Crane monitoring system and prediction-based crane maintenance method

Abstract

The invention provides a crane monitoring system and a crane maintenance method based on prediction, and the method comprises the steps: obtaining first structural data and first vibration data of a target structural member, judging whether the first structural data has a damage feature, and when the first structural data does not have the damage feature, carrying out the maintenance of the target structural member; predicting second structure data and second vibration data in a period of time in the future, and performing corresponding analysis on the second structure data and the second vibration data to judge whether the second structure data and the second vibration data meet a preset corresponding relation or not, and when the second structure data and the second vibration data meet a preset corresponding relationship, determining that the target structural member has a damage risk, calculating a damage risk level of the target structural member according to the second vibration data, generating a maintenance strategy of the target structural member according to the damage risk level, and maintaining the target structural member according to the maintenance strategy. The physical damage of the crane can be accurately predicted, and an effective maintenance strategy is provided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of crane maintenance, in particular to a crane monitoring system and a crane maintenance method based on prediction. BACKGROUND

[0002] The crane, also known as a hoist, is a transport auxiliary device that can horizontally or vertically move heavy objects from one position to another within a certain range, and is widely used in transport ports, factory workshops or construction sites where heavy objects need to be frequently moved. As a large mechanical device, the crane needs to be regularly maintained to ensure its safety during operation. In order to achieve the heavy object moving capacity, the crane is usually designed as a large mechanical device composed of large mechanical structures and corresponding electrical control systems. Due to the fact that the crane often needs to work in some harsh environments, such as some high-temperature and humid indoor and outdoor environments, combined with the frequent changes in load weight, the mechanical components or electrical components of the crane may be damaged due to various internal and external factors. If not timely repaired and maintained, safety accidents may occur when transporting heavy objects, resulting in adverse consequences such as personnel casualties and property losses.

[0003] Among the various components of the crane, the electrical system can usually identify its working state and health state by real-time acquisition of electrical parameters of each electrical component in the electrical system, so the fault identification and daily maintenance scheme of the electrical system of the crane can be more explicit, i.e. the maintenance scheme has stronger directionality and purpose. The maintenance of the mechanical system of the crane is more difficult than the maintenance of its electrical system, because the crane has a large number of mechanical components, each of which is an important component of the crane to realize the hoisting function. These mechanical components may be damaged, such as wear, aging, deformation, and fracture, under the influence of various internal and external factors during long-term operation. Some damage is subtle and difficult to be observed by maintenance personnel in daily maintenance, but long-term damage accumulation may cause physical failure or functional failure of the crane, and in severe cases, it may even lead to major safety accidents. SUMMARY

[0004] The present application is based on the above problems, and proposes a crane monitoring system and a crane maintenance method based on prediction, which can accurately predict the physical damage of the crane and provide effective maintenance strategies.

[0005] Therefore, the first aspect of the present application provides a crane monitoring system, comprising a structure monitoring sensor configured to monitor a structure change of a target structure of a crane, a vibration monitoring sensor configured to monitor a structure stability of the target structure of the crane, and a processing unit in communication connection with the structure monitoring sensor and the vibration monitoring sensor, wherein the processing unit is configured to:

[0006] obtain first structure data of the target structure and first vibration data corresponding to the first structure data, wherein the first structure data comprises internal first structure data, shape monitoring data and / or surface monitoring data of the target structure;

[0007] perform damage feature identification on the first structure data to determine whether the first structure data has a damage feature;

[0008] when the first structure data does not have a damage feature, predict second structure data and second vibration data in a future period of time;

[0009] perform correspondence analysis on the second structure data and the second vibration data to determine whether the second structure data and the second vibration data satisfy a preset correspondence relationship;

[0010] when the second structure data and the second vibration data satisfy the preset correspondence relationship, determine that the target structure has a damage risk;

[0011] calculate a damage risk level of the target structure according to the second vibration data;

[0012] generate a maintenance strategy of the target structure according to the damage risk level.

[0013] The second aspect of the present application provides a crane maintenance method based on prediction, comprising:

[0014] obtain first structure data of the target structure and first vibration data corresponding to the first structure data, wherein the first structure data comprises internal first structure data, shape monitoring data and / or surface monitoring data of the target structure;

[0015] perform damage feature identification on the first structure data to determine whether the first structure data has a damage feature;

[0016] when the first structure data does not have a damage feature, predict second structure data and second vibration data in a future period of time;

[0017] perform correspondence analysis on the second structure data and the second vibration data to determine whether the second structure data and the second vibration data satisfy a preset correspondence relationship;

[0018] determining that the target structure is at risk of damage when the second structure data and the second vibration data satisfy a preset correspondence relationship;

[0019] calculating a damage risk level of the target structure according to the second vibration data;

[0020] generating a maintenance strategy of the target structure according to the damage risk level.

[0021] Further, the first structure data is a structure image of the target structure, and the step of performing damage feature recognition on the first structure data to determine whether the first structure data has a damage feature specifically includes:

[0022] determining whether there is a structure abnormal point in the structure image;

[0023] generating a local structure image of the target structure at the structure abnormal point when the structure image has the structure abnormal point;

[0024] inputting the local structure image into a pre-trained damage recognition model for damage feature recognition to determine whether the first structure data has a damage feature.

[0025] Further, the step of determining whether there is a structure abnormal point in the structure image specifically includes:

[0026] configuring a standard image corresponding to the target structure and the structure image, the standard image being a structure image of the target structure in a damage-free state;

[0027] obtaining a first structure image sequence of the target structure in a past period of time;

[0028] generating a first difference feature sequence according to difference features between each structure image in the first structure image sequence and the standard image;

[0029] generating a first difference feature size parameter sequence corresponding to the first difference feature sequence, the first difference feature size parameter sequence being composed of size parameters in one or more different directions of each difference feature in the first difference feature sequence;

[0030] performing trend analysis on the first difference feature size parameter sequence to determine whether the corresponding difference feature is a structure abnormal point according to a trend analysis result.

[0031] Further, the step of performing trend analysis on the first difference feature size parameter sequence to determine whether the corresponding difference feature is a structure abnormal point according to a trend analysis result specifically includes:

[0032] obtaining each size parameter p1 i (t) in the first difference feature size parameter sequence, where i∈[1, n p ], i is a positive integer, n p is the number of size parameter types of the difference feature, and the first difference feature size parameter sequence is a discrete data sequence constructed in the order of generation time t of the first structure data structure image;

[0033] for the i-th size parameter, fitting a trend linear function f i (t) of the size parameter p1 i (t) with time t as the independent variable, where f i (t) = k i ×t + b i ;

[0034] determining whether the slope k i in the trend linear function f i (t) is greater than 0;

[0035] when there is any i value that satisfies k i >0 in the range of [1, n p ], it is determined that there is a structure abnormal point in the structure image.

[0036] Further, the step of performing corresponding analysis on the second structure data and the second vibration data to determine whether the second structure data and the second vibration data satisfy a preset corresponding relationship comprises:

[0037] generating a first trend feature of the second structure data and a second trend feature of the second vibration data;

[0038] calculating a matching degree of the first trend feature and the second trend feature;

[0039] determining whether the matching degree of the first trend feature and the second trend feature is greater than a preset matching degree threshold;

[0040] when the matching degree of the first trend feature and the second trend feature is greater than the preset matching degree threshold, it is determined that the second structure data and the second vibration data satisfy the preset corresponding relationship.

