A method and system for intelligent management of equipment operation and maintenance

Abstract

The present invention relates to the field of equipment operation and maintenance technology, and specifically discloses an intelligent management method and system for equipment operation and maintenance, wherein the method comprises: obtaining historical operation data of the required operation and maintenance management equipment, wherein the historical operation data comprises operation parameters and operation status; preprocessing the historical operation data; training a neural network model through the preprocessed data, and adjusting the hyperparameters through cross-validation to avoid overfitting; inspecting the equipment, and using the inspection results to train the neural network model; T 1 The operating parameters of the equipment at each moment are input into the neural network model to predict the subsequent T 2 The possible operating state of the equipment at all times, as well as the confidence level of each possible operating state of the equipment; based on the deviation of the parameter value from the normal range under all possible operating states of the equipment and the corresponding confidence level, a comprehensive determination is made as to whether an early warning is needed. The present invention can effectively utilize historical operating data and combine it with real-time inspection data to provide more efficient and accurate equipment status prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment operation and maintenance, and in particular to an intelligent management method and system for equipment operation and maintenance. Background Art

[0002] At present, traditional equipment monitoring methods rely on manual inspection and simple statistical analysis, which has the problems of low efficiency and high misjudgment rate. Although existing machine learning methods such as decision trees and linear regression have been applied in some prediction tasks, they often fail to capture the deep nonlinear relationship in complex equipment status. Therefore, with the improvement of the intelligence level of industrial equipment, there is an urgent need for a method based on deep learning that can effectively utilize historical operation data and combine real-time inspection data to provide more efficient and accurate equipment status prediction. Summary of the invention

[0003] In order to solve the problems existing in the prior art, the present invention provides an intelligent management method and system for equipment operation and maintenance.

[0004] The technical solution adopted by the present invention is:

[0005] An intelligent management method for equipment operation and maintenance comprises the following steps:

[0006] S1. Obtain historical operation data of the required operation and maintenance management equipment, wherein the historical operation data includes operation parameters and operation status;

[0007] S2. Preprocess the historical operation data;

[0008] S3, train the neural network model with the preprocessed data and adjust the hyperparameters through cross-validation to avoid overfitting;

[0009] S4. Inspect the equipment and use the T 1 The operating parameters of the equipment at the moment are preprocessed and input into the neural network model to predict the subsequent T 2 The operating status of the equipment at all times, and the confidence level of each operating status of the equipment;

[0010] S5. Comprehensively determine whether an early warning is needed based on the degree to which the parameter value deviates from the normal range under each operating state of the equipment and the corresponding confidence level.

[0011] Preferably, in step S2, preprocessing the historical operation data includes the following steps:

[0012] S21, remove noise and missing values ​​in historical operation data;

[0013] S22, selecting features that are helpful for prediction from historical operation data;

[0014] S23. Standardize the data corresponding to the features and scale them to the same range to improve the convergence speed of the model.

[0015] Preferably, in step S23, the data corresponding to the features are standardized and scaled to the same range, including normalizing the data, including the following steps:

[0016] The data is linearly converted to the range of 0 to 1. The conversion formula is as follows:

[0017] ;

[0018] in, is the normalized value of the eigenvalue, is the characteristic value, and are the maximum and minimum eigenvalues ​​respectively.

[0019] Preferably, in step S4, performing an inspection on the equipment includes the following steps:

[0020] S41. Combine the plant design drawings and the actual plant scene measurement results to establish a digital twin model of the plant;

[0021] S42. Generate classification codes for each required inspection equipment in the factory area, and establish a unique mapping relationship between the factory area digital twin model and each required inspection equipment in the actual factory area;

[0022] S43. Based on the spatial positioning of each device in the digital twin model of the plant, combined with the inspection demand analysis of the plant, with the shortest inspection path as the goal, a path optimization algorithm is established, and the inspection path is optimized using the path optimization algorithm to automatically calculate the optimal inspection path;

[0023] S44, push the optimal inspection path to the inspection subject, guide the inspection subject to perform inspection according to the optimal inspection path, and judge whether the inspection subject has completed the inspection. If the inspection is completed, the process ends; if the inspection is not completed, execute step S45;

[0024] S45. Determine whether the inspection subject is on the path to the next inspection device in the optimal inspection path. If so, jump to step S44; if not, recalculate the optimal inspection path based on the current position of the inspection subject and the uninspected device and jump to step S44.

[0025] Preferably, in step S42, a unified coding standard is used to generate a classification code for each required inspection equipment in the factory, and the classification code for each required inspection equipment is unique, and the classification code includes basic information, operation and maintenance information, and spatial location information of the required inspection equipment.

