An intelligent inspection and alarm method and system based on panoramic video fusion

Abstract

The application provides an intelligent inspection alarm method and system based on panoramic video fusion, relates to the technical field of intelligent inspection, and obtains device fault first alarm information through a panoramic video monitoring component; constructs a device fault relationship tree based on upstream and downstream topological information of the device according to an abnormal device position number list; extracts a first fault device and a parent node fault device set of the first fault device and a child node fault device set of the first fault device based on the device fault relationship tree; performs fault tracing analysis on the first fault device and the parent node fault device set based on an instrument monitoring state list, and generates a fault source device; adds the fault source device into a fault source device group; and generates an intelligent inspection alarm signal according to the fault source device group. The application solves the technical problem that the prior art cannot accurately trace the fault device, achieves the technical effect of accurately positioning the fault source device and improving the alarm reliability.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent inspection, in particular to an intelligent inspection alarm method and system based on panoramic video fusion. BACKGROUND

[0002] Intelligent inspection alarm technology is a technology that uses artificial intelligence, Internet of Things, big data analysis and other technical means to monitor and automatically inspect equipment or systems. It can automatically identify and analyze the running state of equipment, discover abnormal conditions in a timely manner, and notify relevant personnel for processing through an alarm mechanism, thereby improving inspection efficiency and accuracy and reducing labor costs. However, the existing intelligent inspection technology has insufficient analysis and learning capabilities, and in complex production devices, it cannot exclude the state influence between devices or trace back to the source device of the fault, thereby resulting in a high false alarm rate and affecting fault handling efficiency. SUMMARY

[0003] The present application provides an intelligent inspection alarm method and system based on panoramic video fusion, which solves the technical problem that the prior art cannot accurately trace back to the source device of the fault, realizes accurate positioning of the source device of the fault and alarm, and improves fault handling efficiency.

[0004] In view of the above problems, on the one hand, the present application provides an intelligent inspection alarm method based on panoramic video fusion, which obtains device fault first alarm information through a panoramic video monitoring component, wherein the device fault first alarm information includes an abnormal device position number list and an instrument monitoring state list; a device fault relationship tree is constructed based on upstream and downstream topology information of the device according to the abnormal device position number list; a first fault device and a parent node fault device set of the first fault device and a child node fault device set of the first fault device are extracted based on the device fault relationship tree; fault tracing analysis is performed on the first fault device and the parent node fault device set based on the instrument monitoring state list to generate a fault source device; the fault source device is added to a fault source device group; and an intelligent inspection alarm signal is generated according to the fault source device group.

[0005] On the other hand, the present application also provides an intelligent patrol alarm system based on panoramic video fusion, including: an information acquisition module: the information acquisition module is used to obtain the first-level alarm information of equipment failure through a panoramic video monitoring component, wherein the first-level alarm information of equipment failure includes an abnormal equipment position number list and an instrument monitoring status list; a relationship tree construction module: the relationship tree construction module is used to construct an equipment fault relationship tree based on the abnormal equipment position number list and the upstream and downstream topology information of the equipment; an extraction module: the extraction module is used to extract the first fault device and the parent node fault device set of the first fault device, and the child node fault device set of the first fault device based on the equipment fault relationship tree; an analysis module: the analysis module is used to perform fault tracing analysis on the first fault device and the parent node fault device set based on the instrument monitoring status list to generate a fault source device; a collection module: the collection module is used to add the fault source device to the fault source device group; an alarm module: the alarm module is used to generate an intelligent patrol alarm signal based on the fault source device group.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] Through the panoramic video monitoring component, a list of abnormal device bit numbers and an instrument monitoring status list are obtained; according to the abnormal device bit number list, based on the upstream and downstream topology information of the equipment, a device fault relationship tree is constructed to grasp the connection between upstream and downstream equipment, and based on the device fault relationship tree, the first fault device and the parent node fault device set of the first fault device, and the child node fault device set of the first fault device are extracted; based on the instrument monitoring status list, a fault tracing analysis is performed on the first fault device and the parent node fault device set to generate the fault source device; by analyzing the causal relationship of the equipment status, the fault source device is accurately located, and the fault source device is added to the fault source device group; intelligent patrol alarms are performed based on the fault source device group to improve the accuracy and reliability of intelligent patrol alarms, accurately locate the fault location, facilitate maintenance personnel to deal with the fault in a timely manner, save fault handling time, improve fault handling efficiency, and ensure the safe and stable operation of the entire device.

[0008] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A flowchart of an intelligent inspection and alarm method based on panoramic video fusion provided in an embodiment of the present application;

[0010] Figure 2 A schematic diagram of a process for generating a fault source device in an intelligent inspection and alarm method based on panoramic video fusion provided in an embodiment of the present application;

[0011] Figure 3 A schematic diagram of a process for obtaining a predicted instrument status of a first fault device in an intelligent inspection and alarm method based on panoramic video fusion provided in an embodiment of the present application;

[0012] Figure 4 A structural diagram of an intelligent patrol and alarm system based on panoramic video fusion provided in an embodiment of the present application.

[0013] Description of reference numerals: information acquisition module 10 , relationship tree construction module 20 , extraction module 30 , analysis module 40 , collection module 50 , alarm module 60 . DETAILED DESCRIPTION

[0014] The embodiment of the present application provides an intelligent inspection and alarm method and system based on panoramic video fusion. Through the panoramic video monitoring component, it obtains the first-level alarm information of equipment failure, constructs an equipment fault tree, performs fault tracing analysis based on the number of equipment failures and the instrument monitoring status list, and generates the fault source device. This solves the technical problem that the existing technology cannot accurately trace the fault source device, realizes the precise positioning of the fault source device and issues an alarm, and improves the efficiency of fault handling.

[0015] Example 1, as Figure 1 As shown, the embodiment of the present application provides an intelligent inspection and alarm method based on panoramic video fusion, including:

[0016] Step S1: Obtaining level one alarm information of equipment failure through a panoramic video monitoring component, wherein the level one alarm information of equipment failure includes a list of abnormal equipment position numbers and a list of instrument monitoring status.