[0041] Further, the step of generating the first trend feature of the second structure data comprises:

[0042] generating a second difference feature size parameter sequence of a structure abnormal point in the second structure data;

[0043] obtaining each size parameter p2 in the second difference feature size parameter sequence i (t), where i ∈ [1, n p ], i is a positive integer, n p is the number of size parameter types of the difference feature, and the second difference feature size parameter sequence is a discrete data sequence constructed in order of generation time t of a structure image of the second structure data;

[0044] For the i-th size parameter, a first change curve function C1 i (t) of the size parameter p2 i (t) is fitted with time t as the independent variable;

[0045] A time range set (ts i , te ij ) corresponding to the rising interval of the first change curve function C1 ij (t) is extracted, where j ∈ [1, np i ], np i is the number of rising intervals in the first change curve function C1 i (t), ts ij is the start time of the j-th rising interval in the first change curve function C1 i (t), and te ij is the end time of the j-th rising interval in the first change curve function C1 i (t).

[0046] The time range set (ts ij , te ij ) is determined as the first trend feature of the second structure data.

[0047] Further, the step of generating the second trend feature of the second vibration data specifically comprises:

[0048] obtaining a vibration intensity v(t) in the second vibration data;

[0049] fitting a second change curve function C2(t) of the vibration intensity v(t) with time t as the independent variable;

[0050] extracting a rising inflection point set tip k , where k ∈ [1, n v ], n v is the number of rising inflection points in the second change curve function C2(t), and the rising inflection point is an inflection point where the slope of the second change curve function C2(t) increases;

[0051] the rising inflection point set tipk A second trend feature of the second vibration data is determined.

[0052] Furthermore, the step of calculating the matching degree between the first trend feature and the second trend feature specifically includes:

[0053] determining the number of structural anomalies on the target structural component;

[0054] When the number of structural abnormal points on the target structural part is 1, obtain np corresponding to the structural abnormal point in the second structural data. i A set of time ranges (ts ij ,te ij );

[0055] For each time range set (ts ij ,te ij ), according to the rising inflection point set tip of the second vibration data k The number of items that fall within its time range is used to calculate its matching score s. i ;

[0056] np i The maximum value among the matching scores Determine the target matching degree P between the first trend feature and the second trend feature.

[0057] Furthermore, each structural abnormal point has a corresponding first trend feature. After the step of determining the number of structural abnormal points on the target structural component, the method further includes:

[0058] When the number of structural abnormal points on the target structural part is greater than 1, determine the distance d between each structural abnormal point and the vibration monitoring point. l , where l∈[1,n un ],n un is the number of structural abnormal points on the target structural component;

[0059] According to the distance d between each structural abnormal point and the vibration monitoring point l Configure the influence weight σ of each structural abnormal point on the matching degree between the first trend feature and the second trend feature l ;

[0060] Calculate the matching degree p between the first trend feature and the second trend feature of each structural anomaly point l ;

[0061] According to the influence weight σ of each structural outlier l The matching degree p between the first trend feature and the second trend feature lCalculating a target matching degree for determining whether the second structure data and the second vibration data satisfy a preset corresponding relationship:

[0062]

[0063] The present invention proposes a crane monitoring system and a prediction-based crane maintenance method. By obtaining first structural data and first vibration data of a target structural part, it is determined whether the first structural data has damage characteristics. When the first structural data does not have damage characteristics, the second structural data and second vibration data in a future period of time are predicted, and a correspondence analysis is performed on the second structural data and the second vibration data to determine whether the second structural data and the second vibration data satisfy a preset correspondence relationship. When the second structural data and the second vibration data satisfy the preset correspondence relationship, it is determined that the target structural part has a damage risk. The damage risk level of the target structural part is calculated according to the second vibration data. The maintenance strategy of the target structural part is generated according to the damage risk level. The physical damage of the crane can be accurately predicted and an effective maintenance strategy can be provided. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 is a schematic diagram of a crane monitoring system provided by one embodiment of the present invention;

[0065] Figure 2 This is a flowchart of a prediction-based crane maintenance method provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0066] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, in the absence of conflict, the embodiments of the present application and the features therein can be combined with each other.

[0067] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0068] In the description of the present application, the term "a plurality of" refers to two or more, unless otherwise explicitly specified. The terms "upper", "lower", and the like refer to the orientation or positional relationship shown in the drawings based on the orientation or positional relationship shown in the drawings, and are merely used for the purpose of facilitating the description of the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application. The terms "connected", "mounted", "fixed", and the like should be interpreted broadly, for example, "connected" can be fixed connection, or detachable connection, or integral connection; can be directly connected, or indirectly connected through an intermediate medium. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances. In addition, the terms "first", "second", and the like are used for the purpose of description only, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features limited by "first", "second", etc. can be explicitly or implicitly included one or more features. In the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise specified.

[0069] In the description of the present application, the terms "one embodiment", "some embodiments", "a specific embodiment", and the like mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present description, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0070] A crane monitoring system and a crane maintenance method based on prediction according to some embodiments of the present application are described below with reference to the accompanying drawings.

[0071] As shown in Figure 1 The crane monitoring system includes a structure monitoring sensor for monitoring the structural changes of a target structural member of the crane, a vibration monitoring sensor for monitoring the structural stability of the target structural member of the crane, and a processing unit in communication with the structure monitoring sensor and the vibration monitoring sensor.

[0072] Specifically, the mechanical system of the crane is composed of several parts, such as a boom, a chassis, a leg, and a cockpit, etc., and each component part is composed of several structural members. In the technical solution of the present application, by determining the vulnerable structural members on the crane, such as the main beam, the coupling, the hook, the steel wire rope, etc., and determining them as the target structural members, the structure monitoring sensor and the vibration monitoring sensor are installed thereon to monitor the structure and vibration, so as to formulate the maintenance strategy of the crane according to the monitoring data.

[0073] In the technical scheme of some embodiments of the present application, the structural monitoring sensor can include an ultrasonic sensor for internal flaw detection of the target structure, and an image sensor for appearance monitoring (including surface crack inspection, overall or local deformation inspection, etc.) of the target structure.

[0074] The crane monitoring system further includes a communication unit, which can be a wired communication unit or a wireless communication unit, and the processor establishes a communication connection with the structural monitoring sensor and the vibration monitoring sensor through the communication unit.

[0075] In the technical scheme of the present application, the processing unit is configured to:

[0076] obtain first structural data of the target structure and first vibration data corresponding to the first structural data, wherein the first structural data includes internal first structural data, shape monitoring data and / or surface monitoring data of the target structure;

[0077] perform damage feature identification on the first structural data to determine whether the first structural data has a damage feature;

[0078] when the first structural data does not have a damage feature, predict second structural data and second vibration data in a future period of time;

[0079] perform corresponding analysis on the second structural data and the second vibration data to determine whether the second structural data and the second vibration data satisfy a preset corresponding relationship;

[0080] when the second structural data and the second vibration data satisfy the preset corresponding relationship, determine that the target structure has a damage risk;

[0081] calculate a damage risk level of the target structure according to the second vibration data;

[0082] generate a maintenance strategy for the target structure according to the damage risk level.

