Petroleum filtering device fault detection method and system based on Internet of Things

Abstract

The invention discloses a petroleum filtering device fault detection method and system based on the Internet of Things, and relates to the technical field of data processing. The method comprises the following steps: acquiring attribute information of a filtering layer of a target filtering device; performing simulation modeling on the target filtering device to generate a filtering twinborn model; determining a simulation monitoring index set; performing filtering prediction on the preprocessed water quality information to obtain a prediction monitoring index set; determining a fusion monitoring index set under the plurality of monitoring nodes; performing data acquisition on the N filter layers to obtain a real-time monitoring data set; and judging the real-time monitoring data set, and positioning a filter layer which does not meet the fusion monitoring index to carry out fault early warning. The technical problem that fault positioning and fault early warning cannot be accurately achieved due to the fact that the fineness and the accuracy of fault early warning index setting of a filtering layer of a petroleum fiber filter in the prior art are insufficient is solved, and the technical effects of improving the accuracy and the timeliness of fault detection and achieving accurate early warning of the filtering layer are achieved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and more specifically to a method and system for fault detection of oil filtration devices based on the Internet of Things. Background Technology

[0002] In the oil extraction and processing process, the treatment of oilfield wastewater is a crucial step. Oilfield fiber filters, as a highly efficient wastewater treatment device, are widely used in the oil industry. However, in actual operation, due to the complexity of oilfield fiber filters and the variability of the working environment, early warning of filter layer failures has become a technical challenge. Traditional oilfield fiber filter failure warning methods often rely on fixed threshold settings and simple sensor monitoring. When dealing with complex oilfield wastewater, the precision and accuracy of the warning indicators are often insufficient. This leads to frequent misjudgments or missed judgments in practical applications, making it impossible to accurately locate and warn of faults, thus affecting the efficiency and quality of fault handling. Summary of the Invention

[0003] This application provides a method and system for fault detection of oil filtration devices based on the Internet of Things, which solves the technical problem that the precision and accuracy of the fault warning indicators set for the filter layer of oil fiber filters in the prior art are insufficient, resulting in the inability to accurately locate and warn of faults.

[0004] In view of the above problems, embodiments of this application provide a method and system for fault detection of oil filtration devices based on the Internet of Things.

[0005] A first aspect of this application provides a method for fault detection of an oil filtration device based on the Internet of Things, the method comprising: The process involves: acquiring attribute information of the filter layers in a target filtration device, wherein the target filtration device comprises N filter layers connected in series in sequence; and obtaining attribute information for each filter layer including material type, distribution structure, pore diameter, porosity, filtration method, and location coordinates. Based on a digital twin, a simulation model of the target filtration device is generated according to the material type, distribution structure, pore diameter, porosity, filtration method, and location coordinates. Pre-treated water quality information is collected, and the pre-treated water quality information is simulated for filtration using the filtration twin model to determine a set of simulated monitoring indicators at multiple monitoring nodes, including water quality indicators, flow rate indicators, and pressure indicators. Based on the attribute information and the pre-treated water quality information, historical filtration records are retrieved. The data is used to filter and predict the pre-treated water quality information to obtain a set of predicted monitoring indicators for the multiple monitoring nodes. The simulated monitoring indicator set and the predicted monitoring indicator set are fused according to a preset fusion strategy to determine the fused monitoring indicator set for the multiple monitoring nodes. At each monitoring node, data is collected from the N filter layers through an IoT monitoring unit to obtain a real-time monitoring dataset. The IoT monitoring unit includes N sensor groups, each corresponding to one of the N filter layers and deployed at the filter layer outlet. Each sensor group includes a water quality sensor, a flow sensor, and a pressure sensor. Based on the fused monitoring indicator set, the real-time monitoring dataset for the same monitoring node is judged, and filter layers that do not meet the fused monitoring indicators are identified and given a fault warning.