[0026] Preferably, in step S43, the inspection path is optimized by using a path optimization algorithm, and the automatic calculation of the optimal inspection path includes the following steps:

[0027] S431, discretizing the visible area within the inspection range into continuous rasterized grids, where the edges of the grids form paths, and the intersections between adjacent grids form nodes of the path;

[0028] S432, set the node of the inspection starting point as the origin Establish a two-dimensional coordinate system, determine the coordinate data of all nodes, determine the inspection nodes at the required inspection equipment locations according to the locations of all required inspection equipment, and set the set of all inspection nodes as ,in, Indicates the inspection node. Indicates Inspection nodes;

[0029] S433, set the path value of the node at the inspection starting point to , and set the path values ​​of the remaining nodes to , set the status of the original node to "not reached", add all nodes and their path values ​​and status to the Fibonacci queue, and set the node of the inspection starting point to , the coordinates are ;

[0030] S434. Select a node with a status of "not reached" and a minimum path value from the Fibonacci queue, and record the node as , the coordinates are , and the node The status of is updated to "arrived";

[0031] S435, Settings and Nodes The set of adjacent nodes is , and the set of adjacent nodes The node whose status is "reached" in the Fibonacci queue is not included in the adjacent node set The coordinate set of the midpoint is ,in, represents the adjacent nodes, Indicates neighboring nodes, Indicates The coordinates of adjacent nodes;

[0032] S436, calculate the node from the inspection starting point Departure through the node Arriving at a certain node Neighboring nodes The distance value is the first distance value , calculate the adjacent nodes To the nearest inspection node The distance value is the second distance value , according to the first distance value and the second distance value The sum of the values ​​determines the adjacent nodes The updated path value of The status is "Not Arrived";

[0033] First distance value The calculation formula is as follows:

[0034] ;

[0035] Second distance value The calculation formula is as follows:

[0036] ;

[0037] in, For adjacent nodes The nearest inspection node The coordinates of

[0038] S437, according to the adjacent node set All adjacent nodes in The updated path value is selected from all updated path values, and the adjacent node corresponding to the minimum value in the updated path value is The path value in the Fibonacci queue is replaced with the minimum value among the updated path values;

[0039] S438: Determine the adjacent node corresponding to the minimum value in the updated path value Whether it belongs to the set of all inspection nodes , if it belongs to, then the adjacent node Add inspection path node set , and trace back to the adjacent node , generate a local path, and then execute step S439. If it does not belong to , jump to step S434, where Represents the inspection node in the inspection path node set. Indicates the first node in the inspection path node set nodes;

[0040] S439: Determine the inspection path node set The collection of all inspection nodes Is there a one-to-one correspondence between the nodes included in the inspection? If so, all local paths are traced back to generate a complete path to obtain the optimal inspection path. If not, the node at the inspection starting point is Update to the adjacent node corresponding to the minimum value in the updated path value in step S437 , and jump to step S434.

[0041] Preferably, in step S45, recalculating the optimal inspection path according to the current position of the inspection subject and the uninspected equipment includes the following steps:

[0042] The node where the inspection starts It is updated to the node closest to the current position of the inspection subject, and then jumps to execute step S434, and finally obtains the recalculated optimal inspection path.

[0043] Preferably, in step S45, when recalculating the optimal inspection path based on the current position of the inspection subject and the uninspected equipment, the node at the inspection starting point is Update to the node closest to the current location of the inspection subject. According to the optimal inspection path, set the inspection node corresponding to the inspection device completed in the previous inspection as , the next inspection device to be inspected corresponds to the inspection node , calculate the node of the inspection starting point To inspection node The distance is the third distance value , calculate the node of the inspection starting point To the inspection node The distance is the fourth distance value , obtain the inspection nodes in the optimal inspection path To inspection node The fifth distance , determine the third distance value , the fourth distance value and the fifth distance The relationship between ,and , then calculate the node of the inspection starting point To inspection node The shortest path of the inspection node in the original optimal inspection path The subsequent paths are merged to form a recalculated optimal inspection path, and the process jumps to step S44. Otherwise, the node of the inspection starting point is calculated. To inspection node The shortest path of the inspection node in the original optimal inspection path The subsequent paths are merged to form a recalculated optimal inspection path, and the process jumps to step S44, where: is a positive constant;

[0044] The third distance value The calculation formula is as follows:

[0045] ;

[0046] Fourth distance value The calculation formula is as follows:

[0047] ;

[0048] Fifth distance value The calculation formula is as follows:

[0049] ;

[0050] in, For Node The coordinates of For Node The coordinates of .