[0017] Specifically, the panoramic video monitoring component refers to the monitoring equipment used to monitor the entire device or production line. It is an integral part of the monitoring system and contains multiple cameras. It can capture real-time video of the entire device area or the entire production area, monitor all equipment in the entire device or production line in real time, and obtain the instrument status information of the equipment. The first-level alarm information for equipment failure refers to the highest level of alarm notification issued in the monitoring system when a serious equipment failure is detected. A level one alarm usually means that the equipment failure needs to be paid attention to and handled immediately because it may pose a direct threat to production safety or equipment integrity. The first-level alarm information for equipment failure includes a list of abnormal equipment bit numbers and a list of instrument monitoring status, where the abnormal equipment bit number list refers to an information list composed of the device bit numbers corresponding to all devices that have detected abnormalities, and the instrument monitoring status list refers to an information list composed of the real-time monitoring status of all equipment monitoring instruments, including but not limited to real-time readings of flow meters, pressure gauges, thermometers, etc.

[0018] Through the panoramic video detection component, we can obtain the first-level alarm information of equipment failure, respond to the failure quickly, locate the specific faulty equipment, and facilitate subsequent fault tracing analysis.

[0019] Step S2: constructing a device fault relationship tree based on the abnormal device bit number list and the upstream and downstream topology information of the device.

[0020] Specifically, equipment upstream and downstream topology information refers to information that describes the logical connection and physical layout between upstream and downstream equipment in a production line or process flow, which can help understand and analyze how the entire production line or process flow operates. Upstream equipment refers to equipment that performs tasks before the current equipment in the production line or process flow, and can be equipment that provides raw materials or services to the current equipment, while downstream equipment refers to equipment that performs tasks subsequent to the current equipment in the production line or process flow, and can be equipment that receives the output of the current equipment. The equipment fault relationship tree is a tree-structured tool that describes the cause-and-effect relationship of equipment failures in a production line or process flow, and is used to demonstrate and analyze how equipment failures affect the operation of the entire device or production line. In the equipment fault relationship tree, the root node is usually the equipment where the failure initially occurred, and the branches represent how the failure propagates to other equipment through upstream and downstream relationships.

[0021] Based on the abnormal device number list obtained in step S1 and combined with the upstream and downstream topology information of the devices, the upstream devices are placed at the root node and the downstream devices are used as child nodes to construct a device fault relationship tree. This helps to understand and predict the chain reaction of faults and locate the device where the fault initially occurred.

[0022] Step S3: Based on the device fault relationship tree, extract the first faulty device, the parent node faulty device set of the first faulty device, and the child node faulty device set of the first faulty device.

[0023] Specifically, the first faulty device refers to the device initially detected as faulty, i.e., the device directly identified in the abnormal device number list in the alarm information. The parent node faulty device set refers to the set of faulty devices corresponding to all parent nodes of the first faulty device in the device fault relationship tree. This includes all directly upstream faulty devices that could affect the operating status of the first faulty device. The child node faulty device set refers to the set of faulty devices corresponding to all child nodes of the first faulty device in the device fault relationship tree. This includes all directly downstream faulty devices that could be affected by the operating status of the first faulty device.

[0024] Extracting the first faulty device and the parent node faulty device set of the first faulty device, as well as the child node faulty device set of the first faulty device can more quickly understand the potential propagation path of the fault between all devices, facilitate subsequent rapid positioning of the fault source device through panoramic video monitoring and intelligent analysis, and make accurate intelligent patrol alarms.

[0025] Step S4: Based on the instrument monitoring status list, perform fault tracing analysis on the first faulty device and the parent node faulty device set to generate a fault source device.

[0026] Specifically, the first fault device is first judged. When the number of devices in the parent node fault device set of the first fault source device is equal to 0, the first fault device is the fault source device. When the number of devices in the parent node fault device set is not equal to 0, the instrument monitoring status of the first fault device is predicted according to the instrument monitoring status list and the instrument monitoring status of the devices in the parent node fault device set. The predicted instrument monitoring status of the first fault device is compared with its actual instrument monitoring status. If the difference is too large, the first fault device is considered to be the fault source device. Otherwise, each device in the parent node fault device set is traversed and used as the first fault source device for fault tracing analysis to determine the true fault source device.

[0027] Through the above steps, the fault source device can be accurately located and the device that displays an abnormal status but is actually in a normal state can be eliminated, thereby obtaining accurate equipment fault analysis results, reducing the false alarm rate, and improving the accuracy and reliability of intelligent inspection alarms.

[0028] Step S5: Add the fault source device to the fault source device group.

[0029] Specifically, all fault source devices identified in step S4 are added to a fault source device group. This group is the collection of all fault source devices identified through fault tracing analysis. By adding these devices to the group, all truly faulty devices in the production line or process flow can be located, facilitating the subsequent generation of accurate intelligent inspection alarm signals.

[0030] Step S6: Generate an intelligent inspection alarm signal based on the fault source equipment group.

[0031] Specifically, based on the equipment location characteristics, the fault source equipment group obtained in step S5 is clustered and analyzed to obtain multiple cluster analysis results. Maintenance personnel and routes are matched for each cluster analysis result, and corresponding intelligent patrol alarm signals are generated to remind maintenance personnel, which helps to quickly respond to and handle equipment failures and improve the efficiency of handling equipment failures.

[0032] Furthermore, step S6 of the embodiment of the present application further includes:

[0033] Step S61: performing cluster analysis based on the fault source device group and a distance threshold to generate multiple alarm identification areas.

[0034] Step S62: Obtain the real-time coordinate set of the target group through the maintenance personnel terminal.

[0035] Step S63: traverse the multiple warning identification areas, sort the target population based on the real-time coordinate set of the target population, and obtain a maintenance recommended personnel list and a maintenance recommended route list.

[0036] Step S64: generating the intelligent inspection alarm signal according to the maintenance recommended personnel list, the maintenance recommended route list, and the fault source equipment group and sending the signal to the corresponding maintenance personnel terminal.

