Lifting machinery operation and maintenance method based on digital twin and its application

Abstract

The present invention discloses a crane operation and maintenance method based on digital twin and its application, the method comprising: arranging a plurality of sensors at the target part of the crane to collect the status data of the target part of the crane in real time; preprocessing the collected data, and extracting the first characteristic data that is helpful for the target part fault identification from the preprocessed data; based on REF and combined with the random forest model, filtering out the second characteristic data with the most distinguishing degree for fault identification from the first characteristic data; based on the filtered second characteristic data, analyzing the fault probability and fault type of each target component of the crane; when the fault probability of the target part reaches or exceeds the preset level, executing the early warning prompt of the corresponding fault type of the target part according to the identification result of the fault type. The present invention can realize the real-time monitoring of the crane operation status, accurately identify and warn the crane fault, and can significantly improve the operation and maintenance efficiency and reliability of the crane.

Description

Technical Field

[0001] The present invention relates to the field of crane technology, and in particular to a crane operation and maintenance method based on digital twins and applications thereof. Background Art

[0002] Cranes are multi-action lifting machines that can lift heavy objects vertically and carry them horizontally within a certain range. Cranes have harsh working environments, complex loads, and are prone to various types of damage, which can affect the safe operation of the crane. Therefore, timely maintenance of cranes is of great significance to the safe operation of cranes.

[0003] At present, most cranes are inspected and maintained regularly. Therefore, it is difficult to find and maintain the failure of the crane in time, which may easily lead to safety accidents. Summary of the invention

[0004] To solve the above problems, the present invention provides a crane operation and maintenance method based on digital twins and its application, which can realize real-time monitoring of the crane's operating status, and accurately identify and warn of crane faults, which is conducive to timely maintenance of the crane and significantly improves the operation and maintenance efficiency and reliability of the crane.

[0005] In order to achieve the above technical objectives, the technical solution adopted by the present invention is: a lifting machinery operation and maintenance method based on digital twin, characterized in that it includes:

[0006] S1: Arrange multiple sensors at the target parts of the crane to collect the status data of the target parts of the crane in real time, and associate the acquired data with the corresponding parts;

[0007] S2: preprocessing the collected data, and extracting first characteristic data that is helpful for identifying faults at target parts from the preprocessed data;

[0008] S3: Based on REF (recursive feature elimination method) and combined with random forest model, the second feature data with the best discrimination for fault identification is selected from the first feature data;

[0009] S4: Analyze the failure probability and failure type of each target component of the crane based on the screened second characteristic data;

[0010] S5: When the failure probability of the target part reaches or exceeds the preset level, an early warning prompt of the failure type corresponding to the target part is executed according to the identification result of the failure type.

[0011] As a possible implementation mode, further, the crane operation and maintenance method based on digital twins further includes S6: acquiring characteristic information of the crane, constructing a high-precision three-dimensional model of the crane, and associating each part in the three-dimensional model with a corresponding part in the crane;

[0012] When there is a fault warning, the fault location is rendered in the 3D model and the fault type of the fault location is prompted.

[0013] As a possible implementation manner, further, the target part in step S1 includes a boom, a slewing support, and a braking system; and the sensor includes a vibration sensor, a temperature sensor, a displacement sensor, a stress sensor, and a pressure sensor.

[0014] As a possible implementation, further, in step S2, the collected data is preprocessed, and first characteristic data that is helpful for target part fault identification is extracted from the preprocessed data; specifically, the following steps are included:

[0015] S2.1: De-noise, interpolate and Z-score normalize the collected data;

[0016] S2.2: extracting first characteristic data that is helpful for target part fault identification from the processed data; the first characteristic data includes time domain characteristics, frequency domain characteristics and statistical characteristics;

[0017] The time domain features include mean, variance, skewness, and kurtosis; the frequency domain features include spectrum energy, main frequency, and spectrum kurtosis; and the statistical features include maximum value, minimum value, signal entropy, and autocorrelation coefficient.

