Crane fault monitoring method and system based on data analysis

Abstract

The invention belongs to the technical field of data analysis, and particularly relates to a crane fault monitoring method and system based on data analysis, and the method comprises the steps: obtaining the feature distinction degree of each type of fault monitoring data under each working condition according to the crane fault monitoring data; acquiring a monitoring importance degree according to the feature distinction degree; obtaining the distance between the feature vectors according to the monitoring importance degree of each type of fault monitoring data and the difference between the feature vectors in the fault monitoring data set; obtaining a reference distance according to the distance between the feature vectors, and obtaining a reference number according to the reference distance of each fault monitoring data set; and crane fault monitoring is carried out through the reference distance and the reference number of each fault monitoring data set. According to the invention, the crane fault monitoring has higher reliability and working condition adaptability.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and more particularly to a crane fault monitoring method and system based on data analysis. Background Art

[0002] Cranes often operate in complex environments, carrying enormous weights. Frequent use and heavy loads can make their key components susceptible to fatigue damage, cracks, deformation, and other hidden dangers, which can lead to serious accidents. Crane fault monitoring is necessary to take measures before accidents occur, repair or replace crane components, avoid major accidents such as crane overturning and falling heavy objects, and protect the lives of operators, surrounding personnel, and equipment.

[0003] In the related art, for example, the Chinese patent document with the authorization announcement number CN114358060B discloses a crane equipment fault detection method, which includes the following steps: When a fault occurs, the vibration sensor collects the mechanical parts The vibration signal is uploaded to the host computer; the host computer extracts the average vibration frequency of the vibration signal ; Extract features from vibration signals to form a set of eigenvalues ; Define mechanical parts Fault vibration frequency range and eigenvalue intervals ; Step 2: During operation, the vibration sensor Collecting mechanical parts Vibration signal Upload to the host computer; Step 3: The host computer filters and de-noises the vibration signal and extracts the vibration frequency If you judge Falling into the fault vibration frequency range If the value is within , then go to step 4; Step 4: Extract vibration signal The eigenvalue of ;like Falling into the fault characteristic value interval If the mechanical parts Fault.

[0004] Related technologies ignore the fact that a crane's vibration signal may fluctuate during normal operation due to changes in load, operating speed, or external conditions. If the vibration frequency corresponding to the vibration signal generated during normal operation falls within the fault vibration frequency range, or if the corresponding characteristic value of the vibration signal falls within the fault characteristic value range, a false alarm may be triggered. Therefore, related technologies for crane fault monitoring suffer from insufficient reliability. Summary of the Invention

[0005] In order to solve the above-mentioned technical problem of insufficient reliability of crane fault monitoring due to lack of equipment vibration analysis during normal operation of the crane, the present invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a crane fault monitoring method based on data analysis, comprising: constructing a fault monitoring data set using feature vectors, wherein the feature vectors are vectors obtained from historical fault monitoring data under various working conditions of the crane; taking the fault monitoring data set under any working condition as a target data set, and taking any category of data in the target data set as target category data; obtaining the feature discrimination of the target category data based on the coefficient of variation and information entropy of the target category data; obtaining the monitoring importance of the target category data based on the difference between the feature discrimination of the target category data and the minimum feature discrimination corresponding to the target data set; obtaining the distance between feature vectors based on the monitoring importance of the target category data and the difference between feature vectors in the target data set; obtaining a reference distance based on the distance between feature vectors in the target data set; obtaining a reference number based on a reference number of feature vectors in the target data set, wherein the reference number is the number of feature vectors whose distance from the feature vector is less than the reference distance; using feature vectors in the fault monitoring data set whose distance from a vector to be detected is less than the reference distance as target vectors; in response to the number of target vectors being less than the reference number, making the vector to be detected an abnormal vector, wherein the vector to be detected is obtained from real-time fault monitoring data; and performing crane fault monitoring based on the number of abnormal vectors.