[0037] Preferably, step S63 in the embodiment of the present application further includes:

[0038] Step S631: Obtain the center coordinates of the first warning mark area of ​​the multiple warning mark areas.

[0039] Step S632: obtaining the real-time coordinates of the first target person in the target population real-time coordinate set up to the real-time coordinates of the Nth target person.

[0040] Step S633: Path optimization is performed based on the real-time coordinates of the first target person and the center coordinates of the first warning identification area to generate a first path optimization result.

[0041] Step S634: Path optimization is performed based on the real-time coordinates of the Nth target person and the center coordinates of the first alarm identification area to generate an Nth path optimization result.

[0042] Step S635: selecting the shortest path personnel from the first path optimization result and the Nth path optimization result, setting them as the maintenance recommended personnel for the first alarm identification area, and adding them to the maintenance recommended personnel list.

[0043] Step S636: The path optimization result of the shortest path person is set as the first alarm identification area maintenance recommended route, added to the maintenance recommended route list, and the shortest path person is deleted from the real-time coordinate set of the target population.

[0044] Specifically, a distance threshold is preset, which is used to perform cluster analysis on the fault source device group, and the fault source devices with a distance between devices less than the distance threshold are divided into the same alarm identification area. The distance threshold is set by the technician based on the actual distribution of the equipment and the monitoring and maintenance requirements, and can be 50 meters, 100 meters, etc. The location data of each device in the fault source device group is collected. This data is usually in the form of coordinates, such as longitude and latitude. A clustering algorithm is used to process the location data of the fault source device. According to the distance between the devices, the fault source devices in the fault source device group are clustered based on the set distance threshold to obtain multiple cluster analysis results. The clustering algorithm can use K-means, DBSCAN, etc. Each cluster analysis result contains several fault source devices that are close in physical location. The locations of all fault source devices in the same clustering result are divided into an alarm identification area, and multiple alarm identification areas are obtained.

[0045] The real-time location information of all maintenance personnel is obtained through the terminal carried by the maintenance personnel, and a real-time coordinate set of the target population is generated. The real-time coordinate set of the target population includes the real-time coordinate data of multiple maintenance personnel, and the real-time coordinate data of each maintenance personnel includes the maintenance personnel number and the corresponding real-time location coordinates.

[0046] Determine the center coordinates of each alarm identification area by calculating the average coordinates of all fault source devices within each alarm identification area, obtaining multiple area center coordinates. Select any area center coordinate as the center coordinate of the first alarm identification area. Obtain the real-time coordinates of each target person in the target population real-time coordinate set and randomly name them from the first target person real-time coordinates to the Nth target person real-time coordinates, where N is the total number of target people and is an integer greater than or equal to 1.

[0047] Use the path optimization algorithm to optimize the real-time coordinates of each maintenance personnel and the center coordinates of the first alarm identification area, that is, based on the real-time coordinates of the first target personnel and the center coordinates of the first alarm identification area, the path optimization is performed to generate the first path optimization result, and based on the real-time coordinates of the Nth target personnel and the center coordinates of the first alarm identification area, the path optimization is performed to generate the Nth path optimization result, so as to find the optimal path from the current position of the maintenance personnel to the first alarm identification area, and a total of N path optimization results are obtained. When performing path optimization, factors such as avoiding obstacles and selecting the shortest path are considered. The path optimization algorithm can adopt breadth-first search (BFS), depth-first search (DFS), Dijkstra algorithm, Bellman-Ford algorithm, A * (A-Star) algorithm, etc.

[0048] The N obtained path optimization results are sorted by path length, and the shortest path optimization result is selected. The maintenance personnel corresponding to the shortest path optimization result is designated as the shortest path personnel. This shortest path personnel is set as the recommended maintenance personnel for the first warning identification area and added to the recommended maintenance personnel list. The path optimization result for this shortest path personnel is set as the recommended maintenance route for the first warning identification area and added to the recommended maintenance route list. Simultaneously, the shortest path personnel is deleted from the target population's real-time coordinate set to avoid duplicate assignment. The recommended maintenance personnel list is a list of information consisting of the recommended maintenance personnel corresponding to each warning identification area, and the recommended maintenance route list is a list of information consisting of the optimal travel route for each maintenance personnel in all recommended maintenance personnel lists from their current location to the corresponding warning identification area.

[0049] After determining the recommended maintenance personnel for the first alarm identification area and the recommended maintenance route for the first alarm identification area, randomly select an alarm identification area from the remaining alarm identification areas as the second alarm identification area, repeat steps S631 to S636, determine the recommended maintenance personnel for the second alarm identification area and the recommended maintenance route for the second alarm identification area, and repeat the above process until the corresponding recommended maintenance personnel and recommended maintenance route are matched for each alarm identification area.

[0050] Based on the list of recommended maintenance personnel, the list of recommended maintenance routes, and the group of fault source equipment, an intelligent patrol alarm signal is generated and sent to the corresponding maintenance personnel terminal. Among them, the intelligent patrol alarm signal is in the form of a message push within the system to send the maintenance personnel a fault location reminder and the best travel route information.

[0051] Through the above steps, a corresponding intelligent inspection alarm signal can be generated for each alarm identification area, which makes it easier for the nearest maintenance personnel to take the best route to deal with the fault source equipment, eliminate the fault, save fault handling time, and improve fault handling efficiency.

[0052] Further, the step S2 of the embodiment of the present application further comprises:

[0053] Step S21: obtaining device job flow information, wherein the device job flow information represents a device response sequence of a real-time job task.

[0054] The device upstream and downstream topology information represents a distribution topology of upstream and downstream devices that will interfere with the device state.

[0055] Step S22: rendering the device upstream and downstream topology information according to the device job flow information to obtain job device position number topology information.

[0056] Step S23: constructing the device fault relationship tree according to the job device position number topology information and the abnormal device position number list.