[0006] A second aspect of this application provides an Internet of Things (IoT)-based oil filtration device fault detection system, the system comprising: The system comprises the following modules: an information acquisition module, used to acquire attribute information of the filter layers of a target filtration device, wherein the target filtration device includes N filter layers connected in series in sequence, and the attribute information of each filter layer includes material type, distribution structure, pore diameter, porosity, filtration method, and location coordinates; a modeling module, used to perform simulation modeling of the target filtration device based on digital twins, according to the material type, distribution structure, pore diameter, porosity, filtration method, and location coordinates, generating a filter twin model; a simulation module, used to collect pretreated water quality information, and simulate filtration of the pretreated water quality information using the filter twin model, determining a set of simulated monitoring indicators under multiple monitoring nodes, wherein the monitoring indicators include water quality indicators, flow indicators, and pressure indicators; and a prediction module, used to adjust the system based on the attribute information and the pretreated water quality information. The pretreated water quality information is filtered and predicted using historical filtering data to obtain a set of predicted monitoring indicators for the multiple monitoring nodes; a fusion module is used to fuse the simulated monitoring indicator set and the predicted monitoring indicator set according to a preset fusion strategy to determine a fused monitoring indicator set for the multiple monitoring nodes; a data acquisition module is used to collect data from the N filter layers at the monitoring nodes through an IoT monitoring unit to obtain a real-time monitoring dataset, wherein the IoT monitoring unit includes N sensor groups, each corresponding to one of the N filter layers and deployed at the filter layer outlet, and each sensor group includes a water quality sensor, a flow sensor, and a pressure sensor; a judgment module is used to judge the real-time monitoring dataset for the same monitoring node based on the fused monitoring indicator set, and to locate filter layers that do not meet the fused monitoring indicators and issue a fault warning.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages: First, the attribute information of the filter layers of the target filtration device is obtained. The target filtration device comprises N filter layers, which are connected in series in sequence. The attribute information of each filter layer includes material type, distribution structure, pore diameter, porosity, filtration method, and location coordinates. Next, based on digital twins, a simulation model of the target filtration device is generated according to the material type, distribution structure, pore diameter, porosity, filtration method, and location coordinates. Then, pretreated water quality information is collected, and the pretreated water quality information is simulated for filtration using the filtration twin model to determine a set of simulated monitoring indicators for multiple monitoring nodes. These monitoring indicators include water quality indicators, flow rate indicators, and pressure indicators. Next, based on the attribute information and pretreated water quality information, historical filtration records are used to predict the filtration of the pretreated water quality information, resulting in a set of predicted monitoring indicators for multiple monitoring nodes. Finally, the simulated monitoring indicator set and the predicted monitoring indicator set are fused according to a preset fusion strategy to determine the fused monitoring indicator set for the multiple monitoring nodes. At each monitoring node, data is collected from N filter layers via an IoT monitoring unit to obtain a real-time monitoring dataset. The IoT monitoring unit comprises N sensor groups, each corresponding to one of the N filter layers, deployed at the filter layer outlet. Each sensor group includes a water quality sensor, a flow sensor, and a pressure sensor. Finally, based on a fused monitoring index set, the real-time monitoring dataset at the same monitoring node is analyzed to identify filter layers that do not meet the fused monitoring indexes and provide fault warnings. This solves the technical problem in existing petroleum fiber filter technologies where the precision and accuracy of fault warning index settings for filter layers are insufficient, leading to inaccurate fault location and warning. It improves the accuracy and timeliness of fault detection, achieving precise early warning for filter layers. Attached Figure Description

[0008] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0009] Figure 1 A schematic diagram of the process for a fault detection method for an oil filtration device based on the Internet of Things provided in this application embodiment; Figure 2 This is a schematic diagram of the structure of an Internet of Things-based oil filtration device fault detection system provided in an embodiment of this application.

[0010] Explanation of reference numerals in the attached diagram: Information acquisition module 11, modeling module 12, simulation module 13, prediction module 14, fusion module 15, data acquisition module 16, and judgment module 17. Detailed Implementation

[0011] This application provides a method and system for fault detection of oil filtration devices based on the Internet of Things, which solves the technical problem that the precision and accuracy of the fault warning indicators for the filter layer of oil fiber filters are insufficient in the prior art, resulting in the inability to accurately locate and warn of faults.

[0012] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0013] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to these processes, methods, products, or devices.

[0014] Example 1 like Figure 1 As shown in the embodiment of this application, a fault detection method for an oil filtration device based on the Internet of Things is provided, wherein the method includes: Obtain the attribute information of the filter layer of the target filtration device, wherein the target filtration device includes N filter layers, which are distributed in series in sequence, and the attribute information of each filter layer includes material type, distribution structure, pore diameter, porosity, filtration method and location coordinates.

[0015] By interacting with the target filtration device, the attribute information of the filter layers of the target filtration device is obtained. The target filtration device comprises N filter layers, each connected in series in sequence, and each filter layer possesses specific attribute information. The attribute information of each filter layer includes material type, distribution structure, pore diameter, porosity, filtration method, and location coordinates. Material type describes the material used in the filter layer, such as polyester, nylon, or glass fiber; distribution structure describes the arrangement, number of layers, and density of the materials within the filter layer, such as particle size distribution and layered structure; pore diameter describes the average or specific diameter of the pores in the filter layer, which determines the size of particles that can pass through the filter layer; porosity indicates the proportion of pores in the filter layer, affecting filtration efficiency and flow rate; filtration method describes how the filter layer works, such as interception filtration or adsorption filtration; and location coordinates refer to the specific position of each filter layer within the filtration device.

[0016] Based on digital twins, the target filtration device is simulated and modeled according to the material type, distribution structure, pore diameter, porosity, filtration method, and location coordinates to generate a filtration twin model.

[0017] Based on digital twin technology, a 3D model of the target filtration device is created using modeling software, taking into account all relevant information, including material type, distribution structure, pore diameter, porosity, filtration method, and location coordinates. This 3D model is then imported into simulation software to construct a filtration twin model. The filtration twin model can simulate the physical behavior of the filtration device during actual operation, including fluid flow, filtration efficiency, and pressure distribution.

