Wear prediction and notification system for machine

Abstract

The present disclosure describes, in part, systems and techniques for monitoring a link joint (104) of a machine (102, 106, 108), such as an earth moving machine, using an onboard sensor (138) of the machine (102, 106, 108) for identifying the link joint (104) at risk of wear. The systems and techniques involve acquiring sensor data (124) and determining a wear risk score (210, 212, 214) for each link joint (104). The wear risk score can then be used to alert an operator and / or maintenance system to determine when to take remedial actions and / or perform maintenance, while also avoiding expensive downtime due to wear failure.

Description

TECHNICAL FIELD

[0001] The present disclosure generally relates to monitoring wear of a link pin at a link joint of a machine, such as predicting wear and service life of a link pin using one or more models. BACKGROUND

[0002] Many earthmoving machines, such as, for example, loaders, tractors, and excavators, include link pin assemblies at joints between machine components that move relative to one another during operation of the machine. Such link pin assemblies include link pins that pivotally or rotatably support various structural members and other components of the machine relative to one another and bear shear, tensile, compressive, and torsional stresses exerted on the structural members and other components during operation of the machine performing work. The maintenance costs of these link pins and link pin assemblies often comprise a large percentage of the total costs associated with operating earthmoving machines due to wear from abrasion and impact experienced during use.

[0003] One mode of failure for link pin assemblies involves wearout failure. Wearout failure is known to occur during sliding contact between a link pin and a joint bushing or between a link pin and a machine member in a link pin joint assembly, particularly under high load applications. High load applications, such as incurred on larger heavy machinery, typically reduce the risk of wear by using a sleeve bearing positioned around the outer diameter surface of the pin between the link pin and the joint bushing. When wear occurs, both the link pin and the joint can need to be replaced, which is a time-consuming process, particularly replacing the joint.

[0004] An example system for detecting threaded joint wear is described in Japanese Publication No. 60225045 (hereinafter the ‘045 filing). In particular, the ‘045 filing describes a device for detecting wear of a steel pipe threaded joint by detecting the temperature of the pin and coupling at the joint during tightening of the pin. A sudden temperature change of the pin and / or threaded joint is indicative of wear occurring at the threaded connection.

[0005] While the device described in the ‘045 filing is configured to detect the occurrence of wear at a joint including a threaded connection, the device described in the ‘045 filing is not configured to predict and / or prevent wear at a link pin joint based on sensor data from the machine to avoid wear at the joint, thereby avoiding replacement of the joint and / or pin due to wearout failure. As a result, the device described in the ‘045 filing is not configured to maximize the efficiency of the machine and avoid downtime due to time-consuming repair of worn pins and joints.

[0006] Examples of the present disclosure are directed to overcoming the aforementioned deficiencies. SUMMARY

[0007] One general aspect includes a wear prediction and notification system for a machine. The wearing machine includes a first portion of a first machine member that defines a first bore, a second portion of a second machine member that defines a second bore, and a link pin that pivotally connects the first machine member and the second machine member. The machine also includes a sensor configured to monitor the link pin and / or a joint during operation of the machine. The machine further includes one or more processors and a non-transitory computer-readable medium operably connected to the one or more processors and storing instructions that, when executed by the one or more processors, cause the one or more processors to perform operations. The operations include receiving sensor data from the sensor, determining a first threshold value for the sensor data based on the link pin, the first machine member, and the second machine member, determining a risk score for the link pin, the risk score being associated with a wear risk and based on the sensor data and the threshold value, and generating an alert in response to the risk score being at or above a second threshold value. Other embodiments of this aspect include corresponding computer systems, apparatus, and computer programs recorded on one or more computer storage devices, each configured to perform the operations.

[0008] Another general aspect includes a method for assessing a wear risk of a link joint. The method includes receiving sensor data from a sensor of a machine, the sensor configured to measure the sensor data at a link pin that rotationally connects a first machine member and a second machine member. The method also includes determining, by a computing device of the machine, a threshold value for the sensor data. The method further includes determining, by the computing device of the machine and using an algorithm that receives inputs of the sensor data and the threshold value, a risk score associated with a wear risk at the link pin. The method also includes generating an alert in response to the risk score exceeding a second threshold value. Other embodiments of this aspect include corresponding computer systems, apparatus, and computer programs recorded on one or more computer storage devices, each configured to perform the actions of the method.

[0009] Another general aspect includes a non-transitory computer-readable medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform actions. The actions include receiving sensor data from a sensor of a machine, the sensor configured to measure the sensor data at a link pin that rotationally connects a first machine member and a second machine member, determining, by a computing device of the machine, a threshold value for the sensor data, determining, by the computing device of the machine and using an algorithm that receives inputs of the sensor data and the threshold value, a risk score associated with a wear risk at the link pin, and generating an alert in response to the risk score exceeding a second threshold value. Attached Figure Description

[0010] Detailed descriptions are provided with reference to the accompanying drawings. In the drawings, the leftmost numeral of the reference numeral indicates the drawing in which that numeral first appears. The same reference numerals are used in different drawings to indicate similar or identical items or features.

[0011] Figure 1 An example is provided of a system for determining wear risk predictions and generating machine maintenance notices, based on at least one example.

[0012] Figure 2 An example is illustrated of a process for generating a prediction of the wear risk level of a connecting rod pin based on load data and motion data, according to this disclosure.

[0013] Figure 3 An example is illustrated of a process for generating a prediction of the wear risk level of a connecting rod pin based on load data, temperature data, and motion data, according to this disclosure.

[0014] Figure 4 Charts illustrating load and motion data based on historical data to indicate wear risk limits according to this disclosure are provided.

[0015] Figure 5 A process for generating predictions and alarms of connecting pin wear in a machine, according to this disclosure, is illustrated.

[0016] Figure 6 An example is shown of a display illustrating the prediction and scoring of wear risk in operation of a machine according to this disclosure.