[0083] In the technical solution of the present application, the first structure data and the first vibration data are both monitoring data collected by a structure monitoring sensor and a vibration monitoring sensor installed on the target structure, and the first structure data and the first vibration data have a corresponding relationship in time, that is, at each data sampling time point, the collected monitoring data at least includes a set of first structure data of the target structure and a set of first vibration data of the target structure. Therefore, the first vibration data corresponding to the first structure data refers to the first vibration data corresponding to the first structure data in time for the same target structure.

[0084] In the first structure data, the internal first structure data is internal structure data of the target structure obtained by ultrasonic wave or X-ray, gamma ray, etc., which can be an internal structure image corresponding to a specific view angle of the target structure, or a plurality of internal structure images corresponding to different view angles of the target structure. The shape monitoring data and the surface monitoring data are appearance data of the target structure obtained by imaging sensors such as cameras, etc., which include one or more appearance images reflecting the overall shape and size of the target structure, and surface images of one or more specific surface regions, which can be a plurality of vulnerable regions pre-configured.

[0085] In the structure of the crane, various damage types such as wear, aging, deformation, fracture, etc. are represented as different damage features in the first structure data, and the damage features are quantitative features of various damage types in the first structure data.

[0086] The second structure data is structure data of the target structure in a future period of time obtained by using a traditional prediction algorithm or an artificial intelligence prediction model based on the first structure data. Similarly, the second vibration data is vibration data of the target structure in a future period of time obtained by using a traditional prediction algorithm or an artificial intelligence prediction model based on the first vibration data. Preferably, the second structure data and the second vibration data are both data obtained by using a pre-trained artificial intelligence prediction model, and the artificial intelligence prediction model is an artificial intelligence prediction model trained using historical structure data and historical vibration data of the target structure collected by the structure monitoring sensor and the vibration monitoring sensor as sample data by using machine learning technology.

[0087] Further, in the step of calculating the damage risk level of the target structure according to the second vibration data, the processing unit is configured to:

[0088] obtaining a standard vibration amplitude of the target structure, the standard vibration amplitude being a maximum vibration amplitude measured in a working state of the target structure without physical damage;

[0089] obtaining a damage vibration amplitude of the target structure in the second vibration data, the damage vibration amplitude being a maximum vibration amplitude of the target structure in the second vibration data;

[0090] calculating a difference between the damage vibration amplitude and the standard vibration amplitude;

[0091] determining a damage risk level of the target structure according to the difference between the damage vibration amplitude and the standard vibration amplitude.

[0092] In the technical solution of the above-mentioned embodiment, the mapping relationship between the size of the difference between the damage vibration amplitude and the standard vibration amplitude and the damage risk level is pre-configured in the database, and after the size of the difference between the damage vibration amplitude and the standard vibration amplitude is determined, the damage risk level of the target structure is queried from the database.

[0093] In the technical solution of some embodiments of the present application, the maintenance strategies of each target structure under different damage risk levels are pre-configured in the database, and in the step of generating the maintenance strategy of the target structure according to the damage risk level, the corresponding maintenance strategy is queried from the database according to the type of the target structure and its damage risk level. Further, in the step of predicting the second structure data and the second vibration data in the future period of time, the processing unit is configured to:

[0094] obtaining a pre-configured first time length and a second time length, the first time length being a time length corresponding to the input data of the artificial intelligence prediction model, and the second time length being a time length corresponding to the second structure data and the second vibration data to be predicted, i.e. the output data of the artificial intelligence prediction model;

[0095] determining the time when the first structure data of the target structure and the first vibration data corresponding to the first structure data are obtained as the current time;

[0096] determining a first time range with the first time length ending at the current time and a second time range with the second time length starting at the current time;

[0097] arranging the first structure data collected in the first time range into a first input data sequence of the structure prediction model, and arranging the first vibration data collected in the first time range into a second input data sequence of the vibration prediction model;

[0098] inputting the first input data sequence into a structure prediction model to predict structural data in the second time range as the second structural data;

[0099] inputting the second input data sequence into a vibration prediction model to predict vibration data in the second time range as the second vibration data.

[0100] In the technical solution of the above embodiment, the future period of time is a second time range with the second time length and starting from the current time.

[0101] The structure prediction model is an artificial intelligence prediction model trained using historical structural data of the target structure as sample data, and the vibration prediction model is an artificial intelligence prediction model trained using historical vibration data of the target structure as sample data.

[0102] In the technical solution of some embodiments of the present application, the first time length and the second time length can be the same time length.

[0103] Further, the first structural data is a structural image of the target structure, and in the step of performing damage feature identification on the first structural data to determine whether the first structural data has a damage feature, the processing unit is configured to:

[0104] determine whether there is a structural abnormal point in the structural image;

[0105] when there is a structural abnormal point in the structural image, generate a local structural image of the target structure at the structural abnormal point;

[0106] input the local structural image into a pre-trained damage identification model for damage feature identification to determine whether the first structural data has a damage feature.

[0107] From the perspective of image content, the structural image can be one or more of an internal structural image, an appearance image, and a surface image of the target structure. From the perspective of image generation mode, the structural image can be one or more of an ultrasonic image, an X-ray image, a gamma-ray image, an infrared image, and a visible light image.

[0108] The local structural image is a local image of the structural image at the structural abnormal point. Further, the step of generating a local structural image of the target structure at the structural abnormal point specifically includes:

[0109] obtaining a standard size of an input image of the pre-configured damage identification model;

[0110] determining a center point and a maximum width of the structural abnormal point in the structural image, the center point can be a geometric center of the structural abnormal point;

[0111] cutting a square local image centered at the center point in the structural image with the maximum width as the side length;

[0112] scaling the square local image to the standard size to obtain a local structural image of the target structural member at the structural abnormal point.

[0113] The damage identification model is a pre-trained deep learning model, more specifically, the damage identification model is a classification model of damage types trained using deep learning technology. By collecting and sorting the local structural images of the target structural member under various damage conditions with standard size, labeling each local structural image with a damage type to construct it as training sample data, and using these training sample data for deep learning to obtain the damage identification model. In specific implementation scenarios, independent damage identification models can be trained for different materials of structural members on cranes as needed, or independent damage identification models can be trained for different damage types to improve the accuracy of damage identification.

[0114] Further, in the step of determining whether there is a structural abnormal point in the structural image, the processing unit is configured to:

[0115] configure a standard image corresponding to the target structural member and the structural image, the standard image being a structural image generated by the target structural member in a damage-free state;

[0116] obtain a first sequence of structural images of the target structural member in the past period of time;

[0117] generate a first difference feature sequence according to the difference features of each structural image in the first sequence of structural images and the standard image;

[0118] generate a first difference feature size parameter sequence corresponding to the first difference feature sequence, the first difference feature size parameter sequence being composed of size parameters in one or more different directions of each difference feature in the first difference feature sequence;

[0119] trend analysis is performed on the first difference feature size parameter sequence to determine whether the corresponding difference feature is a structural abnormal point according to the trend analysis result.

[0120] The standard image and the structural image are generated in the same way, and the shooting parameters such as the shooting angle and the shooting distance are the same.