[0018] Pre-treated water quality information is collected, and the pre-treated water quality information is simulated and filtered using the filter twin model to determine a set of simulated monitoring indicators under multiple monitoring nodes. The monitoring indicators include water quality indicators, flow rate indicators, and pressure indicators.

[0019] Pre-treated water quality information refers to the water quality information of petroleum wastewater, including extraction wastewater and refining wastewater. This information includes turbidity, color, pH value, suspended solids content, and dissolved oxygen. Pre-treated water quality information is collected and input into a filtration twin model to simulate filtration, mimicking the process of water flow through each filter layer. Based on the design and actual operational requirements of the filtration device, multiple monitoring nodes are set in the filtration twin model. These nodes are typically located at the outlet or key positions of the filter layers to monitor critical parameters during the filtration process. At each monitoring node, water quality indicators, flow rate indicators, and pressure indicators are monitored to obtain a simulated monitoring indicator set for multiple monitoring nodes. These monitoring indicators reflect the filtration effect and operational status of the filtration device at that node.

[0020] Furthermore, the methods include: The water quality indicators include impurity types and corresponding concentrations for each impurity type. The impurity types include oil impurities, metal particles, inorganic suspended solids, and organic suspended solids.

[0021] Water quality indicators include impurities such as oily impurities, metal particles, inorganic suspended solids, and organic suspended solids, along with their corresponding concentrations in the water. Oily impurities include floating oil, dispersed oil, emulsified oil, and dissolved oil; metal particles include copper and iron filings; inorganic suspended solids include silt and clay; and organic suspended solids include algae and bacteria. At each monitoring point, the concentrations of each type of impurity (oily impurities, metal particles, inorganic suspended solids, and organic suspended solids) in the water are calculated.

[0022] Based on the attribute information and the pretreated water quality information, historical filtration record data is called to perform filtration prediction on the pretreated water quality information, thereby obtaining a set of predicted monitoring indicators for the multiple monitoring nodes.

[0023] Based on the target filtration device's attribute information and pretreated water quality information, historical filtration records are retrieved from big data. These historical records refer to data on the filtration device's operation and monitoring nodes under similar water quality conditions in the past. Based on these historical filtration records, filtration predictions are made for the pretreated water quality information, resulting in a set of predicted monitoring indicators for multiple monitoring nodes.

[0024] Furthermore, based on the attribute information and the pretreated water quality information, historical filtration record data is used to perform filtration prediction on the pretreated water quality information, including: Select the first filter layer among the N filter layers, and obtain the first attribute information of the first filter layer, wherein the first filter layer is the filter layer ranked first among the N filter layers; using the first attribute information and the pretreated water quality information as constraints, retrieve multiple first sample filter record datasets through network retrieval; filter the multiple first sample filter record datasets based on the multiple monitoring nodes to obtain multiple first node filter record datasets; calculate the mean of the node filter record data under the same monitoring node based on the multiple first node filter record datasets, and use the mean calculation result as the predicted monitoring indicator for the corresponding monitoring node to obtain the first predicted monitoring indicator set under the multiple monitoring nodes; construct the predicted monitoring indicator set based on the first predicted monitoring indicator set.

[0025] From the N filter layers of the target filtration device, the highest-ranked filter layer is selected as the first filter layer. The first attribute information of the first filter layer, including material type, pore diameter, and porosity, is obtained. Using the first attribute information of the first filter layer and the pretreated water quality information as constraints, relevant historical filtration record datasets are retrieved online to obtain multiple first sample filtration record datasets. Each first sample filtration record dataset includes continuous records within a time period. The retrieved first sample filtration record datasets are then filtered to select those containing data from specific monitoring nodes. This results in multiple first node filtration record datasets, each containing filtration records from a specific monitoring node. For each monitoring node, the data in its first node filtration record dataset is processed. The mean of the node filtration record data under the same monitoring node is calculated, and the mean result is used as the predicted monitoring indicator for the corresponding monitoring node, thus obtaining a first predicted monitoring indicator set for multiple monitoring nodes. This process is repeated for other filter layers until the Nth predicted monitoring indicator set is obtained. The predicted monitoring indicator sets from multiple filter layers are then integrated to obtain the final predicted monitoring indicator set.

[0026] Furthermore, constructing the predictive monitoring indicator set based on the first predictive monitoring indicator set includes: Select the second filter layer among the N filter layers, obtain the second attribute information of the second filter layer, wherein the second filter layer is an adjacent filter layer of the first filter layer; using the second attribute information and the first predictive monitoring indicator set as constraints, retrieve multiple second sample filter record datasets through network retrieval, and analyze to obtain the second predictive monitoring indicator set under the multiple monitoring nodes; perform iterative analysis on the remaining filter layers of the N filter layers until the Nth predictive monitoring indicator set is obtained; generate the predictive monitoring indicator set based on the first predictive monitoring indicator set, the second predictive monitoring indicator set, and up to the Nth predictive monitoring indicator set.