[0018] As a possible implementation, further, in step S3, based on REF (recursive feature elimination method) and combined with random forest model, the most distinguishing feature data for fault identification is screened out, which specifically includes the following steps:

[0019] S3.1: Obtaining first characteristic data of each fault part of the crane and the corresponding fault type from historical fault events of each target part of the crane;

[0020] S3.2: Based on the first feature data and fault type obtained, construct the initial feature set of each target part of the crane Wherein, Z represents the part of the crane; D z Represents the initialization feature set of the crane Z position; N z represents the total number of failures in crane Z part in historical failure events; zi Indicates crane Z position N z The first characteristic data corresponding to the i-th fault event in the historical fault events, X zi ={x zi1 , x zi2 , x zi3 ....x ziM} is the feature vector, M represents the number of the first feature data, x zi1 ~x ziM Refers to the crane Z part N z Each feature data in the first feature data corresponding to the i-th fault event in the historical fault events; y zi Indicates crane Z position N z The fault type label corresponding to the i-th fault event in the historical fault events;

[0021] S3.3: Taking dataset X zi As the feature input of the first random forest model, y zi As model output, the first random forest model is trained;

[0022] S3.4: Extract the importance score I of each feature j from the first random forest model after training j , j∈X zi ; Among them, the importance score of feature j is I j It is obtained by accumulating the reduction of Gini impurity of node n in all trees that use this feature for splitting. Its expression is as follows:

[0023]

[0024] Where M represents the total number of decision trees; ΔG n,j N is the reduction in Gini impurity when feature j is split at node n; n represents the number of samples in node n; T i represents the i-th tree;

[0025] S3.5: Compare the importance scores I of all features j j , select the feature with the lowest score and remove it from the current feature set X zi Remove the feature and get a new feature set;

[0026] S3.6: Use the new feature set as the current feature set X zi , and repeat steps S3.3 to S3.6 until the model performance improvement ΔPerf is less than the set threshold ε;

[0027] S3.7: When the iteration stops, the feature set X zi The feature type retained in step S2 is used as a reference, and features of corresponding types are screened out from the first feature data obtained in step S2 as second feature data.

[0028] As a possible implementation, further, in step S4, the failure probability and failure type of each target component of the crane are analyzed based on the filtered second characteristic data, as follows:

[0029] Input the screened second feature data into the trained second random forest model;

[0030] When predicting the fault probability and fault type, each decision tree in the model assigns a category label to the input second feature data; wherein the category label includes all fault type labels, the total number of decision trees in the second random forest is an integer multiple of the number of category labels, and each category label has the same proportion in the total number of decision trees;

[0031] By summarizing the prediction results of all decision trees, the prediction probability corresponding to each category is calculated. The specific expression is as follows:

[0032]

[0033] In the above formula, P(y=c|X) represents the predicted probability of category c; X represents the second feature data of the input; M represents the total number of decision trees in the random forest; h i (X) represents the prediction result of the i-th decision tree for the second feature data X; II represents the indicator function. i (X) = 1 when c, otherwise 0;

[0034] The highest predicted probability is taken as the failure probability of the corresponding target part of the crane, and the type label with the highest predicted probability is taken as the failure type of the corresponding target part.

[0035] As a possible implementation method, further, the training method of the second random forest model specifically includes the following steps:

[0036] Obtaining first feature data and corresponding status labels of each target part of the crane from historical data of each target part of the crane;

[0037] Filtering out the second characteristic data with the highest degree of distinction for fault identification from the first characteristic data, and associating the second characteristic data with the state label corresponding to the first characteristic data;

[0038] The second feature data is used as the input of the second random forest model, and the state label associated with the second feature data is used as the output of the model. The second random forest model is trained until the model converges to obtain a trained second random forest model.

[0039] As a possible implementation, further, in step S5, when the failure probability of the target part reaches or exceeds a preset level, an early warning prompt of the failure type corresponding to the target part is executed according to the identification result of the failure type, which specifically includes the following steps:

[0040] Step S5.1: setting different probability threshold intervals for each target part, and different threshold intervals correspond to different fault probability levels;

[0041] Step S5.2: Determine the corresponding fault probability level according to the fault probability of each target part. When the fault probability level of the target part reaches the preset level, execute the early warning prompt corresponding to the fault type of the target part according to the fault type of the target part analyzed and identified in step S4.