[0017] Figure 7 An example system architecture for a computational system associated with a mechanical and / or wear prediction system is illustrated. Detailed Implementation

[0018] In all the accompanying drawings, the same reference numerals will be used as much as possible to denote the same or similar parts.

[0019] Figure 1 A system 100 for determining wear risk predictions and generating machine maintenance notices, based on at least one example, is illustrated. System 100 can be implemented on machines, such as excavators or other such machines. As used herein, the term "machine" refers to a stationary or mobile machine that performs some type of operation associated with a particular industry (such as mining, construction, agriculture, etc.) and operates between or within a work environment (e.g., construction site, mine, power plant, etc.). Non-limiting examples of machines include commercial machinery such as trucks, cranes, excavators, mining vehicles, backhoe excavators, material handling equipment, agricultural equipment, ships, aircraft, and any type of mobile machine operating in a work environment. Figure 1As shown, machines 102, 106, and 108 (collectively referred to as "machine 102") are excavators. Figure 1 The machine types shown are exemplary and not intended to be limiting. It is contemplated that the disclosed embodiments can realize any type of machine having links for connecting components that rotate relative to each other.

[0020] Machines 102, 106, and 108 are illustrated as having a linkage engagement 104 that connects components such as machine parts and allows the machine parts to rotate relative to each other. For example, linkage engagement 104 can connect a first end of a hydraulic piston or other actuating mechanism to a fixed position on the machine, such that linkage engagement 104 can accommodate changes in the angle of the hydraulic piston as it extends. Linkage engagement 104 can be positioned between a first machine part 112 and a second machine part 114, allowing rotational or oscillating motion between the first machine part 112 and the second machine part 114. Linkage engagement 104 includes a linkage pin that rests in an opening or hole defined by the first machine part 112 and the second machine part 114. Linkage pin engagements are common in excavators and other large machinery and equipment. Linkage pin 116 is cylindrical in shape and in some applications is made of hardened steel or other such materials. A connecting rod pin 116 is disposed in a connecting rod engagement 104 where adjacent machine components meet, and the connecting rod pin 116 provides a cylindrical bearing surface about which the components rotate. The connecting rod engagement 104 typically includes the connecting rod pin 116, one or more bearings 122, lubricant, seals 120, and a retaining member 118, which serves to hold the connecting rod engagement 104 assembled during use, but also allows for maintenance and repair of the engagement by removing the retaining member 118.

[0021] One common failure mode among the common failure modes of the connecting rod joint 104 is wear, also known as adhesive wear. Wear failure can occur catastrophically without warning when the connecting rod pin 116 and bearing 122 are subjected to heavy loads during frequent movement. This results in unexpected downtime for the machine owner, as well as additional ownership and operating costs. Operational problems accelerate the risk of wear, including insufficient lubrication and overloading of the connecting rod pin 116 due to operator misuse. Therefore, system 100 is used to automatically assess the wear risk level of various connecting rod joints 104 of machines 102, 106, or 108 in real time during operation, which can be used to trigger preventative measures and avoid costly repairs and downtime.

[0022] Wear failure of the connecting rod pin 116 typically manifests as welding between the connecting rod pin 116, bearing 122, and / or first machine component 112 and / or second machine component 114 due to high pressure and temperature at the connecting rod joint 104. Wear failure can lead to catastrophic failure and unexpected downtime because the entire joint must be removed and replaced. Conversely, if wear can be predicted and anticipated before it occurs, the connecting rod pin 116 can be replaced by removing the retaining member 118 (a quick and inexpensive process) to reduce machine downtime and maintenance costs.

[0023] Bearing 122 may include a sleeve bearing disposed within connecting rod engagement 104 to provide an internal bearing surface between machine components 112 and 114 and connecting rod pin 116. Bearing 122 is tubular in shape and may be press-fitted into connecting rod engagement 104 or secured by other suitable mechanisms. As adjacent components of the connecting rod pivot relative to each other, the outer bearing surface of the connecting rod pin supports against the inner surface of bearing 122. Lubrication is maintained between the connecting rod pin and the inner surface of the bearing by seal 120.

[0024] Bearing 122 serves as a sliding component abutting against connecting rod pin 116. However, under certain load and motion conditions during repetitive machine operation, bearing 122 may become welded to connecting rod pin 116. The bonded portion then separates from the original component, leaving rough surfaces and material on bearing 122 and connecting rod pin 116. This wear failure then accelerates surface damage and causes corresponding damage to adjacent components, rendering the machine unusable until the entire connecting rod joint 104 is repaired. Because the interface between connecting rod pin 116 and bearing 122 is enclosed within connecting rod joint 104, unfortunately, by the time wear failure is detected, significant damage to surrounding structures or hydraulic components has already occurred. Therefore, the system 100 described herein provides early prediction and detection of conditions that may lead to wear and can be used to enable maintenance or replacement of connecting rod pin 116 and / or bearing 122 before wear that could cause machine damage and extensive repairs occurs.

[0025] Machine 102 includes a first computing device 138. In such examples, besides relative to... Figure 7In addition to the other components shown and described, the first computing device 138 further includes a processor, memory, and a user interface. The processor is an electronic controller that operates logically to perform operations, execute control algorithms, store and retrieve data, and perform other desired operations. The processor may include or access memory, auxiliary storage devices, the processor itself, and any other components used to run applications. The memory and auxiliary storage devices may be in the form of read-only memory (ROM) or random access memory (RAM) or integrated circuits accessible by the processor. Various other circuits may be associated with the processor, such as power supply circuits, signal conditioning circuits, driver circuits, and other types of circuitry. The user interface may include an analog display, a digital display, and / or a touchscreen display, and such control interfaces are configured to, for example, at least display machine status (including, for example, relative to...). Figure 6 Information related to the status of the linkage pins shown and described. The user interface may also support other functions, including, for example, sharing various operational data with one or more other machines or computing devices of system 100.