[0121] The past period of time refers to a connected time range with the third time length ending at the current time point. The third time length is a specific time length pre-configured for performing structure abnormal point identification.

[0122] Further, in the step of generating a first difference feature sequence according to the difference features between each structure image in the first structure image sequence and the standard image, the processing unit is configured to:

[0123] determine the position with obvious data difference between the structure image and the standard image as a difference point, wherein the obvious data difference can be the case that the difference between the two is greater than a pre-configured threshold value;

[0124] analyze and identify the existence of each difference point on the structure image in the first structure image sequence;

[0125] when a difference point continuously exists in a plurality of structure images in the first structure image sequence, determine the difference region where the difference point is located as a difference feature, wherein the difference region is a local image region composed of continuous difference points on the structure image;

[0126] construct the difference features at the same position in the first structure image sequence into the first difference feature sequence in time sequence.

[0127] The size parameter of the difference feature can be one or more of the shape parameters such as width, height, area (pixel area), perimeter, and circumscribed circle radius of the difference feature.

[0128] Further, in the step of performing trend analysis on the first difference feature size parameter sequence to determine whether the corresponding difference feature is a structure abnormal point according to the trend analysis result, the processing unit is configured to:

[0129] obtain each size parameter p1 i (t) in the first difference feature size parameter sequence, where i∈[1,n p ], i is a positive integer, n p is the number of size parameter types of the difference feature, and the first difference feature size parameter sequence is a discrete data sequence constructed in time sequence according to the generation time t of the first structure data structure image;

[0130] for the i-th size parameter, fit a trend linear function f i (t) of the size parameter p1 i (t) = k i ×t+b i;

[0131] determining whether a slope k i in the trend linear function f i is greater than 0;

[0132] when there is any i value satisfying k p > 0 in the range of [1, n i ], it is determined that there is an abnormal point in the structure image.

[0133] Specifically, when n p is greater than 1, the first difference feature size parameter sequence includes n p parallel subsequences, each corresponding to a time sequence of one size parameter such as width, height, and area. Therefore, p1 i (t) is the size parameter of the difference feature in the structure image generated at time t in the i-th subsequence of the first difference feature size parameter sequence, or p1 i (t) is the size of the i-th size parameter of the difference feature in the structure image generated at time t.

[0134] In the technical solution of the embodiment, f i (t) is a trend linear function of the size parameter p1 i (t) fitted using a linear fitting algorithm such as least squares, and the trend linear function f i (t) can represent the change trend of the size parameter p1 i (t) in the past period of time.

[0135] Further, in the step of performing corresponding analysis on the second structure data and the second vibration data to determine whether the second structure data and the second vibration data satisfy a preset corresponding relationship, the processing unit is configured to:

[0136] generate a first trend feature of the second structure data and a second trend feature of the second vibration data;

[0137] calculate a matching degree of the first trend feature and the second trend feature;

[0138] determine whether the matching degree of the first trend feature and the second trend feature is greater than a preset matching degree threshold;

[0139] when the matching degree of the first trend feature and the second trend feature is greater than the preset matching degree threshold, it is determined that the second structure data and the second vibration data satisfy the preset corresponding relationship.

[0140] In the technical solution of the above-mentioned embodiment, the first trend feature is a feature reflecting the variation trend of the size parameter of one or more structural abnormal points in the second structural data. Similarly, the second trend feature is a feature reflecting the variation trend of the second vibration data.

[0141] Further, before the step of generating the first trend feature of the second structural data and the second trend feature of the second vibration data, the processing unit is configured to:

[0142] determine the number of structural abnormal points in the second structural data;

[0143] configure the size of the matching degree threshold according to the number of structural abnormal points in the second structural data.

[0144] In the technical solution of the above-mentioned embodiment, the matching degree threshold is not a single fixed threshold, and the size of the matching degree threshold is related to the number of structural abnormal points in the second structural data. For the second structural data with different numbers of structural abnormal points, the matching degree threshold for matching with the second vibration data is different.

[0145] Further, in the step of generating the first trend feature of the second structural data, the processing unit is configured to:

[0146] generate a second difference feature size parameter sequence of the structural abnormal points in the second structural data;

[0147] obtain each size parameter p2 i (t) in the second difference feature size parameter sequence, where i∈[1, n p ], i is a positive integer, n p is the number of size parameter types of the difference feature, and the second difference feature size parameter sequence is a discrete data sequence constructed in the order of the generation time t of the structural image of the second structural data;

[0148] for the i-th size parameter, fit a first variation curve function C1 i (t) of the size parameter p2 i (t) with time t as the independent variable;

[0149] extract a time range set (ts ij , te ij ) corresponding to the rising interval of the first variation curve function C1 i , where j∈[1, np i ], np i is the number of the first variation curve function C1 ij .(t) the number of rising intervals in the first change curve function C1 ij is the first change curve function C1 i (t) the start time of the jth rising interval in the first change curve function C1 ij is the first change curve function C1 i (t) the end time of the jth rising interval in the first change curve function C1

[0150] determines the time range set (ts ij , te ij ) as the first trend feature of the second structure data.

[0151] In the technical solution of some embodiments of the present application, the first change curve function C1 i (t) is a curve function fitted by using a polynomial fitting or other curve fitting algorithm, wherein the size of the size parameter p2 i (t) changes over time. i (t) is used to represent the change of the size of the size parameter p2 i (t) over time in the future period of time.

[0152] The rising interval of the first change curve function C1 i (t) refers to the interval in the first change curve function C1 i (t) with a slope greater than 0, i.e., ts ij and te ij are the time points at which the slope of the first change curve function C1 i (t) is 0.

[0153] Further, in the step of generating the second difference feature size parameter sequence of the structure abnormal point in the second structure data, the processing unit is configured to:

[0154] determine the target position of the structure abnormal point in the target structure based on the first structure data;

[0155] generate a second structure image sequence of the target structure in the future period of time based on the second structure data;

[0156] generate a second difference feature size parameter sequence corresponding to the target position in the second structure image sequence.

[0157] Further, in the step of generating the second trend feature of the second vibration data, the processing unit is configured to:

[0158] obtain the vibration intensity v(t) in the second vibration data;

[0159] fitting a second change curve function C2(t) of the vibration intensity v(t) with time t as the independent variable;

[0160] extracting a set of rising inflection points tip of the second change curve function C2(t) k , where k ∈ [1, n v ], n v is the number of rising inflection points of the second change curve function C2(t), and the rising inflection point is an inflection point of the second change curve function C2(t) where the slope increases;

[0161] determining the set of rising inflection points tip k as the second trend feature of the second vibration data.

[0162] The vibration intensity v(t) is the vibration intensity of the target structure detected by the vibration monitoring sensor on the target structure at time t. The second change curve function C2(t) is a curve function of the size of the vibration intensity v(t) changing with time fitted by using a curve fitting algorithm such as polynomial fitting, and the second change curve function C2(t) is used to represent the change of the size of the vibration intensity v(t) with time in the future period of time.