[0042] The present invention also provides a lifting machinery operation and maintenance system based on digital twins, which applies the above-mentioned lifting machinery operation and maintenance based on digital twins, and the system includes:

[0043] A sensor unit, used to collect status data of target parts of the crane in real time;

[0044] A data preprocessing unit, used for preprocessing the collected data;

[0045] A first characteristic data extraction unit, used to extract first characteristic data that is helpful for target part fault identification from the preprocessed data;

[0046] A second characteristic data screening unit, used to screen out the second characteristic data with the highest degree of distinction for fault identification from the first characteristic data;

[0047] A fault analysis unit, used to analyze the fault probability and fault type of each target component of the crane by using the screened second characteristic data;

[0048] A logic unit is used to associate the data acquired by the sensor with the corresponding part, and to determine the fault probability level of the target part of the crane according to the analysis result output by the fault analysis unit. When the fault probability of the target part reaches or exceeds the preset level, the early warning unit is controlled to issue a corresponding early warning prompt according to the fault type of the target part;

[0049] The early warning unit is used to perform corresponding early warning operations according to the judgment results output by the logic unit.

[0050] The present invention also provides a computer-readable storage medium, in which at least one instruction, at least one program, code set or instruction set is stored. The at least one instruction, at least one program, code set or instruction set is loaded and executed by a processor to implement the above-mentioned digital twin-based crane operation and maintenance method.

[0051] By adopting the above technical solution, the present invention has the following beneficial effects compared with the prior art:

[0052] The operation and maintenance method provided by the present invention can monitor the operation status of the crane in real time, and accurately identify and warn the crane fault, so as to remind the maintenance personnel to perform timely maintenance on the faulty part of the crane, thereby ensuring the safe operation of the crane. The present invention can significantly improve the operation and maintenance efficiency and reliability of the crane, has broad application prospects and market value, and is suitable for further promotion and application. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0054] Figure 1 It is a schematic diagram of a brief implementation process of the lifting machinery operation and maintenance method based on digital twin of the present invention;

[0055] Figure 2 A schematic diagram of a system implementation structure of a lifting machinery operation and maintenance method based on digital twin corresponding to the solution of the present invention;

[0056] Figure 3 A schematic diagram of a computer-readable storage medium provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0057] The present invention will be further described in detail below in conjunction with the accompanying drawings and examples. It is particularly noted that the following examples are only used to illustrate the present invention, but are not intended to limit the scope of the present invention. Similarly, the following examples are only partial embodiments of the present invention rather than all embodiments, and all other embodiments obtained by those of ordinary skill in the art without creative work are within the scope of protection of the present invention.

[0058] Example 1

[0059] Refer to the attached Figure 1 As shown, this embodiment provides a lifting machinery operation and maintenance method based on digital twin, including:

[0060] S1: Arrange multiple sensors (including vibration sensors, temperature sensors, displacement sensors, stress sensors, pressure sensors, etc.) at target parts of the crane (target parts include boom, slewing support, braking system, etc.) to collect status data of target parts of the crane in real time, and associate the acquired data with the corresponding parts.

[0061] In the present invention, fault warning is performed based on the data detected by the sensor. After the acquired data is associated with its corresponding part, when a fault warning is performed, the fault part that needs maintenance can be located by the data that triggers the fault warning, which can help operation and maintenance personnel quickly locate the fault part.

[0062] S2: preprocessing the collected data and extracting first characteristic data that is helpful for target part fault identification from the preprocessed data; specifically comprising the following steps:

[0063] S2.1: De-noising, interpolation and Z-score standardization are performed on the collected data; in this embodiment, moving average filtering is used to denoise the data; data quality can be improved by preprocessing the data, providing a reliable basis for subsequent analysis.

[0064] S2.2: Extracting first characteristic data that is helpful for identifying the fault of the target part from the processed data; in this embodiment, the first characteristic data is extracted from the data within a preset time length T, that is, the first characteristic data is extracted from (T p -T)~T p Extracted from the data within the time period, T p Indicates the current moment.

[0065] The first feature data includes time domain features, frequency domain features and statistical features;

[0066] Time domain features include mean, variance, skewness, and kurtosis;

[0067] The frequency domain feature is obtained by converting the time domain signal into the frequency domain signal through Fourier transform. The expression is as follows:

[0068]

[0069] In the above formula, X(f) is the frequency domain signal, which represents the components of the signal at different frequencies; x(t) is the time domain signal, which represents the signal that changes over time; f is the frequency, in Hz; L is an imaginary number that satisfies L 2 =-1; e -L2πft : Complex exponential function, representing the rotation of the unit vector.

