Crown block fault prediction method and device, storage medium and electronic equipment

Abstract

The invention provides a crown block fault prediction method and device, a storage medium and electronic equipment, and belongs to the technical field of transportation control. The method comprises the following steps: acquiring a first driving feature set of a to-be-detected crown block; calling a preset fault prediction model to perform fault prediction on the to-be-detected crown block, and outputting a fault prediction score; when the fault prediction score is between the first score threshold value and the second score threshold value, positioning a target sky rail section of a first abnormal data subset generated by the to-be-detected crown block in the driving process; acquiring a second driving feature set, wherein the second driving feature set comprises a first abnormal data subset and a second driving feature subset generated by other crown blocks in the driving process of the target sky rail section; performing cross-device correlation analysis on the first abnormal data subset and the second driving feature subset, and calculating an abnormal score of the first abnormal data subset on the target sky track section; and determining whether the to-be-detected crown block has a fault risk based on the abnormal score. The method can improve the accuracy of crown block fault prediction.

Description

Technical Field

[0001] This application relates to the technical field of transportation control, and particularly to a method and device for predicting overhead hoist faults, a storage medium, and an electronic device. Background Art

[0002] In the field of semiconductor manufacturing, an overhead hoist (Overhead Hoist Transport, OHT, aerial transporter, hereinafter referred to as "overhead hoist") is an automated transportation device that runs on an elevated track and is mainly used for precise handling of wafers, raw materials, and finished products in a clean workshop. As a core component of the Automated Material Handling System (AMHS), the overhead hoist works in coordination with the track network, control system, and scheduling software to jointly achieve efficient and continuous material transfer in the semiconductor production process. AMHS aims to optimize the logistics efficiency of the production line, reduce manual intervention, and ensure process stability by integrating automated transportation equipment and intelligent management technology, and is an essential infrastructure for modern semiconductor factories.

[0003] The stability of the overhead hoist's operation is directly related to production efficiency and product quality. Currently, the industry generally adopts a regular inspection and maintenance strategy to ensure the reliability of the overhead hoist, such as checking key components such as the track system, drive motor, and sensors at fixed intervals. However, this maintenance mode has certain limitations: First of all, regular inspections rely on preset time intervals or historical experience and it is difficult to identify sudden abnormalities that may occur to the equipment during the interval between two inspections in real time. For example, during operation, the overhead hoist may experience performance degradation or local failures due to mechanical wear, load fluctuations, or environmental factors (such as temperature and humidity changes), and these dynamic problems may not be detected in time through regular inspections, ultimately leading to unplanned downtime and affecting production continuity.

[0004] Secondly, existing fault detection methods are mostly based on the analysis of a single type of data (such as vibration signals or current monitoring) and lack adaptability to the complex operating scenarios of the overhead hoist. The operating parameters of the overhead hoist vary significantly under different working conditions (such as fully loaded straight running on the sky track, empty turning, etc.), and single-dimensional monitoring is difficult to comprehensively cover the diverse state characteristics of the equipment, which may lead to missed detection or misjudgment of potential faults.

[0005] In addition, as a component of the factory logistics system, the operating state of the overhead hoist is closely related to factors such as the track environment and multi-device coordination. However, the existing technology often limits the analysis of the operating data of the overhead hoist to a single device or a local area, and fails to fully combine the dynamic association between the device and the environment, restricting the comprehensiveness and accuracy of fault prediction.

[0006] Generally speaking, the existing technologies still face problems such as insufficient timeliness in the prediction of overhead crane failures and limited multi-dimensional data integration capabilities. There is an urgent need for more efficient monitoring and analysis methods to improve the accuracy of equipment maintenance and ensure the efficient operation of semiconductor production. Summary of the Invention

[0007] The purpose of this application is to provide an overhead crane failure prediction method, device, storage medium, and electronic device to solve at least one of the above technical problems.

[0008] In the first aspect of this application, an overhead crane failure prediction method is provided. The method includes: Obtain the first driving feature set of the overhead crane to be tested; Based on the first driving feature set, call a preset failure prediction model to perform failure prediction on the overhead crane to be tested and output a failure prediction score; When the failure prediction score is between the first score threshold and the second score threshold, based on the first driving feature set, locate the target overhead track section where the first abnormal data subset is generated during the driving process of the overhead crane to be tested; Obtain a second driving feature set, where the second driving feature set includes the first abnormal data subset and a second driving feature subset generated by other overhead cranes during the driving process on the target overhead track section; Perform cross-device correlation analysis on the first abnormal data subset and the second driving feature subset, and calculate the abnormal score of the first abnormal data subset on the target overhead track section; Based on the abnormal score, determine whether there is a failure risk for the overhead crane to be tested.

[0009] Optionally, the step of based on the first driving feature set, calling a preset failure prediction model to perform failure prediction on the overhead crane to be tested and output a failure prediction score includes: According to the driving condition of the overhead crane and / or the distribution condition of the overhead track section, divide the first driving feature set into first driving feature subsets of multiple driving types; Call the failure prediction model to analyze each first driving feature subset of each driving type, and obtain the type score corresponding to each first driving feature subset; Calculate the failure prediction score based on the type scores.

[0010] Optionally, the second driving feature subset belongs to the same driving type as the first abnormal data subset; The driving condition of the overhead crane includes one or more of the load driving condition and the no-load driving condition, and the distribution condition of the overhead track section includes one or more of the straight overhead track section, the turning overhead track section, the ascending overhead track section, and the descending overhead track section.

[0011] Optionally, the fault prediction model is a multimodal analysis model based on the Transformer architecture, including a multimodal input embedding unit, an encoding processing unit, a feature fusion unit, and a decoding unit; Based on the first driving feature set, calling a preset fault prediction model to perform fault prediction on the overhead crane to be tested, and outputting a fault prediction score, including: Calling the multimodal input embedding unit to convert the operating parameters of different dimensions in the first driving feature set into high-dimensional feature vectors; Calling the encoding processing unit to perform encoding processing and feature analysis on the high-dimensional feature vectors to obtain an analysis result; Calling the feature fusion unit to fuse the analysis results to form fusion information; Calling the decoding unit to calculate a fault prediction score according to the fusion information.

[0012] Optionally, the calling the encoding processing unit to perform encoding processing and feature analysis on the high-dimensional feature vectors to obtain an analysis result includes: Calling the spatio-temporal encoding unit in the encoding processing unit to convert the high-dimensional feature vectors into spatio-temporal joint encoding vectors; Calling the self-attention encoding unit in the encoding processing unit to extract global features and local features from the spatio-temporal joint encoding vectors, and analyze the global features and local features to obtain an analysis result.