[0163] Further, in the step of extracting the set of rising inflection points tip k of the second change curve function C2(t), the processing unit is configured to:

[0164] identify each inflection point in the second change curve function C2(t);

[0165] calculate a first slope at Δt before each inflection point and a second slope at Δt after each inflection point, where Δt is a pre-configured slope calculation interval;

[0166] determine the inflection point with the second slope greater than the first slope as the rising inflection point.

[0167] Further, in the step of calculating the matching degree of the first trend feature and the second trend feature, the processing unit is configured to:

[0168] determine the number of structural abnormal points on the target structure;

[0169] when the number of structural abnormal points on the target structure is 1, obtain np i set of time ranges (ts ij , te ij ) corresponding to the structural abnormal point in the second structure data;

[0170] For each time range set (ts ij ,te ij ), according to the rising inflection point set tip of the second vibration data k The number of items that fall within its time range is used to calculate its matching score s. i ;

[0171] np i The maximum value among the matching scores Determine the target matching degree P between the first trend feature and the second trend feature.

[0172] In the technical solution of the above embodiment, the target matching degree P is a matching degree used to determine whether the second structural data and the second vibration data satisfy a preset correspondence relationship. Specifically, in the step of determining whether the matching degree between the first trend feature and the second trend feature is greater than a preset matching degree threshold, the step specifically involves determining whether the target matching degree P is greater than the preset matching degree threshold.

[0173] Furthermore, for each time range set (ts ij ,te ij ), according to the rising inflection point set tip of the second vibration data k The number of items that fall within its time range is used to calculate its matching score s. i In the step, the processing unit is configured to:

[0174] Count the rising inflection point set tip of the second vibration data k The number of rising inflection points np that fall into the i-th time range set i ;

[0175] The rising inflection point set tip of the second vibration data k The number of rising inflection points np that fall into the i-th time range set i The matching score is determined as follows:

[0176]

[0177] Furthermore, when each structural abnormal point has a corresponding first trend feature, after the step of determining the number of structural abnormal points on the target structural component, the processing unit is configured to:

[0178] When the number of structural abnormal points on the target structural part is greater than 1, determine the distance d between each structural abnormal point and the vibration monitoring point. l , where l∈[1,n un ],n una number of structural abnormal points on the target structure;

[0179] a distance d of each structural abnormal point from the vibration monitoring point l a degree of influence weight σ of each structural abnormal point on the matching degree of the first trend feature and the second trend feature l ;

[0180] a matching degree p of the first trend feature and the second trend feature of each structural abnormal point l ;

[0181] a degree of influence weight σ of each structural abnormal point l and a matching degree p of the first trend feature and the second trend feature of each structural abnormal point l a target matching degree for judging whether the second structure data and the second vibration data satisfy a preset corresponding relationship:

[0182]

[0183] Specifically, the vibration monitoring point is a position on the target structure where the vibration monitoring sensor is installed. The greater the distance from the vibration monitoring point, the smaller the degree of influence weight of the structural abnormal point on the matching degree of the first trend feature and the second trend feature.

[0184] As Figure 2 shown, a second aspect of the present application proposes a crane maintenance method based on prediction, comprising:

[0185] obtaining first structure data of a target structure and first vibration data corresponding to the first structure data, the first structure data including internal first structure data, shape monitoring data and / or surface monitoring data of the target structure;

[0186] performing damage feature recognition on the first structure data to determine whether the first structure data has a damage feature;

[0187] when the first structure data does not have a damage feature, predicting second structure data and second vibration data in a future period of time;

[0188] performing corresponding analysis on the second structure data and the second vibration data to determine whether the second structure data and the second vibration data satisfy a preset corresponding relationship;

[0189] when the second structure data and the second vibration data satisfy the preset corresponding relationship, determining that the target structure has a damage risk;

[0190] calculate a damage risk level of the target structure according to the second vibration data;

[0191] generate a maintenance strategy of the target structure according to the damage risk level.

[0192] In the technical scheme of the present application, the first structure data and the first vibration data are both monitoring data collected by structure monitoring sensors and vibration monitoring sensors installed on the target structure, and the first structure data and the first vibration data have a corresponding relationship in time, that is, at each data sampling time point, the collected monitoring data at least includes a set of first structure data of the target structure and a set of first vibration data of the target structure. Therefore, the first vibration data corresponding to the first structure data refers to the first vibration data corresponding to the first structure data in the collection time for the same target structure.

[0193] In the first structure data, the internal first structure data is internal structure data of the target structure obtained by ultrasonic wave or X-ray, gamma ray, etc., which can be an internal structure image corresponding to a specific view angle of the target structure, or a plurality of internal structure images corresponding to different view angles of the target structure. The shape monitoring data and the surface monitoring data are appearance data of the target structure obtained by image sensors such as cameras, etc., which include one or more appearance images reflecting the overall shape and size of the target structure, and surface images of one or more specific surface regions, which can be a plurality of damage-prone regions pre-configured.

[0194] In the structure of the crane, various damage types such as wear, aging, deformation, fracture, etc. are represented as different damage features in the first structure data, and the damage features are quantitative features of various damage types in the first structure data.

[0195] The second structure data is structure data of the target structure in a future period of time obtained by using a traditional prediction algorithm or an artificial intelligence prediction model to predict based on the first structure data. Similarly, the second vibration data is vibration data of the target structure in a future period of time obtained by using a traditional prediction algorithm or an artificial intelligence prediction model to predict based on the first vibration data. Preferably, the second structure data and the second vibration data are both data obtained by using a pre-trained artificial intelligence prediction model to predict, and the artificial intelligence prediction model is an artificial intelligence prediction model obtained by using historical structure data and historical vibration data of the target structure collected by the structure monitoring sensors and the vibration monitoring sensors as sample data, and using machine learning technology to train.

[0196] Further, the step of calculating the damage risk level of the target structure according to the second vibration data specifically comprises:

[0197] obtaining a standard vibration amplitude of the target structure, the standard vibration amplitude being a maximum vibration amplitude measured in a working state of the target structure without physical damage;

[0198] obtaining a damage vibration amplitude of the target structure in the second vibration data, the damage vibration amplitude being a maximum vibration amplitude of the target structure in the second vibration data;

[0199] calculating a difference value between the damage vibration amplitude and the standard vibration amplitude;

[0200] determining the damage risk level of the target structure according to the size of the difference value between the damage vibration amplitude and the standard vibration amplitude.

[0201] In the technical solution of the above-mentioned embodiment, the mapping relationship between the size of the difference value between the damage vibration amplitude and the standard vibration amplitude and the damage risk level is pre-configured in the database, and after the size of the difference value between the damage vibration amplitude and the standard vibration amplitude is determined, the damage risk level of the target structure is queried from the database.

[0202] In the technical solution of some embodiments of the present application, the maintenance strategies of each target structure under different damage risk levels are pre-configured in the database, and in the step of generating the maintenance strategy of the target structure according to the damage risk level, the corresponding maintenance strategy is queried from the database according to the type of the target structure and its damage risk level.

[0203] Further, the step of predicting the second structure data and the second vibration data in a future period of time specifically comprises:

[0204] obtaining a pre-configured first time length and a second time length, the first time length being a time length corresponding to the input data of the artificial intelligence prediction model, and the second time length being a time length corresponding to the second structure data and the second vibration data to be predicted, i.e. a time length corresponding to the output data of the artificial intelligence prediction model;

[0205] determining the time point of obtaining the first structure data of the target structure and the first vibration data corresponding to the first structure data as the current time;

[0206] determining a first time range with the first time length ending at the current time, and a second time range with the second time length starting at the current time.