[0027] From the N filter layers of the target filtration device, select the adjacent filter layer of the first filter layer as the second filter layer, and obtain the second attribute information of the second filter layer, including material type, pore diameter, porosity, etc. Using the second attribute information of the second filter layer and the first predictive monitoring index set as constraints, search the network for historical filter record datasets related to these conditions to obtain multiple second sample filter record datasets. Filter the second sample filter record datasets based on multiple monitoring nodes to obtain multiple second node filter record datasets. Calculate the mean of the node filter record data under the same monitoring node, and use the mean calculation result as the predictive monitoring index for the corresponding monitoring node, thereby obtaining the second predictive monitoring index set. Repeat the above process, selecting the next filter layer (i.e., the third filter layer, then the fourth filter layer, and so on), and obtaining its attribute information. Using the predictive monitoring index set of the previous filter layer and the attribute information of the current filter layer as constraints, search the network for sample filter record datasets. Analyze the retrieved datasets to calculate the predictive monitoring index set corresponding to the current filter layer. Iterate this process until all N filter layers have been analyzed, obtaining the Nth predictive monitoring index set. The first set of predictive monitoring indicators is integrated with the Nth set of predictive monitoring indicators to obtain the final set of predictive monitoring indicators. This set includes predicted water quality, flow, and pressure indicators for all monitoring nodes at each filtration layer.

[0028] The simulated monitoring indicator set and the predicted monitoring indicator set are fused according to a preset fusion strategy to determine the fused monitoring indicator set under the multiple monitoring nodes.

[0029] The preset fusion strategies include weighted average, maximum value, minimum value, and median. Based on these strategies, corresponding indicators from the simulated and predicted monitoring indicator sets are fused and calculated. For example, using a weighted average as the fusion strategy requires determining the weight of each indicator set and calculating the weighted average as the fused indicator value. The fused indicator data from all monitoring nodes are then integrated to generate the final fused monitoring indicator set. This fused monitoring indicator set combines information from both the simulated and predicted monitoring indicator sets, providing more comprehensive and accurate monitoring results.

[0030] Furthermore, the simulated monitoring indicator set and the predicted monitoring indicator set are fused according to a preset fusion strategy, including: The simulation accuracy of the filtering twin model is obtained, and the deviation of the simulation accuracy is calculated according to the preset simulation accuracy to determine the simulation accuracy deviation. Based on the coefficient of variation method, the simulation confidence weight is set according to the simulation accuracy deviation, and the predicted confidence weight is calculated based on the simulation confidence weight, wherein the sum of the simulation confidence weight and the predicted confidence weight is 1. The preset fusion strategy is generated according to the simulation confidence weight and the predicted confidence weight.

[0031] The simulation accuracy of the simulated monitoring indicator set is obtained from the filtered twin model. Simulation accuracy represents the error or accuracy between the model's predicted values ​​and the actual values. A preset simulation accuracy is set based on historical averages, industry standards, or expert opinions, and the deviation between the preset simulation accuracy and the actual simulation accuracy is calculated. The coefficient of variation (COP) is the ratio of the standard deviation to the mean, used to measure the dispersion of the data. The COP of the simulation accuracy deviation is calculated, and simulation confidence weights are set based on the COP. A function (such as an inverse proportional function or an exponential function) can be used to map the COP to a weight value between 0 and 1. The smaller the COP, the higher the stability or reliability of the simulated data, and therefore the larger the simulation confidence weight. The sum of the simulation confidence weight and the prediction confidence weight is 1, and the prediction confidence weight is obtained by subtracting the simulation confidence weight from 1. The simulated monitoring indicator set and the predicted monitoring indicator set are merged based on the simulation confidence weight and the prediction confidence weight as a fusion strategy.

[0032] At the monitoring node, data is collected from the N filter layers through the Internet of Things (IoT) monitoring unit to obtain a real-time monitoring dataset. The IoT monitoring unit includes N sensor groups, which correspond one-to-one with the N filter layers and are deployed at the outlet of the filter layers. Each sensor group includes a water quality sensor, a flow sensor, and a pressure sensor.

[0033] The IoT monitoring unit comprises N sensor groups, each corresponding one-to-one with one of the N filter layers. These N sensor groups are deployed at the outlet of each filter layer to ensure accurate capture of various data points related to the effluent. Each sensor group includes a water quality sensor, a flow sensor, and a pressure sensor. The water quality sensor monitors the quality of the effluent, including but not limited to temperature, pH, turbidity, dissolved oxygen, and concentrations of specific pollutants. The flow sensor measures the flow rate of the effluent, and the pressure sensor monitors pressure changes within the filter layer. At the monitoring node, the IoT monitoring unit collects data from the N filter layers in real time, obtaining a real-time monitoring dataset.

[0034] Based on the fusion monitoring index set, the real-time monitoring dataset under the same monitoring node is judged, and the filter layer that does not meet the fusion monitoring index is located for fault warning.

[0035] After obtaining the fusion monitoring index set, the fusion monitoring index set is used to judge the real-time monitoring dataset under the same monitoring node in order to locate the filter layer that does not meet the fusion monitoring index and to issue a fault warning.