[0013] Optionally, the cross-device correlation analysis of the first abnormal data subset and the second driving feature subset, and calculating the abnormal score of the first abnormal data subset on the target overhead rail section includes: Calculating the correlation values between the first abnormal data subset and each second driving feature subset with the same driving type according to a preset correlation calculation model; Performing weighted summation on each calculated correlation value, and using the obtained weighted summation value as the abnormal score; Determining whether there is a fault risk for the overhead crane to be tested based on the abnormal score, including: when the abnormal score exceeds the third score threshold, determining that the overhead crane to be tested has a fault risk.

[0014] Optionally, the method further includes: when the fault prediction score is less than the first score threshold, determining that the overhead crane to be tested has no fault risk; When the fault prediction score is greater than the second score threshold, determining that the overhead crane to be tested has a fault risk.

[0015] In the second aspect of the present application, an overhead crane fault prediction device is provided, and the device includes: The first feature acquisition module is configured to acquire a first driving feature set of the overhead crane to be tested; The first scoring module is configured to, based on the first driving feature set, call a preset fault prediction model to perform fault prediction on the overhead crane to be tested and output a fault prediction score; The target track section positioning module is configured to, when the fault prediction score is between a first scoring threshold and a second scoring threshold, locate the target track section of the overhead crane to be tested where a first abnormal data subset is generated during the driving process based on the first driving feature set; The second feature acquisition module is configured to acquire a second driving feature set, where the second driving feature set includes the first abnormal data subset and a second driving feature subset generated by other overhead cranes during the driving process on the target track section; The second scoring module is configured to perform cross-device correlation analysis on the first abnormal data subset and the second driving feature subset, and calculate an abnormal score of the first abnormal data subset on the target track section; The fault prediction module is configured to determine whether there is a fault risk for the overhead crane to be tested based on the abnormal score.

[0016] In a third aspect of the present application, there is provided a computer-readable storage medium, on which executable instructions are stored, and when the executable instructions are executed by a processor, the processor is caused to execute the method as described in any one of the embodiments of the present application.

[0017] In a fourth aspect of the present application, there is provided an electronic device, including: one or more processors; a memory for storing one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors are caused to execute the method as described in any one of the embodiments of the present application.

[0018] In the overhead crane fault prediction method, device, storage medium and electronic device of the present application, by performing fault prediction according to the first driving feature set of the overhead crane to be tested, when the fault prediction score based on the first driving feature set is between a first scoring threshold and a second scoring threshold, a first abnormal data subset that can reflect the abnormal situation of the overhead crane to be tested is further determined from the first driving feature set, and the target track section corresponding to the first abnormal data subset is located. Then, in combination with the first abnormal data subset and the second driving feature subset of other overhead cranes when driving on the target track section, correlation analysis is performed on the two to obtain an abnormal score; and finally, it is determined whether there is a fault risk for the overhead crane to be tested. The present application locates the target track section where the overhead crane to be tested is operating (the track section where the overhead crane travels that can reflect the abnormal situation of the overhead crane to be tested), and incorporates factors such as multi-device collaborative operation (reflected by the driving data of other overhead cranes on this track section) into the analysis scope, so as to achieve more comprehensive and accurate fault prediction. Description of the Drawings

[0019] To more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present application and should not be regarded as limiting the scope of the present application.

[0020] Figure 1 It is a schematic diagram of the scenario of the overhead crane fault prediction method in one embodiment; Figure 2 It is a schematic flowchart of the overhead crane fault prediction method in one embodiment; Figure 3 It is a schematic flowchart of calling a preset fault prediction model to perform fault prediction on a to-be-tested overhead crane based on a first driving feature set and outputting a fault prediction score in one embodiment; Figure 4 It is a schematic flowchart of calling a preset fault prediction model to perform fault prediction on a to-be-tested overhead crane based on a first driving feature set and outputting a fault prediction score in another embodiment; Figure 5 It is a schematic structural diagram of the overhead crane fault prediction device in one embodiment; Figure 6 It is a schematic structural diagram of an electronic device in one embodiment. Detailed Embodiments

[0021] In order to make the objectives, technical solutions and advantages of the present application more clear and understandable, the present application will be further described in detail below in conjunction with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0022] All terms used in the present application (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0023] For example, terms such as "first" and "second" used in the present application may be used herein to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from another element.

[0024] For another example, terms such as "including" and "comprising" used in the present application indicate the presence of features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0025] The overhead crane fault prediction method in the present application can be applied to, for example, Figure 1in the scenario shown. In combination with Figure 1 As shown, in the AMHS, a large number of overhead cranes 120 travel on the overhead rail to deliver items (such as wafer boxes) and deliver the items to the corresponding destinations. Among them, the overhead rail can be divided into a straight overhead rail section 110-A and a turning overhead rail section 110-B according to the distribution, and can be further divided into an ascending overhead rail section and a descending overhead rail section, etc. The overhead crane 120 can include unloaded travel and loaded travel on the overhead rail. Corresponding sensing devices are distributed on both the overhead rail and / or the overhead crane, which can real-time locate the position of the overhead crane on the overhead rail and measure relevant travel data such as the travel speed, power consumption, vibration, etc. of the overhead crane at each position on the overhead rail.

[0026] In one embodiment, a method for predicting overhead crane faults is provided, and this method can be applied to Figure 1 the scenario shown. As Figure 2 shown, this method includes: Step 210, obtaining a first set of travel characteristics of the overhead crane to be tested.

[0027] In this embodiment, the first set of travel characteristics is a set of all or part of the travel characteristics generated during the travel of the overhead crane to be tested within a preset time period. The preset time period can be from the last fault detection to the present, or can be any suitable duration such as the most recent one day, two days, one week, etc.

[0028] These travel characteristics include one or more of the travel speed, power consumption, vibration, friction coefficient, load state, etc. when the overhead crane travels at each position on the overhead rail. The power consumption can be one or more of the power, current, etc. of the overhead crane.

[0029] One or more detection devices such as a vibration sensor, a speed sensor, an acceleration sensor, a temperature sensor, a current sensor, an optical encoder, a laser rangefinder, etc. can be integrated on the overhead crane. Each detection device can collect the above-mentioned travel data such as vibration data, travel speed, power consumption, and friction coefficient according to the corresponding preset acquisition frequency, and these travel data can be transmitted to the electronic device in the AMHS. The electronic device can specifically be a relevant server. For the received relevant travel data within the preset time period, it can be segmented according to the preset sliding time window to form initial travel characteristics, and these initial travel characteristics are preprocessed such as normalized, and the preprocessed initial travel characteristics are aggregated and summarized to form a first set of travel characteristics.