[0051] An intelligent equipment operation and maintenance management system, used to execute the intelligent equipment operation and maintenance management method described above, characterized in that it includes the following modules:

[0052] The data preprocessing module is used to obtain the historical operation data of the required operation and maintenance management equipment and the T 1 Preprocess the operating parameters of the equipment at all times so as to be used for neural network training and prediction of equipment operating status;

[0053] The neural network module is connected to the data preprocessing module and is used to receive the preprocessed historical operation data of the required operation and maintenance management equipment and the T 1 The operating parameters of the equipment at all times in order to conduct training and predict the equipment operating status;

[0054] The plant visualization modeling module is used to combine the plant design drawings and the actual plant scene measurement results to establish a plant digital twin model;

[0055] The inspection equipment coding module is connected to the plant visualization modeling module to generate the classification code of each required inspection equipment in the plant and establish a unique mapping relationship between the plant digital twin model and each required inspection equipment in the actual plant;

[0056] The inspection path planning module is connected to the plant visualization modeling module and the inspection equipment coding module. It is used to establish a path optimization algorithm based on the spatial positioning of each device in the plant digital twin model, combined with the inspection demand analysis of the plant, with the shortest inspection path as the goal, and use the path optimization algorithm to optimize the inspection path and automatically calculate the optimal inspection path;

[0057] The inspection path guidance module, together with the plant visualization modeling module and the inspection path planning module, is used to push the optimal inspection path to the inspection subject, guide the inspection subject to conduct inspections according to the optimal inspection path, and determine whether the inspection subject has completed the inspection;

[0058] The inspection progress monitoring module is connected with the plant visualization modeling module and the inspection path guidance module. It is used to determine whether the inspection subject is on the path to the next inspection device in the optimal inspection path when the inspection is not completed. If not, the inspection path planning module is requested to recalculate the optimal inspection path according to the current position of the inspection subject and the uninspected equipment.

[0059] The inspection data collection module is connected to the data preprocessing module and is used to collect inspection data so that the neural network can predict the operating status of the inspection equipment after the data is processed by the data preprocessing module.

[0060] The beneficial effects of the present invention are:

[0061] The present invention processes the historical operation data of the equipment, uses the historical operation data of the equipment to train a neural network and verify it, and then uses the trained neural network in combination with the current inspection data of the equipment to predict the next operation status of the equipment. Compared with traditional equipment monitoring methods that rely more on manual inspection and simple statistical analysis, the prediction efficiency is higher and the error rate is lower. The trained neural network can more easily capture the deep nonlinear relationship in the complex equipment status, and can effectively use historical operation data in combination with real-time inspection data to provide more efficient and accurate equipment status prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a schematic diagram of the process of the present invention;

[0063] Figure 2 A flow chart of equipment inspection in the method of the present invention;

[0064] Figure 3 A flow chart of calculating the optimal inspection path in the method of the present invention;

[0065] Figure 4 It is a schematic diagram of the structure of the system of the present invention. DETAILED DESCRIPTION

[0066] The present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0067] Example:

[0068] like Figure 1 The figure shows a flow chart of an intelligent management method for equipment operation and maintenance, which includes the following steps:

[0069] S1. Obtain historical operation data of the required operation and maintenance management equipment, wherein the historical operation data includes operation parameters and operation status.

[0070] S2. Preprocess the historical operation data.

[0071] Specifically, preprocessing the historical operation data includes the following steps:

[0072] S21, remove noise and missing values ​​in historical operation data;

[0073] S22, selecting features that are helpful for prediction from historical operation data;

[0074] S23. Standardize the data corresponding to the features and scale them to the same range to improve the convergence speed of the model.

[0075] Specifically, in step S23, the data corresponding to the features are standardized and scaled to the same range, including normalizing the data, including the following steps:

[0076] The data is linearly converted to the range of 0 to 1. The conversion formula is as follows:

[0077] ;

[0078] in, is the normalized value of the eigenvalue, is the characteristic value, and are the maximum and minimum eigenvalues ​​respectively.

[0079] S3. Train the neural network model using preprocessed data and adjust hyperparameters through cross-validation to avoid overfitting.

[0080] S4. Inspect the equipment, pre-process the operating parameters of the equipment at time T1 obtained from the inspection, and input them into the neural network model to predict and deduce the operating status of the equipment at the subsequent time T2, as well as the confidence level of each operating status of the equipment.

[0081] Specifically, Figure 2 The following is the process of inspecting the equipment:

[0082] S41. Build a digital twin model of the factory by combining the factory design drawings and actual factory measurement results.

[0083] Specifically, by combining the factory design drawings and the actual factory scene measurement results, real-time rendering technology is used to establish a highly simulated factory digital twin model in the 3D rendering engine, which includes all the equipment required for inspection, such as environmental equipment, production equipment, etc. Environmental equipment includes lighting, air conditioning, ventilation, drainage and related power supply, while production equipment is generally various machines directly involved in production.

[0084] S42. Generate classification codes for each required inspection equipment in the factory, and establish a unique mapping relationship between the factory digital twin model and each required inspection equipment in the actual factory.