[0036] Furthermore, based on the fused monitoring indicator set, the real-time monitoring dataset under the same monitoring node is judged, including: Obtain the real-time monitoring nodes of the real-time monitoring dataset; Based on the fusion monitoring indicator set, associated fusion monitoring indicators are obtained by matching the real-time monitoring nodes. The filtering influence of each of the N filter layers is analyzed, and N tolerance intervals are set based on the influence analysis results. The tolerance intervals are inversely proportional to the influence analysis results. The correlation fusion monitoring indicators are updated according to the N tolerance intervals to obtain the correlation fusion monitoring threshold, and the real-time monitoring dataset is judged based on the correlation fusion monitoring threshold.

[0037] Extract the currently monitored node information from the real-time monitoring dataset. Based on the extracted real-time monitoring node information, search for related fusion monitoring indicators in the fusion monitoring indicator set, ensuring that the found related fusion monitoring indicators are applicable to the current real-time monitoring node and can accurately reflect the node's operating status. Perform filtration impact analysis on N filter layers, that is, evaluate the degree of impact on overall wastewater treatment. Optionally, determine key indicators for assessing the impact, such as filtration efficiency, treatment capacity, energy consumption, and failure frequency. Based on the expert team's evaluation of the key indicators for each filter layer, calculate an impact score for each filter layer using a weighted scoring method. Set N tolerance intervals based on the impact analysis results. The tolerance interval is a threshold that allows the filter layer monitoring indicators to fluctuate within a certain range. Calculate a specific tolerance interval based on the impact score of each filter layer. The size of the tolerance interval is inversely proportional to the impact analysis results; that is, the greater the impact of the filter layer, the smaller its tolerance interval should be to ensure the stability and safety of the system. For example, consider two filter layers, A and B, where A has an impact score of 80 (out of 100) and B has an impact score of 60. The water quality tolerance range for these two filter layers is set to [90, 100] (using the water quality index as an example, a higher value indicates better water quality). Because A has a higher impact score, a narrower tolerance range is set for it, such as [95, 100]. Because B has a lower impact score, a wider tolerance range is set for it, such as [90, 100]. Based on the set N tolerance ranges, the associated fusion monitoring indicators are updated to obtain the associated fusion monitoring threshold. The associated fusion monitoring threshold reflects the normal operating status of each filter layer within its tolerance range and triggers a fault warning when the tolerance range is exceeded. Using the updated associated fusion monitoring threshold, each monitoring indicator in the real-time monitoring dataset is judged. If the monitoring indicator of a filter layer exceeds its corresponding associated fusion monitoring threshold, the filter layer is considered to have a fault or abnormal situation, and the corresponding fault warning mechanism is triggered.

[0038] Furthermore, based on the aforementioned correlation fusion monitoring threshold, the real-time monitoring dataset is judged, and the process further includes: The associated fusion monitoring threshold includes N associated fusion monitoring intervals, wherein each associated fusion monitoring interval includes a water quality monitoring interval, a flow monitoring interval, and a pressure monitoring interval; Based on the N associated fusion monitoring intervals, the real-time monitoring dataset is mapped and judged to determine multiple abnormal monitoring indicators, and the filtering layers of the abnormal monitoring indicators are mapped and located to obtain multiple abnormal filtering layers. The abnormal monitoring indicators are input into the fault analysis database for matching to determine multiple predicted fault types. The multiple abnormal filtering layers are mapped and identified according to the multiple predicted fault types, and a fault warning signal is generated based on the multiple identified abnormal filtering layers and their corresponding location coordinates and sent to the fault repair unit.

[0039] The correlated fusion monitoring threshold comprises N correlated fusion monitoring intervals, each including a water quality monitoring interval, a flow monitoring interval, and a pressure monitoring interval. The water quality monitoring interval includes multiple impurity types and their corresponding concentration thresholds. Each data point in the real-time monitoring dataset (including water quality, flow, and pressure data) is mapped to its corresponding correlated fusion monitoring interval to determine if the real-time monitoring data falls within that interval. If it does not, the data point is considered an abnormal monitoring indicator. Abnormal monitoring indicators are then mapped to their corresponding filter layers, resulting in multiple abnormal filter layers. This means that for each data point exceeding the correlated fusion monitoring interval, the filter layer that generated the data point can be traced back to it. The abnormal monitoring indicators are then input into a fault analysis database for matching. This database includes multiple sample monitoring indicators and their corresponding fault types, with one-to-one or one-to-many relationships. For example, a rapid increase in pressure drop within a short period indicates severe blockage of the filter layer, requiring cleaning or replacement; or a decline in water filtration quality may be due to excessive impurity accumulation or a leak in the fiber layer. Through matching, the predicted fault type that best matches the abnormal monitoring indicator can be determined. Based on the predicted fault type, the abnormal filtering layers are mapped and identified. Based on multiple identified abnormal filtering layers and their corresponding location coordinates, a fault early warning signal is generated so that the fault repair unit can quickly locate and handle the fault. The generated fault early warning signal is sent to the fault repair unit, which can respond quickly and arrange personnel for on-site repair based on the received signal.

