Crane full life cycle evaluation method based on digital twinning

Abstract

The invention discloses a crane full life cycle evaluation method based on digital twinning, and relates to the technical field of cranes, the method comprises the following steps: determining a primary loss structural component set and a secondary loss structural component set; constructing an initial digital twin structure model; performing time sequence updating on the initial digital twinning structure model to obtain a digital twinning structure monitoring model sequence; performing multi-receptive-field space-time state interaction enhancement on the digital twin structure monitoring model sequence, and determining digital twin space-time state features of the target crane; and carrying out life evaluation based on the digital twin space-time state characteristics of the target crane, and determining a full life cycle stage and a life evaluation result of the target crane. The technical problems that in the prior art, crane operation data analysis depth is not enough, so that full life cycle evaluation is not comprehensive, and hysteresis exists are solved. The technical effects of carrying out multi-level component interaction analysis on the crane and improving the accuracy of full-life-cycle evaluation are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cranes, in particular to a crane full life cycle evaluation method based on digital twinning. BACKGROUND

[0002] At present, the life cycle evaluation of cranes mainly relies on traditional experience-based inspection and regular maintenance plans, and the state information of crane components cannot be obtained in real time. Although there are some intelligent monitoring systems, these systems mostly focus on the monitoring of certain specific components, and the overall performance and life evolution of the crane structure are not fully considered, especially the comprehensive performance under dynamic load, environmental influence and various working conditions. How to comprehensively monitor and accurately evaluate the life and state of each component of the crane, and then optimize the maintenance plan and reduce sudden failures, has become an important research direction of intelligent management of cranes.

[0003] In the life evaluation process of the existing technology, the main dependence is on the experience-based wear prediction model, which cannot accurately capture the synergistic evolution effect of multiple levels and components, especially the change law under the influence of complex load and environmental factors. In addition, the traditional evaluation method mainly relies on manual inspection and periodic data collection, and cannot accurately reflect the dynamic performance and real-time state of the crane in operation in real time, which makes the prediction of faults and equipment maintenance have a certain lag.

[0004] The existing technology has the technical problem of insufficient depth of crane operation data analysis, resulting in incomplete full life cycle evaluation and lag. SUMMARY

[0005] The present application provides a crane full life cycle evaluation method based on digital twinning, which is used to solve the technical problem of insufficient depth of crane operation data analysis in the prior art, resulting in incomplete full life cycle evaluation and lag.

[0006] In view of the above problems, the present application provides a crane full life cycle evaluation method based on digital twinning, which comprises: Obtaining the structure parameters, structure component attributes and working condition boundaries of the target crane, performing historical full life cycle wear data mining based on the structure parameters, structure component attributes and working condition boundaries of the target crane, identifying the primary wear structure components and secondary wear structure components in the structure component set according to the obtained historical full life cycle wear data set, and determining the primary wear structure component set and the secondary wear structure component set. construct an initial digital twin structure model by taking the first-level loss structure component set as a first-level digital twin node set, taking the second-level loss structure component set as a second-level digital twin node set, and representing the physical association relationship between the structure components in the first-level loss structure component set and the second-level loss structure component set as edges; deploy the initial digital twin structure model on a digital twin platform, the digital twin platform receives operation monitoring data collected from the first-level loss structure component set and the second-level loss structure component set at a preset monitoring frequency and performs time sequence updating on the initial digital twin structure model according to the operation monitoring data, and obtains a digital twin structure monitoring model sequence; perform multi-receptive field space-time state interaction enhancement on the digital twin structure monitoring model sequence to determine a target crane digital twin space-time state feature; perform life assessment based on the target crane digital twin space-time state feature to determine a full life cycle stage and a target crane life assessment result.

[0007] The one or more technical solutions provided in the present application have at least the following technical effects or advantages: The present application obtains the structure parameters, structure component attributes and working condition boundaries of the target crane, performs historical full life cycle loss data mining based on the structure parameters, structure component attributes and working condition boundaries of the target crane, identifies the first-level loss structure component and the second-level loss structure component in the structure component set according to the obtained historical full life cycle loss data set, and determines the first-level loss structure component set and the second-level loss structure component set. The first-level loss structure component set is taken as a first-level digital twin node set, the second-level loss structure component set is taken as a second-level digital twin node set, and the physical association relationship between the structure components in the first-level loss structure component set and the second-level loss structure component set is represented as edges to construct an initial digital twin structure model. The initial digital twin structure model is deployed on a digital twin platform, the digital twin platform receives operation monitoring data collected from the first-level loss structure component set and the second-level loss structure component set at a preset monitoring frequency and performs time sequence updating on the initial digital twin structure model according to the operation monitoring data, and obtains a digital twin structure monitoring model sequence. The digital twin structure monitoring model sequence is subjected to multi-receptive field space-time state interaction enhancement to determine a target crane digital twin space-time state feature. Life assessment is performed based on the target crane digital twin space-time state feature to determine a full life cycle stage and a target crane life assessment result. The technical effect of multi-level component interaction analysis of the crane and improvement of the full life cycle assessment accuracy is achieved. BRIEF DESCRIPTION OF DRAWINGS

[0008] ATTACHMENT Figure 1This is a flow chart of a crane life cycle assessment method based on digital twins provided in an embodiment of the present invention.