[0085] Specifically, a unified coding standard is used to generate classification codes for each required inspection equipment in the factory area, and the classification codes of each required inspection equipment are unique, and the classification codes contain the basic information, operation and maintenance information, and spatial location information of the required inspection equipment. Classification can adopt multiple levels of categories, from large categories to small categories, each category corresponds to a code, until all required inspection equipment can be clearly distinguished, and each required inspection equipment corresponds to a unique classification code.

[0086] S43. Based on the spatial positioning of each device in the digital twin model of the factory, combined with the inspection demand analysis of the factory, a path optimization algorithm is established with the shortest inspection path as the goal, and the path optimization algorithm is used to optimize the inspection path and automatically calculate the optimal inspection path.

[0087] Specifically, Figure 3 The figure shows a flow chart of optimizing the inspection path by using the path optimization algorithm in step S43 to automatically calculate the optimal inspection path. The path optimization algorithm includes the following steps:

[0088] S431, discretizing the visible area within the inspection range into continuous rasterized grids, the edges of the grids forming paths, and the intersections between adjacent grids forming nodes of the path.

[0089] S432, set the node of the inspection starting point as the origin Establish a two-dimensional coordinate system, determine the coordinate data of all nodes, determine the inspection nodes at the required inspection equipment locations according to the locations of all required inspection equipment, and set the set of all inspection nodes as .

[0090] in, Indicates the inspection node. Indicates Inspection nodes. Generally speaking, the inspection tool room or the main control room can be used as the origin. During the initial calculation, the node of the inspection starting point can be set at the origin, and an XY two-dimensional coordinate system is established based on the origin. The directions of the X-axis and the Y-axis can refer to a certain wall and be parallel to the wall so that the XY two-dimensional coordinate system conforms to people's observation habits. Inspection nodes can be selected from the nodes around the inspected equipment, and generally the nodes that are most suitable for inspecting or observing the inspection equipment are selected.

[0091] S433, set the path value of the node at the inspection starting point to , and set the path values ​​of the remaining nodes to , set the status of the original node to "not reached", add all nodes and their path values ​​and status to the Fibonacci queue, and set the node of the inspection starting point to , the coordinates are .

[0092] Among them, the path value of the defined node is According to the previous description, during the initial calculation, the node at the starting point of the inspection That is the origin.

[0093] S434. Select a node with a status of "not reached" and a minimum path value from the Fibonacci queue, and record the node as , the coordinates are , and the node The status of is updated to "Arrived".

[0094] Among them, during the initial calculation, the first node with the smallest path value selected must be the origin.

[0095] S435, Settings and Nodes The set of adjacent nodes is , and the set of adjacent nodes The node whose status is "reached" in the Fibonacci queue is not included in the adjacent node set The coordinate set of the midpoint is .

[0096] in, represents the adjacent nodes, Indicates neighboring nodes, Indicates The coordinates of adjacent nodes, the set of adjacent nodes Only nodes with the status of “not reached” in the Fibonacci queue are included in the adjacent node set The coordinate set of nodes with the status of "not reached" is .

[0097] S436, calculate the node from the inspection starting point Departure through the node Arriving at a certain node Neighboring nodes The distance value is the first distance value , calculate the adjacent nodes To the nearest inspection node The distance value is the second distance value , according to the first distance value and the second distance value The sum of the values ​​determines the adjacent nodes The updated path value.

[0098] Specifically, the nearest inspection node Also for nodes in the Fibonacci queue whose status is "Not reached", the first distance value The calculation formula is as follows:

[0099] ;

[0100] Second distance value The calculation formula is as follows:

[0101] ;

[0102] in, For adjacent nodes The nearest inspection node The calculation formula for updating the path value is as follows:

[0103] ;

[0104] in, To update the path value.

[0105] S437, according to the adjacent node set All adjacent nodes in The updated path value is selected from all updated path values, and the adjacent node corresponding to the minimum value in the updated path value is The path value in the Fibonacci queue is replaced by the minimum value among the updated path values.

[0106] Specifically, from the calculated updated path values ​​of each adjacent node, select the adjacent node corresponding to the one with the smallest updated path value, and replace the path value of the adjacent node in the Fibonacci queue with the value of the adjacent node in the Fibonacci queue. Replaced with the value of the smallest of the update path values.

[0107] S438: Determine the adjacent node corresponding to the minimum value in the updated path value Whether it belongs to the set of all inspection nodes , if it belongs to, then the adjacent node Add inspection path node set , and trace back to the adjacent node The predecessor node generates a local path, and then executes step S439. If it does not belong to, jump to step S434.