[0040] In summary, the embodiments of this application have at least the following technical effects: First, the attribute information of the filter layers of the target filtration device is obtained. The target filtration device comprises N filter layers, which are connected in series in sequence. The attribute information of each filter layer includes material type, distribution structure, pore diameter, porosity, filtration method, and location coordinates. Next, based on digital twins, a simulation model of the target filtration device is generated according to the material type, distribution structure, pore diameter, porosity, filtration method, and location coordinates. Then, pretreated water quality information is collected, and the pretreated water quality information is simulated for filtration using the filtration twin model to determine a set of simulated monitoring indicators for multiple monitoring nodes. These monitoring indicators include water quality indicators, flow rate indicators, and pressure indicators. Next, based on the attribute information and pretreated water quality information, historical filtration records are used to predict the filtration of the pretreated water quality information, resulting in a set of predicted monitoring indicators for multiple monitoring nodes. Finally, the simulated monitoring indicator set and the predicted monitoring indicator set are fused according to a preset fusion strategy to determine the fused monitoring indicator set for the multiple monitoring nodes. At each monitoring node, data is collected from N filter layers via an IoT monitoring unit to obtain a real-time monitoring dataset. The IoT monitoring unit comprises N sensor groups, each corresponding to one of the N filter layers, deployed at the filter layer outlet. Each sensor group includes a water quality sensor, a flow sensor, and a pressure sensor. Finally, based on a fused monitoring index set, the real-time monitoring dataset at the same monitoring node is analyzed to identify filter layers that do not meet the fused monitoring indexes and provide fault warnings. This solves the technical problem in existing petroleum fiber filter technologies where the precision and accuracy of fault warning index settings for filter layers are insufficient, leading to inaccurate fault location and warning. It improves the accuracy and timeliness of fault detection, achieving precise early warning for filter layers.

[0041] Example 2 Based on the same inventive concept as the IoT-based oil filtration device fault detection method in the foregoing embodiments, such as Figure 2 As shown, this application provides an Internet of Things-based fault detection system for oil filtration devices. The system and method embodiments in this application are based on the same inventive concept. The system includes: The system comprises: an information acquisition module 11, used to acquire attribute information of the filter layers of a target filtration device, wherein the target filtration device includes N filter layers connected in series in sequence, and the attribute information of each filter layer includes material type, distribution structure, pore diameter, porosity, filtration method, and location coordinates; a modeling module 12, used to perform simulation modeling of the target filtration device based on digital twin, according to the material type, distribution structure, pore diameter, porosity, filtration method, and location coordinates, to generate a filter twin model; a simulation module 13, used to collect pretreated water quality information, and simulate filtration of the pretreated water quality information through the filter twin model to determine a set of simulated monitoring indicators under multiple monitoring nodes, wherein the monitoring indicators include water quality indicators, flow indicators, and pressure indicators; and a prediction module 14, used to predict based on the attribute information and the pretreated water quality information. The system calls historical filtration record data to perform filtration prediction on the pretreated water quality information, obtaining a predicted monitoring index set under the multiple monitoring nodes; a fusion module 15 is used to fuse the simulated monitoring index set and the predicted monitoring index set according to a preset fusion strategy to determine the fused monitoring index set under the multiple monitoring nodes; a data acquisition module 16 is used to collect data from the N filter layers through an IoT monitoring unit under the monitoring nodes to obtain a real-time monitoring dataset, wherein the IoT monitoring unit includes N sensor groups, each of which corresponds to one of the N filter layers and is deployed at the outlet of the filter layer, and each sensor group includes a water quality sensor, a flow sensor, and a pressure sensor; a judgment module 17 is used to judge the real-time monitoring dataset under the same monitoring node based on the fused monitoring index set, and locate filter layers that do not meet the fused monitoring indicators to provide fault warnings.

[0042] Furthermore, the simulation module 13 is used to perform the following methods: The water quality indicators include impurity types and corresponding concentrations for each impurity type. The impurity types include oil impurities, metal particles, inorganic suspended solids, and organic suspended solids.

[0043] Furthermore, the prediction module 14 is used to perform the following method: Select the first filter layer among the N filter layers, and obtain the first attribute information of the first filter layer, wherein the first filter layer is the filter layer ranked first among the N filter layers; using the first attribute information and the pretreated water quality information as constraints, retrieve multiple first sample filter record datasets through network retrieval; filter the multiple first sample filter record datasets based on the multiple monitoring nodes to obtain multiple first node filter record datasets; calculate the mean of the node filter record data under the same monitoring node based on the multiple first node filter record datasets, and use the mean calculation result as the predicted monitoring indicator for the corresponding monitoring node to obtain the first predicted monitoring indicator set under the multiple monitoring nodes; construct the predicted monitoring indicator set based on the first predicted monitoring indicator set.