[0009] Attachment Figure 2 It is a flow chart of determining a primary loss structural component set and a secondary loss structural component set in a crane life cycle assessment method based on digital twins provided by an embodiment of the present invention.

[0010] Attachment Figure 3 This is a flow chart of determining the spatiotemporal state characteristics of a target crane digital twin in a crane life cycle assessment method based on digital twins provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0011] Below in conjunction with specific embodiment, further set forth the present invention.Should be understood that these embodiments are only used to illustrate the present invention and are not used in limiting the scope of the present invention.In addition, should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall equally within the scope limited by the appended claims of the application.

[0012] It should be noted that the terms "including" and "having" are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products or devices.

[0013] Examples, such as the attached Figure 1 As shown, the present application provides a crane full life cycle assessment method based on digital twins, wherein the method includes: Step S1000: Acquire structural parameters, structural component attributes, and operating condition boundaries of a target crane, perform historical lifecycle loss data mining based on the structural parameters, structural component attributes, and operating condition boundaries of the target crane, identify primary loss structural components and secondary loss structural components in a structural component set based on the historical lifecycle loss data set obtained through mining, and determine a primary loss structural component set and a secondary loss structural component set; Further, such as Figure 2 As shown, the structural parameters, structural component attributes, and operating condition boundaries of the target crane are obtained, historical life cycle loss data mining is performed based on the structural parameters, structural component attributes, and operating condition boundaries of the target crane, and the primary loss structural components and secondary loss structural components in the structural component set are identified based on the historical life cycle loss data set obtained by mining, and the primary loss structural component set and the secondary loss structural component set are determined. In this embodiment of the application, step S1000 further includes: Step S1100: data extraction is performed on the historical full life cycle loss data set based on the loss structure component, the loss degree and the use time length, to obtain M loss structure components, M loss degree sets and M loss structure component use time length sets; Step S1200: mean value tracking screening is performed on the M loss degree sets and the M loss structure component use time length sets respectively, to determine M loss degree screening values and M loss structure component use time length screening values; Step S1300: the first level loss structure component and the second level loss structure component in the structure component set are identified according to the M loss degree screening values and the M loss structure component use time length screening values, to determine a first level loss structure component set and a second level loss structure component set.

[0014] In one possible embodiment, the structure parameter represents the physical characteristics of each component of the crane, such as size, weight, material properties, load bearing capacity, etc. The structure component attribute represents the functional characteristics, service life, manufacturing material, durability, etc. of each component of the crane. The working condition boundary represents the range of external conditions that the crane is subjected to during use, including load, environmental factors, work cycle, etc. The first level loss structure component represents the component that experiences higher loss in the full life cycle of the crane, which is usually the key component of the crane that bears a large load (such as the boom, tower, etc.). The second level loss structure component is usually a component that bears a smaller load or wears slower in use (such as the electrical system, control system, etc.).

[0015] Preferably, the structure parameter, the structure component attribute and the working condition boundary are used as indexes to search in the big data, to obtain the loss condition of the crane under the same condition as the target crane, that is, the historical full life cycle loss data set. Then, the loss structure component, the loss degree and the use time length are used as indexes to search in the historical full life cycle loss data set, to obtain the corresponding data in the loss data, that is, M loss structure components, M loss degree sets and M loss structure component use time length sets. Wherein, M is a positive integer.

[0016] Further, the obtained data needs to be screened to determine the most representative data in the M loss degree sets and the M loss structure component use duration sets. Preferably, the M loss degree screening values and the M loss structure component use duration screening values are obtained by mean tracking screening of the M loss degree sets and the M loss structure component use duration sets. The loss degree screening value reflects the general use loss degree of each loss structure component. The loss structure component use duration screening value reflects the general use duration of each loss structure component when loss occurs. By analyzing the M loss degree screening values and the M loss structure component use duration screening values, the structure components that have experienced more loss, i.e., the structure components that need to be closely monitored, are determined as primary loss structure components, and further, in combination with the structure component set, the remaining structure components are determined as secondary loss structure components.

[0017] Further, based on the loss structure components, the loss degrees, and the use durations, data extraction is performed on the historical full-life-cycle loss data set to obtain M loss structure components, M loss degree sets, and M loss structure component use duration sets. The step S1100 of the embodiment of the present application further includes: Step S1101: Based on the loss structure components, the loss degrees, and the use durations, data extraction is performed on the historical full-life-cycle loss data set to obtain a loss structure component set, a loss degree set, and a use duration set. Step S1102: The loss structure component set is aggregated by the same type to obtain M loss structure components. Step S1103: Based on the M loss structure components, the loss degree set and the use duration set are mapped and aggregated to obtain the M loss degree sets and the M loss structure component use duration sets.

