Crane operation and maintenance method based on machine learning, electronic equipment and computer readable storage medium

Abstract

The invention provides a crane operation and maintenance method based on machine learning, which comprises the following steps of: arranging sensor groups on a metal structure, an electrical control element and a driving mechanism of a crane to monitor the intrinsic safety of the crane to form data groups of key parts; using safety real-time monitoring is conducted on the state of the crane in the using process, and a using process data set is formed; performing data processing on the key part data set and the use process data set to extract feature data, establishing a machine learning model by using a neural network, and performing model training; real-time data are collected and input into the model for judgment, when the prediction probability is higher than a certain threshold value, fault alarms are given out, meanwhile, the fault alarms are classified through the convolutional neural network, and maintenance work is conducted according to different types of fault alarms. The problem that the equipment fault condition cannot be effectively predicted in the operation and maintenance management process in the prior art is solved. The invention further provides electronic equipment and a computer readable storage medium.

Description

Technical Field

[0001] The present invention relates to the technical field of crane operation and maintenance management, and in particular to a crane operation and maintenance method based on machine learning, an electronic device and a computer-readable storage medium. Background Art

[0002] During the operation and use of the crane, its working environment may be relatively harsh. The traditional operation and maintenance methods may have certain human subjectivity and timeliness, which may easily lead to waste of resources. At the same time, due to the need for high-altitude operations, frequent operation and maintenance plans will increase the risk of high-altitude operations for maintenance personnel, affecting the efficiency and coverage of inspections. It is difficult to implement manual inspections in some important safety parts, so the intervals between operation and maintenance also need to be appropriately extended, which increases the equipment failure rate and safety hazards. During the operation and maintenance interval, if the equipment failure cannot be effectively predicted, it may lead to frequent use of the equipment, resulting in long-term "disease operation" of the equipment, which will cause equipment failure and shutdown, resulting in longer downtime and greater losses. Therefore, it is necessary to improve the efficiency and accuracy of crane operation and maintenance management. Summary of the invention

[0003] In view of the deficiencies in the prior art, the present invention provides a crane operation and maintenance method based on machine learning, which solves the problem in the prior art that equipment failure conditions cannot be effectively predicted during the operation and maintenance management process.

[0004] According to an embodiment of the present invention, a crane operation and maintenance method based on machine learning includes:

[0005] Install sensor groups on the metal structure, electrical control components, and drive mechanisms of the crane to monitor the inherent safety of the crane and form data groups for each key part;

[0006] Real-time monitoring of the crane's status during use to ensure safety and form a data set of the use process;

[0007] Process the key parts data set and the usage process data set to extract feature data and use neural networks to build machine learning models and perform model training;

[0008] Real-time data is collected and input into the model for judgment. When the predicted probability is higher than a certain threshold, a fault alarm is issued. At the same time, a convolutional neural network is used to classify the fault alarms, and maintenance work is performed according to different types of fault alarms.

[0009] In this embodiment, through the training of the machine learning model, the model can make more accurate fault alarms and classify fault alarms, perform targeted operation and maintenance management according to different types of fault alarms, improve operation and maintenance efficiency and the accuracy of operation and maintenance management, and avoid the situation where the equipment "operates with illness".

[0010] Furthermore, data processing includes data cleaning of the data set, and data cleaning includes one or more of filling in missing values, smoothing noisy data, and identifying or deleting outliers.

[0011] Furthermore, data processing also includes adjusting the cleaned data into a format suitable for machine learning algorithms, combining data from multiple data groups and storing them uniformly, establishing a data warehouse, and performing data integration.

[0012] Furthermore, the data processing also includes:

[0013] Extract the original data samples from the data warehouse, and perform feature extraction on the original data samples as original signal features;

[0014] Fault features, including time domain features, frequency domain features, and time-frequency domain features, are extracted from the public crane fault database for fault diagnosis.