[0044] Furthermore, the prediction module 14 is used to perform the following method: Select the second filter layer among the N filter layers, obtain the second attribute information of the second filter layer, wherein the second filter layer is an adjacent filter layer of the first filter layer; using the second attribute information and the first predictive monitoring indicator set as constraints, retrieve multiple second sample filter record datasets through network retrieval, and analyze to obtain the second predictive monitoring indicator set under the multiple monitoring nodes; perform iterative analysis on the remaining filter layers of the N filter layers until the Nth predictive monitoring indicator set is obtained; generate the predictive monitoring indicator set based on the first predictive monitoring indicator set, the second predictive monitoring indicator set, and up to the Nth predictive monitoring indicator set.

[0045] Furthermore, the fusion module 15 is used to perform the following methods: The simulation accuracy of the filtering twin model is obtained, and the deviation of the simulation accuracy is calculated according to the preset simulation accuracy to determine the simulation accuracy deviation. Based on the coefficient of variation method, the simulation confidence weight is set according to the simulation accuracy deviation, and the predicted confidence weight is calculated based on the simulation confidence weight, wherein the sum of the simulation confidence weight and the predicted confidence weight is 1. The preset fusion strategy is generated according to the simulation confidence weight and the predicted confidence weight.

[0046] Furthermore, the determination module 17 is used to perform the following method: Obtain the real-time monitoring nodes of the real-time monitoring dataset; based on the fusion monitoring indicator set, obtain the associated fusion monitoring indicators by matching the real-time monitoring nodes; perform filtering impact analysis on the N filtering layers respectively, and set N tolerance intervals based on the impact analysis results, wherein the tolerance intervals and impact analysis results are inversely proportional; update the associated fusion monitoring indicators according to the N tolerance intervals to obtain the associated fusion monitoring thresholds, and judge the real-time monitoring dataset based on the associated fusion monitoring thresholds.

[0047] Furthermore, the determination module 17 is used to perform the following method: The associated fusion monitoring threshold includes N associated fusion monitoring intervals, where each associated fusion monitoring interval includes a water quality monitoring interval, a flow monitoring interval, and a pressure monitoring interval. Based on the N associated fusion monitoring intervals, the real-time monitoring dataset is mapped and judged to determine multiple abnormal monitoring indicators, and the filter layers of the abnormal monitoring indicators are mapped and located to obtain multiple abnormal filter layers. The abnormal monitoring indicators are input into the fault analysis database for matching to determine multiple predicted fault types. The multiple abnormal filter layers are mapped and identified according to the multiple predicted fault types, and a fault early warning signal is generated based on the multiple identified abnormal filter layers and their corresponding location coordinates and sent to the fault repair unit.

[0048] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0049] 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.

[0050] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A fault detection method for oil filtration devices based on the Internet of Things, characterized in that, The method includes: Obtain the attribute information of the filter layer of the target filtration device, wherein the target filtration device includes N filter layers, which are distributed in series in sequence, and the attribute information of each filter layer includes material type, distribution structure, pore diameter, porosity, filtration method and location coordinates; Based on digital twins, the target filtration device is simulated and modeled according to the material type, distribution structure, pore diameter, porosity, filtration method and location coordinates to generate a filtration twin model. Collect pretreated water quality information, simulate filtering the pretreated water quality information through the filter twin model, and determine the set of simulated monitoring indicators under multiple monitoring nodes, wherein the monitoring indicators include water quality indicators, flow rate indicators and pressure indicators; Based on the attribute information and the pretreated water quality information, historical filtration record data is called to perform filtration prediction on the pretreated water quality information to obtain a set of predicted monitoring indicators under the multiple monitoring nodes. The simulated monitoring indicator set and the predicted monitoring indicator set are fused according to a preset fusion strategy to determine the fused monitoring indicator set under the multiple monitoring nodes; At the monitoring node, data is collected from the N filter layers through the Internet of Things (IoT) monitoring unit to obtain a real-time monitoring dataset. The IoT monitoring unit includes N sensor groups, which correspond one-to-one with the N filter layers and are deployed at the outlet of the filter layers. Each sensor group includes a water quality sensor, a flow sensor, and a pressure sensor. Based on the fusion monitoring index set, the real-time monitoring dataset under the same monitoring node is judged, and the filter layer that does not meet the fusion monitoring index is located for fault warning.

2. The method according to claim 1, characterized in that, The method further includes: The water quality indicators include impurity types and corresponding concentrations for each impurity type. The impurity types include oil impurities, metal particles, inorganic suspended solids, and organic suspended solids.

3. The method according to claim 1, characterized in that, Based on the attribute information and the pretreated water quality information, historical filtration record data is used to perform filtration prediction on the pretreated water quality information, including: Select the first filter layer among the N filter layers, and obtain the first attribute information of the first filter layer, wherein the first filter layer is the filter layer that is ranked first among the N filter layers. Using the first attribute information and the pretreated water quality information as constraints, multiple first sample filter record datasets are obtained through network retrieval; Based on the multiple monitoring nodes, the multiple first sample filtered record datasets are filtered to obtain multiple first node filtered record datasets; Based on the multiple first node filtered record datasets, the mean of the node filtered record data under the same monitoring node is calculated, and the mean calculation result is used as the predicted monitoring index of the corresponding monitoring node to obtain the first predicted monitoring index set under the multiple monitoring nodes. The predictive monitoring index set is constructed based on the first predictive monitoring index set.