[0007] The present invention arranges sensors at the trolley, car and hook of the crane to collect multi-dimensional historical fault monitoring data, and analyzes the data in combination with the operating characteristics of the crane under different working conditions, so that the analysis results can comprehensively reflect the operating status of the crane, avoid the influence of information loss caused by single-dimensional data on crane fault monitoring, and improve the accuracy of fault monitoring; through feature discrimination, it can accurately reflect the fault discrimination ability of each category of data under the corresponding working condition, and then obtain the monitoring importance through feature discrimination, which can make the subsequent calculation of the distance between feature vectors more targeted, so that the feature vectors can be better applied to crane fault monitoring; through the distribution characteristics of the feature vectors in the fault monitoring data set of each working condition, the benchmark distance and benchmark number are obtained, which provides reliable support for the fault monitoring of the crane under each working condition, and improves the effectiveness of crane fault monitoring under various working conditions.

[0008] Preferably, the feature discrimination satisfies the relationship: Where, For the Under the working condition The feature discrimination of class fault monitoring data, For the Under the working condition The coefficient of variation of the class fault monitoring data, For the The sum of the coefficients of variation of all types of fault monitoring data under a working condition, For the Under the working condition Information entropy of fault-like monitoring data, For the The maximum value of the information entropy of all types of fault monitoring data under a working condition, is an exponential function with a natural constant as its base.

[0009] Preferably, the feature discrimination satisfies the relationship: Where, For the Under the working condition The feature discrimination of class fault monitoring data, For the Under the working condition The coefficient of variation of the class fault monitoring data, For the The maximum value of the coefficient of variation of all types of fault monitoring data under the working conditions, For the Under the working condition Information entropy of fault-like monitoring data, For the The maximum value of the information entropy of all types of fault monitoring data under a working condition, is an exponential function with a natural constant as its base.

[0010] The present invention calculates the monitoring importance of each type of fault monitoring data through feature discrimination, ensuring that in crane fault monitoring, data with high feature discrimination obtains more references and reduces the impact of data with low feature discrimination on crane fault monitoring, thereby making crane fault detection more targeted and providing more reliable support for crane fault monitoring.

[0011] Preferably, the monitoring importance satisfies the relationship: Where, For the Under the working condition The monitoring importance of the fault monitoring data, For the Under the working condition The feature discrimination of class fault monitoring data, For the The minimum value of the feature discrimination of all types of fault monitoring data under the working condition, is the normalized exponential function.

[0012] The present invention highlights the feature categories with high feature discrimination through the calculation of monitoring importance, so that the data of the categories with high feature discrimination receive more attention in the crane fault monitoring process. At the same time, under different working conditions of the crane, the feature categories with high feature discrimination are also different. By monitoring the importance, the crane fault monitoring under different working conditions can be made more adapted to the crane operation characteristics under different working conditions, so that the crane fault monitoring has a better effect.

[0013] Preferably, the distance between the feature vectors satisfies the relationship: Where, For the The characteristic vector 𝑎 in the fault monitoring data set corresponding to the working condition and the characteristic vector The distance between For the The first feature vector 𝑎 in the fault monitoring data set corresponding to the working condition The value corresponding to the class fault monitoring data, For the The first feature vector 𝑏 in the fault monitoring data set corresponding to the working condition The value corresponding to the class fault monitoring data, For the Under the working condition The monitoring importance of the fault monitoring data, The number of fault monitoring data categories.

[0014] Preferably, the method for obtaining the reference distance includes: calculating the distance between any two eigenvectors in the target data set, and for each eigenvector, obtaining the minimum value of the distance between the eigenvector and other eigenvectors in the target data set, taking the minimum value as the reference distance of the eigenvector, and taking the average of the reference distances of all eigenvectors in the target data set as the reference distance.

[0015] Preferably, the method for obtaining the benchmark number includes: for any feature vector in the target data set, obtaining the number of feature vectors in the target data set whose distance from the feature vector is less than the benchmark distance, taking this number as the reference number, and making the average of the reference numbers of all feature vectors in the target data set the benchmark number.

[0016] Preferably, the method for obtaining the benchmark number includes: for any feature vector in the target data set, obtaining the number of feature vectors in the target data set whose distance from the feature vector is less than the benchmark distance, taking this number as the reference number, and making the maximum value of the reference number of all feature vectors in the target data set the benchmark number.