[0018] In one embodiment, by extracting the loss degree and use duration data related to the structure components from the historical loss data set, the loss structure component set, the loss degree set, and the use duration set are obtained, which can quantify the loss condition of each structure component and form a comprehensive loss data set to support the identification of primary and secondary loss components and subsequent evaluation. Components with similar loss characteristics, working conditions, or structural characteristics are classified into the same category, and M loss structure components are screened out.

[0019] The M loss structure components are mapped and aggregated with their corresponding loss degree sets and use duration sets. Through this mapping, the system can integrate the loss degree and use duration information of each loss structure component together to form the corresponding aggregated data. Through aggregated analysis, complete loss information can be provided for each structure component, and it is used as the basis for subsequent loss screening and evaluation.

[0020] Further, according to the M loss degree screening values and M loss structure component use time screening values, a first loss structure component and a second loss structure component in the structure component set are identified, a first loss structure component set and a second loss structure component set are determined, and the embodiment of the present application further includes the following steps S1300: Step S1301: The reciprocals of the M loss degree screening values and the M loss structure component use time screening values are respectively weighted and fused to determine M loss structure component attention coefficients; Step S1302: The loss structure components corresponding to the loss structure component attention coefficients greater than or equal to a preset coefficient threshold in the M loss structure component attention coefficients are added to the first loss structure component set. Step S1303: The loss structure components corresponding to the loss structure component attention coefficients less than the preset coefficient threshold in the M loss structure component attention coefficients are added to the second loss structure component set.

[0021] In one embodiment, the loss degree screening value set is first subjected to a reciprocal operation, so that the components with lighter loss have a larger weight, and then the use time screening value of each loss structure component is combined, and the two values are weighted and combined to obtain a loss structure component attention coefficient. By calculating the attention coefficient, different weights are given to different components in the loss evaluation, so that important loss components are given sufficient attention in the subsequent screening process.

[0022] A preset coefficient threshold is set by a person skilled in the art, and then the components with attention coefficients greater than or equal to the threshold are identified as first loss structure components, and these components are added to the first loss structure component set. The first loss structure components screened out are removed from the structure component set, and the remaining components are second loss structure components. These second loss structure components are usually lighter in loss, but still play a certain role in the operation of the entire system. By accurately dividing the loss levels of the components, more accurate component classification is provided for subsequent health evaluation, life prediction and maintenance planning. Thus, more attention is focused on the first loss components that have the greatest impact on the system, while ensuring that the evaluation of the second loss components is not ignored.

[0023] Further, the M loss degree sets and the M loss structure component use time sets are respectively subjected to mean tracking screening to determine M loss degree screening values and M loss structure component use time screening values, and the embodiment of the present application further includes the following steps S1200: Step S1201: A first loss degree set is extracted from the M loss degree sets. Step S1202: calculate the mean value of the first loss degree set, and take the mean value as a first mean value tracking screening value and construct a first mean value tracking screening center neighborhood; Step S1203: randomly extract a first loss degree from the edge of the first mean value tracking screening center neighborhood as a first iterative tracking screening value, and construct a first iterative tracking screening center neighborhood of the first iterative tracking screening value; Step S1204: compare the first iterative tracking screening center neighborhood and the first mean value tracking screening center neighborhood to determine the mean value tracking screening direction and the first stage tracking screening value; Step S1205: iteratively screen the first stage tracking screening value in the first loss degree set according to the mean value tracking screening direction until a preset number of iterative screenings is met to determine a first loss degree screening value; Step S1206: iteratively screen the M loss degree sets and the M loss structure component usage time sets using the mean value tracking screening to determine M loss degree screening values and M loss structure component usage time screening values.

[0024] Further, comparing the first iterative tracking screening center neighborhood and the first mean value tracking screening center neighborhood to determine the mean value tracking screening direction and the first stage tracking screening value, the step S1204 of the embodiments of the present application further comprises: Step S1204-1: calculate the neighborhood density of the first iterative tracking screening center neighborhood and the first mean value tracking screening center neighborhood respectively to obtain the first iterative tracking screening center neighborhood density and the first mean value tracking screening center neighborhood density; Step S1204-2: determine whether the first iterative tracking screening center neighborhood density is greater than or equal to the first mean value tracking screening center neighborhood density, if yes, take the direction from the first mean value tracking screening value to the first iterative tracking screening value as the mean value tracking screening direction, and take the first iterative tracking screening value as the first stage tracking screening value; Step S1204-3: if not, take the direction from the first iterative mean value tracking screening value to the first mean value tracking screening value as the mean value tracking screening direction, and take the first mean value tracking screening value as the first stage tracking screening value.