[0108] Specifically, determine the adjacent nodes of the replaced path value in step S437 With a patrol node Whether they are the same coordinates, that is, to determine the adjacent nodes of the replaced path value Is it a patrol node? If yes, add the adjacent node to the new set ,in, Represents the nodes in the new set, that is, the inspection nodes in the inspection path node set. Indicates the first node in the inspection path node set nodes. There are already 2 nodes in and , then according to and its predecessor nodes The first local path is generated according to and its predecessor nodes The second partial path is generated, newly added to the collection The node is , then according to and its predecessor nodes What is generated is the third partial path.

[0109] S439: Determine the inspection path node set The collection of all inspection nodes Is there a one-to-one correspondence between the nodes included in the inspection? If so, all local paths are traced back to generate a complete path to obtain the optimal inspection path. If not, the node at the inspection starting point is Update to the adjacent node corresponding to the minimum value in the updated path value in step S437 , and jump to step S434.

[0110] Specifically, such as the inspection path node set The collection of all inspection nodes If the nodes included in are one-to-one corresponding, it means that the path calculation has completed the calculation of all inspection nodes, and all local paths can be traced back to generate a complete path, that is, the optimal inspection path. For example, the first local path, the second local path and the third local path mentioned above are integrated to generate a complete path, that is, the optimal inspection path. If the inspection path node set The collection of all inspection nodes If the nodes included in are not one-to-one corresponding, it means that there are inspection nodes that have not been calculated, then the starting point is updated to the node of the inspection starting point Update to the adjacent node corresponding to the minimum value in the updated path value in step S437 , and jump to step S434.

[0111] S44, push the optimal inspection path to the inspection subject, guide the inspection subject to perform inspection according to the optimal inspection path, and judge whether the inspection subject has completed the inspection. If the inspection is completed, end; if not, execute step S45.

[0112] Specifically, if the inspection subject is a robot, the optimal inspection path is directly pushed to the robot to guide the robot to inspect along the optimal inspection path. If the inspection subject is an inspection technician, the inspection technician is generally equipped with an intelligent terminal device, and the optimal inspection path can be pushed to the intelligent terminal device to guide the inspection technician to inspect along the optimal inspection path. The inspection progress of the inspection subject can be judged based on the position of the inspection subject. By default, the inspection node passed by the inspection subject is completed. Of course, the inspection subject can also upload inspection information to determine whether a certain inspection node has completed the inspection. There are many existing technologies to realize whether the inspection of the inspection node has been completed, which will not be repeated. If it is determined that the inspection subject has completed the inspection of all inspection nodes, the process ends, otherwise execute step S45.

[0113] S45. Determine whether the inspection subject is on the path to the next inspection device in the optimal inspection path. If so, jump to step S44; if not, recalculate the optimal inspection path based on the current position of the inspection subject and the uninspected device and jump to step S44.

[0114] Specifically, in order to facilitate the positioning of the inspection subject to determine whether the inspection subject is on the optimal inspection path, a UWB module can be installed on the robot or smart terminal device, and the UWB indoor positioning technology can be used for positioning. Of course, positioning technologies such as Bluetooth or WIFI can also be used to obtain the position of the inspection subject in real time. If the inspection subject is on the optimal inspection path and on the path to the next inspection device (inspection node), that is, there are inspection devices that have not completed the inspection, then it can be considered that the inspection is not completed, and the inspection subject has not deviated from the optimal inspection path, then jump to execute step S44, keep pushing the optimal inspection path and continue to guide the inspection subject, and continue to detect whether the inspection is completed. If the inspection subject is not on the optimal inspection path, it is determined that the inspection subject has deviated from the optimal inspection path, and the node at the starting point of the inspection is moved. Update to the node closest to the current position of the inspection subject, and recalculate the optimal inspection path, then jump to execute step S44, recalculate the optimal inspection path to update the optimal inspection path according to the current position of the inspection subject, and re-push and guide. Among them, in order to reduce resource usage, a range that deviates from the optimal inspection path can be set, with the node of the previous inspection device as the center, and a circular electronic fence with a preset radius. The optimal inspection path is recalculated only when it is detected that the inspection subject exceeds the electronic fence. When it deviates from the optimal inspection path not far, especially when the technician is inspecting, it can return to the original optimal inspection path by itself. When the robot is inspecting, short-distance automatic positioning and path finding are also relatively mature existing technologies. For example, existing sweeping robots or walking service robots have adopted related technologies, which will not be repeated. Among them, the robot can directly record the equipment parameters during the inspection, and the technician can record the equipment parameters through the intelligent terminal device during the inspection.