[0017] The present invention constructs a fault monitoring data set by constructing a feature vector through historical data, calculates the reference distance and reference quantity corresponding to the fault monitoring data set under different working conditions, and performs crane fault monitoring based on the reference distance and reference quantity, so that the fault monitoring of the crane under different working conditions is adapted to the characteristics of the data generated under different working conditions of the crane, making the crane fault monitoring under different working conditions more adaptable.

[0018] Preferably, the crane fault monitoring by the number of abnormal vectors includes: if the number of vectors to be detected that are abnormal vectors within a preset time exceeds a preset threshold, it is considered that the crane has an operational fault.

[0019] The present invention comprehensively considers the differences in crane operating conditions through multi-source data, avoiding the problems of false alarms and missed alarms caused by judging the crane operating status through a single threshold, making crane fault monitoring more reliable. At the same time, the judgment mechanism of continuous abnormal vectors further enhances the robustness of crane fault detection and improves the reliability of crane fault monitoring.

[0020] In a second aspect, the present invention provides a crane fault monitoring system based on data analysis, comprising a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned crane fault monitoring method based on data analysis is implemented.

[0021] By adopting the above technical solution, the above-mentioned crane fault monitoring method based on data analysis is generated into a computer program and stored in a memory so as to be loaded and executed by a processor, thereby making a terminal device based on the memory and the processor for easy use.

[0022] The beneficial effects of the present invention are: the present invention constructs fault monitoring data sets for the crane trolley operating conditions, car operating conditions and lifting conditions respectively, and calculates the feature discrimination of each type of fault monitoring data through the coefficient of variation and information entropy, fully considering the impact of the differences between crane working conditions on the distribution of fault monitoring data, making fault monitoring more targeted and discriminative, thereby improving the reliability of crane fault monitoring; weighting the distance between feature vectors based on the monitoring importance, so that features with stronger correlation with faults under different working conditions occupy a more important position in data analysis, thereby improving the sensitivity and accuracy of anomaly detection; through the benchmark distance and benchmark number, a data distribution reference standard under normal conditions is established, so that fault monitoring can adapt to the data characteristics under different working conditions, avoiding the limitations of static thresholds, and providing strong support for crane fault monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 FIG. 1 is a flow chart schematically illustrating a crane fault monitoring method based on data analysis in the present invention. DETAILED DESCRIPTION

[0024] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work shall fall within the scope of protection of the present invention.

[0025] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0026] The embodiment of the present invention discloses a crane fault monitoring method based on data analysis, referring to Figure 1 , including steps S1 to S7: S1. Obtain historical crane fault monitoring data.

[0027] Specifically, vibration sensors are arranged at the crane trolley, car, and hook to collect historical data at each moment when the crane is operating normally under different working conditions. The historical data at each moment includes the vibration data at the trolley, car, and hook at each moment when the crane is operating normally, as well as the height and load of the hook.

[0028] It should be noted that the structure of a crane includes a trolley, a small car, and a hook. The vibration of the crane varies under different operating conditions. For example, in the lifting condition, the hook is raised and lowered by rotating the wire rope drum via an electric motor. In the trolley operating condition, the entire crane moves along the guide rails under the drag of the trolley moving mechanism, while in the trolley operating condition, the trolley moves along the guide rails on the crane main beam. Obviously, the vibration state of the crane is different under different operating conditions. At the same time, when the hook is at different heights, the wire rope connected to the hook has different lengths, which will also cause the vibration of the hook at different heights to be affected by the swing amplitude of the wire rope. Moreover, the vibration data of the hook under different loads will also vary to a certain extent due to the influence of the wire rope length. Therefore, the present invention collects vibration data from the trolley, the trolley, and the hook, and combines the height and load of the hook to monitor crane faults.

[0029] S2. Obtain the characteristic discrimination degree of each category of fault monitoring data under each working condition based on the crane fault monitoring data.

[0030] Specifically, the mean and standard deviation of each type of fault monitoring data under each operating condition are calculated, and the information entropy of each type of fault monitoring data under each operating condition is calculated. For example, for the fault monitoring data under a certain operating condition, the mean and standard deviation of the vibration data at the trolley at each moment, as well as the information entropy of the vibration data at the trolley at each moment are calculated, the mean and standard deviation of the vibration data at the trolley at each moment, as well as the information entropy of the vibration data at the trolley at each moment are calculated, and so on.