[0026] A processor can be a single processor or other device, or it can include more than one controller or processor configured to control various functions and / or features of the control system 100. As used herein, the term "processor" in its broadest sense means including one or more controllers, processors, and / or microprocessors associated with and capable of cooperating with the control system 100 to control various functions and operations of components of the control system 100 (e.g., machine 102). The functionality of the processor can be implemented in hardware and / or software, without regard to the functionality itself. The processor relies on data that can be stored in memory associated with the processor, such as sensor data related to the operating conditions of the machine 102.

[0027] The first computing device 138 is configured to facilitate interoperability between its included components and to receive data sensed from sensors of machine 102, such as temperature sensors, pressure sensors, and motion sensors (e.g., speed, acceleration, etc.). In some examples, the processor is configured to operate according to instructions stored in memory, and in some cases, the processor is configured to cause a user interface to be displayed and / or otherwise provide information. In some examples, the second machine 108 includes a second computing device. In such examples, the second computing device includes components that are the same as, similar to, and / or different from those of the first computing device 138. Each machine in machine 102 may have one or more dedicated computing devices that may operate in the same, similar to, and / or different ways than the first computing device 138 described herein.

[0028] The second computing device 140 may be a fleet management computing device, which has a relative to the fleet management computing device. Figure 7 The computing system 700 has a similar or identical structure to the one described. In some examples, the second computing device 140 may include a job site controller computing device that manages one or more operations at a specific job site. In some examples, the second computing device 140 may include a fleet management device configured to track and maintain records and status information of a fleet of machines. For example, the second computing device 140 may be used to store maintenance records, wear prediction scores, usage information, and other such data that can be used to identify when maintenance is required for the machines 102 in the fleet.

[0029] exist Figure 1 In the example shown, the first computing device 138 and the second computing device 140 communicate with each other and / or are otherwise connected via network 126. Network 126 may be a local area network (“LAN”), a larger network (such as a wide area network (“WAN”)), or a collection of networks (such as the Internet). Network communication protocols (such as TCP / IP) may be used to implement network 126. Although the implementation is described herein as using network 126 (such as the Internet), other distribution technologies for transmitting information via memory cards, flash memory, or other portable storage devices may be implemented. Network 126 may implement or utilize any desired system or protocol, including any of the various communication standards. The desired protocol will allow communication between the various processors and computing devices of the components in system 100. Examples of wireless communication systems or protocols that system 100 may use herein include wireless personal area networks (such as Bluetooth RTM (e.g., IEEE 802.15)), local area networks (such as IEEE 802.11b or 802.11g), cellular networks, or any other system or protocol for data transmission. Other wireless communication systems and configurations are contemplated. In some cases, wireless communication can be transmitted and received directly between components of system 100 (e.g., between machines). In other cases, communication can be automatically routed without requiring retransmission by remote personnel.

[0030] In a further example, the computing devices of two or more machines in machine 102 communicate with each other and / or are otherwise connected via network 126. In some further examples, network 126 further connects one or more additional computing devices and / or one or more satellites. Thus, as described herein, network 126 is configured to facilitate communication between any one or more machines located at and / or away from the work site and / or machine 102, computing platforms at the work site, other work sites and / or material handling and / or storage plants, and / or away from the work site, other work sites and / or material handling and / or storage plants, etc. Additional computing devices may also include components of system 100. For example, such additional computing devices may include a mobile phone, laptop computer, desktop computer, and / or tablet computer of a project manager (e.g., a foreman) who oversees daily paving operations at the work site and / or paving material plant. Such additional computing devices may include and / or be configured to access one or more processors, microprocessors, memories, or other components. In such examples, the additional computing device has similar and / or the same components and / or functions as the first computing device 138.

[0031] like Figure 1 As shown, system 100 also includes a computing platform 140 that communicates with and / or is otherwise connected to network 126. In such examples, computing platform 140 includes a user interface 130 and memory 132. User interface 130 provides the ability to perform actions (i.e., manual, semi-autonomous, and / or fully autonomous) in a centralized location (i.e., in the background), including data input, determination of object classification (e.g., object type, object identity, etc.), determination of remediation procedures, and determination of machine alerts and / or instructions to perform remediation. Memory 132 may store computer-readable media, machine learning algorithm instructions, compiled sensor data, machine work schedule data, field data, etc. In some further examples, computing platform 140 includes processor 134, which includes a pin management engine 136. Pin management engine 136 may include software components of processor 134 and / or associated with processor (e.g., executable software components stored in memory 132) or hardware components of processor 134. In some examples, processor 134 performs tasks provided by instructions stored in memory 132, including but not limited to activating pin management engine 136, which determines a risk score and prediction of wear at linkage joint 104 of machine 102 from one or more sensors located on one or more machines.

[0032] System 100 uses sensors of machine 102 to acquire sensor data 124. This sensor data can be consumed by the pin management engine 136, embodied in the first computing device 138 or computing platform 140, to detect operating conditions at the connecting pins and thereby determine the risk of wear at each connecting rod joint 104. In some examples, the sensor data may include temperature data, load data (e.g., pressure related to the hydraulic piston associated with the connecting rod joint 104), motion data (e.g., rotational rate or speed at the connecting rod joint 104), and other such data.

[0033] In some examples, the sensor may include a thermocouple located at or near the link joint 104, which is used to monitor temperature data as part of sensor data 124. In some examples, a thermal imager may be used to monitor temperature data of one or more link joints 104 of the machine 102. For example, a thermal imager may be attached to the cab of the machine 102 and pointed towards the front of the machine to capture thermal data of the front of the machine. Positioning systems, such as a link tracking system of the machine 102 that tracks the location and position of the link joints 104 and / or machine learning algorithms trained for object recognition, may be used to identify portions of the temperature data associated with the various link joints 104. Other temperature sensors may be implemented, and it is envisioned to acquire temperature data at or adjacent to the link joint 104 to estimate or determine the temperature of the link pin 116.