[0207] arranging the first structural data collected in the first time range into a first input data sequence of a structural prediction model, and arranging the first vibration data collected in the first time range into a second input data sequence of a vibration prediction model;

[0208] inputting the first input data sequence into the structural prediction model to predict the structural data in the second time range as the second structural data;

[0209] inputting the second input data sequence into the vibration prediction model to predict the vibration data in the second time range as the second vibration data.

[0210] In the technical solution of the above-mentioned embodiments, the future period of time is a second time range with the second time length starting from the current time.

[0211] The structural prediction model is an artificial intelligence prediction model trained using historical structural data of the target structural member as sample data, and the vibration prediction model is an artificial intelligence prediction model trained using historical vibration data of the target structural member as sample data.

[0212] In the technical solution of some embodiments of the present application, the first time length and the second time length can be the same time length.

[0213] Further, the first structural data is a structural image of the target structural member, and the step of performing damage feature recognition on the first structural data to determine whether the first structural data has a damage feature specifically includes:

[0214] determining whether there is a structural abnormal point in the structural image;

[0215] when there is a structural abnormal point in the structural image, generating a local structural image of the target structural member at the structural abnormal point;

[0216] inputting the local structural image into a pre-trained damage recognition model for damage feature recognition to determine whether the first structural data has a damage feature.

[0217] From the perspective of image content, the structural image can be one or more of an internal structural image, an appearance image, and a surface image of the target structural member. From the perspective of image generation mode, the structural image can be one or more of an ultrasonic image, an X-ray image, a gamma-ray image, an infrared image, and a visible light image.

[0218] The local structure image is a local image of the structure image at the structure abnormal point. Further, the step of generating the local structure image of the target structure at the structure abnormal point specifically comprises:

[0219] obtaining a standard size of an input image of the damage identification model pre-configured;

[0220] determining a center point and a maximum width of the structure abnormal point in the structure image, the center point being a geometric center of the structure abnormal point;

[0221] cutting a square local image centered at the center point in the structure image with the maximum width as a side length;

[0222] scaling the square local image to the standard size to obtain the local structure image of the target structure at the structure abnormal point.

[0223] The damage identification model is a pre-trained deep learning model, and more specifically, the damage identification model is a classification model of damage types trained using deep learning technology. By collecting and organizing local structure images of the target structure under various damage conditions with a standard size, each local structure image is labeled with a damage type to construct training sample data, and deep learning is performed using these training sample data to obtain the damage identification model. In specific implementation scenarios, independent damage identification models can be trained for different materials of the structure on the crane as needed, or independent damage identification models can be trained for different damage types to improve the accuracy of damage identification.

[0224] Further, the step of judging whether there is a structure abnormal point in the structure image specifically comprises:

[0225] configuring a standard image corresponding to the target structure and the structure image, the standard image being a structure image of the target structure in a damage-free state;

[0226] obtaining a first structure image sequence of the target structure in a past period of time;

[0227] generating a first difference feature sequence according to difference features of each structure image in the first structure image sequence and the standard image;

[0228] generating a first difference feature size parameter sequence corresponding to the first difference feature sequence, the first difference feature size parameter sequence being composed of size parameters in one or more different directions of each difference feature in the first difference feature sequence;

[0229] The first difference feature size parameter sequence is subjected to trend analysis, so as to determine whether the corresponding difference feature is a structural abnormal point according to the trend analysis result.

[0230] The standard image and the structural image are generated in the same way and have the same shooting parameters such as shooting angle and shooting distance.

[0231] The past period of time refers to a connected time range with the third time length ending at the current time point. The third time length is a specific time length pre-configured for performing structural abnormal point identification.

[0232] Further, the step of generating a first difference feature sequence according to the difference feature between each structural image in the first structural image sequence and the standard image specifically includes:

[0233] The position with obvious data difference between the structural image and the standard image is determined as a difference point. The obvious data difference can be that the difference between the two is greater than a pre-configured threshold value.

[0234] The existence of each difference point on the structural image in the first structural image sequence is analyzed and identified;

[0235] When a difference point continuously exists in a plurality of structural images in the first structural image sequence, a difference region where the difference point is located is determined as a difference feature. The difference region is a local image region composed of continuous difference points on the structural image.

[0236] The difference features at the same position in the first structural image sequence are constructed in time sequence as the first difference feature sequence.

[0237] The size parameter of the difference feature can be one or more of shape parameters such as width, height, area (pixel area), perimeter, and circumscribed circle radius of the difference feature.

[0238] Further, the step of performing trend analysis on the first difference feature size parameter sequence to determine whether the corresponding difference feature is a structural abnormal point according to the trend analysis result specifically includes:

[0239] Each size parameter p1 i (t) in the first difference feature size parameter sequence is obtained, where i∈[1, n p ], i is a positive integer, n p is the number of size parameter types of the difference feature, and the first difference feature size parameter sequence is a discrete data sequence constructed in the order of generation time t of the first structural data structural image.

[0240] fitting the size parameter pi i (t) with time t as the independent variable i (t) = k i ×t + b i ;

[0241] determining whether the slope k i in the trend line function f i (t) is greater than 0;

[0242] when there is any i value satisfying k p > 0 in the range of [1, n i ], it is determined that there is a structural abnormal point in the structural image.

[0243] Specifically, when n p is greater than 1, the first difference feature size parameter sequence includes n p sub-sequences in parallel, each of which corresponds to a time sequence of one size parameter such as width, height, and area. Therefore, pi i (t) is the size parameter of the difference feature in the structural image generated at time t in the i-th sub-sequence of the first difference feature size parameter sequence, or in other words, pi i (t) is the size of the i-th size parameter of the difference feature in the structural image generated at time t.

[0244] In the technical solution of the embodiment, f i (t) is a trend line function of the size parameter pi i (t) fitted using a linear fitting algorithm such as least squares method, and the trend line function f i (t) can represent the change trend of the size parameter pi i (t) in the past period of time.

[0245] Further, the step of performing corresponding analysis on the second structural data and the second vibration data to determine whether the second structural data and the second vibration data satisfy a preset corresponding relationship specifically includes:

[0246] generating a first trend feature of the second structural data and a second trend feature of the second vibration data;

[0247] calculating a matching degree of the first trend feature and the second trend feature;

[0248] determining whether the matching degree of the first trend feature and the second trend feature is greater than a preset matching degree threshold;

[0249] When the matching degree between the first trend feature and the second trend feature is greater than a preset matching degree threshold, it is determined that the second structure data and the second vibration data satisfy a preset corresponding relationship.

[0250] In the technical solution of the above embodiment, the first trend feature is a feature reflecting the change trend of the size parameters of one or more structural abnormal points in the second structural data. Similarly, the second trend feature is a feature reflecting the change trend of the second vibration data.

[0251] Furthermore, before the step of generating the first trend feature of the second structural data and the second trend feature of the second vibration data, the method further includes:

[0252] determining the number of structural anomalies in the second structural data;

[0253] The size of the matching degree threshold is configured according to the number of structural abnormal points determined in the second structural data.

