponding node is determined to be reliable completed.

6. The method for estimating the remaining time of multi-channel sampling and detection according to claim 1, characterized in that, In step S6, the shared resource queuing constraint calculation logic is as follows: if the resources required by the target waterway are in an idle state and there is no additional waiting time; if the resources are occupied by other waterways, the resource release time is determined based on the remaining detection time of the occupied task, and the resource release time is used as the earliest start time of the target waterway node, which is then added to the remaining time of the node.

7. A multi-channel sampling and detection remaining time estimation system, characterized in that, The method for estimating the remaining time of multi-channel sampling and detection as described in any one of claims 1 to 6 includes: The status acquisition module is used to acquire real-time status data such as flow rate, pressure, liquid level, pump and valve feedback, instrument operating conditions, detection data, and communication interaction within the multi-channel sampling and testing device, and generate a status time series. The heterogeneous event graph construction module is used to construct a full-process heterogeneous event graph containing multiple types of nodes and five types of associated edges, and to establish a logical and physical relationship model between various detection links, devices, and tasks. The trusted completion determination module has built-in trusted completion determination rules to determine whether each node meets the completion conditions and filter out completed nodes. The abnormal time cost calculation module identifies various equipment, fluid, data, and communication anomalies, quantifies the time loss caused by the anomalies, and outputs the abnormal time cost. The node remaining time estimation module includes a basic time calculation unit, an anomaly overlay unit, and a graph time series prediction unit, which respectively output the uncorrected and corrected remaining time and uncertainty range of each node. The shared resource queuing calculation module identifies shared equipment resources across multiple waterways, calculates resource queuing waiting time, and corrects node startup times. The remaining time fusion module for the entire process traverses all execution paths in the heterogeneous event graph, merges the time consumed by serial, parallel, waiting, and abnormal processes, and solves the total estimated remaining time. The critical impact chain identification module traverses all detection paths, locates the most time-consuming critical path, and outputs the anomalies, waiting, and device nodes on the path. The probabilistic result output module outputs the estimated remaining time, conservative remaining time, confidence interval, key impact chain, abnormal blocking nodes, and shared waiting nodes. The historical sample update module collects full-process time-series data after a single detection and updates the time consumption distribution, anomaly recovery distribution, resource waiting distribution, and graph time-series prediction model parameters.

8. The multi-channel sampling and detection remaining time estimation system according to claim 7, characterized in that, The system communicates with the controller, human-machine interface, and DCS host system of the multi-channel online sampling and detection device, and pushes the probabilistic remaining time and key impact chain to the human-machine interface for display in real time. At the same time, it uploads the data to the host system for intelligent scheduling of multi-channel detection tasks.

9. The method for estimating the remaining time of multi-channel sampling and detection according to claim 1, characterized in that, It is applicable to two types of operating conditions: independent detection of a single water path and time-sharing sharing of instruments / pumps and valves for multiple water paths. In the case of time-sharing sharing of multiple water paths, the waiting time of shared resources is added sequentially according to the task queue order, and the remaining time of the entire process of each water sample detection task is calculated independently, and the corresponding key influence chain of each path is identified.

10. The method for estimating the remaining time of multi-channel sampling and detection according to claim 1, characterized in that, When a serious fault or anomaly is determined to exist that cannot be quantified in terms of recovery time, the system blocks the output of the remaining time value and only pushes the location of the abnormal node, the type of the anomaly, and the fault prompt to the human-machine interface. The remaining time iterative calculation is paused, and the entire process estimation logic is restarted after the fault is eliminated and the status returns to normal.

Citation Information

Patent Citations

  • A method for predicting the remaining time of a business process based on multi-layer machine learning

    CN111178623B

  • Identification method for online monitoring data exception of water pollution source

    CN112287988A

  • Small multi-parameter water quality on-line monitoring system and method of integrated flow path and application of small multi-parameter water quality on-line monitoring system

    CN114965922A