[0057] Specifically, the device job flow information and the device upstream and downstream topology information are obtained through information interaction, wherein the device job flow information refers to a device response sequence of all devices in a production line or device when a job task is completed, and the device upstream and downstream topology information describes a distribution of upstream and downstream devices that will interfere with the device state.

[0058] According to the device job flow information, the device upstream and downstream topology information is rendered, the device job flow information and the device upstream and downstream topology information are converted into a graphical representation that is intuitive and easy to understand, and a device position number corresponding to each job device is added to generate job device position number topology information, which is helpful to understand the interaction between devices and quickly locate the fault source device.

[0059] According to the job device position number topology information and the abnormal device position number list, a relationship tree is constructed, the fault device is placed in the middle position, for each fault device, the direct upstream device is added to the relationship tree as a parent node, these upstream devices may be the cause or influencing factor of the fault. The direct downstream device of each fault device is added to the relationship tree as a child node, these downstream devices may be affected by the state of the fault device. The upstream and downstream relationship between devices is represented by an arrow or a connection line, the arrow points from the upstream device to the downstream device. If the upstream device also has a fault, these devices can also be added to the relationship tree as fault devices, and their upstream devices are further added. Similarly, if the downstream device has a fault, they can also be added to the relationship tree as child nodes, and their downstream devices are considered, and finally the device fault relationship tree is obtained. For example, assume that device C has a fault, and through the upstream and downstream topology information, it is known that device B is the upstream device of device C, and device D is the downstream device of device C. In the device fault relationship tree, device C is located in the center as a fault device. Device B is located above device C as a parent node, because its state may have affected device C. Device D is located below device C as a child node, because it may be affected by the fault of device C. If device B also has a fault, device B will become a new fault device, and its upstream devices will be added to the relationship tree, and so on.

[0060] In this way, the device fault relationship tree can be constructed, the propagation path of the fault and the mutual dependence relationship between devices are mastered, and the fault tracing analysis is more effectively completed.

[0061] Further, as shown in Figure 2 the step S4 further includes:

[0062] Step S41: When the number of devices in the parent node fault device set is equal to 0, set the first fault device as the fault source device.

[0063] Step S42: When the number of devices in the parent node fault device set is not equal to 0, obtain a first instrument monitoring state set of a parent node device set from the instrument monitoring state list, and obtain a second instrument monitoring state of the first fault device from the instrument monitoring state list, wherein the parent node device set is all devices of the next level that will have a state impact on the first fault device, and the parent node fault device set belongs to the parent node device set.

[0064] Step S43: According to the first instrument monitoring state set, execute fuzzy data mining to generate a first fault device predicted instrument state.

[0065] Step S44: When the first deviation coefficient of the first fault device predicted meter state and the second meter monitoring state is greater than or equal to a first deviation coefficient threshold, the first fault device is set as the fault source device.

[0066] Step S45: When the first deviation coefficient of the first fault device predicted meter state and the second meter monitoring state is less than the first deviation coefficient threshold, the parent node fault device set is traversed for fault tracing analysis, and the fault source device is generated.

[0067] Specifically, the number of devices in the parent node fault device set is checked. If this set is empty, i.e., no upstream fault device that can affect the state of the first fault device is found, it can be inferred that the first fault device is the source of the fault and no other device has an impact on it. In this case, the first fault device is set as the fault source device.

[0068] If the parent node fault device set is not empty, i.e., there are upstream fault devices that can affect the state of the first fault device, further analysis of the states of these devices is needed. From the list of meter monitoring states, the current meter monitoring states of all parent node devices are selected to form a first meter monitoring state set, and the current meter monitoring state of the first fault device is selected from the list of meter monitoring states to obtain a second meter monitoring state, which is used for comparative analysis with the states of the parent node devices.

[0069] According to the first meter monitoring state set, fuzzy data mining is performed to generate a first fault device predicted meter state, which refers to the meter state of the first fault device under normal circumstances predicted by fuzzy data mining technology through analysis of historical data.

[0070] The first deviation coefficient between the first fault device predicted meter state and the second meter monitoring state is calculated. When the first deviation coefficient between the first fault device predicted meter state and the second meter monitoring state is greater than or equal to a preset first deviation coefficient threshold, it means that the difference between the predicted state and the actual state exceeds the allowed range, indicating that the first fault device itself is in an abnormal state, and in this case, the first fault device is determined as the fault source device.

[0071] If the first deviation coefficient between the first fault device predicted meter state and the second meter monitoring state is less than the first deviation coefficient threshold, it means that the difference between the predicted state and the actual state is within the allowed range, indicating that the first fault device itself is in a normal state, i.e., the first fault device is not the fault source device.

[0072] At this time, the parent node fault device set is traversed, and each fault device in the parent node fault device set is treated as the first fault device and fault source analysis is performed. The above analysis steps are repeated until all fault source devices have been found.

[0073] By analyzing each faulty device individually and comparing its actual status with the expected impact of its parent node device through the above steps, you can more accurately determine the true source of the fault and eliminate devices that display abnormal status but are actually in normal status.

[0074] Preferably, step S43 in the embodiment of the present application further includes:

[0075] Step S431: performing normalization processing and weighting according to the first instrument monitoring state set to construct a first spatial coordinate.

[0076] Step S432: Based on the first spatial coordinates and a preset fault tolerance distance, a state mining fuzzy space is constructed.

[0077] Step S433: extracting a first fault equipment recording instrument state set belonging to the state mining fuzzy space from the historical operation records.

[0078] Step S434: performing frequency analysis on the first faulty device recorded instrument state set to obtain the first faulty device predicted instrument state.

[0079] Specifically, the data from the first instrument monitoring status set is normalized and converted to the same standard. For example, all pressure data is converted to megapascals, and all temperature data is converted to kelvins. By analyzing historical data, different data types are assigned different weights based on their importance. For example, if temperature fluctuations are consistently the main cause of equipment failures, the temperature data might be given a higher weight. A spatial coordinate is constructed to establish a location for each device status in multidimensional space.