[0070] Frequency domain features include spectrum energy, main frequency, and spectrum kurtosis;

[0071] The spectrum energy E(f) is expressed as follows:

[0072] E(f)=|X(f)| 2

[0073] Main frequency (f peak ): The frequency with the largest energy in the spectrum;

[0074] The spectral kurtosis S is expressed as follows:

[0075]

[0076] Among them, μ f is the mean of the frequencies.

[0077] Statistical features include maximum value, minimum value, signal entropy, and autocorrelation coefficient.

[0078] S3: Based on REF (recursive feature elimination method) and combined with random forest model, the second feature data with the best discrimination for fault identification is selected from the first feature data; specifically, the following steps are included:

[0079] S3.1: Obtaining first characteristic data of each fault part of the crane and the corresponding fault type from historical fault events of each target part of the crane;

[0080] S3.2: Based on the first feature data and fault type obtained, construct the initial feature set of each target part of the crane Wherein, Z represents the part of the crane; D z Represents the initialization feature set of the crane Z position; N z represents the total number of failures in crane Z part in historical failure events; zi Indicates crane Z position N z The first characteristic data corresponding to the i-th fault event in the historical fault events, X zi ={x zi1 , χ zi2 , χ zi3 ....x ziM} is the feature vector, M represents the number of the first feature data, χ zi1 ~x ziM Refers to the crane Z part N z Each feature data in the first feature data corresponding to the i-th fault event in the historical fault events; y zi Indicates crane Z position N z The fault type label corresponding to the i-th fault event in the historical fault events;

[0081] S3.3: Taking dataset X zi As the feature input of the first random forest model, y zi As model output, the first random forest model is trained;

[0082] S3.4: Extract the importance score I of each feature j from the first random forest model after training j , j∈X zi; Among them, the importance score of feature j is I j It is obtained by accumulating the reduction of Gini impurity of node n in all trees that use this feature for splitting. Its expression is as follows:

[0083]

[0084] Where M represents the total number of decision trees; ΔG n,j N is the reduction in Gini impurity when feature j is split at node n; n represents the number of samples in node n; T i represents the i-th tree;

[0085] S3.5: Compare the importance scores I of all features j j , select the feature with the lowest score and remove it from the current feature set X zi Remove the feature and get a new feature set;

[0086] S3.6: Use the new feature set as the current feature set X zi , and repeat steps S3.3 to S3.6 until the model performance improvement ΔPerf is less than the set threshold ε;

[0087] S3.7: When the iteration stops, the feature set X zi The feature type retained in step S2 is used as a reference, and the features of the corresponding type are selected from the first feature data obtained in step S2 as the second feature data. For example, when the iteration stops, the feature set X zi The feature types retained in the model are: skewness, kurtosis, spectrum energy, main frequency, spectrum kurtosis, maximum value, minimum value, and autocorrelation coefficient. Then, "skewness, kurtosis, spectrum energy, main frequency, spectrum kurtosis, maximum value, minimum value, and autocorrelation coefficient data" are selected from the first feature data as the second feature data. By selecting the most discriminative feature data for fault identification for model training, the accuracy and sensitivity of model fault analysis can be effectively improved.

[0088] S4: Based on the screened second characteristic data, the failure probability and failure type of each target component of the crane are analyzed; the details are as follows:

[0089] Input the screened second feature data into the trained second random forest model;

[0090] When predicting the fault probability and fault type, each decision tree in the model assigns a category label to the input second feature data; wherein the category label includes all fault type labels, the total number of decision trees in the second random forest is an integer multiple of the number of category labels, and each category label has the same proportion in the total number of decision trees;

[0091] By summarizing the prediction results of all decision trees, the prediction probability corresponding to each category is calculated. The specific expression is as follows:

[0092]

[0093] In the above formula, P(y=c|X) represents the predicted probability of category c; X represents the second feature data of the input; M represents the total number of decision trees in the random forest; h i (X) represents the prediction result of the i-th decision tree for the second feature data X; II represents the indicator function. i (X) = 1 when c, otherwise 0;

[0094] The highest predicted probability is used as the failure probability of the corresponding target part of the crane, and the type label with the highest predicted probability is used as the failure type of the corresponding target part. For example, through model analysis, the failure type of the slewing bearing is "wear", and the highest predicted probability output is A. Then A is used as the failure probability of the slewing bearing, and "wear" is used as the failure type of the slewing bearing.