[0031] Furthermore, the characteristic discrimination of the crane under each working condition is obtained according to the mean, standard deviation and information entropy of each type of fault monitoring data under each working condition.

[0032] In one embodiment, the feature discrimination of the crane satisfies the relationship: ; Where, For the Under the working condition The feature discrimination of class fault monitoring data, For the Under the working condition The coefficient of variation of the class fault monitoring data, For the The sum of the coefficients of variation of all types of fault monitoring data under a working condition, For the Under the working condition Information entropy of fault-like monitoring data, For the The maximum value of the information entropy of all types of fault monitoring data under a working condition, is an exponential function with a natural constant as its base; , When it is 1, it represents the trolley operating condition. When it is 2, it represents the trolley operating condition. When it is 3, it represents the lifting condition; , When it is 1 to 5, it represents the vibration data at the trolley, the vibration data at the car, the vibration data at the hook, the height of the hook and the load of the hook respectively.

[0033] It should be noted that the coefficient of variation is a concept in statistics. Since the fault monitoring data of different categories have different dimensions, the coefficient of variation can effectively eliminate the impact of the dimension differences of different categories of data on the evaluation of data dispersion. The larger the Under the working condition The more discrete the distribution of the fault monitoring data is, the more likely the crane will operate normally. Under the working condition The wider the distribution range of the type of fault monitoring data, the more difficult it is to distinguish the data generated when the crane fails from the data generated when the crane is operating normally. Under the working condition The smaller the feature discrimination of class fault monitoring data; The smaller the Under the working condition The more concentrated the distribution of fault monitoring data is, the Under the working condition The smaller the distribution range of the type of fault monitoring data, the easier it is to distinguish the data generated when the crane fails from the data generated when the crane is operating normally. Under the working condition The greater the feature discrimination of the fault monitoring data, the Divide by It is for Normalization is performed to facilitate subsequent calculations.

[0034] It needs to be further explained that Represents the Under the working condition The relative size of the information entropy of the fault-like monitoring data, The larger the Under the working condition The more evenly and irregularly the distribution of the type-fault monitoring data is, the wider the data coverage may be when the crane is operating normally, and the more likely it is that the distribution of normal data and fault data will be highly overlapped. The data generated when the crane fails is more difficult to distinguish from the data generated when the crane is operating normally. Under the working condition The smaller the feature discrimination of class fault monitoring data; The smaller the Under the working condition The more concentrated the distribution of the type fault monitoring data is, the more likely it is that there will be a clear difference in the distribution of normal data and fault data when the crane is operating normally. The easier it is to distinguish the data generated when the crane fails from the data generated when the crane is operating normally, the more likely it is that the type fault monitoring data will be concentrated when the crane is operating normally. Under the working condition The greater the feature discrimination of the class fault monitoring data.

[0035] In another embodiment, the characteristic differentiation of the crane satisfies the relationship: ; Where, For the Under the working condition The feature discrimination of class fault monitoring data, For the Under the working condition The coefficient of variation of the class fault monitoring data, For the The maximum value of the coefficient of variation of all types of fault monitoring data under the working conditions, For the Under the working condition Information entropy of fault-like monitoring data, For the The maximum value of the information entropy of all types of fault monitoring data under a working condition, is an exponential function with a natural constant as its base.

[0036] In which, by dividing Way to Normalize and get the normalized value Compared to Bigger, use Calculating feature discrimination can improve the sensitivity of subsequent crane fault detection and is more suitable for monitoring scenarios where cranes often fail.

[0037] S3. Obtain the monitoring importance of each category of fault monitoring data under each working condition of the crane based on the characteristic discrimination of the fault monitoring data.

[0038] Specifically, the monitoring importance of the fault monitoring data is obtained according to the difference between the characteristic discrimination degree of the fault monitoring data under each working condition and the minimum value of the characteristic discrimination degrees of each fault monitoring data under the working condition.