[0080] Based on the first spatial coordinate constructed in step S431, a fault tolerance distance is preset, and a state mining fuzzy space is constructed according to the preset fault tolerance distance. The fault tolerance distance refers to the state difference range of the instrument monitoring state of the first fault device compared to the first spatial coordinate. The instrument monitoring state within this range is considered normal. The above-mentioned state mining fuzzy space refers to the possible instrument monitoring state range of the first fault device under normal conditions determined based on the first instrument monitoring state set. For example, the ideal operating temperature of a certain device is 1300 Kelvin, and the fault tolerance is set to ±50 Kelvin. Then, the temperature of the device between 750 Kelvin and 850 Kelvin is considered to be a normal temperature.

[0081] All recorded instrument states of the first faulty device belonging to the state mining fuzzy space are extracted from historical operation records to form the first faulty device recorded instrument state set. The first faulty device recorded instrument state set is the collection of all normal instrument monitoring states that occurred during the first faulty device's historical operation, as determined by the state mining fuzzy space. This set is used to predict the most likely instrument monitoring state of the first faulty device under the influence of all devices in the parent node device set, i.e., the predicted instrument state of the first faulty device.

[0082] Perform frequency analysis on the extracted instrument status set recorded for the first faulty device, count the most frequently occurring instrument monitoring status, and use it as the predicted instrument status of the first faulty device for fault tracing analysis and determining the fault source device.

[0083] Through the above steps, the impact of upstream equipment on the first faulty equipment can be better understood, and the possible instrument monitoring status range of the first faulty equipment can be predicted, thereby accurately determining the fault source equipment and avoiding unnecessary maintenance activities.

[0084] Preferably, Figure 3 As shown, step S434 of the embodiment of the present application also includes:

[0085] Step S434-1: traverse the state attributes of the first faulty device and configure the state attribute deviation threshold.

[0086] Step S434-2: comparing the proportion of the number of attributes of any two recording instruments in the first faulty device recording instrument state set whose state deviation is greater than or equal to the state attribute deviation threshold, and setting the proportion as a second deviation coefficient.

[0087] Step S434 - 3 : When the second deviation coefficient is less than or equal to the second deviation coefficient threshold, the corresponding two recording instrument states are added into the same cluster; otherwise, they are added into different clusters.

[0088] Step S434-4: cyclically analyze to obtain multiple clusters of first faulty device recording instrument states, extract a random recording instrument state of the largest cluster among the multiple clusters of first faulty device recording instrument states, and set it as the predicted instrument state of the first faulty device.

[0089] Specifically, each state attribute of the first faulty device is traversed and a deviation threshold is configured for each state attribute. The deviation threshold is the allowable range of difference between the actual state attribute value and the expected state attribute value, typically preset by engineers and system designers based on historical data and device performance. If the difference between the actual state attribute value and the expected state attribute value exceeds the deviation threshold, the attribute is considered to have deviated. Setting the deviation threshold ensures that the system can distinguish between normal fluctuations and potential faults, thereby improving the accuracy and reliability of fault diagnosis.

[0090] Compare the deviations of any two recording instrument states in the first faulty device's recording instrument state set for each state attribute. If the deviation between the two states on a particular attribute is greater than or equal to the state attribute deviation threshold, the two recording instrument states are considered to be deviated on that attribute. By calculating the ratio of the number of deviating attributes to the total number of attributes, we can obtain a second deviation coefficient, which reflects the degree of deviation between the two recording instrument states.

[0091] The first faulty device recording instrument state set is classified according to the second deviation coefficient and the second deviation coefficient threshold. If the second deviation coefficient is less than or equal to the second deviation coefficient threshold, then the two recording instrument states are similar in most attributes and should be classified into the same cluster. Otherwise, they should be classified into different clusters.

[0092] Perform a cyclic analysis, selecting any two recording instrument states from the remaining recording instrument states in the first faulty device's recording instrument state set, calculating their second deviation coefficients, and classifying them to obtain multiple clusters of recording instrument states for the first faulty device. Then, extract a random recording instrument state from the largest of these clusters and set it as the predicted instrument monitoring state for the first faulty device.

[0093] For example, consider a faulty device A with three key state attributes: temperature (T), pressure (P), and speed (V). For each attribute, set a deviation threshold: temperature (T): ±5 degrees Celsius, pressure (P): ±10 Pascals, and speed (V): ±10%. Set the second deviation coefficient threshold to 0.67.

[0094] Get the recording instrument status set of device A and select any two recording instrument statuses from it:

[0095] State 1: Temperature 500 degrees Celsius, Pressure 100 Pascals, Speed ​​1000 rpm

[0096] State 2: Temperature 505 degrees Celsius, Pressure 105 Pascals, Speed ​​1010 rpm

[0097] In terms of temperature attributes, the deviation between state 1 and state 2 is 5 degrees Celsius, which is greater than the temperature deviation threshold. In terms of pressure attributes, the deviation between state 1 and state 2 is 5 Pascals, which is less than the pressure deviation threshold. In terms of speed attributes, the deviation between state 1 and state 2 is 1%, which is less than the speed deviation threshold.

[0098] Calculate the ratio of all deviating attributes to the total number of attributes. The second deviation coefficient for states 1 and 2 is 0.33, which is less than the second deviation coefficient threshold. Therefore, states 1 and 2 are grouped together. Randomly select two more recorder states from the recorder state set for faulty device A that are different from states 1 and 2, and repeat the above steps.

[0099] By calculating the second deviation coefficient, the recording instrument states in the recording instrument state set of the first faulty device are divided into multiple clusters, thereby obtaining the most likely instrument monitoring state of the first faulty device, helping to determine whether the first faulty device is actually faulty.

[0100] Optionally, step S434 in this embodiment of the application further includes:

[0101] Step S434-5: comparing the first fault device predicted instrument state and the second instrument monitored state, and determining the proportion of attributes whose state deviation is greater than or equal to the state attribute deviation threshold, and setting the proportion as the first deviation coefficient.