[0095] In this embodiment, the training method of the second random forest model specifically includes the following steps:

[0096] From the historical data of each target part of the crane, the first characteristic data and the corresponding status label of each target part of the crane are obtained; wherein the status label includes a normal label (the normal label indicates that the state of the target component of the crane is normal) and a fault type label.

[0097] Filter out the second feature data with the highest degree of discrimination for fault identification from the first feature data (the specific screening method is the same as step S3, removing the feature with the lowest importance score until the model performance improvement ΔPerf is less than the set threshold ε), and associate the second feature data with the state label corresponding to the first feature data;

[0098] The second feature data is used as the input of the second random forest model, and the state label associated with the second feature data is used as the output of the model. The second random forest model is trained until the model converges to obtain a trained second random forest model. The model is trained using the second feature data that is most discriminative for fault identification, which can effectively improve the accuracy and sensitivity of the model fault analysis.

[0099] S5: When the failure probability of the target part reaches or exceeds the preset level, an early warning prompt corresponding to the failure type of the target part is executed according to the identification result of the failure type; specifically, the following steps are included:

[0100] Step S5.1: Set different probability threshold intervals for each target part, and different threshold intervals correspond to different fault probability levels. For ease of understanding, an example is given here to illustrate, for example: three threshold intervals are set for the slewing bearing: ① less than 0.5; ② greater than or equal to 0.5 and less than 0.8; ③ greater than or equal to 0.8. Among them, the fault probability level corresponding to threshold interval ① is "low"; the fault probability level corresponding to threshold interval ② is "medium"; and the fault probability level corresponding to threshold interval ③ is "high".

[0101] Step S5.2: Determine the corresponding fault probability level according to the fault probability of each target part. When the fault probability level of the target part reaches the preset level, execute the early warning prompt corresponding to the fault type of the target part according to the fault type of the target part analyzed and identified in step S4.

[0102] For ease of understanding, an example is given here to illustrate, for example: when the "wear" failure probability level of the slewing bearing reaches or exceeds "medium", an early warning is issued. After model analysis and identification, the fault type of the slewing bearing is wear, and the failure probability is 0.6; the failure probability of the slewing bearing falls into the threshold interval ②, so the failure probability level of the slewing bearing is "medium". Since the failure probability level of the slewing bearing reaches "medium", an early warning is issued; when an early warning is issued, an early warning prompt is issued that the "wear" failure probability of the slewing bearing is high and maintenance is required.

[0103] In addition, if the failure probability of the slewing bearing is 0.9 and the corresponding failure probability level is "high", an early warning will also be issued; if the failure probability level of the slewing bearing is "low", no early warning will be issued.

[0104] S6: Acquire characteristic information of the crane, construct a high-precision three-dimensional model of the crane, and associate each part in the three-dimensional model with a corresponding part in the crane;

[0105] When there is a fault warning, the fault part is rendered in the 3D model (in this embodiment, the fault part is rendered in red to remind maintenance personnel of the fault part of the crane), and the fault type of the fault part is prompted. In this embodiment, the fault type is prompted in text or voice. In this way, it can help operation and maintenance personnel quickly locate the fault part and determine the fault type of the fault part, which helps to improve the operation and maintenance efficiency of the crane.

[0106] Example 2

[0107] See attached Figure 2 As shown, an embodiment of the present invention provides a lifting machinery operation and maintenance system based on digital twins, which applies the lifting machinery operation and maintenance method based on digital twins in the above embodiment 1, and the system includes:

[0108] A sensor unit, used to collect status data of target parts of the crane in real time;

[0109] A data preprocessing unit, used for preprocessing the collected data;

[0110] A first characteristic data extraction unit, used to extract first characteristic data that is helpful for target part fault identification from the preprocessed data;

[0111] A second characteristic data screening unit, used to screen out the second characteristic data with the highest degree of distinction for fault identification from the first characteristic data;

[0112] A fault analysis unit, used to analyze the fault probability and fault type of each target component of the crane by using the screened second characteristic data;

[0113] A logic unit is used to associate the data acquired by the sensor with the corresponding part, and to determine the fault probability level of the target part of the crane according to the analysis result output by the fault analysis unit. When the fault probability of the target part reaches or exceeds the preset level, the early warning unit is controlled to issue a corresponding early warning prompt according to the fault type of the target part;

[0114] The early warning unit is used to perform corresponding early warning operations according to the judgment results output by the logic unit.