[0030] Among them, the time length of the sliding time window can be a fixed length or a variable length. For example, it can be default set to default lengths such as 3 seconds and 5 seconds. It can also be adjusted in real time according to one or more of the distribution status of the delivery status, driving speed of the overhead crane, and the distribution status of the overhead track. For example, the basic window length is set to 5 seconds. When the driving speed of the overhead crane > 2m / s, the window is shortened to 3 seconds to capture rapid changes; when the speed < 0.5m / s (such as the start-stop stage), the window is extended to 10 seconds to improve data stability. By adjusting the time length of each sliding time window in real time, the amount of data in each piece of driving data is kept consistent.

[0031] For these initial driving data, data synchronization and alignment such as timestamp calibration and spatial coordinate matching can be performed to ensure that each data point corresponds to an accurate spatial position (such as track segment number, three-dimensional coordinates). And through noise filtering, signal enhancement, normalization processing, missing data processing, outlier removal, etc., the initial driving features after preprocessing are obtained.

[0032] Step 220, based on the first driving feature set, call a preset fault prediction model to perform fault prediction on the overhead crane to be measured, and output a fault prediction score.

[0033] In this embodiment, the fault prediction model can be a pre-trained model for overhead crane fault prediction. Specifically, the model can be a combination of one or more models such as support vector machine, random forest, long short-term memory network, and Transformer architecture. The fault prediction model analyzes and calculates the input first driving feature set, and according to its internal algorithm and trained parameters, outputs a fault prediction score.

[0034] The fault prediction score is a numerical index used to quantify the possibility of the overhead crane having a fault. The larger the value. This score is based on various driving feature data of the overhead crane, such as speed, acceleration, vibration, temperature, power consumption, load, etc., and combines the data patterns and rules learned by the fault prediction model in the normal operation and fault states to evaluate the proximity of the current operating state of the overhead crane to the fault state.

[0035] The range of the score can be any suitable numerical range set, for example, 0 - 100 points. The higher the score, the greater the possibility of the overhead crane having a fault. Through the fault prediction score, the staff can quickly understand the potential fault risk degree of the overhead crane, so as to take corresponding measures, such as further inspection, maintenance, or preparing repair spare parts in advance, etc., so as to ensure the safe operation of the overhead crane and reduce production interruptions and safety accidents caused by faults.

[0036] Specifically, the electronic device can iteratively train the set model with the pre-collected driving feature sets of normal overhead cranes (negative samples) and the driving feature sets of faulty overhead cranes (positive samples), and finally form a trained fault prediction model. The fault prediction model can output a fault prediction score reflecting the probability of the overhead crane having a fault, as well as one or more pieces of driving feature data (i.e., abnormal data) in the first driving feature set that best reflect the abnormality of the overhead crane.

[0037] Taking the fault prediction model as a multi-modal analysis model based on the Transformer architecture as an example, the model can include a multi-modal input embedding unit, an encoding processing unit, and a decoding unit.

[0038] The multi-modal input embedding layer is used to convert each piece of driving feature data into a feature vector through one-dimensional convolution and LSTM; the encoding processing unit can perform temporal encoding and spatial encoding on the formed feature vectors, and perform feature analysis to obtain an analysis result; the decoding unit can encode the analysis result to finally obtain the corresponding fault prediction score.

[0039] Step 230, when the fault prediction score is between the first score threshold and the second score threshold, based on the first driving feature set, locate the target track section where the first abnormal data subset is generated during the driving of the overhead crane to be tested.

[0040] In this embodiment, the electronic device has preset the first score threshold and the second score threshold. The first score threshold and the second score threshold can be values custom-set according to historical fault data, or values manually set based on operating experience. The first score threshold is less than the second score threshold. When the fault prediction score of the overhead crane to be tested is within the interval formed by the first score threshold and the second score threshold, it indicates that the overhead crane to be tested may have a potential fault risk and further analysis is required.

[0041] Specifically, the first score threshold and the second score threshold are key values used to define the interval in which the overhead crane fault prediction score is located, so as to judge the degree of potential fault risk of the overhead crane. The first score threshold is a relatively low score limit, such as 40 points. When the fault prediction score is higher than this threshold, it means that the operating state of the overhead crane begins to deviate from the normal range and enters the area where there may be a potential fault risk. It is a warning signal, reminding the staff that the overhead crane has shown some signs worthy of attention and further evaluation and analysis are needed, but it does not mean that the overhead crane must have a serious fault.

[0042] The second scoring threshold is a relatively high score boundary, for example, set at 70 points. When the fault prediction score is lower than this threshold, it indicates that although the overhead crane has potential risks, it has not reached the high-risk state where a fault is very likely to occur. If the score is between the first scoring threshold and the second scoring threshold, for example, when the output fault prediction score is 60 points, it means that the overhead crane is in an intermediate fuzzy zone, and subsequent steps such as locating the target track section of the overhead crane, obtaining and analyzing the second set of driving characteristics, etc. are needed to further accurately judge its fault risk to determine whether maintenance and other measures are required.

[0043] When in this interval, the fault prediction model further outputs one or more driving characteristics in the first set of driving characteristics that can best reflect the suspected faults of the overhead crane to be tested. The set composed of the one or more output driving characteristics (i.e., abnormal data) is the first abnormal data subset described below.

[0044] For each piece of abnormal data output, the corresponding track section of the overhead crane (i.e., the target track section) can be determined by combining the time stamp and / or position data corresponding to this data. The first set of driving characteristics is divided according to the driving track and time of the overhead crane, and corresponding abnormal data is located by means of data analysis such as outlier detection. The track sections of the overhead crane associated with these abnormal data become the target track sections. For example, when it is located that a piece of abnormal data is generated when the overhead crane travels to track section A–5, then track section A-5 is identified as the target track section. After determining the target track section, further analysis can be performed on the abnormal data on this track section of the overhead crane, and combined with the driving data of other overhead cranes on this section, it provides a basis for accurately judging the fault risk of the overhead crane.

[0045] In one embodiment, when the fault prediction score is less than the first scoring threshold, it is determined that the overhead crane to be tested has no fault risk; when the fault prediction score is greater than the second scoring threshold, it is determined that the overhead crane to be tested has a fault risk.

[0046] Specifically, if the output fault prediction score is less than the first scoring threshold, it means that the risk probability of the overhead crane to be tested having a fault is relatively low, and it can be considered that it has no fault risk. If it is higher than the second scoring threshold, it means that the risk probability of the overhead crane to be tested having a fault is very high, and it can be directly determined that it has a fault risk.