[0015] Furthermore, the data processing also includes:

[0016] The extracted features are screened and calculated as follows:

[0017] S1: Calculate the internal distance of feature x of class = y using the formula expressed in the equation,

[0018]

[0019] Where N is the number of samples, F is the number of features, the number of classes is represented by C, px, ycand-px, yn are the x features of C samples of the yth class and the x features of N samples of the yth class respectively;

[0020] S2: Calculate the average of the distances within class y for feature x, expressed as:

[0021]

[0022] S3: Calculate the average value of feature x for N samples in class y, expressed as:

[0023]

[0024] S4: Calculate the average distance between C classes of feature x using the equation:

[0025]

[0026] S5: Calculate the evaluation factor of feature x using the equation:

[0027]

[0028] The features are sorted based on the value of x and then the number of features is increased.

[0029] Furthermore, the parameters of the convolutional neural network for fault type classification include: accuracy, sensitivity, specificity, precision value and cross entropy.

[0030] Furthermore, the model makes judgments including:

[0031] Deploy the trained model to the actual crane, monitor sensor data in real time, and use the model for fault diagnosis;

[0032] The failure probability is determined based on the predicted situation. If the failure probability is repeatedly shown to be higher than the set threshold, a failure alarm is triggered and the operation and maintenance management system is connected to dispatch operation and maintenance tasks.

[0033] Furthermore, the operation and maintenance management system includes safety management, intrinsic safety management, and operation and maintenance process management; safety management is used to supervise the real-time monitoring of safety monitoring data, and also includes the monitoring of peripheral driving safety, to realize display, alarm, abnormal data output, etc.; intrinsic safety management is used to supervise the monitoring of key components; operation and maintenance process management is used to publish the maintenance plan, maintenance tasks, and standard situation module information for the predicted equipment conditions to relevant personnel.

[0034] According to an embodiment, an electronic device is also provided, which includes: one or more processors; a memory; one or more applications, wherein the one or more applications are stored in the memory and are configured to be executed by the one or more processors, and the one or more applications are configured to: execute the above-mentioned high-altitude work dangerous area detection method based on deep learning.

[0035] According to an embodiment, a computer-readable storage medium is also provided, which includes: a computer program that can be loaded by a processor and execute the above-mentioned high-altitude work dangerous area detection method based on deep learning.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] It can improve the efficiency and accuracy of operation and maintenance management. Through the training of machine learning models, the model can make more accurate fault alarms and classify fault alarms, and carry out targeted operation and maintenance management according to different types of fault alarms. It solves the problem that equipment failures cannot be effectively predicted in the operation and maintenance management process in the existing technology, and avoids the situation where equipment "operates with illness". BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a schematic diagram of an operation and maintenance management system according to an embodiment of the present invention; DETAILED DESCRIPTION

[0039] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0040] In an exemplary embodiment, Figure 1 As shown, this embodiment provides a crane operation and maintenance method based on machine learning, which includes:

[0041] Install sensor groups on the metal structure, electrical control components, and drive mechanisms of the crane to monitor the inherent safety of the crane and form data groups for each key part;

[0042] Real-time monitoring of the crane's status during use to ensure safety and form a data set of the use process;

[0043] Process the key parts data set and the usage process data set to extract feature data and use neural networks to build machine learning models and perform model training;

[0044] Real-time data is collected and input into the model for judgment. When the predicted probability is higher than a certain threshold, a fault alarm is issued. At the same time, a convolutional neural network is used to classify the fault alarms, and maintenance work is performed according to different types of fault alarms.

[0045] Specifically, the operation and maintenance management system includes usage safety management, intrinsic safety management, and operation and maintenance process management; usage safety management is used to supervise the real-time monitoring of usage safety monitoring data, and also includes the monitoring of peripheral driving safety, to realize display, alarm, abnormal data output, etc.; intrinsic safety management is used to supervise the monitoring of key components; operation and maintenance process management is used to publish the maintenance plan, maintenance tasks, and standard situation module information for the predicted equipment conditions to relevant personnel.