[0039] Preferably, the monitoring importance satisfies the relationship: ; Where, For the Under the working condition The monitoring importance of the fault monitoring data, For the Under the working condition The feature discrimination of class fault monitoring data, For the The minimum value of the feature discrimination of all types of fault monitoring data under the working condition, is the normalized exponential function.

[0040] It should be noted that the input data and output data of the normalized exponential function are both sequence data. For example, Using the normalized exponential function we get , then in this example In fact, .

[0041] It should be further explained that the same category of fault monitoring data has different reference significance for crane fault detection under different working conditions. For example, in the trolley operation condition or the trolley operation condition, the hook will produce a certain vibration with the trolley or the trolley. In this case, the vibration of the hook is a normal phenomenon, but in the lifting condition, the vibration of the hook may indicate that there may be a mechanical failure or safety hazard at the hook. As a result, in different working conditions, even if the vibration data of the hook is the same, the crane status it represents is different. Therefore, the present invention obtains the monitoring importance of each type of fault monitoring data through the feature discrimination of the fault monitoring data.

[0042] The larger the Under the working condition The easier it is to classify whether a crane has a fault based on the fault monitoring data, the better Under the working condition The greater the monitoring importance of the type of fault monitoring data; The smaller it is, the more Under the working condition The more difficult it is to classify whether the crane has a fault based on the fault monitoring data, the Under the working condition At the same time, in order to make each category of fault monitoring data better applied to crane fault monitoring, the present invention uses a normalized exponential function to Perform weighting.

[0043] S4. Construct a crane feature vector through the fault monitoring data of each historical moment under each working condition of the crane, and construct a fault monitoring data set under each working condition through the feature vector.

[0044] Specifically, the crane feature vector is constructed by the normalized values ​​of the vibration data at the trolley, the normalized values ​​of the vibration data at the trolley, the normalized values ​​of the vibration data at the hook at each historical moment under each working condition, as well as the normalized values ​​of the hook height and the normalized values ​​of the hook load; the normalization method for each category of data is linear normalization.

[0045] For example, the crane feature vector at a moment is ,in is the normalized value of the vibration data at the vehicle at that moment, is the normalized value of the vibration data at the car at that moment, is the normalized value of the vibration data at the hook at that moment, is the normalized value of the hook height at that moment, It is the normalized value of the hook load at that moment.

[0046] Furthermore, all feature vectors corresponding to the trolley operating conditions constitute a trolley operating condition fault monitoring data set, all feature vectors corresponding to the car operating conditions constitute a trolley operating condition fault monitoring data set, and all feature vectors corresponding to the lifting conditions constitute a lifting condition fault monitoring data set.

[0047] S5. Obtain the distance between the feature vectors according to the monitoring importance of each type of fault monitoring data and the difference between the feature vectors in the fault monitoring data set corresponding to each working condition.

[0048] Preferably, the distance between the feature vectors satisfies the relationship: ; Where, For the The characteristic vector 𝑎 in the fault monitoring data set corresponding to the working condition and the characteristic vector The distance between For the The first feature vector 𝑎 in the fault monitoring data set corresponding to the working condition The value corresponding to the class fault monitoring data, For the The first feature vector 𝑏 in the fault monitoring data set corresponding to the working condition The value corresponding to the class fault monitoring data, For the Under the working condition The monitoring importance of the fault monitoring data, The number of fault monitoring data categories.

[0049] It is easy to understand that The larger the Under the working condition The more important the type of fault monitoring data is to crane fault monitoring, the more important it is in calculating the The distance between the feature vectors in the fault monitoring data set corresponding to the working condition is Gaps in Class Fault Monitoring Data The greater the proportion, The smaller the Under the working condition The less important the type of fault monitoring data is to crane fault monitoring, the more important it is in calculating the The distance between the feature vectors in the fault monitoring data set corresponding to the working condition is Gaps in Class Fault Monitoring Data In order to better monitor the crane faults under various working conditions, the present invention uses As The weights are used to calculate the distance between feature vectors.

[0050] S6. Obtain a reference distance based on the distance between the feature vectors in the fault monitoring data set corresponding to each working condition, and obtain a reference quantity based on the reference distance of each fault monitoring data set.