[0047] Step 240, obtain the second set of driving characteristics.

[0048] In this embodiment, the second set of driving characteristics includes the first abnormal data subset and the second driving characteristic subset generated during the driving process of other overhead cranes on the target track section.

[0049] For the determined target overhead rail section, the electronic device can obtain the driving characteristic data generated by other overhead cranes in the AHSM when driving in the target overhead rail section, aggregate and summarize them to form a second driving characteristic subset, and the second driving characteristic subset and the first abnormal data subset constitute the second driving characteristic set. Optionally, the data in the second driving characteristic subset and the data in the first driving characteristic set are both generated when the overhead crane is driving on the target overhead rail section, and the generation time of the two belongs to the same time period.

[0050] After locating the target overhead rail section, the corresponding second time period is further determined according to the time corresponding to the driving characteristic data in the first abnormal data subset, and the driving characteristic data of other overhead cranes driving under the target overhead rail section in the second time period are obtained and aggregated into a second driving characteristic subset.

[0051] Specifically, for each piece of abnormal data, the corresponding target overhead rail segment and the corresponding timestamp are determined, and the corresponding second time period is determined based on the timestamp, and the driving characteristic data of other overhead cranes except the overhead crane to be tested that travel under the target overhead rail segment in the second time period are obtained. After the driving characteristic data of other overhead cranes corresponding to each piece of abnormal data are obtained, they are aggregated to form a second driving characteristic subset.

[0052] Furthermore, the load conditions of other overhead cranes may be considered so that the load conditions of other overhead cranes are consistent with the load conditions of the overhead crane to be tested. For example, if one of the abnormal data of the overhead crane to be tested is generated when it is running unloaded on the overhead rail section A-1 in a certain time period, then the driving characteristic data generated by other overhead cranes running unloaded on the overhead rail section A-1 in a second time period may be obtained and used as the second driving characteristic data corresponding to the abnormal data.

[0053] Step 250 , performing cross-device correlation analysis on the first abnormal data subset and the second driving feature subset, and calculating an abnormality score of the first abnormal data subset on the target sky track segment.

[0054] In this embodiment, the correlation between the driving feature data in the first abnormal data subset and the driving feature data in the second driving feature subset can be analyzed by a relevant correlation calculation model or a cluster analysis model. The correlation is used to reflect the degree of difference between the two. The smaller the correlation, the greater the difference between the two.

[0055] When in the same time period and the same overhead rail section, the greater the difference between the first abnormal data subset and the second driving characteristic subset, the greater the difference in driving data between the overhead crane to be tested and other overhead cranes, which indicates that the possibility of abnormality in the overhead crane to be tested is greater.

[0056] Optionally, the driving characteristic data of other overhead cranes corresponding to each piece of abnormal data can be subjected to correlation analysis, and the corresponding abnormal score can be calculated according to each correlation analysis. Optionally, each correlation analysis can calculate a corresponding correlation value, and the corresponding abnormal score can be finally calculated based on each correlation value.

[0057] Step 260, determine whether there is a risk of failure for the overhead crane to be tested based on the abnormal score.

[0058] In this embodiment, after obtaining the abnormal score and the fault prediction score, the two can be combined to comprehensively determine whether there is a risk of failure for the overhead crane to be tested. For example, when the fault prediction score is within the above range, if the abnormal score is higher than a certain value, it can be determined that there is a risk of failure for the overhead crane to be tested; if the abnormal score is lower than a certain value, it is determined that there is no risk of failure for the overhead crane to be tested.

[0059] In the overhead crane fault prediction method of the present application, by obtaining the first driving characteristic set of the overhead crane to be tested in real time, covering multi-dimensional real-time data such as speed, acceleration, vibration, and temperature, the running state of the overhead crane can be continuously monitored. Based on these real-time data, a preset fault prediction model is called for fault prediction and a fault prediction score is output, which can quickly determine whether there is a potential risk of failure for the overhead crane at present. Compared with traditional regular maintenance, this method can detect abnormalities at an early stage before the occurrence of a fault, greatly improving the timeliness of fault prediction, effectively avoiding unplanned downtime caused by failure to detect potential faults in time, and strongly guaranteeing the continuity of production.

[0060] Secondly, when obtaining the first driving characteristic set in the present application, measurement data of multiple dimensions are integrated, comprehensively reflecting the running state of the overhead crane under different working conditions. For example, when the overhead crane is driving straight with full load or turning with no load, parameters such as speed and vibration will change significantly, and these multi-dimensional data are all included in the first driving characteristic set. When the fault prediction score is within a specific range, a second driving characteristic set is further obtained, including the first abnormal data subset of the overhead crane to be tested and the second driving characteristic subset generated when other overhead cranes travel on the target track section, and cross-device correlation analysis is performed on them. This deep integration and analysis of multi-dimensional data can comprehensively cover the diverse state characteristics of the overhead crane, greatly reducing the missed detection or misjudgment of potential faults, and significantly improving the accuracy of fault prediction.

[0061] In addition, when determining the target overhead track section, by comprehensively considering the abnormal data generated when the overhead crane to be measured travels in this overhead track section and the travel data of other overhead cranes in the same target overhead track section, factors such as the track environment (such as the characteristics of different overhead track sections) where the overhead crane to be measured operates and the collaborative operation of multiple devices (reflected by the travel data of other overhead cranes in this overhead track section) are incorporated into the analysis scope, thereby achieving more comprehensive and accurate fault prediction and effectively overcoming the deficiencies of the prior art in the dynamic correlation analysis of equipment and environment.

[0062] In one embodiment, as Figure 3 shown, step 220 includes: Step 310, dividing the first travel feature set into first travel feature subsets of multiple travel types according to the travel condition of the overhead crane and / or the distribution condition of the overhead track section.

[0063] In this embodiment, during the operation of the overhead crane, there are different travel states such as full-load travel, no-load travel, accelerating travel, decelerating travel, high-speed travel, and low-speed travel; the overhead track section can be divided into types such as straight overhead track section, turning overhead track section, ascending overhead track section, and descending overhead track section according to shape and function. Through the position sensor of the overhead crane and the map information of the overhead track system, the type of the overhead track section where the overhead crane is located is determined; through the speed sensor and / or weight sensor of the overhead crane, the type of the travel state of the overhead crane can be determined.