[0034] In some examples, the sensor may include a pressure sensor or a load sensor configured to detect cylinder pressure of a hydraulic cylinder associated with the link joint 104 as part of sensor data 124 (e.g., a hydraulic cylinder or other actuating device coupled to an adjacent machine component engaged at the link joint 104). In some examples, the load sensor may be used to determine the pressure at the link joint 104. The positioning and orientation of the cylinder may be used to determine the positioning of the link of the computing machine 102. The calculation may be performed by a first computing device 138.

[0035] In some examples, the sensor may include a motion sensor used to determine motion data as part of the sensor data for the link joint 104. The motion sensor may include a positioning sensor, an encoder, or other such sensor capable of determining the rate of change of rotational positioning at the link joint 104. In some examples, the positioning of the link over time may be used to calculate the motion. For example, the lifting, tilting, angulating, and / or positioning of the hydraulic cylinder actuating the link joint 104 may be used to calculate the motion at the link joint. The calculation may be performed by the first computing device 138.

[0036] System 100 can use sensor data 124 transmitted to computing platform 140 via network 126 to determine a wear risk score for each link joint 104. In some examples, computing platform 140 may include a remote computing device separate from and remote from machine 102. In some examples, computing platform 140 may include computing devices of machine 102. Pin management engine 136 is configured to determine a wear risk score for each link joint 104 of machine 102 and to report and / or induce additional actions based on the wear risk score. The wear risk score can be determined using sensor data 124 and based on pressure, motion, and / or temperature limits for each link joint, as determined by historical data. Historical data may include indications of wear and temperature data, load data, and / or motion data associated with the indications of wear. In some examples, historical data may be used to determine threshold limits for pressure, motion, and / or temperature data. Threshold limits may be, for example, in... Figure 4 The data can be represented as a two-dimensional chart or a three-dimensional surface. When sensor data 124 approaches and / or exceeds a two-dimensional or three-dimensional threshold, the pin management engine 136 can generate a wear alarm, calculate a wear risk score based on sensor data 124, and / or adjust the operation of machine 102 to reduce the wear risk until the linkage joint 104 can be investigated and / or maintained (e.g., by replacing the linkage pin 116).

[0037] A system 100 temperature sensor in machine 102 is used to detect changes in the temperature of the link joint 104. In addition to pressure and motion data, the temperature sensor can also provide additional confidence relative to the risk of wear, as a rising temperature of the link joint 104 may be a direct precursor to impending wear failure. In some examples, temperature can also indicate the occurrence of wear and can therefore be used to identify when the link joint 104 should be taken out of service for maintenance. Thus, using load data, motion data, and temperature data makes it possible to accurately predict the wear risk of each link joint 104 on machine 102.

[0038] In some examples, a first threshold may be established at which system 100 generates a wear risk alarm. A second threshold may be established at which the wear risk exceeds the threshold, making wear expected in the near future. In some cases, the pin management engine 136 may generate alarm data in response to approaching (e.g., based on the rate of change) the first threshold and / or reaching or exceeding the threshold. The pin management engine 136 may further generate additional alarms in response to approaching and / or reaching the second threshold. In some examples, the pin management engine 136 may terminate or reduce activity at the identified link joint 104 in response to reaching or exceeding the second threshold. For example, the pin management engine may prevent rotation and / or limit the maximum rotational speed and / or pressure at the link joint 104 to prevent damage to the link joint 104 until a replacement link pin 116 can be installed. This avoids costly repairs to the link joint 104, while other operations of the machine 102 can continue. Therefore, the machine can continue to be used without causing damage to the specific link joint 104.

[0039] In some examples, the sales management engine 136 can implement one or more algorithms and / or machine learning models to determine a wear risk score based on sensor data 124. The machine learning model can determine the wear risk score and also a confidence score associated with it. Thus, the machine learning model can determine the wear risk score using, for example, pressure sensor data and motion sensor data, but in response to a confidence score below a threshold, can use additional sensor data (e.g., from a second time period and / or from different sensors or different types of sensors) to determine a second wear risk score. The machine learning model can then update the confidence score accordingly.

[0040] In some examples, remediation engine 146 includes or has access to one or more machine learning models (e.g., data models) configured to determine a remediation plan for object 119. As part of determining the remediation plan, the one or more machine learning models can be used for object detection feature estimation, object classification, or other tasks described herein. Machine learning can take empirical data, such as data from machine sensors, as well as object data and machine location, as input. Machine learning systems can utilize data to capture features of interest with unknown underlying probability distributions or weights. Machine learning can be used to identify possible relationships between observed variables. Machine learning can also be used to identify complex patterns and make machine decisions based on input data. In some examples, machine learning systems can generalize from available data to produce useful outputs, such as when the amount of available data is too large to be used effectively or practically. When applied to this technique, machine learning can be used to learn which performance features to retain during the localization process and to validate the localized content while performance features are retained.

[0041] A wide variety of methods can be used to perform machine learning, such as contrastive learning, supervised learning, unsupervised learning, temporal difference learning, reinforcement learning, etc. Some unrestricted examples of supervised learning that can be used with this technique include AODE (Average Single Dependency Estimator), artificial neural networks, backpropagation, Bayesian statistics, Naive Bayes classifiers, Bayesian networks, Bayesian knowledge bases, case-based reasoning, decision trees, inductive logic programming, Gaussian process regression, gene expression programming, grouping methods for data processing (GMDH), learning automata, learning vector quantization, minimum message length (decision trees, decision graphs, etc.), lazy learning, instance-based learning, nearest neighbor algorithms, analogy modeling, possibly approximating correct (PAC) learning, ripple descent rules, knowledge acquisition methods, symbolic machine learning algorithms, sub-symbolic machine learning algorithms, support vector machines, random forests, classifier ensembles, bootstrap aggregation (bagging), boosting (meta-algorithms), ordered classification, regression analysis, information fuzzy networks (IFN), statistical classification, linear classifiers, Fisher's linear discriminant method, logistic regression, perceptrons, support vector machines, quadratic classifiers, k-nearest neighbors, hidden Markov models, and boosting. Some non-limiting examples of unsupervised learning that can be used with this technique include artificial neural networks, data clustering, expectation maximization, self-organizing maps, radial basis function networks, vector quantization, generative terrain mapping, information bottleneck methods, IBSEAD (Interaction Based on Distributed Autonomous Entity Systems), association rule learning, prior algorithms, Eclat algorithm, FP-growth algorithm, hierarchical clustering, single-linked clustering, concept clustering, partitioned clustering, K-means algorithm, fuzzy clustering, and reinforcement learning. Some non-limiting examples of temporal difference learning can include Q-learning and learning automata. Another example of machine learning includes data preprocessing. Specific details regarding any example of generally known supervised, unsupervised, temporal difference, or other machine learning examples described in this paragraph are also considered to be within the scope of this disclosure. Support vector machines (SVM) and regression are several specific examples of machine learning that can be used with this technique.