[0115] Specifically, in step S45, if the inspection subject deviates far from the optimal inspection path and needs to recalculate the optimal inspection path, the node at the inspection starting point is recalculated according to the current position of the inspection subject and the uninspected equipment. Update to the node closest to the current location of the inspection subject. According to the original optimal inspection path, set the inspection node corresponding to the inspection device completed in the previous inspection as , the next inspection device to be inspected corresponds to the inspection node , calculate the node of the inspection starting point To the inspection node The distance is the third distance value , calculate the node of the inspection starting point To the inspection node The distance is the fourth distance value , obtain the inspection nodes in the optimal inspection path To inspection node The fifth distance , determine the third distance value , the fourth distance value and the fifth distance Size relationship, such as ,and , then calculate the node of the inspection starting point To inspection node The shortest path of the inspection node in the original optimal inspection path The subsequent paths are merged to form a recalculated optimal inspection path, otherwise the process jumps to step S434 and finally recalculates the optimal inspection path. ,node and nodes Generally speaking, it will be a triangle. According to the principle that the sum of the two sides of a triangle is greater than the third side, the node Passage Node To Node The distance must be greater than the node To the node Even if the three points are in a straight line, there are only nodes At the node Away from the node When the node Passage Node To Node The distance is equal to the node To Node distance, and when the node At the node and nodes When the node Passage Node To Node The distance is also smaller than the node To Node Based on the principle of the shortest path, the computing node will be selected. With Node The shortest path between the nodes is then compared with the original optimal inspection path. The subsequent paths are integrated to form a new optimal inspection path. However, considering the use of resources, when the third distance value Less than the fourth distance value When the fourth distance value The third distance value The difference is greater than the fifth distance When the ratio is certain, due to the fourth distance value The third distance value The complexity of the path calculation of the difference distance leads to the resource cost and time cost of the calculation for the timeliness of the calculation, which is higher than the fourth distance value. Compared to the third distance value Distance to the fifth The path cost saved by the sum value is not necessary to take a shorter node With Node The path between ,and When , the node of the inspection starting point can be calculated To inspection node The shortest path of the inspection node in the original optimal inspection path The subsequent paths are merged to form a recalculated optimal inspection path, and the process jumps to step S44. Otherwise, the node of the inspection starting point is calculated. To inspection node The shortest path of the inspection node in the original optimal inspection path The subsequent paths are merged to form a recalculated optimal inspection path, and the process jumps to step S44. is a positive number, which to some extent reflects the weight relationship between the resources required for calculation and the resources required for the movement of the inspection subject. If you pay more attention to saving the resources required for calculation, you can increase , otherwise it decreases . Calculate the node where the inspection starts To inspection node The shortest path and the node where the inspection starts are calculated To inspection node The shortest path can use the path optimization algorithm in step S43, which will not be described in detail here.

[0116] Specifically, the third distance value The calculation formula is as follows:

[0117] ;

[0118] Fourth distance value The calculation formula is as follows:

[0119] ;

[0120] Fifth distance value The calculation formula is as follows:

[0121] ;

[0122] in, For Node The coordinates of For Node The coordinates of .

[0123] S5. Based on the deviation of the parameter value from the normal range under each operating state of the equipment (obtained from historical operating data statistics or user experience knowledge) and the corresponding confidence level, a comprehensive determination is made as to whether an early warning is needed.

[0124] Specifically, for example, the expected value of the sum of the deviation degree of each state parameter multiplied by the corresponding confidence level may be used to determine whether a threshold requiring an early warning has been reached.

[0125] Note that the uncertainty in the model predictions is also related to T 1 and T 2 Time difference About, among which, ,therefore, It is also a configuration parameter that needs to be given in advance.

[0126] like Figure 4 The figure shows a schematic diagram of the structure of an intelligent management system for equipment operation and maintenance, which is used to execute the intelligent management method for equipment operation and maintenance described above, and includes the following modules:

[0127] The data preprocessing module is used to obtain the historical operation data of the required operation and maintenance management equipment and the T 1 Preprocess the operating parameters of the equipment at all times so as to be used for neural network training and prediction of equipment operating status;

[0128] The neural network module is connected to the data preprocessing module and is used to receive the preprocessed historical operation data of the required operation and maintenance management equipment and the T 1 The operating parameters of the equipment at all times in order to conduct training and predict the equipment operating status;

[0129] The plant visualization modeling module is used to combine the plant design drawings and the actual plant scene measurement results to establish a plant digital twin model;

[0130] The inspection equipment coding module is connected to the plant visualization modeling module to generate the classification code of each required inspection equipment in the plant and establish a unique mapping relationship between the plant digital twin model and each required inspection equipment in the actual plant;

[0131] The inspection path planning module is connected to the plant visualization modeling module and the inspection equipment coding module. It is used to establish a path optimization algorithm based on the spatial positioning of each device in the plant digital twin model, combined with the inspection demand analysis of the plant, with the shortest inspection path as the goal, and use the path optimization algorithm to optimize the inspection path and automatically calculate the optimal inspection path;

[0132] The inspection path guidance module is connected with the plant visualization modeling module and the inspection path planning module to push the optimal inspection path to the inspection subject, guide the inspection subject to conduct inspections according to the optimal inspection path, and determine whether the inspection subject has completed the inspection;

[0133] The inspection progress monitoring module is connected with the plant visualization modeling module and the inspection path guidance module. It is used to determine whether the inspection subject is on the path to the next inspection device in the optimal inspection path when the inspection is not completed. If not, the inspection path planning module is requested to recalculate the optimal inspection path according to the current position of the inspection subject and the uninspected equipment.