[0064] For example, when the speed is between 0 - 1 m / s and the weight sensor shows that the load reaches 80% or more of the rated load of the overhead crane, it is determined as the full-load low-speed travel state; when the speed is between 1 - 3 m / s and the load is less than 20% of the rated load, it is determined as the no-load medium-speed travel state. When the position sensor data combined with the map information shows that the overhead crane travels on a track with no curvature change, it is determined to be in a straight overhead track section; when it is detected that the travel direction of the overhead crane changes continuously by an angle, it is determined to be in a turning overhead track section. Based on these conditions, multiple different travel types can be divided, and each travel type corresponds to a condition (one travel type corresponds to the same travel condition of the overhead crane and the distribution condition of the overhead track section).

[0065] In one embodiment, the traveling condition of the overhead crane includes one or more of the loaded traveling condition and the unloaded traveling condition, and the distribution condition of the overhead rail section includes one or more of the straight overhead rail section, the turning overhead rail section, the ascending overhead rail section, and the descending overhead rail section. For example, from the perspective of the traveling condition of the overhead crane, it can be divided into 2 cases such as the loaded traveling condition and the unloaded traveling condition. From the perspective of the distribution condition of the overhead rail section, it can be divided into 4 cases such as the straight overhead rail section, the turning overhead rail section, the ascending overhead rail section, and the descending overhead rail section. Based on the combination of the above two-dimensional divisions, a total of 8 traveling types can be obtained. For example, several of the traveling types are the loaded traveling type under the straight overhead rail section, the unloaded traveling type under the straight overhead rail section, the loaded traveling type under the turning overhead rail section, the unloaded traveling type under the turning overhead rail section, the loaded traveling type under the ascending overhead rail section, the unloaded traveling type under the ascending overhead rail section, the loaded traveling type under the descending overhead rail section, and the unloaded traveling type under the descending overhead rail section.

[0066] The traveling state type can be divided according to the traveling data generated during a complete delivery process of the overhead crane. A complete delivery process can be the process of the overhead crane delivering the goods from the picking point to the destination and then traveling unloaded to the next picking point. For the real-time acquired traveling data, according to the traveling path and the load condition, the traveling data generated during a complete delivery process can be respectively divided into one or more of the above 8 traveling types. For the traveling data of each traveling type, it is cut according to the corresponding sliding time window to form multiple initial traveling data.

[0067] Correspondingly, the second subset of traveling characteristics and the first subset of abnormal data belong to the same traveling type.

[0068] For example, when the traveling type to which a certain piece of abnormal data in the first subset of abnormal data belongs is the loaded traveling type under the straight overhead rail section, then find out the traveling characteristic data of other overhead cranes in the load state on the target overhead rail section (the target overhead rail section belongs to the straight overhead rail section) during the corresponding time period, and use these data as the first subset of abnormal data corresponding to this abnormal data.

[0069] By combining the traveling condition of the overhead crane and / or the distribution condition of the overhead rail section, the data in the first set of traveling characteristics corresponding to different overhead rail section types and / or different traveling conditions are divided to form a first subset of traveling characteristics based on the distribution condition of the overhead rail section. Each first subset of traveling characteristics corresponds to one traveling type. The traveling characteristic data in each first subset of traveling characteristics can also be divided according to the above-mentioned sliding time window, and each piece of traveling data divided can simultaneously include the relevant data such as the traveling speed, power consumption, and vibration mentioned above.

[0070] Step 320: Invoke the fault prediction model to analyze each first driving feature subset of each driving type, and obtain the type score corresponding to each first driving feature subset.

[0071] In this embodiment, the fault prediction model can independently analyze each first driving feature subset of each driving type to obtain the type score corresponding to one driving type.

[0072] Step 330: Calculate the fault prediction score based on the type score.

[0073] For the type scores of each driving type, weighted summation can be performed according to the preset type score weights, and finally the corresponding fault prediction score can be calculated. Among them, the weights of each driving type can be the same or different, and the electronic device can set the weights according to the actual situation.

[0074] In one embodiment, the fault prediction model is a multi-modal analysis model based on the Transformer architecture, including a multi-modal input embedding unit, an encoding processing unit, a feature fusion unit, and a decoding unit. As Figure 4 shown, step 220 includes: Step 410: Invoke the multi-modal input embedding unit to convert the operating parameters of different dimensions in the first driving feature set into high-dimensional feature vectors.

[0075] In this embodiment, the first driving feature set includes one or more types of operating parameters such as vibration data, driving speed, power consumption, friction, load, temperature, etc. For these multiple types of operating parameters, the multi-modal input embedding unit uses a specific embedding function to map each type of operating parameter into a vector space of a corresponding dimension. For example, for speed data, each speed value is mapped into a 100-dimensional vector space to carry more semantic information. Suppose the speed value 1.2 is converted into a vector [0.23, 0.12, -0.05, …, 0.08] (example of a 100-dimensional vector) after being converted by the embedding function. Similarly, for data such as acceleration, vibration, and temperature, similar embedding operations are performed to convert the operating parameters of different dimensions into high-dimensional feature vectors for subsequent processing.

[0076] Step 420: Invoke the encoding processing unit to perform encoding processing and feature analysis on the high-dimensional feature vectors to obtain the analysis result.

[0077] The encoding processing unit is based on the encoder module in the Transformer architecture. It consists of multiple multi-head attention mechanisms and feed-forward neural network layers. Taking the high-dimensional feature vector sequence of speed data as an example, the multi-head attention mechanism can simultaneously focus on vector information at different positions, capturing the global and local features of speed changes. Suppose the speed vector sequence is [V1, V2, …, V10] (each Vi is a 100-dimensional vector). Through the multi-head attention mechanism, different heads focus on different vector relationships at different positions. For example, head 1 focuses on the change trend of adjacent vectors, and head 2 focuses on the difference between vectors separated by two positions, etc. The data processed by the multi-head attention mechanism then undergoes a non-linear transformation through the feed-forward neural network layer to further extract features. For example, the feed-forward neural network layer performs a series of linear transformations and activation function operations (such as the ReLU function) on the speed vectors processed by the attention mechanism to obtain a new feature representation. Similarly, similar encoding processing and feature analysis are performed on the high-dimensional feature vectors of other dimensional data such as vibration, power consumption, friction, load, and temperature, and finally, the analysis results of each dimensional data are obtained.

[0078] Specifically, the spatio-temporal encoding unit in the encoding processing unit is called to convert the high-dimensional feature vector into a spatio-temporal joint encoding vector; the self-attention encoding unit in the encoding processing unit is called to extract the global feature and local feature from the spatio-temporal joint encoding vector, and the global feature and local feature are analyzed to obtain the analysis result.