[0042] In some examples, the machine learning model may include access to or versions of multiple different machine learning models, which may be implemented and / or trained according to the techniques described herein. For example, machine learning algorithms may include, but are not limited to, regression algorithms (e.g., Ordinary Least Squares Regression (OLSR), Linear Regression, Logistic Regression, Stepwise Regression, Multivariate Adaptive Regression Splines (MARS), Local Estimation Scatter Graph Smoothing (LOESS)), instance-based algorithms (e.g., Ridge Regression, Minimum Absolute Shrinkage and Selection Operator (LASSO), Elastic Net, Minimum Angular Regression (LARS)), decision tree algorithms (e.g., Classification and Regression Tree (CART), Iterative Dichotomizer 3 (ID3), Chi-Square Automatic Interaction Detection (CHAID), Decision Stump, Conditional Decision Tree), Bayesian algorithms (e.g., Naive Bayes, Gaussian Naive Bayes, Multinomial Naive Bayes, Average Single Dependency Estimator (AODE), Bayesian Belief Network (BNN), Bayesian Network), clustering algorithms (e.g., k-means, k-median, Expectation-Maximization (EM), Hierarchical Clustering), and association rule learning. Algorithms (e.g., perceptron, backpropagation, Hopfield network, radial basis function network (RBFN)), deep learning algorithms (e.g., deep Boltzmann machine (DBM), deep belief network (DBN), convolutional neural network (CNN), stacked autoencoder), dimensionality reduction algorithms (e.g., principal component analysis (PCA), principal component regression (PCR), partial least squares regression (PLSR), Sammon map, multidimensional scaling (MDS), projection tracking, linear discriminant analysis (LDA), mixture discriminant analysis (MDA), quadratic discriminant analysis (QDA), flexible discriminant analysis (FDA)), ensemble algorithms (e.g., boosting, bootstrap aggregation (bagging), AdaBoost, stacked generalization (blending), gradient boosting machine (GBM), gradient boosting regression tree (GBRT), random forest), SVM (support vector machine), supervised learning, unsupervised learning, semi-supervised learning, etc.). Additional examples of architectures include neural networks such as ResNet50, ResNet101, VGG, DenseNet, PointNet, etc.

[0043] Training examples can be used to train one or more machine learning algorithms. Training examples may include example inputs from sensor data from multiple sensors, and the expected output associated with the input, identifying the object, obstacle, or object type. Using training examples to train a machine learning algorithm generates a trained machine learning algorithm, which can then be used to estimate the output for inputs not included in the training examples. In some examples, engineers, scientists, processes, and machines training the machine learning algorithm may further use validation examples and / or test examples. For instance, validation examples and / or test examples may include example inputs and the expected outputs corresponding to those example inputs. The trained machine learning algorithm and / or intermediately trained machine learning algorithm can be used to estimate the outputs of the example inputs to the validation examples and / or test examples. The estimated outputs can be compared with the corresponding expected outputs, and the trained machine learning algorithm and / or intermediately trained machine learning algorithm can be evaluated based on the comparison results.

[0044] Figure 2 An example of process 200 for generating a prediction of link pin wear risk level based on load data and motion data according to this disclosure is illustrated. In some examples, process 200 may be... Figure 1 The pin management engine 136 implements this. Process 200 can be performed by one or more computing devices communicating with or onboard to the machine 102. Process 200 can be used to determine a wear risk score and an associated confidence score. The wear risk score can then be used to identify the risk level of a specific linkage joint 104 of the machine 102. Process 200 can be used to generate a wear risk assessment 208 based on load data 202 and motion data 204. In some examples, the wear risk assessment 208 can be based on only one sensor data input from the sensor data inputs.

[0045] Wear risk assessment 208 can be the result of inputs from load data 202 and motion data 204. Load data 202 can include loads at the joint, such as measurements based on hydraulic actuator pressure or other loads associated with the joint. Motion data 204 can include the rotational speed of the joint, movement relative to the ground, directional movement relative to gravity (e.g., co-gravity or anti-gravity), and acceleration data. Inputs can include sensor data 124 and can be acquired from one or more sensor systems of machine 102. Wear risk assessment 208 can also be generated based on wear prediction engine 206. Wear prediction engine 206 can consume sensor data 124 (such as load data 202 and motion data 204) to generate a prediction of wear risk for link joint 104. In some examples, wear prediction engine 206 can include graphs (such as those relative to...) Figure 4(As shown and described), the graph illustrates wear indicators that vary with motion and load data. In some examples, the wear prediction engine 206 can use load data 202 and motion data 204 to identify areas of the graph 400 and determine wear risk based on the areas where data points are located. As described above, the wear prediction engine can also implement machine learning methods to determine wear risk. For example, the wear prediction engine 206 can be a component of the pin management engine 136.

[0046] In some examples, wear risk can be categorized into different risk levels. Wear risk assessment 208 can output a binary indication, based on risk scores above or below a threshold and / or motion and load data indicating areas of graph 400 indicating no wear with wear risk, low wear risk, or no wear risk. In some examples, wear risk assessment 208 can output risk levels, such as risk level A 210, risk level B 212, and risk level C 214. Risk levels can be based on thresholds of risk scores, where a first risk score span is assigned risk level A 210, a second risk score span is assigned risk level B 212, and a third risk score span is assigned risk level C 214.