[0046] like Figure 1 As shown, in a further exemplary scheme, the intrinsic safety management collects data from the crane's steel structure, electrical components, wire ropes, brakes, reducers, wheels, motors, etc., and transmits the collected data to the system. The system provides display, storage, query and other functions for the data, and pushes abnormal items that have been judged to be faulty to the main system interface for display.

[0047] like Figure 1 As shown, in a further exemplary scheme, safety management is used to monitor the crane using video monitoring, thermal imaging monitoring, usage status, load status, wind speed status, track status, wheel pressure status, etc. The monitoring data also supports display, storage, query and other functions.

[0048] like Figure 1As shown, in a further exemplary scheme, operation and maintenance process management includes modules such as maintenance plan, maintenance tasks, maintenance management, and standard management; formulate periodic inspection and maintenance plans to ensure that all systems and equipment are inspected and maintained as necessary within the scheduled time, and use historical data and predictive analysis to optimize maintenance plans and reduce unexpected downtime; clarify the specific content, person in charge, required tools and materials of each maintenance task, track the progress of maintenance tasks, and ensure that they are completed on time and meet quality standards; establish a fault response mechanism to quickly respond to system or equipment failures, and record the maintenance process and results, including the cause of the failure, maintenance measures and subsequent preventive measures.

[0049] In a further exemplary embodiment, data processing includes data cleaning of a data group, wherein data cleaning includes filling in missing values, smoothing noise data, identifying or deleting one or more of outliers; data processing also includes adjusting the cleaned data into a format suitable for a machine learning algorithm by smoothing aggregation, data generalization, normalization, etc., combining data from multiple data groups and storing them uniformly, establishing a data warehouse, and performing data integration;

[0050] In a further exemplary embodiment, the data processing further comprises:

[0051] Extract the original data samples from the data warehouse, and perform feature extraction on the original data samples as original signal features;

[0052] Fault features, including time domain features, frequency domain features, and time-frequency domain features, are extracted from the public crane fault database for fault diagnosis.

[0053] In a further exemplary embodiment, the data processing further comprises:

[0054] The extracted features are screened and calculated as follows:

[0055] S1: Calculate the internal distance of feature x of class = y using the formula expressed in the equation,

[0056]

[0057] Where N is the number of samples, F is the number of features, the number of classes is represented by C, px, ycand-px, yn are the x features of C samples of the yth class and the x features of N samples of the yth class respectively;

[0058] S2: Calculate the average of the distances within class y for feature x, expressed as:

[0059]

[0060] S3: Calculate the average value of feature x for N samples in class y, expressed as:

[0061]

[0062] S4: Calculate the average distance between C classes of feature x using the equation:

[0063]

[0064] S5: Calculate the evaluation factor of feature x using the equation:

[0065]

[0066] The features are sorted based on the value of x and then the number of features is increased.

[0067] In a further exemplary embodiment, when the convolutional neural network performs classification, the initial network layer is responsible for sensing information from the fault, the convolutional filter is further responsible for finding fault features that can identify and distinguish faults with different faults, and the output layer provides the classification results obtained by the feature information obtained at the fully connected layer; the parameters of the convolutional neural network for fault type classification include: accuracy, sensitivity, specificity, precision value and cross entropy; wherein,

[0068] Accuracy (ACC): It defines the percentage of accurate correction results obtained among the total number of events;

[0069] Sensitivity (SEN.): The percentage of accurately indicated error category cases that actually have the error symptoms is expressed by sensitivity;

[0070] Specificity (SEP.): Specificity provides a higher probability of truly indicating a fault-free class without giving a false positive result;

[0071] Precision value: This value indicates the accurate result obtained correctly or precisely from the total number of positive predictions;

[0072] Cross Entropy: The cross entropy loss provides the correct classification ability of the neural network in the range of 0 to 1.

[0073] In a further exemplary embodiment, the model determines:

[0074] Deploy the trained model to the actual crane, monitor sensor data in real time, and use the model for fault diagnosis;

[0075] The failure probability is determined based on the predicted situation. If the failure probability is repeatedly shown to be higher than the set threshold, a failure alarm is triggered and the operation and maintenance management system is connected to dispatch operation and maintenance tasks.