[0079] The spatio-temporal encoding unit uses a preset specific encoding algorithm. For example, it can construct a spatio-temporal encoding structure based on the idea of a convolutional neural network. For the speed vector sequence, each vector at each time step (for example, speed data is collected once per second, corresponding to one time step) is regarded as a "position" in space. By designing a two-dimensional convolutional kernel, the width of which corresponds to the dimension of the high-dimensional feature vector (100 dimensions), and the length corresponds to the number of time steps (here are 10 time steps). The convolutional kernel slides on the speed vector sequence and performs a convolutional operation on the vector at each time step. Suppose the convolutional kernel weight is K and the bias is b. For the vector Vi at the i-th time step, a new feature vector Si is obtained after the convolutional operation. For example, taking the ReLU function as the activation function, then Si = ReLU (K × Vi + b). This new feature vector Si not only integrates the features of the speed data itself at this time step but also contains the correlation information of the speed data at adjacent time steps, realizing the information integration in the time dimension. Meanwhile, considering the position information of the overhead crane on the overhead rail (assuming the overhead rail is divided into 10 different regions, and the position of the overhead crane is obtained in real time through a position sensor), the position information can also be encoded into the feature vector. For example, convert the number of the overhead rail region where the overhead crane is currently located into a 10-dimensional one-hot encoded vector P, and then concatenate P with Si to obtain the final spatio-temporal joint encoding vector Ti = [Si, P]. For the high-dimensional feature vectors of other dimensional data, a similar method is also used for spatio-temporal encoding to obtain their respective spatio-temporal joint encoding vectors.

[0080] The self-attention encoding unit is based on the self-attention mechanism in the Transformer architecture. The self-attention mechanism determines the degree of attention of each vector to other vectors by calculating the attention scores between different position vectors. Taking the spatio-temporal joint encoding vector sequence [T1, T2, …, T10] of speed data as an example, the specific calculation process is as follows: First, pass each spatio-temporal joint encoding vector Ti through three linear transformation matrices Wq, Wk, and Wv respectively to obtain the query vector Qi = Ti × Wq, the key vector Ki = Ti × Wk, and the value vector Vi = Ti × Wv. Then, calculate the attention scores. For any two vectors Ti and Tj, the attention score eij = (Qi × Kj^T) / sqrt(d), where d is the dimension of the query vector and the key vector (assumed to be 50 dimensions here), and Kj^T represents the transpose of Kj. The attention score eij reflects the degree of attention of vector Ti to vector Tj. Next, perform Softmax normalization on the attention scores to obtain the normalized attention score hij = Softmax(eij). Finally, calculate the output vector according to the normalized attention scores. For the i-th vector Ti, its output vector O i = Σ(h ij × Vj), j = 1, 2, …… 10. In this way, the self-attention mechanism can capture the global features of speed data at different time steps and overhead rail positions. To extract local features, the scope of attention can be restricted on the basis of the self-attention mechanism. For example, only calculate the attention scores between adjacent two time-step vectors, that is, for the i-th vector Ti, only calculate eij (j = i - 1, i, i + 1), and then calculate the local feature vector Li according to the above steps. After obtaining the global feature vector Oi and the local feature vector Li, further analysis is performed on them. For example, through a fully connected neural network layer, the global feature vector Oi and the local feature vector Li can be concatenated as the input, and after a series of linear transformations and activation function operations (such as the ReLU function), the analysis results regarding the speed data are obtained. For the spatio-temporal joint encoding vectors of other dimensional data, the same self-attention encoding unit operation is also adopted to extract global features and local features and perform analysis. Finally, the analysis results of each dimensional data are obtained, and these analysis results will be used as the input of the subsequent feature fusion unit for further calculating the fault prediction score.

[0081] Step 430: Invoke the feature fusion unit to fuse the analysis results to form fusion information.

[0082] The feature fusion unit fuses the analysis results of different dimensional data obtained by the encoding processing unit. Optionally, the fusion is performed by concatenation. These feature vectors are concatenated in sequence into a new vector F, and its dimension can be the sum of the dimensions of each dimensional data.

[0083] By performing feature fusion, it is possible to comprehensively integrate the overhead crane operation state information contained in different dimensional data, form more comprehensive fusion information, and provide a basis for accurately calculating the fault prediction score subsequently.

[0084] Step 440: Invoke the decoding unit to calculate the fault prediction score based on the fusion information.

[0085] The decoding unit is also based on the decoder module in the Transformer architecture. It receives the fusion information vector F output by the feature fusion unit. The decoding unit contains a fully connected neural network layer inside. The fusion information vector F is input into the fully connected neural network layer, and this network layer performs linear transformations and activation function operations (such as the Sigmoid function) through a series of weight matrices and bias terms. Assume that the weight matrix W and the bias term b obtained after the fully connected neural network layer is trained. After operations (such as score = Sigmoid (W × F + b)), a fault prediction score is finally output. For example, the calculated fault prediction score is 60 (assuming that a full score of 100 indicates the highest fault possibility and 0 indicates no fault). This score reflects the degree of the possibility of a fault in the current operation state of the overhead crane and is used for subsequent judgments on whether the overhead crane needs further inspection and maintenance, etc.

[0086] In one embodiment, when the first set of driving characteristics is divided into multiple first subsets of driving characteristics, each first subset of driving characteristics can be used as the input of the multimodal analysis model based on the Transformer architecture. According to the process from step 410 to step 440 above, each first subset of driving characteristics is analyzed, and the type score corresponding to each first subset of driving characteristics is output in step 440. Then, by combining each type score, the fault prediction score is finally calculated. Details are not described herein again.

[0087] In one embodiment, step 250 includes: calculating the correlation value between the first subset of abnormal data and each second subset of driving characteristics with the same driving type according to a preset correlation calculation model; performing weighted summation on each calculated correlation value, and using the obtained value after weighted summation as the abnormality score.

[0088] In this embodiment, for each piece of abnormal data, the corresponding second subset of driving characteristics is determined, and the correlation value between it and the corresponding second subset of driving characteristics is calculated. After calculating the correlation value corresponding to each piece of abnormal data, weighted summation is performed according to a preset weight, and the obtained value after summation is used as the corresponding abnormality score.

[0089] Specifically, the calculation of the correlation value can be performed using the Pearson correlation coefficient calculation model. Taking the data of the load driving type under a certain straight trolley track section as an example, assume that a piece of abnormal data in the first subset of abnormal data is a sequence A = [a1, a2,..., an], and its corresponding second subset of driving characteristics is the sequence B = [b1, b2,..., bn]. The calculation formula of the Pearson correlation coefficient is

[0090] where is the average value of the sequence A, is the average value of the sequence B. The correlation value r between the abnormal data and the corresponding second subset of driving characteristics is calculated through this formula. Weighted summation is performed on the correlation value r corresponding to each piece of abnormal data, and the final abnormality score can be obtained.