[0047] In an illustrative example, risk level A 210 may indicate a high wear risk based on a risk score higher than a threshold score (such as above 75 points (out of 100) on a risk scoring scale or some other threshold). Risk level A 210 may cause the computing device of machine 102 to take a first set of remedial actions, including alerting the operator, alerting the maintenance system, and / or slowing down or terminating the operation of the identified machine 102 and / or linkage joint 104. Risk level B 212 may be associated with low risk (such as a risk score range between 25 and 75 points). In response to risk level B 212, the computing device may generate a status alarm to the operator and / or maintenance system. Risk level C 214 may be associated with no wear risk (such as a risk score range below 25 points). The computing device may take no action in response to risk level C 214. In some examples, when determining wear risk assessment 208, regardless of the risk level, the computing system may report the risk score to the maintenance or logging system to maintain records related to the operation of machine 102.

[0048] Figure 3 An example is illustrated of a process 300 for generating a prediction of connecting rod pin wear risk level based on load data, temperature data, and motion data, according to this disclosure. In some examples, process 300 may be... Figure 1The pin management engine 136 is used to implement this. Process 300 can be performed by one or more computing devices that communicate with or are onboard to the machine 102. Process 300 can be used to determine a wear risk score and an associated confidence score. The wear risk score can then be used to identify the risk level of a specific linkage joint 104 of the machine 102. Process 300 can be used to determine a wear risk assessment 312 using load data 302, motion data 304, and temperature data 306. In some examples, the wear risk assessment 312 can be based on one, two, or all three sensor data inputs. In some examples, additional inputs, such as operating time and environmental conditions, can also be provided as inputs to process 300.

[0049] Wear risk assessment 312 can be the result of inputs from load data 302, motion data 304, and temperature data 306. Inputs may include sensor data 124, which can be acquired from one or more sensor systems of machine 102. Wear risk assessment 312 can also be generated based on joint limits 310 of specific joints of machine 102. For example, pin joint database 308 may include data on different link pin configurations, arrangements, materials, diameters, and other such differences that may cause wear predictions for different link joints. Joint limits 310 of a specific link joint 104 can be accessed from pin joint database 308. Pin management engine 136 can consume sensor data 124 and joint limits 310 to generate wear risk predictions for link joint 104. In some examples, pin management engine 136 may include charts (such as relative to...) Figure 4 (As shown and described) (e.g., a three-dimensional graph with a threshold surface), the graph shows an indication of wear as motion data, temperature data, and load data change.

[0050] In some examples, the wear risk determined at wear risk assessment 312 can be categorized into different risk levels. Wear risk assessment 312 can output a binary indication, indicating areas of graph 400 with no wear but wear risk, low wear risk, or no wear risk, based on risk scores above or below thresholds and / or motion and load data. In some examples, wear risk assessment 312 can output risk levels (such as risk level A 314, risk level B 316, risk level C 318) and an indication of failure 320. Risk levels can be based on thresholds of risk scores and / or directly from sensor data, where a first risk score span is assigned risk level A 314, a second risk score span is assigned risk level B 316, and a third risk score span is assigned risk level C 318. Failure 320 can also be determined, for example, at the third risk score span and / or based on sensor data (including temperature data indicating a temperature rise above the steady-state temperature at the link joint 104).

[0051] Risk levels and / or indications of failure can be consumed by onboard and / or offboard systems associated with machine 102. Examples of ways in which risk levels can be consumed include: alarms to operators and maintenance systems; changes to equipment operation, failure codes, onboard displays; adjustments to lubrication systems or settings associated with linkage 104; or other such actions. In some examples, where machine 102 is configured with an automatic lubrication system (which is configured to deliver lubricant to different systems of machine 102 such as linkage 104), the calculation system can respond to various risk levels by increasing the amount or rate of lubricant delivered to linkage 104.

[0052] Figure 4 A graph 400 illustrating motion data 402 and load data 404, based on historical data according to this disclosure, is shown. The graph 400 is divided into specific regions based on the collected historical data related to the indication of wear. For example, when wear is detected and / or determined, the motion data 402 and load data 404 at the time of wear occurrence can be plotted on the annotated graph 400. Annotations for load data and motion data that do not cause wear can also be added to the graph 400. Thus, regions of various risks can be determined based on the annotations (e.g., indicating wear versus indicating no wear). Thresholds 406 and 408 can be determined. When data is collected, for each data point along the motion data 402 axis and the load data 404 axis, the data points can be labeled or annotated using an indication of wear and / or a wear risk score. The set of data points can be collected and identified in regions A 410, B 412, and C 414.

[0053] Threshold 406 can indicate the engagement limit of the link engagement 104. The engagement limit can be a limit (at a specific pressure level) at which the link pin 116 is expected to engage based on load, independent of motion data 402. Therefore, region C 414 can indicate wear failure due to exceeding the engagement limit of the link engagement 104. The engagement limit can be determined based on historical data. Therefore, data points falling within region C 414 can be identified as indicating wear failure, independent of motion data 402.

[0054] Threshold 408 may indicate a load motion threshold defining the boundary between region A 410 and region B 412. Region A 410 may indicate no wear present, detected, or predicted, while region B 412 may be a set of points (which may include portions of region C 414) where wear is present, detected, or predicted for the link joint 104. Although thresholds 406 and 408 are illustrated as linear, they may have other shapes or profiles based on the specific distribution of data points in regions A 410, B 412, and C 414.