[0051] Specifically, for any fault monitoring data set, the distance between any two eigenvectors in the fault monitoring data set is calculated. For each eigenvector, the minimum value of the distance between the eigenvector and other eigenvectors in the fault monitoring data set is obtained as the reference distance of the eigenvector, and the mean of the reference distances of all eigenvectors is taken as the benchmark distance.

[0052] Indicatively, if the vehicle operation condition fault monitoring data set is { , , , }, then calculate and 、 and 、 and 、 and 、 and 、 and If the distance between The minimum distance between the vectors is ,like The minimum distance between the vectors is , The minimum distance between the vectors is , The minimum distance between the vectors is , then the benchmark distance of the vehicle operation condition fault monitoring data set is .

[0053] Furthermore, for any feature vector in the fault monitoring data set, the number of feature vectors whose distance to the feature vector is less than a reference distance is obtained, and this number is used as a reference number, so that each feature vector has a reference number.

[0054] In one embodiment, a mean value of the reference quantities of all feature vectors in the fault monitoring dataset is calculated, and the mean value is used as the benchmark quantity of the fault monitoring dataset.

[0055] In another embodiment, a maximum value of reference quantities corresponding to all feature vectors in the fault monitoring data set is obtained, and the maximum value is used as the benchmark quantity of the fault monitoring data set.

[0056] It should be noted that the reference distance and reference quantity of the present invention are obtained through the data distribution characteristics of each working condition fault monitoring data set. The fault monitoring data sets of different working conditions have different reference distances and reference quantities, so that the crane fault monitoring is applicable to different working conditions of the crane.

[0057] S7. Perform crane fault monitoring based on the benchmark distance and benchmark quantity of each fault monitoring data set.

[0058] Specifically, during the crane fault monitoring process, real-time fault monitoring data of the crane is obtained, and the real-time fault monitoring data is constructed into a feature vector according to the method in step S4. The vector is used as the vector to be detected, and the distance between the vector to be detected and all feature vectors in the corresponding fault monitoring data set is calculated. The number of feature vectors whose distance to the vector to be detected is less than the benchmark distance of the working condition is obtained. If the number is less than the benchmark number, the vector to be detected is an abnormal vector; for example, if the vector to be detected is collected under the trolley operation condition, the distance between the vector to be detected and all feature vectors in the trolley operation condition fault monitoring data set is calculated; if the vector to be detected is collected under the trolley operation condition, the distance between the vector to be detected and all feature vectors in the trolley operation condition fault monitoring data set is calculated.

[0059] Preferably, when the number of abnormal vectors among the vectors to be detected generated within a preset time period exceeds a preset threshold, it is considered that the crane has an operational fault. In this embodiment, the length of the preset time period is 30 seconds, and the size of the preset threshold is half of the number of vectors to be detected generated within the preset time period. For example, if 30 vectors to be detected are generated within 30 seconds, if more than 15 of the vectors to be detected are abnormal vectors, it is considered that the crane has an operational fault.

[0060] Furthermore, when an operational failure occurs in the crane, an alarm is sounded through the crane alarm system to remind relevant staff to troubleshoot the problem.

[0061] An embodiment of the present invention further discloses a crane fault monitoring system based on data analysis, comprising a processor and a memory, wherein the memory stores computer program instructions. When the computer program instructions are executed by the processor, the crane fault monitoring method based on data analysis according to the present invention is implemented.

[0062] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface. The configuration and functions of these components are known in the art and will not be described in detail here.

Claims

1. A crane fault monitoring method based on data analysis, characterized in that: include: Constructing a fault monitoring data set using feature vectors, wherein the feature vectors are vectors obtained from historical fault monitoring data under various working conditions of the crane; The fault monitoring data set under any working condition is taken as the target data set, and any category of data in the target data set is taken as the target category data; Obtain the characteristic discrimination of the target category data based on the coefficient of variation and information entropy of the target category data; Obtain the monitoring importance of the target category data based on the gap between the feature discrimination of the target category data and the minimum feature discrimination corresponding to the target data set; Obtaining the distance between feature vectors based on the monitoring importance of the target category data and the gap between feature vectors in the target data set; obtaining a reference distance based on the distance between feature vectors in the target data set; obtaining a reference number based on the reference number of feature vectors in the target data set, where the reference number is the number of feature vectors whose distance from the feature vector is less than the reference distance; a feature vector in the fault monitoring data set whose distance to the vector to be detected is less than a reference distance is used as a target vector; in response to the number of target vectors being less than the reference number, the vector to be detected is considered an abnormal vector, wherein the vector to be detected is obtained from the real-time fault monitoring data; Crane fault monitoring by the number of abnormal vectors.