[0091] The weight for the correlation value of each driving type can be set according to the importance of different driving types during the operation of the overhead crane and the analysis of historical fault data. For example, when the overhead crane is fully loaded, the load is relatively large and the risk of failure is relatively high. Therefore, the weights for load and turning section driving are set relatively high; the weights for no-load and straight section driving are set relatively low.

[0092] In one embodiment, when the lengths of sequence A and sequence B are different, the Dynamic Time Warping (DTW) algorithm can be used to calculate the correlation value. Through the DTW algorithm, the two sequences are elastically matched on the time axis to find the optimal alignment path between the two sequences, so that the sum of the distances between the two sequences under this path is minimized. In this way, the similarity or correlation between them can be measured without requiring the sequences to have equal lengths.

[0093] Specifically, a two-dimensional matrix D can be created first, with a size of m×n, where m is the length of sequence A and n is the length of sequence B. Each element D(i,j) of the matrix represents the distance between the i-th element of sequence A and the j-th element of sequence B. Usually, the Euclidean distance or Manhattan distance is used to calculate the distance between two elements.

[0094] Dynamic programming to calculate the shortest path: Starting from the upper left corner D(1,1) of the matrix, the shortest path to the lower right corner D(m,n) is calculated by the method of dynamic programming. For each position D(i,j), its value is calculated by the following formula: where d(i,j) is the distance between the i-th element of sequence A and the j-th element of sequence B, means taking the minimum value among the adjacent three positions. In this way, by continuously comparing and selecting the minimum distance, the shortest path from D(1,1) to D(m,n) can be found.

[0095] The sum of the distances on the shortest path is the DTW distance between sequence A and sequence B. To obtain the correlation value, the DTW distance can be normalized, for example, using the formula , so that the obtained correlation value ranges from 0 to 1, and the larger the value, the more similar or correlated the two sequences are.

[0096] In one embodiment, step 260 includes: when the anomaly score exceeds the third scoring threshold, it is determined that the overhead crane to be tested has a risk of failure.

[0097] The third scoring threshold can be a suitable value set according to relevant experience. For example, the third scoring threshold can be set to 0.5. When the calculated anomaly score S = 0.39 is less than the third scoring threshold 0.5, it is determined that the overhead crane to be tested currently has no risk of failure. If the anomaly score exceeds the third scoring threshold, it is determined that the overhead crane to be tested has a risk of failure. At this time, it is necessary to arrange maintenance personnel to inspect and maintain the overhead crane in time to avoid the occurrence of failures.

[0098] In one embodiment, as Figure 5 shown, an overhead crane fault prediction device is provided, and the device includes: The first feature acquisition module 510 is configured to acquire a first driving feature set of the overhead crane to be measured.

[0099] The first scoring module 520 is configured to, based on the first driving feature set, call a preset fault prediction model to perform fault prediction on the overhead crane to be measured, and output a fault prediction score.

[0100] The target track section positioning module 530 is configured to, when the fault prediction score is between a first scoring threshold and a second scoring threshold, based on the first driving feature set, locate the target track section of the overhead crane to be measured where the first abnormal data subset is generated during driving.

[0101] The second feature acquisition module 540 is configured to acquire a second driving feature set, where the second driving feature set includes a first abnormal data subset and a second driving feature subset generated by other overhead cranes during driving on the target track section.

[0102] The second scoring module 550 is configured to perform cross-device correlation analysis on the first abnormal data subset and the second driving feature subset, and calculate an abnormal score of the first abnormal data subset on the target track section.

[0103] The fault prediction module 560 is configured to determine whether there is a fault risk for the overhead crane to be measured based on the abnormal score.

[0104] In one embodiment, the first scoring module 520 is further configured to divide the first driving feature set into first driving feature subsets of multiple driving types according to the driving condition of the overhead crane and / or the distribution condition of the track sections; call the fault prediction model to analyze each first driving feature subset of each driving type, and obtain a type score corresponding to each first driving feature subset; calculate the fault prediction score based on the type scores.

[0105] In one embodiment, the second driving feature subset and the first abnormal data subset belong to the same driving type.

[0106] In one embodiment, the driving condition of the overhead crane includes one or more of a loaded driving condition and an unloaded driving condition, and the distribution condition of the track sections includes one or more of a straight track section, a turning track section, an ascending track section, and a descending track section.

[0107] In one embodiment, the fault prediction model is a multi-modal analysis model based on the Transformer architecture, including a multi-modal input embedding unit, an encoding processing unit, a feature fusion unit, and a decoding unit.

[0108] The first scoring module 520 is further configured to call the multi-modal input embedding unit to convert the operating parameters of different dimensions in the first driving feature set into high-dimensional feature vectors; call the encoding processing unit to perform encoding processing and feature analysis on the high-dimensional feature vectors to obtain an analysis result; call the feature fusion unit to fuse the analysis result to form fusion information; call the decoding unit to calculate a fault prediction score according to the fusion information.

[0109] In one embodiment, the first scoring module 520 is further configured to call the spatio-temporal encoding unit in the encoding processing unit to convert the high-dimensional feature vectors into spatio-temporal joint encoding vectors; call the self-attention encoding unit in the encoding processing unit to extract global features and local features from the spatio-temporal joint encoding vectors, and analyze the global features and local features to obtain an analysis result.

[0110] In one embodiment, the second scoring module 550 is further configured to calculate the correlation values between the first abnormal data subset and each second driving feature subset with the same driving type according to a preset correlation calculation model; perform weighted summation on each calculated correlation value, and use the obtained weighted summation value as the abnormal score; determine whether the overhead crane to be measured has a fault risk based on the abnormal score, including: when the abnormal score exceeds the third scoring threshold, it is determined that the overhead crane to be measured has a fault risk.

[0111] In one embodiment, the fault prediction module 560 is further configured to determine that the overhead crane to be measured has no fault risk when the fault prediction score is less than the first scoring threshold; determine that the overhead crane to be measured has a fault risk when the fault prediction score is greater than the second scoring threshold.

[0112] In one embodiment, a computer-readable storage medium is provided, on which executable instructions are stored, and when the instructions are executed by a processor, the processor executes the steps in the above method embodiments.

[0113] In one embodiment, an electronic device is further provided, including one or more processors; a memory, and one or more programs are stored in the memory, wherein when the one or more programs are executed by the one or more processors, the one or more processors execute the steps in the above method embodiments. The electronic device may be a device deployed with the above distribution control system, such as a background server that controls the operation of the overhead crane or sends a driving path to the overhead crane to communicate with the overhead crane.