[0115] Example 3

[0116] See attached Figure 3 As shown, based on the same inventive concept, this embodiment provides a computer-readable storage medium, which stores computer instructions. When the computer instructions are run on a computer, the computer executes the lifting machinery operation and maintenance method based on digital twins as described in Example 1.

[0117] In the specific implementation process, computer-readable storage media include: Universal Serial Bus flash drive (USB), mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, and other storage media that can store program codes.

[0118] The device embodiments described above are merely illustrative, wherein the units / modules described as separate components may or may not be physically separated, and the components displayed as units / modules may or may not be physical units / modules, i.e., they may be located in one place, or they may be distributed over multiple network units / modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without creative effort.

[0119] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0120] The above descriptions are only some embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Any equivalent device or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for operation and maintenance of lifting machinery based on digital twin, characterized in that: include: S1: Arrange multiple sensors at the target parts of the crane to collect the status data of the target parts of the crane in real time, and associate the acquired data with the corresponding parts; S2: preprocessing the collected data, and extracting first characteristic data that is helpful for identifying faults at target parts from the preprocessed data; S3: Based on REF and combined with the random forest model, the second feature data with the best discrimination for fault identification is selected from the first feature data; S4: Analyze the failure probability and failure type of each target component of the crane based on the screened second characteristic data; S5: When the failure probability of the target part reaches or exceeds the preset level, an early warning prompt of the failure type corresponding to the target part is executed according to the identification result of the failure type.

2. The method for operation and maintenance of a lifting machinery based on digital twin according to claim 1, characterized in that: It also includes S6: acquiring characteristic information of the crane, constructing a high-precision three-dimensional model of the crane, and associating each part in the three-dimensional model with a corresponding part in the crane; When there is a fault warning, the fault location is rendered in the 3D model and the fault type of the fault location is prompted.

3. The method for operation and maintenance of a lifting machinery based on digital twin according to claim 1 is characterized in that: The target parts in step S1 include a boom, a slewing support, and a braking system; the sensors include a vibration sensor, a temperature sensor, a displacement sensor, a stress sensor, and a pressure sensor.

4. The method for operation and maintenance of a lifting machinery based on digital twin according to claim 1, characterized in that: In step S2, the collected data is preprocessed, and first characteristic data that is helpful for target part fault identification is extracted from the preprocessed data; specifically, the following steps are included: S2.1: De-noise, interpolate and Z-score normalize the collected data; S2.2: extracting first characteristic data that is helpful for target part fault identification from the processed data; the first characteristic data includes time domain characteristics, frequency domain characteristics and statistical characteristics; The time domain features include mean, variance, skewness, and kurtosis; the frequency domain features include spectrum energy, main frequency, and spectrum kurtosis; and the statistical features include maximum value, minimum value, signal entropy, and autocorrelation coefficient.

5. The method for operation and maintenance of a lifting machinery based on digital twin according to claim 4 is characterized in that: In step S3, based on REF and combined with the random forest model, the most distinguishing feature data for fault identification is screened out, which specifically includes the following steps: S3.1: Obtaining first characteristic data of each fault part of the crane and the corresponding fault type from historical fault events of each target part of the crane; S3.2: Based on the first feature data and fault type obtained, construct the initial feature set of each target part of the crane Wherein, Z represents the part of the crane; D z Represents the initialization feature set of the crane Z position; N z represents the total number of failures in crane Z part in historical failure events; zi Indicates crane Z position N z The first characteristic data corresponding to the i-th fault event in the historical fault events, X zi ={ zi1 , χ zi2 , χ zi3 ....x ziM } is the feature vector, M represents the number of the first feature data, χ zi1 ~x ziM Refers to the crane Z part N z Each feature data in the first feature data corresponding to the i-th fault event in the historical fault events; y zi Indicates crane Z position N z The fault type label corresponding to the i-th fault event in the historical fault events; S3.3: Taking dataset X zi As the feature input of the first random forest model, y zi As model output, the first random forest model is trained; S3.4: Extract the importance score I of each feature j from the first random forest model after training j , j∈X zi ; Among them, the importance score of feature j is I j It is obtained by accumulating the reduction of Gini impurity of node n in all trees that use this feature for splitting. Its expression is as follows: Where M represents the total number of decision trees; ΔG n,j N is the reduction in Gini impurity when feature j is split at node n; n represents the number of samples in node n; T i represents the i-th tree; S3.5: Compare the importance scores I of all features j j , select the feature with the lowest score and remove it from the current feature set X zi Remove the feature and get a new feature set; S3.6: Use the new feature set as the current feature set X zi , and repeat steps S3.3 to S3.6 until the model performance improvement ΔPerf is less than the set threshold ε; S3.7: When the iteration stops, the feature set X zi The feature type retained in step S2 is used as a reference, and features of corresponding types are screened out from the first feature data obtained in step S2 as second feature data.