[0025] Further, take the first mean value tracking screening value as the screening center, take the preset mean value tracking screening bandwidth as the screening radius, and add the first loss degree with a distance to the screening center within the screening radius to the first mean value tracking screening center neighborhood.

[0026] In one embodiment, a first wear degree set is randomly extracted from the M wear degree sets as an example. The mean of the first wear degree set is calculated and the mean is taken as a first mean tracking screening value. The first mean tracking screening value reflects the general wear condition of the first wear degree set. A first mean tracking screening central neighborhood is constructed based on the first mean tracking screening value. The first mean tracking screening central neighborhood reflects the wear degree condition gathered around the first mean tracking screening value.

[0027] Preferably, with the first mean tracking screening value as the screening center and a preset mean tracking screening bandwidth preset by those skilled in the art as the screening radius, if the distance of the first wear degree to the screening center is within the screening radius range, it indicates that it is gathered near the screening center, and it is added to the first mean tracking screening central neighborhood. Assuming that the first wear degree of a component is 9, the distance between it and the first mean tracking screening value 10 is calculated to be 1, which is within the screening radius 2 range, so the wear degree of the component will be considered to be gathered near the screening center and added to the screening central neighborhood. For example, assuming that the first wear degree of another component is 7, the distance between it and the screening center 10 is calculated to be 3, which exceeds the screening radius 2, so the component will not be added to the screening central neighborhood.

[0028] A first wear degree is randomly extracted from the edge of the first mean tracking screening central neighborhood and taken as a first iterative tracking screening value. A first iterative tracking screening central neighborhood of the first iterative tracking screening value is constructed based on the same principle as obtaining the first mean tracking screening central neighborhood. Further, by comparing the neighborhood density of the first iterative tracking screening central neighborhood and the first mean tracking screening central neighborhood, the more representative value between the first iterative tracking screening value and the first mean tracking screening value is taken as the first stage tracking screening value.

[0029] Preferably, the data amount in the first mean tracking screening central neighborhood is counted and the data amount is divided by 2 times the screening radius to obtain the first iterative tracking screening central neighborhood density. The first iterative tracking screening central neighborhood density reflects the data gathering condition around the first iterative tracking screening value. Based on the same obtaining principle, the first mean tracking screening central neighborhood density is calculated.

[0030] Further, it is judged whether the first iterative tracking screening central neighborhood density is greater than or equal to the first mean tracking screening central neighborhood density. If yes, the first iterative tracking screening value is more representative, the direction from the first mean tracking screening value to the first iterative tracking screening value is taken as the mean tracking screening direction, and the first iterative tracking screening value is taken as the first stage tracking screening value.

[0031] If not, the first mean tracking filter value is more representative, and the direction from the first iteration mean tracking filter value to the first mean tracking filter value is taken as the mean tracking filter direction, and the first mean tracking filter value is taken as the first stage tracking filter value.

[0032] The first stage tracking filter value is iteratively filtered in the first loss degree set according to the mean tracking filter direction until a preset iteration filtering number, such as 30 times, 50 times, etc., is met, and a first loss degree filter value is determined. Based on the same principle of obtaining the first loss degree filter value, the M loss degree sets and the M loss structure component use time length sets are mean tracked, and M loss degree filter values and M loss structure component use time length filter values are determined.

[0033] Step S2000: taking the primary loss structure component set as the primary digital twin node set, taking the secondary loss structure component set as the secondary digital twin node set, and taking the physical association relationship between the structure components in the primary loss structure component set and the secondary loss structure component set as the edge to construct an initial digital twin structure model; In one embodiment, the filtered primary loss structure components and secondary loss structure components are taken as digital twin nodes respectively, and the physical association relationship between these nodes is taken as the edge to construct a preliminary digital twin structure model. First, the primary loss structure components and the secondary loss structure components are taken as the primary and secondary node sets of the digital twin model respectively, and these nodes represent the components in the physical crane. Then, based on the physical relationship between the components (such as bearing, transmission, movement, etc.), corresponding edges are created in the model to show how these components interact and cooperate.

[0034] After constructing the initial digital twin structure model, the model will have a virtual copy of the crane, which can reflect the state, behavior and interaction of each component in real time. The role of this step is to provide an accurate virtual structure basis for subsequent dynamic monitoring, loss evaluation, life prediction, etc., and to provide support for real-time updating and feedback of the digital twin model. In this way, the virtual copy of the crane can be continuously updated synchronously with the physical system to realize comprehensive life cycle monitoring and intelligent decision support.