4. The method according to claim 3, characterized in that, The predictive monitoring index set is constructed based on the first predictive monitoring index set, including: Select the second filter layer among the N filter layers, and obtain the second attribute information of the second filter layer, wherein the second filter layer is the adjacent filter layer of the first filter layer; Using the second attribute information and the first predictive monitoring indicator set as constraints, multiple second sample filtered record datasets are obtained through network retrieval, and the second predictive monitoring indicator set under the multiple monitoring nodes is obtained through analysis. The remaining filter layers of the N filter layers are iteratively analyzed until the Nth set of predictive monitoring indicators is obtained. The predictive monitoring indicator set is generated based on the first predictive monitoring indicator set, the second predictive monitoring indicator set, and up to the Nth predictive monitoring indicator set.

5. The method according to claim 1, characterized in that, The simulated monitoring indicator set and the predicted monitoring indicator set are fused according to a preset fusion strategy, including: Obtain the simulation accuracy of the filtered twin model, calculate the deviation of the simulation accuracy based on the preset simulation accuracy, and determine the simulation accuracy deviation; Based on the coefficient of variation method, a simulation confidence weight is set according to the simulation accuracy deviation, and a predicted confidence weight is calculated based on the simulation confidence weight, wherein the sum of the simulation confidence weight and the predicted confidence weight is 1. The preset fusion strategy is generated based on the simulated confidence weights and the predicted confidence weights.

6. The method according to claim 1, characterized in that, Based on the fused monitoring indicator set, the real-time monitoring dataset under the same monitoring node is judged, including: Obtain the real-time monitoring nodes of the real-time monitoring dataset; Based on the fusion monitoring indicator set, associated fusion monitoring indicators are obtained by matching the real-time monitoring nodes. The filtering influence of each of the N filter layers is analyzed, and N tolerance intervals are set based on the influence analysis results. The tolerance intervals are inversely proportional to the influence analysis results. The correlation fusion monitoring indicators are updated according to the N tolerance intervals to obtain the correlation fusion monitoring threshold, and the real-time monitoring dataset is judged based on the correlation fusion monitoring threshold.

7. The method according to claim 6, characterized in that, Based on the aforementioned correlation and fusion monitoring threshold, the real-time monitoring dataset is judged, and then the process further includes: The associated fusion monitoring threshold includes N associated fusion monitoring intervals, wherein each associated fusion monitoring interval includes a water quality monitoring interval, a flow monitoring interval, and a pressure monitoring interval; Based on the N associated fusion monitoring intervals, the real-time monitoring dataset is mapped and judged to determine multiple abnormal monitoring indicators, and the filtering layers of the abnormal monitoring indicators are mapped and located to obtain multiple abnormal filtering layers. The abnormal monitoring indicators are input into the fault analysis database for matching to determine multiple predicted fault types. The multiple abnormal filtering layers are mapped and identified according to the multiple predicted fault types, and a fault warning signal is generated based on the multiple identified abnormal filtering layers and their corresponding location coordinates and sent to the fault repair unit.

8. A fault detection system for oil filtration devices based on the Internet of Things, characterized in that, For implementing the Internet of Things-based oil filtration device fault detection method according to any one of claims 1-7, the system comprises: The information acquisition module is used to acquire the attribute information of the filter layer of the target filtration device. The target filtration device includes N filter layers, which are connected in series in sequence. The attribute information of each filter layer includes material type, distribution structure, pore diameter, porosity, filtration method and location coordinates. The modeling module is used to simulate and model the target filtration device based on digital twin, according to the material type, distribution structure, pore diameter, porosity, filtration method and location coordinates, to generate a filtration twin model. The simulation module is used to collect pre-treated water quality information, simulate filtering the pre-treated water quality information through the filter twin model, and determine a set of simulated monitoring indicators under multiple monitoring nodes, wherein the monitoring indicators include water quality indicators, flow rate indicators and pressure indicators. The prediction module is used to perform filtering prediction on the pretreated water quality information by calling historical filtration record data based on the attribute information and the pretreated water quality information, so as to obtain the predicted monitoring index set under the multiple monitoring nodes. A fusion module is used to fuse the simulated monitoring indicator set and the predicted monitoring indicator set according to a preset fusion strategy to determine the fused monitoring indicator set under the multiple monitoring nodes. The data acquisition module is used to acquire data from the N filter layers through the Internet of Things (IoT) monitoring unit at the monitoring node to obtain a real-time monitoring dataset. The IoT monitoring unit includes N sensor groups, which correspond one-to-one with the N filter layers and are deployed at the outlet of the filter layers. Each sensor group includes a water quality sensor, a flow sensor, and a pressure sensor. The judgment module is used to judge the real-time monitoring dataset under the same monitoring node based on the fusion monitoring index set, and to locate the filter layer that does not meet the fusion monitoring index for fault warning.

Images