6. The method for operation and maintenance of a lifting machinery based on digital twin according to claim 1, characterized in that: In step S4, based on the filtered second characteristic data, the failure probability and failure type of each target component of the crane are analyzed, as follows: Input the screened second feature data into the trained second random forest model; When predicting the fault probability and fault type, each decision tree in the model assigns a category label to the input second feature data; wherein the category label includes all fault type labels, the total number of decision trees in the second random forest is an integer multiple of the number of category labels, and each category label has the same proportion in the total number of decision trees; By summarizing the prediction results of all decision trees, the prediction probability corresponding to each category is calculated. The specific expression is as follows: In the above formula, P(y=c|X) represents the predicted probability of category c; X represents the second feature data of the input; M represents the total number of decision trees in the random forest; h i (X) represents the prediction result of the i-th decision tree for the second feature data X; II represents the indicator function. i (X) = 1 when c, otherwise 0; The highest predicted probability is taken as the failure probability of the corresponding target part of the crane, and the type label with the highest predicted probability is taken as the failure type of the corresponding target part.

7. The method for operating and maintaining a lifting machinery based on digital twins according to claim 6 is characterized in that: The training method of the second random forest model specifically includes the following steps: Obtaining first feature data and corresponding status labels of each target part of the crane from historical data of each target part of the crane; Filtering out the second characteristic data with the highest degree of distinction for fault identification from the first characteristic data, and associating the second characteristic data with the state label corresponding to the first characteristic data; The second feature data is used as the input of the second random forest model, and the state label associated with the second feature data is used as the output of the model. The second random forest model is trained until the model converges to obtain a trained second random forest model.

8. The method for operating and maintaining a lifting machinery based on digital twins according to claim 6, characterized in that: In step S5, when the failure probability of the target part reaches or exceeds the preset level, an early warning prompt of the failure type corresponding to the target part is executed according to the identification result of the failure type, which specifically includes the following steps: Step S5.1: setting different probability threshold intervals for each target part, and different threshold intervals correspond to different fault probability levels; Step S5.2: Determine the corresponding fault probability level according to the fault probability of each target part. When the fault probability level of the target part reaches the preset level, execute the early warning prompt corresponding to the fault type of the target part according to the fault type of the target part analyzed and identified in step S4.

9. A lifting machinery operation and maintenance system based on digital twin, characterized in that: Applying the operation and maintenance method according to any one of claims 1 to 8, the system comprises: A sensor unit, used to collect status data of target parts of the crane in real time; A data preprocessing unit, used for preprocessing the collected data; A first characteristic data extraction unit, used to extract first characteristic data that is helpful for target part fault identification from the preprocessed data; A second characteristic data screening unit, used to screen out the second characteristic data with the highest degree of distinction for fault identification from the first characteristic data; A fault analysis unit, used to analyze the fault probability and fault type of each target component of the crane by using the screened second characteristic data; A logic unit is used to associate the data acquired by the sensor with the corresponding part, and to determine the fault probability level of the target part of the crane according to the analysis result output by the fault analysis unit. When the fault probability of the target part reaches or exceeds the preset level, the early warning unit is controlled to issue a corresponding early warning prompt according to the fault type of the target part; The early warning unit is used to perform corresponding early warning operations according to the judgment results output by the logic unit.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the lifting machinery operation and maintenance method based on digital twin as described in any one of claims 1 to 8 is implemented.

Images