[0102] Specifically, based on the predicted instrument state and the second instrument monitoring state obtained in step S42 for the first faulty device, the differences between the two in various state attributes are compared. For each state attribute, if the difference between the predicted instrument state and the second instrument monitoring state is greater than or equal to a state attribute deviation threshold, the attribute is considered to have deviated. The proportion of all deviating attributes to the total number of attributes is calculated to obtain a first deviation coefficient, which reflects the degree of deviation between the actual instrument monitoring state and the predicted instrument monitoring state of the first faulty device. For example, assume that the predicted instrument state of a device includes four state attributes: temperature, pressure, flow rate, and voltage. The predicted state is 400 degrees Celsius, pressure 500 Pascals, flow rate 600 cubic meters / hour, and voltage 100 volts. The actual state is 405 degrees Celsius, pressure 505 Pascals, flow rate 605 cubic meters / hour, and voltage 101 volts. The state attribute deviation thresholds are set to ±5 degrees Celsius, ±5 Pascals, ±5 cubic meters / hour, and ±1 volt.

[0103] Comparing the predicted and actual states, we found that the deviations for temperature and voltage exceeded the state attribute deviation threshold, while the deviations for pressure and flow did not exceed the threshold. Therefore, two attributes exhibited deviations, accounting for half of the total number of attributes. The first deviation coefficient was 0.5.

[0104] By calculating the first deviation coefficient, the degree of deviation between the predicted state and the actual state of the first faulty device can be determined, thereby determining whether the first faulty device has failed.

[0105] In summary, the intelligent inspection and alarm method based on panoramic video fusion provided by the embodiment of the present application has the following technical effects:

[0106] Through the panoramic video monitoring component, a list of abnormal equipment numbers and a list of instrument monitoring status are obtained; based on the abnormal equipment number list and equipment operation process information, the upstream and downstream topology information of the equipment is rendered, the operating equipment number topology information is obtained, and an equipment fault relationship tree is constructed to understand the fault propagation path and the interdependence between equipment, thereby completing fault tracing analysis more effectively.

[0107] Based on the device fault relationship tree, the first fault device and its parent node fault device set, as well as the child node fault device set of the first fault device, are extracted. Based on the instrument monitoring status list, a state mining fuzzy space is constructed to obtain the predicted instrument status of the first fault device and the second instrument monitoring status. Fault tracing analysis is performed on the first fault device and its parent node fault device set to generate the fault source device. Each fault device is analyzed individually, its actual status compared with the expected impact of its parent node device, and the causal relationship between device status is analyzed to more accurately determine the true source of the fault and eliminate devices that display abnormal status but are actually in normal status. The identified fault source devices are added to the fault source device group. Alarm identification areas are divided according to the fault source device group, and a list of recommended maintenance personnel and recommended maintenance routes is generated to perform intelligent patrol alarms, allowing the nearest maintenance personnel to use the optimal travel route to handle the fault source device, eliminate the fault, save fault handling time, and improve fault handling efficiency.

[0108] As described above, the embodiments of the present application monitor equipment based on panoramic video fusion technology, deeply mine and analyze historical data, accurately judge faulty equipment, identify the real fault source equipment, and generate accurate intelligent patrol alarm signals, thereby improving the accuracy and reliability of intelligent patrol alarms, saving fault handling time, avoiding unnecessary maintenance activities, and improving equipment fault handling efficiency.

[0109] Example 2, as Figure 4 As shown, the embodiment of the present application provides an intelligent inspection and alarm system based on panoramic video fusion, including:

[0110] Information acquisition module 10: The information acquisition module 10 is used to obtain the first-level alarm information of equipment failure through the panoramic video monitoring component, wherein the first-level alarm information of equipment failure includes a list of abnormal equipment bit numbers and a list of instrument monitoring status.

[0111] Relationship tree construction module 20: The relationship tree construction module 20 is used to construct a device fault relationship tree according to the abnormal device bit number list and based on the upstream and downstream topology information of the device.

[0112] Extraction module 30: The extraction module 30 is used to extract the first faulty device, the parent node faulty device set of the first faulty device, and the child node faulty device set of the first faulty device based on the device fault relationship tree.

[0113] Analysis module 40: The analysis module 40 is used to perform fault tracing analysis on the first faulty device and the parent node faulty device set based on the instrument monitoring status list to generate a fault source device.

[0114] Aggregation module 50: The aggregation module 50 is used to add the fault source device into the fault source device group.

[0115] Alarm module 60: The alarm module 60 is used to generate an intelligent inspection alarm signal according to the fault source equipment group.

[0116] Furthermore, the step alarm module 60 of the embodiment of the present application further includes:

[0117] Alarm identification area generation unit: The alarm identification area generation unit is used to perform cluster analysis based on the fault source device group and a distance threshold to generate multiple alarm identification areas.

[0118] Real-time coordinate set acquisition unit: The real-time coordinate set acquisition unit is used to obtain the real-time coordinate set of the target group through the maintenance personnel terminal.

[0119] Recommendation list acquisition unit: The recommendation list acquisition unit is used to traverse the multiple warning identification areas, sort the target population based on the real-time coordinate set of the target population, and obtain a maintenance recommended personnel list and a maintenance recommended route list.

[0120] Alarm signal generating unit: The alarm signal generating unit is used to generate the intelligent inspection alarm signal according to the maintenance recommended personnel list, the maintenance recommended route list, and the fault source equipment group, and send it to the corresponding maintenance personnel terminal.

[0121] Preferably, the recommendation list obtaining unit is further configured to perform the following steps:

[0122] Obtain the center coordinates of a first warning identification area of ​​the multiple warning identification areas.

[0123] The real-time coordinates of the first target person in the target population real-time coordinate set are obtained, up to the real-time coordinates of the Nth target person.

[0124] Path optimization is performed based on the real-time coordinates of the first target person and the center coordinates of the first warning identification area to generate a first path optimization result.

[0125] Path optimization is performed based on the real-time coordinates of the Nth target person and the center coordinates of the first warning identification area to generate an Nth path optimization result.