2. The crane fault monitoring method based on data analysis according to claim 1, characterized in that: The feature discrimination satisfies the relationship: ； Where, For the Under the working condition The feature discrimination of class fault monitoring data, For the Under the working condition The coefficient of variation of the class fault monitoring data, For the The sum of the coefficients of variation of all types of fault monitoring data under a working condition, For the Under the working condition Information entropy of fault-like monitoring data, For the The maximum value of the information entropy of all types of fault monitoring data under a working condition, is an exponential function with a natural constant as its base.

3. The crane fault monitoring method based on data analysis according to claim 1, characterized in that: The feature discrimination satisfies the relationship: ； Where, For the Under the working condition The feature discrimination of class fault monitoring data, For the Under the working condition The coefficient of variation of the class fault monitoring data, For the The maximum value of the coefficient of variation of all types of fault monitoring data under the working conditions, For the Under the working condition Information entropy of fault-like monitoring data, For the The maximum value of the information entropy of all types of fault monitoring data under a working condition, is an exponential function with a natural constant as its base.

4. The crane fault monitoring method based on data analysis according to claim 1, characterized in that: The monitoring importance satisfies the relationship: ； Where, For the Under the working condition The monitoring importance of the fault monitoring data, For the Under the working condition The feature discrimination of class fault monitoring data, For the The minimum value of the feature discrimination of all types of fault monitoring data under the working condition, is the normalized exponential function.

5. The crane fault monitoring method based on data analysis according to claim 1, characterized in that: The distance between the eigenvectors satisfies the relationship: ； Where, For the The characteristic vector 𝑎 in the fault monitoring data set corresponding to the working condition and the characteristic vector The distance between For the The first feature vector 𝑎 in the fault monitoring data set corresponding to the working condition The value corresponding to the class fault monitoring data, For the The first feature vector 𝑏 in the fault monitoring data set corresponding to the working condition The value corresponding to the class fault monitoring data, For the Under the working condition The monitoring importance of the fault monitoring data, The number of fault monitoring data categories.

6. The crane fault monitoring method based on data analysis according to claim 1, characterized in that: The method for obtaining the reference distance includes: Calculate the distance between any two eigenvectors in the target data set. For each eigenvector, obtain the minimum value of the distance between the eigenvector and other eigenvectors in the target data set. Use the minimum value as the reference distance of the eigenvector, and use the mean of the reference distances of all eigenvectors in the target data set as the benchmark distance.

7. The crane fault monitoring method based on data analysis according to claim 1, characterized in that: The method for obtaining the benchmark quantity includes: For any eigenvector in the target data set, the number of eigenvectors in the target data set whose distance to the eigenvector is less than the reference distance is obtained, and this number is used as the reference number, so that the average of the reference numbers of all eigenvectors in the target data set is the reference number.

8. The crane fault monitoring method based on data analysis according to claim 1, characterized in that: The method for obtaining the benchmark quantity includes: For any feature vector in the target data set, the number of feature vectors in the target data set whose distance to the feature vector is less than the reference distance is obtained, and this number is used as the reference number, so that the maximum value of the reference number of all feature vectors in the target data set is the reference number.

9. The crane fault monitoring method based on data analysis according to claim 1, characterized in that: The crane fault monitoring is performed by using the number of abnormal vectors, including: If the number of vectors to be detected that are abnormal vectors within a preset time exceeds a preset threshold, it is considered that the crane has an operating failure.

10. The crane fault monitoring system based on data analysis is characterized by: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the crane fault monitoring method based on data analysis according to any one of claims 1 to 9 is implemented.

Images