[0134] The inspection data collection module is connected to the data preprocessing module and is used to collect inspection data. The inspection data collection module can be built into the inspection body (inspection robot or inspection intelligent terminal device) to obtain equipment operating parameters so that the neural network can predict the operating status of the inspection equipment after the data is processed by the data preprocessing module.

[0135] The above-mentioned embodiments only express the specific implementation of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the present invention. It should be pointed out that for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A method for intelligent management of equipment operation and maintenance, characterized in that: The steps include: S1. Obtain historical operation data of the required operation and maintenance management equipment, wherein the historical operation data includes operation parameters and operation status; S2. Preprocess the historical operation data; S3, train the neural network model with the preprocessed data and adjust the hyperparameters through cross-validation to avoid overfitting; S4, inspect the equipment, pre-process the operating parameters of the equipment at time T1 obtained from the inspection, and input them into the neural network model to predict and deduce the operating status of the equipment at the subsequent time T2, as well as the confidence level of each operating status of the equipment; The equipment inspection includes the following steps: S41. Combine the plant design drawings and the actual plant scene measurement results to establish a digital twin model of the plant; S42. Generate classification codes for each required inspection equipment in the factory area, and establish a unique mapping relationship between the factory area digital twin model and each required inspection equipment in the actual factory area; S43. Based on the spatial positioning of each device in the digital twin model of the plant, combined with the inspection demand analysis of the plant, with the shortest inspection path as the goal, a path optimization algorithm is established, and the inspection path is optimized using the path optimization algorithm to automatically calculate the optimal inspection path; Using the path optimization algorithm to optimize the inspection path, the automatic calculation of the optimal inspection path includes the following steps: S431, discretizing the visible area within the inspection range into continuous rasterized grids, where the edges of the grids form paths, and the intersections between adjacent grids form nodes of the path; S432, set the node of the inspection starting point as the origin Establish a two-dimensional coordinate system, determine the coordinate data of all nodes, determine the inspection nodes at the required inspection equipment locations according to the locations of all required inspection equipment, and set the set of all inspection nodes as ,in, Indicates the inspection node. Indicates Inspection nodes; S433, set the path value of the node at the inspection starting point to , and set the path values ​​of the remaining nodes to , set the status of the original node to "not reached", add all nodes and their path values ​​and status to the Fibonacci queue, and set the node of the inspection starting point to , the coordinates are ; S434: Select a node with a status of "not reached" and a minimum path value from the Fibonacci queue, and record the node as , the coordinates are , and the node The status of is updated to "arrived"; S435, Settings and Nodes The set of adjacent nodes is , and the set of adjacent nodes The node whose status is "reached" in the Fibonacci queue is not included in the adjacent node set The coordinate set of the midpoint is ,in, represents the adjacent nodes, Indicates neighboring nodes, Indicates The coordinates of adjacent nodes; S436, calculate the node from the inspection starting point Departure through the node Arriving at a certain node Neighboring nodes The distance value is the first distance value , calculate the adjacent nodes To the nearest inspection node The distance value is the second distance value , according to the first distance value and the second distance value The sum of the values ​​determines the adjacent nodes The updated path value of The status is "Not Arrived"; First distance value The calculation formula is as follows: ； Second distance value The calculation formula is as follows: ； in, For adjacent nodes The nearest inspection node The coordinates of S437, according to the adjacent node set All adjacent nodes in The updated path value is selected from all updated path values, and the adjacent node corresponding to the minimum value in the updated path value is The path value in the Fibonacci queue is replaced with the minimum value among the updated path values; S438: Determine the adjacent node corresponding to the minimum value in the updated path value Whether it belongs to the set of all inspection nodes , if it belongs to, then the adjacent node Add inspection path node set , and trace back to the adjacent node , generate a local path, and then execute step S439. If it does not belong to , jump to step S434, where Represents the inspection node in the inspection path node set. Indicates the first node in the inspection path node set nodes; S439: Determine the inspection path node set The collection of all inspection nodes Is there a one-to-one correspondence between the nodes included in the inspection? If so, all local paths are traced back to generate a complete path to obtain the optimal inspection path. If not, the node at the inspection starting point is Update to the adjacent node corresponding to the minimum value in the updated path value in step S437 , and jump to step S434; S44, push the optimal inspection path to the inspection subject, guide the inspection subject to perform inspection according to the optimal inspection path, and judge whether the inspection subject has completed the inspection. If the inspection is completed, the process ends; if the inspection is not completed, execute step S45; S45, determining whether the inspection subject is on the path to the next inspection device in the optimal inspection path, if so, jumping to step S44; if not, recalculating the optimal inspection path according to the current position of the inspection subject and the uninspected device and jumping to step S44; When recalculating the optimal inspection path based on the current location of the inspection subject and the uninspected equipment, the node at the inspection starting point is Update to the node closest to the current location of the inspection subject. According to the optimal inspection path, set the inspection node corresponding to the inspection device completed in the previous inspection as , the next inspection device to be inspected corresponds to the inspection node , calculate the node of the inspection starting point To inspection node The distance is the third distance value , calculate the node of the inspection starting point To the inspection node The distance is the fourth distance value , obtain the inspection nodes in the optimal inspection path To inspection node The fifth distance , determine the third distance value , the fourth distance value and the fifth distance The relationship between ,and , then calculate the node of the inspection starting point To inspection node The shortest path of the inspection node in the original optimal inspection path The subsequent paths are merged to form a recalculated optimal inspection path, and the process jumps to step S44. Otherwise, the node of the inspection starting point is calculated. To inspection node The shortest path of the inspection node in the original optimal inspection path The subsequent paths are merged to form a recalculated optimal inspection path, and the process jumps to step S44, where: is a positive constant; The third distance value The calculation formula is as follows: ； Fourth distance value The calculation formula is as follows: ； Fifth distance value The calculation formula is as follows: ； in, For Node The coordinates of For Node The coordinates of S5. Comprehensively determine whether an early warning is needed based on the degree to which the parameter value deviates from the normal range under each operating state of the equipment and the corresponding confidence level.