[0126] The shortest path personnel of the first path optimization result and the Nth path optimization result are selected, set as the maintenance recommended personnel of the first alarm identification area, and added into the maintenance recommended personnel list.

[0127] The path optimization result of the shortest path personnel is set as the first alarm identification area maintenance recommended route, added to the maintenance recommended route list, and the shortest path personnel is deleted from the real-time coordinate set of the target population.

[0128] Furthermore, the relationship tree construction module 20 in the embodiment of the present application further includes:

[0129] Process information acquisition unit: The process information acquisition unit is used to obtain device operation process information, wherein the device operation process information represents the device response sequence of the real-time operation task.

[0130] Rendering unit: The rendering unit is used to render the upstream and downstream topology information of the equipment according to the equipment operation process information, and obtain the operation equipment position number topology information.

[0131] Construction unit: The construction unit is used to construct the equipment fault relationship tree according to the operating equipment bit number topology information and the abnormal equipment bit number list.

[0132] Furthermore, the analysis module 40 in the embodiment of the present application further includes:

[0133] First judgment unit: The first judgment unit is configured to set the first faulty device as the fault source device when the number of devices in the parent node faulty device set is equal to 0.

[0134] An instrument monitoring status acquisition unit: The instrument monitoring status acquisition unit is used to obtain a first instrument monitoring status set of a parent node device set from the instrument monitoring status list when the number of devices in the parent node fault device set is not equal to 0, and to obtain a second instrument monitoring status of the first fault device from the instrument monitoring status list, wherein the parent node device set is all upper-level devices that may affect the status of the first fault device, and the parent node fault device set belongs to the parent node device set.

[0135] Fuzzy data mining unit: The fuzzy data mining unit is used to perform fuzzy data mining according to the first instrument monitoring state set to generate a first fault device predicted instrument state.

[0136] The second determining unit is configured to set the first fault device as the fault source device when a first deviation coefficient of the first fault device predicted instrument state and the second instrument monitored state is greater than or equal to a first deviation coefficient threshold.

[0137] The third determining unit is configured to perform fault tracing analysis on the parent node fault device set to generate the fault source device when the first deviation coefficient of the first fault device predicted instrument state and the second instrument monitored state is less than the first deviation coefficient threshold.

[0138] Preferably, the fuzzy data mining unit is further configured to perform the following steps:

[0139] The first instrument monitored state set is normalized and weighted to construct a first space coordinate.

[0140] Based on the first space coordinate, a state mining fuzzy space is constructed according to a preset fault tolerance distance.

[0141] First fault device record instrument state sets belonging to the state mining fuzzy space are extracted from historical operation records.

[0142] Frequency analysis is performed on the first fault device record instrument state set to obtain the first fault device predicted instrument state.

[0143] Preferably, the fuzzy data mining unit is further configured to perform the following steps:

[0144] The first fault device state attribute is traversed to configure a state attribute deviation threshold.

[0145] The proportion of the number of attributes of any two record instrument states of the first fault device record instrument state set, whose deviation is greater than or equal to the state attribute deviation threshold, is set as a second deviation coefficient.

[0146] When the second deviation coefficient is less than or equal to a second deviation coefficient threshold, the corresponding two record instrument states are added to the same cluster, otherwise, they are added to different clusters.

[0147] Cyclic analysis is performed to obtain a multi-cluster first fault device record instrument state, and a random record instrument state of a largest cluster of the multi-cluster first fault device record instrument state is extracted as the first fault device predicted instrument state.

[0148] Optionally, the fuzzy data mining unit is further configured to perform the following steps:

[0149] The proportion of the number of attributes whose state deviation between the first fault device prediction instrument state and the second instrument monitoring state is greater than or equal to the state attribute deviation threshold value is set as the first deviation coefficient.

[0150] Through the foregoing detailed description of the intelligent inspection alarm method based on panoramic video fusion, those skilled in the art can clearly understand an intelligent inspection alarm system based on panoramic video fusion in the embodiment. For the system disclosed in Embodiment Two, since it corresponds to the method disclosed in Embodiment One, it has corresponding functional modules and beneficial effects, and the related parts can be referred to the method part.

[0151] The above description of disclosed embodiments enables a person skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the application. Therefore, the present application will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An intelligent inspection and alarm method based on panoramic video fusion, characterized in that: include: Obtaining level 1 equipment failure alarm information through the panoramic video monitoring component, wherein the level 1 equipment failure alarm information includes a list of abnormal equipment bit numbers and a list of instrument monitoring status; According to the abnormal device number list, based on the upstream and downstream topology information of the device, a device fault relationship tree is constructed; Extracting a first faulty device, a parent node faulty device set of the first faulty device, and a child node faulty device set of the first faulty device based on the device fault relationship tree; Based on the instrument monitoring status list, perform fault tracing analysis on the first faulty device and the parent node faulty device set to generate a fault source device; Adding the fault source device to the fault source device group; Generate an intelligent inspection alarm signal based on the fault source equipment group; Based on the instrument monitoring status list, a fault tracing analysis is performed on the first faulty device and the parent node faulty device set to generate a fault source device, which includes: When the number of devices in the parent node fault device set is equal to 0, the first faulty device is set as the fault source device; When the number of devices in the parent node fault device set is not equal to 0, obtaining a first instrument monitoring status set of the parent node device set from the instrument monitoring status list, and obtaining a second instrument monitoring status of the first faulty device from the instrument monitoring status list, wherein the parent node device set is all upper-level devices that may affect the status of the first faulty device, and the parent node fault device set belongs to the parent node device set; Performing fuzzy data mining based on the first instrument monitoring state set to generate a first fault device predicted instrument state; When a first deviation coefficient between the predicted instrument state of the first faulty device and the monitored state of the second instrument is greater than or equal to a first deviation coefficient threshold, setting the first faulty device as the fault source device; When the first deviation coefficient between the predicted instrument state of the first faulty device and the monitored state of the second instrument is less than the first deviation coefficient threshold, the parent node faulty device set is traversed to perform fault tracing analysis and generate the fault source device.