[0035] Step S3000: deploying the initial digital twin structure model to a digital twin platform, the digital twin platform receiving operation monitoring data collected by the primary loss structure component set and the secondary loss structure component set according to a preset monitoring frequency and updating the initial digital twin structure model in time sequence according to the operation monitoring data to obtain a digital twin structure monitoring model sequence; In one possible embodiment, the initial digital twin structural model represents a virtual model constructed based on the crane's structure, component attributes, and physical connection relationships, representing the complete structure and state of the crane. The operational monitoring data represents real-time data collected from various parts of the crane, such as the boom, wire rope, tower, etc., typically including temperature, stress, vibration, load, and other sensor data. The preset monitoring frequency represents the data collection frequency set on the digital twin platform, such as every minute, every hour, every day, etc., used to define the time interval for data updates.

[0036] The digital twin platform starts receiving real-time operational monitoring data from the primary loss structure component set (such as the boom, tower) and the secondary loss structure component set (such as the electrical system, sensors, etc.) according to the preset monitoring frequency (e.g., collecting data every hour). These data may include various real-time sensor data such as temperature, vibration, stress, load, etc. Assuming at a certain moment, the digital twin platform receives stress sensor data of the boom (e.g., the load borne by the boom is 500 kg) and vibration frequency data of the wire rope (e.g., the vibration frequency is 20 Hz). The received operational monitoring data is input into the initial digital twin structural model and updated in time sequence according to the latest data. Each time the platform updates, it calculates the state of each component at the current moment and compares it with the previous state, forming a dynamic time sequence update model.

[0037] Assuming that at the previous moment, the stress of the boom is 490 kg and the vibration frequency of the wire rope is 19 Hz. At the next moment, the data collection shows that the stress changes to 500 kg and the vibration frequency changes to 20 Hz. The digital twin platform updates the model in time sequence, updates the state of the boom and wire rope through comparison and weighting processing, and ensures that the model reflects the current state.

[0038] After each time sequence update, the digital twin platform saves the current update result as new time point data and forms a sequence of digital twin structural monitoring models. This sequence represents the health status and performance changes of each component of the crane at different time points. Over time, the monitoring model sequence grows and provides in-depth analysis of the long-term health status of the crane.

[0039] Over time, the digital twin platform accumulates data at multiple time points. For example, at time point t1, the stress of the boom is 490 kg; at time point t2, the stress of the boom is 500 kg; at time point t3, the stress of the boom is 510 kg. After each data update, the platform generates a new monitoring model and records it, forming a state evolution trajectory of the crane.

[0040] By deploying the initial digital twin structure model to the digital twin platform and continuously updating it through the collected operation monitoring data, a sequence of digital twin structure monitoring models is finally obtained, which provides comprehensive dynamic state feedback for the crane.

[0041] Step S4000: Multi-receptive field spatio-temporal state interactive enhancement is performed on the sequence of digital twin structure monitoring models to determine the target crane digital twin spatio-temporal state feature. Further, as shown in Figure 3 Step S4000: Multi-receptive field spatio-temporal state interactive enhancement is performed on the sequence of digital twin structure monitoring models to determine the target crane digital twin spatio-temporal state feature. Step S4100: Long-term receptive field spatial state analysis is performed on the sequence of digital twin structure monitoring models with the first-level digital twin node as an index to determine a first-level digital twin node spatial state feature set. Step S4200: Short-term receptive field temporal state analysis is performed on the sequence of digital twin structure monitoring models with the second-level digital twin node as an index to determine a second-level digital twin node temporal state feature set. Step S4300: The first-level digital twin node spatial state feature set and the second-level digital twin node temporal state feature set are respectively preprocessed to obtain a first-level digital twin node preprocessed spatial state feature and a second-level digital twin node preprocessed temporal state feature. Step S4400: Spatio-temporal state interactive enhancement is performed on the first-level digital twin node preprocessed spatial state feature and the second-level digital twin node preprocessed temporal state feature to determine the target crane digital twin spatio-temporal state feature.

[0042] Further, spatio-temporal state interactive enhancement is performed on the first-level digital twin node preprocessed spatial state feature and the second-level digital twin node preprocessed temporal state feature to determine the target crane digital twin spatio-temporal state feature, and the step S4400 of the embodiment of the application further includes: Step S4401: Similarities of the first-level digital twin node preprocessed spatial state feature and the second-level digital twin node preprocessed temporal state feature are calculated to determine a spatio-temporal state feature similarity set. Step S4402: Normalization processing is performed on the spatio-temporal state feature similarity set to construct a spatio-temporal adjacency matrix. Step S4403: Interactive enhancement is performed on the first-level digital twin node preprocessed spatial state feature by using the spatio-temporal adjacency matrix to determine the target crane digital twin spatio-temporal state feature.

[0043] In one embodiment, first, the long-term receptive field spatial state analysis is performed on the digital twin structure monitoring model sequence indexed by the primary digital twin nodes (such as components with heavy load), and a state feature set related to the spatial position of the key components is extracted. Second, the short-term receptive field time state analysis is performed on the digital twin structure monitoring model sequence indexed by the secondary digital twin nodes (such as auxiliary components or components with light load), and a feature related to the short-term dynamic state of the components is extracted. The short-term receptive field can help capture the rapidly changing working conditions, such as sudden load changes, short-term wear caused by frequent operation, etc. Then, the extracted primary digital twin node spatial state feature set and secondary digital twin node time state feature set are subjected to mean normalization processing to standardize the data, and finally the preprocessed primary digital twin node spatial state feature and preprocessed secondary digital twin node time state feature are obtained.