[0254] In the technical solution of the above-mentioned embodiment, the matching degree threshold is not a single fixed threshold. The size of the matching degree threshold is related to the number of structural anomalies in the second structural data. For second structural data with different numbers of structural anomalies, the matching degree threshold for matching with the second vibration data is different.

[0255] Furthermore, the step of generating the first trend feature of the second structural data specifically includes:

[0256] generating a second difference characteristic size parameter sequence of structural abnormal points in the second structural data;

[0257] Get each size parameter p2 in the second difference feature size parameter sequence i (t), where i∈[1, n p ], i is a positive integer, n p is the number of size parameter types of the difference feature, and the second difference feature size parameter sequence is a discrete data sequence constructed in the order of generation time t of the structural image of the second structural data;

[0258] For the i-th size parameter, the size parameter p2 is fitted with time t as the independent variable. i The first variation curve function C1 of (t) i (t);

[0259] Extract the first change curve function C1 i The time range set (ts) corresponding to the rising interval of (t) ij ,teij ), wherein j ∈ [1, np i ], np i is a number of rising intervals in the first change curve function C1 i (t), ts ij is a start time of the jth rising interval in the first change curve function C1 i (t), te ij is an end time of the jth rising interval in the first change curve function C1 i (t).

[0260] The time range set (ts ij , te ij ) is determined as a first trend feature of the second structure data.

[0261] In the technical solution of some embodiments of the present application, the first change curve function C1 i (t) is a curve function fitted by using a curve fitting algorithm such as polynomial fitting, and the size of the size parameter p2 i (t) changes over time, the first change curve function C1 i (t) is used to represent the change of the size of the size parameter p2 i (t) over time in the future period of time.

[0262] The rising interval of the first change curve function C1 i (t) refers to an interval in which the slope of the first change curve function C1 i (t) is greater than 0, that is, ts ij and te ij are both the time points at which the slope of the first change curve function C1 i (t) is 0.

[0263] Further, the second structure data is a structure image of the target structure, and the step of generating the second difference feature size parameter sequence of the structure abnormal point in the second structure data specifically includes:

[0264] Determining a target position of the structure abnormal point in the target structure based on the first structure data;

[0265] Generating a second structure image sequence of the target structure in the future period of time based on the second structure data;

[0266] Generating a second difference feature size parameter sequence corresponding to the target position in the second structure image sequence.

[0267] Further, the step of generating the second trend feature of the second vibration data specifically comprises:

[0268] obtaining a vibration intensity v(t) in the second vibration data;

[0269] fitting a second change curve function C2(t) of the vibration intensity v(t) with time t as the independent variable;

[0270] extracting a set of rising inflection points tip k of the second change curve function C2(t), where k∈[1, n v ], n v is the number of rising inflection points of the second change curve function C2(t), and the rising inflection point is an inflection point with a slope increasing in the second change curve function C2(t);

[0271] determining the set of rising inflection points tip k as the second trend feature of the second vibration data.

[0272] The vibration intensity v(t) is the vibration intensity of the target structure detected by the vibration monitoring sensor on the target structure at time t. The second change curve function C2(t) is a curve function of the size of the vibration intensity v(t) changing with time, which is fitted by using a curve fitting algorithm such as polynomial fitting. The second change curve function C2(t) is used to represent the change of the size of the vibration intensity v(t) with time in the future.

[0273] Further, the step of extracting the set of rising inflection points tip k of the second change curve function C2(t) specifically comprises:

[0274] identifying each inflection point in the second change curve function C2(t);

[0275] calculating a first slope at a first time interval Δt before each inflection point and a second slope at a second time interval Δt after each inflection point, where Δt is a preconfigured slope calculation interval;

[0276] determining an inflection point with the second slope greater than the first slope as the rising inflection point.

[0277] Further, the step of calculating the matching degree of the first trend feature and the second trend feature specifically comprises:

[0278] determining the number of structural abnormal points on the target structure;

[0279] when the number of structural abnormal points on the target structure is 1, obtaining npi A set of time ranges (ts ij ,te ij );

[0280] For each time range set (ts ij ,te ij ), according to the rising inflection point set tip of the second vibration data k The number of items that fall within its time range is used to calculate its matching score s. i ;

[0281] np i The maximum value among the matching scores Determine the target matching degree P between the first trend feature and the second trend feature.

[0282] In the technical solution of the above embodiment, the target matching degree P is a matching degree used to determine whether the second structural data and the second vibration data satisfy a preset correspondence relationship. Specifically, in the step of determining whether the matching degree between the first trend feature and the second trend feature is greater than a preset matching degree threshold, the step specifically involves determining whether the target matching degree P is greater than the preset matching degree threshold.

[0283] Furthermore, for each time range set (ts ij ,te ij ), according to the rising inflection point set tip of the second vibration data k The number of items that fall within its time range is used to calculate its matching score s. i The steps specifically include:

[0284] Count the rising inflection point set tip of the second vibration data k The number of rising inflection points np that fall into the i-th time range set i ;

[0285] The rising inflection point set tip of the second vibration data k The number of rising inflection points np that fall into the i-th time range set i The matching score is determined as follows:

[0286]

[0287] Furthermore, each structural abnormal point has a corresponding first trend feature. After the step of determining the number of structural abnormal points on the target structural component, the method further includes:

[0288] When the number of structural abnormal points on the target structural part is greater than 1, determine the distance d between each structural abnormal point and the vibration monitoring point. l , where l∈[1,n un ],n un is the number of structural abnormal points on the target structural component;

[0289] According to the distance d between each structural abnormal point and the vibration monitoring point l Configure the influence weight σ of each structural abnormal point on the matching degree between the first trend feature and the second trend feature l ;

[0290] Calculate the matching degree pl of the first trend feature and the second trend feature of each structural abnormal point ；

[0291] According to the influence weight σ of each structural outlier l The matching degree p between the first trend feature and the second trend feature l Calculating a target matching degree for determining whether the second structure data and the second vibration data satisfy a preset corresponding relationship:

[0292]

[0293] Specifically, the vibration monitoring point is a location on the target structural component where the vibration monitoring sensor is installed. The greater the distance from the vibration monitoring point, the smaller the influence weight of the structural abnormality on the matching degree between the first trend feature and the second trend feature.

[0294] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus comprising the element.

[0295] In accordance with the practices of the present invention, these embodiments have been described in relation to the above-described embodiments, which are intended to be illustrative only and not restrictive of the invention. Obviously, many modifications and variations of this invention can be effected without departing from the scope of the novel concept of the disclosure. No limitation with respect to the specific implementation techniques and applications presented thereby should be inferred into the scope of the invention, as understood by those skilled in the art. The specification and drawings should be regarded as illustrative only and in no way limiting of the scope of the invention as defined by the appended claims and equivalents thereof.