[0055] Figure 5 A flowchart illustrating a process 500 for generating a prediction and alarm of linkage pin wear for a machine, according to this disclosure, is shown. Process 500 is illustrated as a set of steps in a logic flowchart, representing operations that can be implemented in hardware, software, or a combination thereof. In the context of software, steps represent computer-executable instructions stored in memory. When such instructions are executed by, for example, computing platform 140, these instructions cause first computing device 138 and / or computing platform and / or additional computing devices to typically receive instructions corresponding to generating at least a wear risk prediction based on sensor data from sensors of machine 102. Such computer-executable instructions may include routines, programs, objects, components, data structures, etc., that perform a particular function or implement a particular abstract data type. The order in which the operations are described is not intended to be construed as limiting, and the process can be implemented in any order and / or in parallel combination of any number of the described steps.

[0056] At 502, process 500 includes: a computing device (e.g., computing platform 140) initiating monitoring of the linkage joint 104 of machine 102. This initiation may be in response to the start of machine operation and / or in response to an operator or other user instructing the system to begin monitoring. Process 500 includes: receiving load data 504, motion data 506, temperature data 508, and other sensor data from the machine's sensors during operation. The sensor data may be provided over time and / or at predetermined time intervals. A first threshold (e.g., a bite threshold) may be associated with load data 504. A second threshold (load-motion threshold) may be associated with load data 504 and motion data 506, such as... Figure 4 As shown and described. The third threshold can be based on temperature data 508.

[0057] At 510, process 500 includes: a pin management engine 136 determining the wear risk of the linkage joint 104. The wear risk may be determined based on sensor data and using one or more of the processes described herein, such as load data and / or motion data and / or temperature data, and also using machine learning algorithms or manual processes (such as lookups based on charts of data and / or risk scores) or other such processes. If the wear risk is not present (e.g., below a first threshold level), process 500 may return to 502 to continue monitoring during machine operation. If the wear risk is above the first threshold (e.g., wear risk is present), process 500 generates a first alarm at 512 by a computing device. The first alarm may identify the linkage joint 104 associated with the alarm and may also identify the risk level. The first alarm may be communicated to an operator, remote station, maintenance system, or other systems associated with the machine.

[0058] At 514, process 500 includes: the computing device determining whether the wear risk is higher than a second threshold. The second threshold is higher than a first threshold. If the risk is lower than the second threshold, process 500 may include: the computing device continuing to monitor operation, wherein the first alarm generated at 512 persists. If the wear risk is higher than the second threshold, the computing device may determine at 516 whether wear has occurred and / or is occurring at the joint. The determination of failure may be based on an increase in temperature data above the steady-state temperature or other indications of wear failure. If wear is not determined to have occurred, a second alarm may be generated at 518, indicating that the linkage pin joint has an increased risk of wear compared to the risk associated with the first alarm. In some examples, the second alarm may indicate that wear is imminent and may prompt the operator to take remedial action and / or the computing device to determine to terminate operation at the linkage joint 104. If wear failure is determined at 516, process 500 may include: generating a third alarm at 520 and / or stopping machine operation to prevent damage to or further damage to the machine.

[0059] Figure 6 An illustration 600, according to this disclosure, shows a prediction and scoring of wear risk in operation of a machine. The display 600 may be a representation of a display that can be presented to the operator and / or other systems associated with the machine during monitoring of linkage joints.

[0060] In display 600, the representation of machine 602 is shown using markers 604 indicating the location of linkage joints on machine 602. In some examples, multiple views or other representations may be presented to illustrate the status and / or location of different linkage joints monitored by the system. Scale 606 within the display may be used to indicate the risk or status associated with each linkage joint and / or the machine as a whole. For example, display 600 may include a touchscreen display that a user can use to select markers and cause scale 606 to reflect a specific wear risk score associated with the selected linkage joint. Scale 606 is illustrated as areas 608, 610, 612, and 614 indicating increasing wear risk scores. Thus, the operator can view the status of the linkage joints at a glance without stopping operation to inspect each joint individually.

[0061] Figure 7 An example system architecture of a computing system 700 associated with computing platform 140 is shown. In some examples, other computing systems (such as the first computing device 138 or other onboard or offboard computing systems associated with machine 102) may have similar architectures. Figure 7The system architecture shown is similar to that of the system. The computing system 700 may include one or more servers or other computing devices, which include one or more processors 702, memory 704, and communication interface 706.

[0062] Processor 702 can operate to perform a variety of functions as set forth herein. Processor 702 may include one or more chips, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs) and / or other programmable circuits, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), and / or other processing units or components known in the art. In some examples, processor 702 may have one or more arithmetic logic units (ALUs) that perform arithmetic and logical operations and / or one or more control units (CUs) that fetch instructions and stored contents from processor cache and execute such instructions by invocation on the ALUs during program execution. Processor 702 may also access contents and computer-executable instructions stored in memory 704 and execute such computer-executable instructions.

[0063] Memory 704 may be a volatile and / or non-volatile computer-readable medium, including integrated or removable storage devices, including random access memory (RAM), read-only memory (ROM), flash memory, hard disk drive or other disk drive, memory card, optical storage device, magnetic storage device, and / or any other computer-readable medium. The computer-readable medium may be a non-transitory computer-readable medium. The computer-readable medium may be configured to store computer-executable instructions that can be executed by processor 702 to perform the operations described herein.

[0064] For example, memory 704 may include drive units and / or other elements including machine-readable media. The machine-readable media may store one or more sets of instructions, such as software or firmware, embodying any one or more of the methods or functions described herein. The instructions may also reside wholly or at least partially within processor 702 and / or communication interface 706 during execution by computing system 700. For example, processor 702 may have local memory that may also store program modules, program data, and / or one or more operating systems.

[0065] Memory 704 may store data and / or computer-executable instructions associated with the pin management engine 136, wear prediction engine 206, pin joint database 308, joint limit 310, and / or other elements as described herein. Memory 704 may also store other modules and data that can be used by computing system 700 to perform or enable the performance of any actions taken by computing system 700. For example, other modules and data 708 may include a platform, operating system, and / or applications, as well as data used by the platform, operating system, and / or applications.