2. The method according to claim 1, wherein Generating an intelligent inspection alarm signal according to the fault source equipment group, including: Perform cluster analysis based on the distance threshold according to the fault source device group to generate multiple alarm identification areas; Obtain the real-time coordinate set of the target population through the maintenance personnel terminal; Traversing the plurality of warning identification areas, sorting the target population based on the real-time coordinate set of the target population, and obtaining a maintenance recommended personnel list and a maintenance recommended route list; According to the maintenance recommended personnel list, the maintenance recommended route list, and the fault source equipment group, the intelligent inspection alarm signal is generated and sent to the corresponding maintenance personnel terminal.

3. The method according to claim 1, wherein According to the abnormal device tag list and based on the device upstream and downstream topology information, a device fault relationship tree is constructed, including: Obtaining device operation process information, wherein the device operation process information represents a device response sequence of a real-time operation task; The upstream and downstream topology information of the device represents the distribution topology of upstream and downstream devices that may interfere with the device status; Render the upstream and downstream topology information of the equipment according to the equipment operation process information to obtain the operation equipment position number topology information; The equipment fault relationship tree is constructed according to the operating equipment bit number topology information and the abnormal equipment bit number list.

4. The method according to claim 1, wherein Performing fuzzy data mining based on the first instrument monitoring state set to generate a first fault device predicted instrument state includes: Normalizing and weighting the first instrument monitoring state set to construct a first spatial coordinate; Based on the first spatial coordinates and a preset fault tolerance distance, a state mining fuzzy space is constructed; Extracting a first fault equipment recording instrument state set belonging to the state mining fuzzy space from historical operation records; Perform frequency analysis on the first faulty device recorded instrument state set to obtain the first faulty device predicted instrument state.

5. The method according to claim 4, wherein Performing frequency analysis on the first faulty device recorded instrument state set to obtain the first faulty device predicted instrument state, including: Traversing the state attributes of the first faulty device and configuring a state attribute deviation threshold; Comparing the attribute quantity ratio of any two recording instruments in the first faulty device recording instrument state set whose state deviation is greater than or equal to the state attribute deviation threshold, the ratio is set as a second deviation coefficient; When the second deviation coefficient is less than or equal to a second deviation coefficient threshold, the corresponding two recording instrument states are added into the same cluster; otherwise, they are added into different clusters; Circular analysis is performed to obtain multiple clusters of first fault device recording instrument states, and a random recording instrument state of the largest cluster among the multiple clusters of first fault device recording instrument states is extracted and set as the predicted instrument state of the first fault device.

6. The method according to claim 5, wherein The first deviation coefficient calculation process includes: The first deviation coefficient is set as the proportion of the number of attributes whose state deviation is greater than or equal to the state attribute deviation threshold when comparing the first fault device predicted instrument state and the second instrument monitored state.

7. The method according to claim 2, wherein Traversing the plurality of warning identification areas, sorting the target population based on the real-time coordinate set of the target population, and obtaining a maintenance recommended personnel list and a maintenance recommended route list, including: Obtaining the center coordinates of a first warning identification area of ​​the plurality of warning identification areas; Obtain the real-time coordinates of the first target person in the target population real-time coordinate set up to the real-time coordinates of the Nth target person; Performing path optimization based on the real-time coordinates of the first target person and the center coordinates of the first warning identification area to generate a first path optimization result; Performing path optimization based on the real-time coordinates of the Nth target person and the center coordinates of the first warning identification area to generate an Nth path optimization result; Selecting the shortest path personnel from the first path optimization result and the Nth path optimization result, setting them as the maintenance recommended personnel for the first alarm identification area, and adding them to the maintenance recommended personnel list; The path optimization result of the shortest path personnel is set as the first alarm identification area maintenance recommended route, added to the maintenance recommended route list, and the shortest path personnel is deleted from the real-time coordinate set of the target population.

8. An intelligent inspection and alarm system based on panoramic video fusion, characterized in that: include: Information acquisition module: The information acquisition module is used to obtain the first-level alarm information of equipment failure through the panoramic video monitoring component, wherein the first-level alarm information of equipment failure includes a list of abnormal equipment bit numbers and a list of instrument monitoring status; Relationship tree construction module: The relationship tree construction module is used to construct a device fault relationship tree based on the abnormal device number list and the device upstream and downstream topology information; Extraction module: The extraction module is used to extract the first faulty device and the parent node faulty device set of the first faulty device and the child node faulty device set of the first faulty device based on the device fault relationship tree; Analysis module: The analysis module is used to perform fault tracing analysis on the first faulty device and the parent node faulty device set based on the instrument monitoring status list to generate a fault source device; Aggregation module: The aggregation module is used to add the fault source device into the fault source device group; Alarm module: The alarm module is used to generate an intelligent inspection alarm signal according to the fault source equipment group; The analysis module also includes: A first judgment unit: The first judgment unit is configured to set the first faulty device as the fault source device when the number of devices in the parent node faulty device set is equal to 0; An instrument monitoring status acquisition unit: The instrument monitoring status acquisition unit is configured to, when the number of devices in the parent node faulty device set is not equal to 0, obtain a first instrument monitoring status set of the parent node device set from the instrument monitoring status list, and obtain a second instrument monitoring status of the first faulty device from the instrument monitoring status list, wherein the parent node device set is all upper-level devices that may affect the status of the first faulty device, and the parent node faulty device set belongs to the parent node device set; Fuzzy data mining unit: the fuzzy data mining unit is used to perform fuzzy data mining based on the first instrument monitoring state set to generate a first fault device predicted instrument state; The second judgment unit is configured to set the first faulty device as the fault source device when a first deviation coefficient between the predicted instrument state of the first faulty device and the monitored state of the second instrument is greater than or equal to a first deviation coefficient threshold; The third judgment unit is used to traverse the parent node fault device set to perform fault tracing analysis and generate the fault source device when the first deviation coefficient between the predicted instrument state of the first fault device and the monitored state of the second instrument is less than the first deviation coefficient threshold.

Images