[0044] For example, the long-term receptive field spatial state analysis is performed on the boom (primary digital twin node), and the gradual fatigue state of the material is identified by analyzing the stress change of the boom under load for a long time. For example, the load bearing capacity of the boom may gradually decrease after a long time of operation, resulting in higher load stress. The short-term receptive field time state analysis is performed on the electrical control system (such as a secondary digital twin node), and the load fluctuation of the system in a short time is analyzed. For example, current transient overload may cause short-term wear of electrical components, which cannot be discovered by long-term observation and needs to be concerned about transient fluctuations.

[0045] Preferably, the spatial feature extraction is performed on the monitoring data sequence corresponding to the primary digital twin node in the digital twin structure monitoring model sequence according to the long-term receptive field preset by the person skilled in the art, and the model constructed by the neural network with the long-term receptive field is optionally used for extraction, to obtain the primary digital twin node spatial state feature set. Further, the model constructed by the neural network with the short-term receptive field is used to extract the monitoring data sequence corresponding to the secondary digital twin node in the digital twin structure monitoring model sequence, to obtain the secondary digital twin node time state feature set.

[0046] Further, the mean values of the primary digital twin node spatial state feature set and the secondary digital twin node time state feature set are calculated respectively to determine the state of the primary and secondary components in the structure component set of the target crane, and the preprocessed primary digital twin node spatial state feature and preprocessed secondary digital twin node time state feature are obtained.

[0047] Further, the preprocessed primary digital twin node spatial state feature and the preprocessed secondary digital twin node time state feature will be enhanced by space-time state interaction. The collaborative changes of each component under different time and space conditions are captured by the way of interaction enhancement.

[0048] Preferably, the similarity of the primary digital twin node pretreatment space state feature and the secondary digital twin node pretreatment time state feature is calculated to determine a spatio-temporal state feature similarity set. The spatio-temporal state feature similarity set reflects the similarity degree of the features between the primary digital twin node and the secondary digital twin node. Then, the spatio-temporal state feature similarity set is normalized by using a softmax function, and the processed data is added to an initially empty matrix to construct a spatio-temporal adjacency matrix.

[0049] The target crane digital twin spatio-temporal state feature is determined by using a graph convolutional neural network model to interactively analyze the spatio-temporal adjacency matrix and the primary digital twin node pretreatment space state feature. The spatio-temporal adjacency matrix reflects the similarity between the components of the primary digital twin node and the secondary digital twin node. By using graph data processing methods such as graph convolution, the dependence between nodes is enhanced, and the extraction of spatio-temporal state features is optimized. For example, if there is a high similarity between the boom and the electrical control system, their state features will be more closely linked, ensuring that the final output of the spatio-temporal state feature can accurately reflect the overall health status of the crane.

[0050] Step S5000: Based on the target crane digital twin spatio-temporal state feature, life assessment is performed to determine the full life cycle stage and the target crane life assessment result.

[0051] Further, a full life cycle evaluator is pre-constructed, and the full life cycle evaluator is used to perform life assessment on the target crane digital twin spatio-temporal state feature to determine the full life cycle stage and the target crane life assessment result.

[0052] In one possible embodiment, the full life cycle evaluator includes an LSTM layer, a CNN layer, and a fully connected layer, the input data is the target crane digital twin spatio-temporal state feature, and the output data is the full life cycle stage and the target crane life assessment result. The LSTM layer is used to process time series data and extract important features in the time dimension. The CNN layer is used to process spatial features and extract patterns in local regions. The fully connected layer maps the extracted features to the final output, including life assessment and full life cycle stage.

[0053] The Adam optimizer is used to adaptively adjust the learning rate during the learning process. The dataset is obtained, wherein the dataset includes a sample target crane digital twin spatiotemporal state feature set, a sample full life cycle stage set and a sample target crane life assessment result set. The dataset is divided into a training set and a validation set. Generally, 80% of the data can be used for training, and 20% of the data can be used for verifying the effect of the model. The training set and the validation set are used to train the full life cycle evaluator in combination with the Adam optimizer. Preferably, during the training process, small batch gradient descent is used for training, and the batch size is generally set to 32 or 64. If the loss of the validation set no longer decreases in continuous rounds of training, the training is stopped in advance.