[0066] Communication interface 706 may include a transceiver, modem, interface, antenna, and / or other components that can transmit and / or receive data via a network or other data connection. In some examples, communication interface 706 may be a wireless communication interface that computing platform 140 can use to send sensor data, alarms, and wear risk data, as well as receive data.

[0067] Industrial applicability

[0068] This disclosure provides systems and methods for monitoring linkage pins on machinery during operation. Such systems and methods can be used to achieve better performance in the operation of one or more machines by reducing downtime and associated repair costs. Additionally, such systems and methods can be used to improve inefficiencies in machine operation. As described above relative to... Figure 1 The system 100 can use sensor data to identify wear risks associated with specific link joints 104, and thus preventatively address or provide maintenance for the link joints, rather than requiring expensive and time-consuming repairs to worn joints. Using sensor data from the machine, the machine's onboard computer can monitor the condition of each joint and alert the operator when conditions that could lead to wear failure are anticipated. This allows the machine to remain operational with minimal downtime, thereby improving workplace efficiency.

[0069] The computing platform 140 is also configured to provide such information to additional computing devices used by, for example, maintenance systems at the work site and / or to computing devices located at, for example, remote maintenance systems. Such information can be used by, for example, maintenance systems to improve repair efficiency by ensuring that parts are ordered proactively, at the initial prediction of wear conditions, but before replacement may be necessary. Therefore, the example systems and methods described above can provide substantial cost and time savings, and reduce the time and labor required for various activities at the work site, as well as other things that become apparent to those skilled in the art.

[0070] While various aspects of this disclosure have been specifically shown and described with respect to the embodiments described above, those skilled in the art will understand that various additional embodiments can be contemplated by modifications to the disclosed machines, systems, and methods without departing from the spirit and scope of the disclosure. Such embodiments should be understood to fall within the scope of this disclosure as determined by the claims and any equivalents.

Claims

1. A wear prediction and notification system (100) for a machine (102), the wear prediction and notification system comprising: A first portion of a first machine component (112) defines a first hole; The second part of the second machine component (114) defines the second hole; Linkage pin (116), which pivotally connects the first machine component and the second machine component; Sensor (138), the sensor being configured to monitor the link pin and / or engagement during operation of the machine; One or more processors (134); and A non-transitory computer-readable medium (132) operatively connected to the one or more processors (134) and storing instructions that, when executed by the one or more processors (134), cause the one or more processors (134) to: Receive sensor data (124) from the sensor (138); A first threshold (406, 408) of the sensor data (124) is determined based on the connecting pin (116), the first machine component (112), and the second machine component (114). Determine a risk score (210, 212, 214) for the connecting pin (116), the risk score being correlated with wear risk and based on the sensor data (124) and the threshold; and An alarm (600) is generated in response to the risk score being at or above the second threshold.

2. The wear prediction and notification system according to claim 1, wherein: The sensor (138) includes a temperature sensor configured to acquire temperature data (306) associated with the connecting rod pin (116); and The threshold is based on: The steady-state operating temperature of the connecting rod pin (116); or The rate of change limit of the connecting pin (116).

3. The wear prediction and notification system of claim 1, wherein the instruction includes an additional instruction that, when executed by the one or more processors, causes the one or more processors to additionally: Determine the third threshold (406) of the sensor data (124); and In response to the sensor data exceeding the third threshold, one or more operations of the machine (102) are terminated to stop the movement at the linkage pin (116).

4. A method (500), the method comprising: Sensor data (504, 506, 508) are received from a sensor (138) of the machine (102), the sensor (102) being configured to measure sensor data (124) at a link pin (116) rotatably connecting the first machine component (112) and the second machine component (114). The threshold (406, 408) of the sensor data (124) is determined by the computing device (140) of the machine. The computing device (140) of the machine and the algorithm (208, 312) that receives the sensor data (124) and the threshold (406, 408) as input determine the risk score (210, 212, 214) associated with the wear risk at the link pin (216). as well as An alarm (600) is generated in response to the risk score exceeding the second threshold.

5. The method according to claim 4, wherein the sensor data (124) includes temperature data (306), and wherein the threshold includes the steady-state operating temperature of the connecting pin (116) or the threshold change rate of the temperature data (306).

6. The method of claim 4, wherein the sensor (138) comprises: A pressure sensor configured to measure load data (202, 302) associated with an actuation device connecting the first machine component (112) and the second machine component (114); A motion sensor configured to measure motion data (204, 304) associated with rotation at the link pin (116), wherein the threshold is based on the load data (202, 302) and the motion data (204, 304).

7. The method of claim 4, wherein the algorithm (208, 312) includes a machine learning algorithm configured to receive the sensor data (124) and output risk score data (210, 212, 214), the machine learning algorithm being trained using historical sensor data (400) annotated with indications of wear at the linkage pins of the machine.

8. A non-transitory computer-readable medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform operations, the operations including: Sensor data (124) is received from a sensor (138) of the machine (102), the sensor (138) being configured to measure the sensor data (124) at a link pin (116) that rotatably connects the first machine component (112) and the second machine component (114). The threshold (406, 408) of the sensor data (124) is determined by the computing device (140) of the machine (102). The computing device (140) of the machine uses an algorithm that receives the sensor data (124) and the thresholds (406, 408) as input to determine a risk score (210, 212, 214) associated with the wear risk at the link pin (116). as well as An alarm (600) is generated in response to the risk score exceeding the second threshold.

9. The non-transient computer-readable medium of claim 8, wherein the sensor data (124) includes temperature data, and wherein the threshold includes the steady-state operating temperature of the connecting pin (116) or the threshold rate of change of the temperature data.

10. The non-transitory computer-readable medium of claim 8, wherein the sensor data (124) comprises: Load data (202), which is associated with an actuation device connecting the first machine component (112) and the second machine component (114); and Motion data (204) is associated with the rate of rotation at the link pin (116), and wherein the threshold is based on historical load data, historical motion data and historical indications of wear at the link pin.

Images