[0076] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A crane operation and maintenance method based on machine learning, characterized in that: include: Install sensor groups on the metal structure, electrical control components, and drive mechanisms of the crane to monitor the inherent safety of the crane and form data groups for each key part; Real-time monitoring of the crane's status during use to ensure safety and form a data set of the use process; Process the key parts data set and the usage process data set to extract feature data and use neural networks to build machine learning models and perform model training; Real-time data is collected and input into the model for judgment. When the predicted probability is higher than a certain threshold, a fault alarm is issued. At the same time, a convolutional neural network is used to classify the fault alarms, and maintenance work is performed according to different types of fault alarms.

2. The crane operation and maintenance method based on machine learning according to claim 1, characterized in that: The data processing includes data cleaning of the data set, and data cleaning includes one or more of filling in missing values, smoothing noise data, and identifying or deleting outliers.

3. The crane operation and maintenance method based on machine learning according to claim 2, characterized in that: The data processing also includes adjusting the cleaned data into a format suitable for machine learning algorithms, combining and uniformly storing data in multiple data groups, establishing a data warehouse, and performing data integration.

4. The crane operation and maintenance method based on machine learning according to claim 3, characterized in that: The data processing also includes: Extract the original data samples from the data warehouse, and perform feature extraction on the original data samples as original signal features; Fault features, including time domain features, frequency domain features, and time-frequency domain features, are extracted from the public crane fault database for fault diagnosis.

5. The crane operation and maintenance method based on machine learning according to claim 4, characterized in that: The data processing also includes: The extracted features are screened and calculated as follows: S1: Calculate the internal distance of feature x of class = y using the formula expressed in the equation, Where N is the number of samples, F is the number of features, the number of classes is represented by C, px, ycand-px, yn are the x features of C samples of the yth class and the x features of N samples of the yth class respectively; S2: Calculate the average of the intra-class distances of the x feature, expressed as: S3: Calculate the average value of feature x for N samples in class y, expressed as: S4: Calculate the average distance between C classes of feature x using the equation: S5: Calculate the evaluation factor of feature x using the equation: The features are sorted based on the value of x and then the number of features is increased.

6. The crane operation and maintenance method based on machine learning according to any one of claims 1 to 5, characterized in that: The parameters of the convolutional neural network for fault type classification include: accuracy, sensitivity, specificity, precision value and cross entropy.

7. The crane operation and maintenance method based on machine learning according to claim 6, characterized in that: The model makes judgments including: Deploy the trained model to the actual crane, monitor sensor data in real time, and use the model for fault diagnosis; The failure probability is determined based on the predicted situation. If the failure probability is repeatedly shown to be higher than the set threshold, a failure alarm is triggered and the operation and maintenance management system is connected to dispatch operation and maintenance tasks.

8. The crane operation and maintenance method based on machine learning according to claim 7, characterized in that: The operation and maintenance management system includes safety management, intrinsic safety management, and operation and maintenance process management; the safety management is used to supervise the real-time monitoring of safety monitoring data, and also includes the monitoring of peripheral driving safety, to achieve display, alarm, abnormal data output, etc.; the intrinsic safety management is used to supervise the monitoring of key components; the operation and maintenance process management is used to publish the maintenance plan, maintenance tasks, and standard situation module information for the predicted equipment conditions to relevant personnel.

9. An electronic device, characterized in that: include: one or more processors; Memory; One or more applications, wherein one or more applications are stored in a memory and configured to be executed by one or more processors, and one or more applications are configured to: execute the high-altitude work dangerous area detection method based on deep learning according to any one of claims 1-8.

10. A computer-readable storage medium, characterized in that: include: A computer program is stored which can be loaded by a processor and execute a method for detecting dangerous areas in aerial work based on deep learning as described in any one of claims 1 to 8.

Images