[0054] Illustratively, the target crane digital twin spatiotemporal state feature is input into the trained full life cycle evaluator, and the life assessment result output by the model is that the remaining service life is 2000 hours, the full life cycle stage is the gradual decline stage, and the health status of the boom and the control system needs to be focused on. By extracting the health information of the equipment from the digital twin spatiotemporal state feature, the remaining service life and the life cycle stage of the equipment are predicted, and the technical effect of reliably evaluating the full life cycle of the crane and improving the evaluation accuracy is achieved.

[0055] In summary, the embodiments of the present application have at least the following technical effects: 1. The present application can accurately identify the key components (such as booms, towers, etc.) that have a greater impact on the performance and life of the crane by identifying the primary and secondary loss components based on historical full life cycle loss data. The technical effect of providing high-quality data support for subsequent digital twin model construction and life assessment is achieved.

[0056] 2. The present application can achieve accurate remaining life prediction and full life cycle stage determination by receiving real-time operation monitoring data from the primary and secondary loss components through the timing update mechanism of the digital twin platform, updating the initial digital twin structure model according to these data, and performing life assessment according to the digital twin spatiotemporal state feature of the target crane. The technical effects of obtaining the latest state of the target crane and providing full life cycle evaluation reliability and timeliness are achieved.

[0057] It should be noted that the above sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. The above describes specific embodiments of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or can be advantageous.

[0058] The above description is only the preferred embodiment of the present application, and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0059] The specification and drawings are only exemplary and illustrative of the present application and are to be considered within the scope of the present application. Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and equivalent technology, the present application is intended to include these modifications and variations.

Claims

1. A crane life cycle assessment method based on digital twin, characterized by: The method comprises: Acquire the structural parameters, structural component attributes, and operating condition boundaries of the target crane, conduct historical life cycle loss data mining based on the structural parameters, structural component attributes, and operating condition boundaries of the target crane, identify primary loss structural components and secondary loss structural components in the structural component set based on the historical life cycle loss data set obtained through mining, and determine the primary loss structural component set and the secondary loss structural component set; The first-level loss structure component set is used as the first-level digital twin node set, the second-level loss structure component set is used as the second-level digital twin node set, and the physical association relationship between each structural component in the first-level loss structure component set and the second-level loss structure component set is represented as an edge to construct an initial digital twin structure model; The initial digital twin structure model is deployed on a digital twin platform. The digital twin platform receives operation monitoring data collected from the first-level loss structure component set and the second-level loss structure component set according to a preset monitoring frequency, and updates the initial digital twin structure model in a time series according to the operation monitoring data to obtain a digital twin structure monitoring model sequence. Performing multi-receptive field spatiotemporal state interactive enhancement on the digital twin structure monitoring model sequence to determine the spatiotemporal state characteristics of the target crane digital twin; A lifespan assessment is performed based on the spatiotemporal state characteristics of the digital twin of the target crane to determine the full lifecycle stages and the lifespan assessment results of the target crane.

2. A crane life cycle assessment method based on digital twins according to claim 1, characterized in that: Obtain the structural parameters, structural component attributes and working condition boundaries of the target crane, perform historical life cycle loss data mining based on the structural parameters, structural component attributes and working condition boundaries of the target crane, identify the first-level loss structural components and the second-level loss structural components in the structural component set based on the historical life cycle loss data set obtained by mining, and determine the first-level loss structural component set and the second-level loss structural component set, including: Extracting data from the historical full life cycle loss data set based on lossy structural components, loss degrees, and usage durations to obtain M lossy structural components, M loss degree sets, and M lossy structural components usage duration sets; Performing mean tracking screening on the M loss degree sets and the M loss structure component usage time sets respectively to determine M loss degree screening values ​​and M loss structure component usage time screening values; According to the M loss degree screening values ​​and the M loss structure component usage time screening values, the first-level loss structure components and the second-level loss structure components in the structure component set are identified, and the first-level loss structure component set and the second-level loss structure component set are determined.

3. A crane life cycle assessment method based on digital twins according to claim 2, characterized in that: Data extraction is performed on the historical full life cycle loss data set based on loss structural components, loss degrees, and usage duration to obtain M loss structural components, M loss degree sets, and M loss structural components usage duration sets, including: Extracting data from the historical full life cycle loss data set based on loss structural components, loss degree, and usage duration to obtain a loss structural component set, a loss degree set, and a usage duration set; Aggregate the lossy structural component set by the same type to obtain M lossy structural components; The loss degree sets and the usage duration sets are mapped and aggregated based on the M loss structure components to obtain the M loss degree sets and the M loss structure component usage duration sets.

4. A crane life cycle assessment method based on digital twins according to claim 2, characterized in that: Identifying the primary lossy structural components and the secondary lossy structural components in the set of structural components according to the M loss degree screening values ​​and the M lossy structural component usage time screening values, and determining the primary lossy structural component set and the secondary lossy structural component set, including: Performing weighted fusion on the reciprocals of the M loss degree screening values ​​and the M loss structure component usage time screening values ​​respectively to determine the M loss structure component attention coefficients; Adding the lossy structural component corresponding to the lossy structural component attention coefficient greater than or equal to the preset coefficient threshold among the M lossy structural component attention coefficients into the first-level lossy structural component set; The lossy structural component corresponding to the lossy structural component attention coefficient less than the preset coefficient threshold among the M lossy structural component attention coefficients is added to the secondary lossy structural component set.