Claims

1. A crane monitoring system, characterized in that: The system comprises a structure monitoring sensor for monitoring structural changes of a target structural member of a crane, a vibration monitoring sensor for monitoring the structural stability of the target structural member of the crane, and a processing unit communicatively connected to the structure monitoring sensor and the vibration monitoring sensor, wherein the processing unit is configured to: Acquiring first structural data of a target structural component and first vibration data corresponding to the first structural data, wherein the first structural data includes first internal structural data, shape monitoring data, and / or surface monitoring data of the target structural component; performing damage feature recognition on the first structural data to determine whether the first structural data has damage features; When the first structural data does not have damage features, predicting second structural data and second vibration data within a future period of time; performing a correspondence analysis on the second structure data and the second vibration data to determine whether the second structure data and the second vibration data satisfy a preset correspondence relationship; When the second structural data and the second vibration data satisfy a preset corresponding relationship, determining that the target structural component has a damage risk; calculating a damage risk level of the target structural component according to the second vibration data; A maintenance strategy for the target structural component is generated according to the damage risk level.

2. A crane maintenance method based on prediction, characterized in that: include: Acquiring first structural data of a target structural component and first vibration data corresponding to the first structural data, wherein the first structural data includes first internal structural data, shape monitoring data, and / or surface monitoring data of the target structural component; performing damage feature recognition on the first structural data to determine whether the first structural data has damage features; When the first structural data does not have damage features, predicting second structural data and second vibration data within a future period of time; performing a correspondence analysis on the second structure data and the second vibration data to determine whether the second structure data and the second vibration data satisfy a preset correspondence relationship; When the second structural data and the second vibration data satisfy a preset corresponding relationship, determining that the target structural component has a damage risk; calculating a damage risk level of the target structural component according to the second vibration data; A maintenance strategy for the target structural component is generated according to the damage risk level.

3. The prediction-based crane maintenance method according to claim 2, characterized in that: The first structural data is a structural image of the target structural part, and the step of performing damage feature recognition on the first structural data to determine whether the first structural data has damage features specifically includes: Determining whether there are structural abnormal points in the structural image; When there is a structural abnormal point in the structural image, generating a local structural image of the target structural component at the structural abnormal point; The local structure image is input into a pre-trained damage recognition model to perform damage feature recognition to determine whether the first structure data has damage features.

4. The prediction-based crane maintenance method according to claim 3, characterized in that: The step of determining whether there are structural abnormal points in the structural image specifically includes: configuring a standard image corresponding to the target structural part and the structural image, wherein the standard image is a structural image generated when the target structural part is in an undamaged state; Acquire a first structural image sequence of the target structural component within a past period of time; generating a first difference feature sequence according to difference features between each structural image in the first structural image sequence and the standard image; generating a first difference feature size parameter sequence corresponding to the first difference feature sequence, wherein the first difference feature size parameter sequence is composed of one or more size parameters in different directions of each difference feature in the first difference feature sequence; A trend analysis is performed on the first difference feature size parameter sequence to determine whether the corresponding difference feature is a structural abnormal point according to the trend analysis result.

5. The prediction-based crane maintenance method according to claim 4, characterized in that: The step of performing trend analysis on the first difference feature size parameter sequence to determine whether the corresponding difference feature is a structural abnormal point according to the trend analysis result specifically includes: Get each size parameter p1 in the first difference feature size parameter sequence i (t), where o∈[1, n p ], i is a positive integer, n p is the number of size parameter types of the difference feature, and the first difference feature size parameter sequence is a discrete data sequence constructed in the order of generation time t of the first structure data structure image; For the i-th size parameter, the size parameter p1 is fitted with time t as the independent variable. i (t) trend line function f i (t) = k i ×t+b i ; Determine the trend line function f i The slope k in (t) i Is it greater than 0? In [1, n p ], when there is any i value that satisfies k i >0, it is determined that there is a structural abnormal point in the structural image.

6. The method for crane maintenance based on prediction according to claim 4, characterized in that: The step of performing correspondence analysis on the second structure data and the second vibration data to determine whether the second structure data and the second vibration data satisfy a preset correspondence relationship specifically includes: generating a first trend feature of the second structural data and a second trend feature of the second vibration data; Calculating a matching degree between the first trend feature and the second trend feature; Determining whether a matching degree between the first trend feature and the second trend feature is greater than a preset matching degree threshold; When the matching degree between the first trend feature and the second trend feature is greater than a preset matching degree threshold, it is determined that the second structure data and the second vibration data satisfy a preset corresponding relationship.

7. The method for crane maintenance based on prediction according to claim 6, characterized in that: The step of generating the first trend feature of the second structural data specifically includes: generating a second difference characteristic size parameter sequence of structural abnormal points in the second structural data; Get each size parameter p2 in the second difference feature size parameter sequence i (t), where i∈[1, n p ], i is a positive integer, n p is the number of size parameter types of the difference feature, and the second difference feature size parameter sequence is a discrete data sequence constructed in the order of generation time t of the structural image of the second structural data; For the i-th size parameter, the size parameter p2 is fitted with time t as the independent variable. i The first variation curve function C1 of (t) i (t); Extract the first change curve function C1 i The time range set (ts) corresponding to the rising interval of (t) ij ,te ij ), where j∈[1, np i ], np i The first change curve function C1 i The number of rising intervals in (t), ts ij The first change curve function C1 i The starting time of the jth rising interval in (t), te ij The first change curve function C1 i The end time of the jth rising interval in (t); The time range set (ts ij ,te ij ) is determined as the first trend feature of the second structural data.

8. The method for crane maintenance based on prediction according to claim 7, characterized in that: The step of generating a second trend feature of the second vibration data specifically includes: Acquire vibration intensity v(t) in the second vibration data; fitting a second variation curve function C2(t) of the vibration intensity v(t) with time t as an independent variable; Extract the rising inflection point set tip of the second change curve function C2(t) k , where k∈[1tn v ],n v is the number of rising inflection points in the second change curve function C2(t), where the rising inflection point is the inflection point where the slope of the second change curve function C2(t) becomes larger; The rising inflection point set tip k A second trend feature of the second vibration data is determined.

9. The prediction-based crane maintenance method according to claim 8, characterized in that: The step of calculating the matching degree between the first trend feature and the second trend feature specifically includes: determining the number of structural anomalies on the target structural component; When the number of structural abnormal points on the target structural part is 1, obtain np corresponding to the structural abnormal point in the second structural data. i A set of time ranges (ts ij ,te ij ); For each time range set (ts ij ,te ij ), according to the rising inflection point set tip of the second vibration data k The number of items that fall within its time range is used to calculate its matching score s. i ; np i The maximum value among the matching scores Determine the target matching degree P between the first trend feature and the second trend feature.

10. The prediction-based crane maintenance method according to claim 9, characterized in that: Each structural abnormal point has a corresponding first trend feature. After the step of determining the number of structural abnormal points on the target structural component, the method further includes: When the number of structural abnormal points on the target structural part is greater than 1, determine the distance d between each structural abnormal point and the vibration monitoring point. l , where l∈[1,n un ],n un is the number of structural abnormal points on the target structural component; According to the distance d between each structural abnormal point and the vibration monitoring point l Configure the influence weight σ of each structural abnormal point on the matching degree between the first trend feature and the second trend feature l ; Calculate the matching degree p between the first trend feature and the second trend feature of each structural anomaly point l ; According to the influence weight σ of each structural outlier l The matching degree p between the first trend feature and the second trend feature l Calculating a target matching degree for determining whether the second structure data and the second vibration data satisfy a preset corresponding relationship:

Images