2. The intelligent management method for equipment operation and maintenance according to claim 1 is characterized in that: In step S2, preprocessing the historical operation data includes the following steps: S21, remove noise and missing values ​​in historical operation data; S22, selecting features that are helpful for prediction from historical operation data; S23. Standardize the data corresponding to the features and scale them to the same range to improve the convergence speed of the model.

3. The intelligent management method for equipment operation and maintenance according to claim 2 is characterized in that: In step S23, the data corresponding to the features are standardized and scaled to the same range, including normalizing the data, including the following steps: The data is linearly converted to the range of 0 to 1. The conversion formula is as follows: ； in, is the normalized value of the eigenvalue, is the characteristic value, and are the maximum and minimum eigenvalues ​​respectively.

4. The intelligent management method for equipment operation and maintenance according to claim 3 is characterized in that: In step S42, a unified coding standard is used to generate a classification code for each required inspection equipment in the factory area, and the classification code for each required inspection equipment is unique. The classification code includes basic information, operation and maintenance information, and spatial location information of the required inspection equipment.

5. An intelligent equipment operation and maintenance management system, used to execute the intelligent equipment operation and maintenance management method according to any one of claims 1 to 4, characterized in that: Includes the following modules: The data preprocessing module is used to preprocess the historical operation data of the required operation and maintenance management equipment and the operation parameters of the equipment at time T1 obtained by inspection, so as to be used for neural network training and prediction of equipment operation status; The neural network module is connected to the data preprocessing module and is used to receive the preprocessed historical operation data of the required operation and maintenance management equipment and the operation parameters of the equipment at time T1 obtained by inspection, so as to perform training and predict the operation status of the equipment; The plant visualization modeling module is used to build a digital twin model of the plant by combining the plant design drawings and the actual plant scene measurement results; The inspection equipment coding module is connected to the plant visualization modeling module to generate the classification code of each required inspection equipment in the plant and establish a unique mapping relationship between the plant digital twin model and each required inspection equipment in the actual plant; The inspection path planning module is connected to the plant visualization modeling module and the inspection equipment coding module. It is used to establish a path optimization algorithm based on the spatial positioning of each device in the plant digital twin model, combined with the inspection demand analysis of the plant, with the shortest inspection path as the goal, and use the path optimization algorithm to optimize the inspection path and automatically calculate the optimal inspection path; The inspection path guidance module, together with the plant visualization modeling module and the inspection path planning module, is used to push the optimal inspection path to the inspection subject, guide the inspection subject to conduct inspections according to the optimal inspection path, and determine whether the inspection subject has completed the inspection; The inspection progress monitoring module is connected with the plant visualization modeling module and the inspection path guidance module. It is used to determine whether the inspection subject is on the path to the next inspection device in the optimal inspection path when the inspection is not completed. If not, the inspection path planning module is requested to recalculate the optimal inspection path according to the current position of the inspection subject and the uninspected equipment. The inspection data collection module is connected to the data preprocessing module and is used to collect inspection data so that the neural network can predict the operating status of the inspection equipment after the data is processed by the data preprocessing module.

Images