5. The crane life cycle assessment method based on digital twin according to claim 2, characterized in that: Performing mean tracking screening on the M loss degree sets and the M loss structure component usage time sets respectively to determine M loss degree screening values ​​and M loss structure component usage time screening values, including: Extracting a first loss degree set from the M loss degree sets; Calculating the mean of the first loss degree set, taking the mean as a first mean tracking screening value, and constructing a first mean tracking screening center neighborhood; Randomly extracting a first loss degree from the edge of the first mean tracking and filtering center neighborhood as a first iterative tracking and filtering value, and constructing a first iterative tracking and filtering center neighborhood of the first iterative tracking and filtering value; Comparing the first iterative tracking and screening center neighborhood with the first mean tracking and screening center neighborhood to determine a mean tracking and screening direction and a first-stage tracking and screening value; Iteratively screening the first-stage tracking and screening value in the first loss degree set according to the mean tracking and screening direction until a preset number of iterative screening times is met, and determining a first loss degree screening value; Similarly, mean tracking screening is performed on the M loss degree sets and the M loss structure component usage time sets to determine M loss degree screening values ​​and M loss structure component usage time screening values.

6. A crane life cycle assessment method based on digital twins according to claim 5, characterized in that: Comparing the first iterative tracking and filtering center neighborhood with the first mean tracking and filtering center neighborhood to determine the mean tracking and filtering direction and the first stage tracking and filtering value includes: Calculating the neighborhood density of the first iterative tracking and filtering center neighborhood and the first mean tracking and filtering center neighborhood respectively to obtain the first iterative tracking and filtering center neighborhood density and the first mean tracking and filtering center neighborhood density; Determine whether the first iterative tracking and screening center neighborhood density is greater than or equal to the first mean tracking and screening center neighborhood density; if so, use the direction from the first mean tracking and screening value to the first iterative tracking and screening value as the mean tracking and screening direction, and use the first iterative tracking and screening value as the first stage tracking and screening value; If not, the direction from the first iteration mean tracking screening value to the first mean tracking screening value is used as the mean tracking screening direction, and the first mean tracking screening value is used as the first stage tracking screening value.

7. A crane life cycle assessment method based on digital twins according to claim 6, characterized in that: Taking the first mean tracking filtering value as the filtering center and the preset mean tracking filtering bandwidth as the filtering radius, the first loss degree whose distance to the filtering center is within the filtering radius is added to the neighborhood of the first mean tracking filtering center.

8. The crane life cycle assessment method based on digital twin according to claim 1, characterized in that: The digital twin structure monitoring model sequence is subjected to multi-receptive field spatiotemporal state interactive enhancement to determine the spatiotemporal state characteristics of the target crane digital twin, including: Taking the first-level digital twin node as the index, a long-term receptive field spatial state analysis is performed on the digital twin structure monitoring model sequence to determine the first-level digital twin node spatial state feature set; Taking the secondary digital twin node as the index, a short-term receptive field time state analysis is performed on the digital twin structure monitoring model sequence to determine the time state feature set of the secondary digital twin node; Preprocessing the spatial state feature set of the first-level digital twin node and the temporal state feature set of the second-level digital twin node respectively to obtain the preprocessed spatial state feature of the first-level digital twin node and the preprocessed temporal state feature of the second-level digital twin node; The spatial state characteristics of the first-level digital twin node preprocessing and the time state characteristics of the second-level digital twin node preprocessing are interactively enhanced to determine the spatial and temporal state characteristics of the target crane digital twin.

9. A crane life cycle assessment method based on digital twins according to claim 8, characterized in that: Performing spatiotemporal state interactive enhancement on the pre-processed spatial state features of the first-level digital twin node and the pre-processed temporal state features of the second-level digital twin node to determine the spatiotemporal state features of the target crane digital twin, including: Calculate the similarity between the pre-processed spatial state features of the first-level digital twin node and the pre-processed temporal state features of the second-level digital twin node, and determine a spatiotemporal state feature similarity set; Normalizing the spatiotemporal state feature similarity set to construct a spatiotemporal adjacency matrix; The spatiotemporal adjacency matrix is ​​used to interactively enhance the pre-processed spatial state characteristics of the first-level digital twin node to determine the spatiotemporal state characteristics of the target crane digital twin.

10. The crane life cycle assessment method based on digital twin according to claim 1, characterized in that: A full life cycle evaluator is pre-built, and the life cycle evaluator is used to evaluate the spatiotemporal state characteristics of the digital twin of the target crane to determine the full life cycle stage and the life evaluation results of the target crane.

Images