[0114] In one embodiment, as Figure 6As shown, it shows a schematic structural diagram of an electronic device for implementing an embodiment of the present application. The electronic device 600 includes a central processing unit (CPU) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage section 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the electronic device 600 are also stored. The CPU 601, ROM 602, and RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0115] The following components are connected to the I / O interface 605: an input section 606 including a keyboard, a mouse, etc.; an output section 607 including, for example, a cathode ray tube (CRT), a liquid crystal display (LCD), etc. and a speaker, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, a modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the I / O interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed so that a computer program read from it can be installed into the storage section 608 as needed.

[0116] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product including a computer-readable medium carrying instructions. In such an embodiment, the instructions can be downloaded and installed from a network through the communication section 609, and / or installed from the removable medium 611. When the instructions are executed by the central processing unit (CPU) 601, the various method steps described in the present application are executed.

[0117] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

[0118] In addition, those skilled in the art will understand that although some embodiments herein include certain features included in other embodiments but not others, the combination of features of different embodiments is meant to be within the scope of this application and forms different embodiments. For example, all of the above embodiments can be used in any combination. The information disclosed in this background section is only intended to enhance the understanding of the overall background of this application and should not be regarded as an admission or any form of implication that this information constitutes prior art already known to those skilled in the art.

Claims

1. A method for predicting overhead crane failures, characterized in that: The method comprises: Obtaining a first driving feature set of the overhead crane to be tested; Based on the first driving feature set, calling a preset fault prediction model to perform fault prediction on the overhead travelling crane to be tested, and outputting a fault prediction score; When the fault prediction score is between a first score threshold and a second score threshold, locating a target overhead rail section of the overhead crane to be tested that generates a first abnormal data subset during driving based on the first driving feature set; Acquire a second driving feature set, wherein the second driving feature set includes the first abnormal data subset and a second driving feature subset generated by other overhead traveling vehicles during driving of the target overhead traveling rail segment; Performing a cross-device correlation analysis on the first abnormal data subset and the second driving feature subset, and calculating an abnormality score of the first abnormal data subset on the target sky track segment; It is determined whether the overhead travelling crane to be tested has a failure risk based on the abnormality score.

2. The method for predicting overhead crane failure according to claim 1, characterized in that: The method of calling a preset fault prediction model to perform fault prediction on the overhead travelling vehicle to be tested based on the first driving feature set and outputting a fault prediction score includes: Dividing the first driving feature set into a plurality of first driving feature subsets of driving types according to the driving condition of the overhead crane and / or the distribution condition of the overhead rail segments; Calling the fault prediction model to analyze the first driving feature subset of each driving type to obtain a type score corresponding to each first driving feature subset; The fault prediction score is calculated based on the type score.

3. The method for predicting overhead crane failure according to claim 2, characterized in that: The second driving feature subset and the first abnormal data subset belong to the same driving type; The overhead crane driving condition includes one or more of a loaded driving condition and an unloaded driving condition, and the overhead rail segment distribution condition includes one or more of a straight overhead rail segment, a turning overhead rail segment, an ascending overhead rail segment, and a descending overhead rail segment.

4. The method for predicting overhead crane failure according to claim 1, characterized in that: The fault prediction model is a multimodal analysis model based on the Transformer architecture, including a multimodal input embedding unit, an encoding processing unit, a feature fusion unit and a decoding unit; The method of calling a preset fault prediction model to perform fault prediction on the overhead travelling vehicle to be tested based on the first driving feature set and outputting a fault prediction score includes: Calling the multimodal input embedding unit to convert the operating parameters of different dimensions in the first driving feature set into a high-dimensional feature vector; Calling the encoding processing unit to perform encoding processing and feature analysis on the high-dimensional feature vector to obtain an analysis result; Calling the feature fusion unit to fuse the analysis results to form fusion information; The decoding unit is called to calculate a fault prediction score according to the fusion information.

5. The method for predicting overhead crane failure according to claim 4, characterized in that: The calling of the encoding processing unit to perform encoding processing and feature analysis on the high-dimensional feature vector to obtain an analysis result includes: Calling the spatiotemporal coding unit in the coding processing unit to convert the high-dimensional feature vector into a spatiotemporal joint coding vector; The self-attention encoding unit in the encoding processing unit is called to extract global features and local features from the spatiotemporal joint encoding vector, and the global features and local features are analyzed to obtain analysis results.

6. The method for predicting overhead crane failure according to claim 2, characterized in that: The performing cross-device correlation analysis on the first abnormal data subset and the second driving feature subset to calculate the abnormality score of the first abnormal data subset on the target sky track segment includes: Calculate the correlation value between the first abnormal data subset and each second driving feature subset having the same driving type according to a preset correlation calculation model; Performing a weighted sum on each calculated correlation value, and using the obtained weighted sum value as the anomaly score; The determining whether the overhead travelling crane to be tested has a failure risk based on the abnormality score includes: when the abnormality score exceeds a third score threshold, determining that the overhead travelling crane to be tested has a failure risk.

7. The method for predicting overhead crane failure according to any one of claims 1 to 6, characterized in that: The method further comprises: When the fault prediction score is less than a first score threshold, it is determined that the overhead travelling crane to be tested does not have a fault risk; When the fault prediction score is greater than a second score threshold, it is determined that the overhead travelling crane to be tested has a fault risk.

8. A crane fault prediction device, characterized in that: The device comprises: A first feature acquisition module, used to acquire a first travel feature set of the overhead crane to be tested; A first scoring module, configured to call a preset fault prediction model to perform fault prediction on the overhead travelling vehicle to be tested based on the first driving feature set, and output a fault prediction score; a target overhead rail segment positioning module, configured to locate a target overhead rail segment of the overhead crane to be tested that generates a first abnormal data subset during driving based on the first driving feature set when the fault prediction score is between a first scoring threshold and a second scoring threshold; A second feature acquisition module is used to acquire a second driving feature set, wherein the second driving feature set includes the first abnormal data subset and a second driving feature subset generated by other overhead traveling vehicles during driving of the target overhead traveling section; a second scoring module, configured to perform cross-device correlation analysis on the first abnormal data subset and the second driving feature subset, and calculate an abnormality score of the first abnormal data subset on the target sky track segment; A fault prediction module is used to determine whether the overhead travelling crane to be tested has a fault risk based on the abnormality score.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores executable instructions, and when the executable instructions are executed by a processor, the processor is enabled to perform the method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: include: one or more processors; A memory for storing one or more programs, which, when executed by the one or more processors, causes the one or more processors to perform the method according to any one of claims 1 to 7.

Images