Predictive fault detection system for a mobile work machine

Abstract

Device for online diagnosis of faults in drive train components of a mobile working machine, comprising: o an anomaly detection unit (101) designed to detect an atypical behavior of at least one operating parameter (OP) of the powertrain components; a first test unit (102) connected to the anomaly detection unit (101), which is configured to check whether the detected atypical behavior can be attributed to a known fault behavior; characterized by a second test unit (104) connected to the first test unit (102), which, if the fault behavior is classified as unknown, is configured to check in a fault database whether data on this unknown fault exists; and o a recording and storage unit (106) connected to the second test unit (104), which is designed to store the course of the anomalous operating size (BG) in the fault database if no data is yet available for the unknown fault.

Description

TECHNICAL AREA

[0001] The invention relates to the field of predictive maintenance for vehicles, in particular for electrically powered mobile machinery such as electric excavators or electric bulldozers. It relates to a device and a method for online diagnostics of faults in components of the electric powertrain. STATE OF THE ART

[0002] Electrically powered mobile machinery, often used in rough terrain, places high demands on the reliability of its drivetrain components, such as the high-voltage battery, power electronics, or electric motors. An unforeseen failure of any of these components not only leads to costly downtime but also to logistical challenges in recovering the machine.

[0003] Diagnostic systems in the current state of the art are often limited to detecting faults whose patterns and effects were already known during the machine's development and implemented in the control units. Therefore, a diagnosis is often only possible when a fault condition exceeds a predefined threshold. This frequently occurs only when the fault is already far advanced, making early, predictive assessment impossible.

[0004] In particular, the current state of the art has a diagnostic gap for errors that were unknown at the time of development (Start of Production, SOP) and only first appear during operation. No predefined models or thresholds exist for such unknown errors. Current systems are unable to interpret these new anomalies, assess their criticality, or make predictive statements. This can lead to sudden failures, especially with new machine generations or in the case of gradual, previously unknown degradation processes.

[0005] The object of the present invention is therefore to provide a device and a method that enable comprehensive online diagnostics capable of predictively detecting, evaluating, and managing not only known but also previously unknown faults. The aim is to create a self-learning system that continuously improves its diagnostic capabilities through the use of fleet data, thereby maximizing the reliability and service life of mobile machinery. SUMMARY

[0006] The problem is solved by a device having the features of claim 1. Advantageous embodiments are the subject of the dependent claims.

[0007] The device according to the invention provides a comprehensive solution that enables predictive diagnosis for both known and unknown faults. The core of the invention is a two-stage diagnostic pathway that is activated after the detection of an anomaly.

[0008] As a first step, the anomaly is compared with patterns of known errors. If it is a known error, a specific analysis module for this error type is activated, which makes a detailed prediction about the severity of the error and the remaining operating time of the component.

[0009] The invention's particular advantage unfolds, however, when a fault is identified as unknown. In this case, the device activates a second diagnostic pathway. First, it checks whether the unknown fault has already occurred in other vehicles within the fleet. If so, this historical data is used to make a statistically sound assessment of the fault's criticality. If the fault is entirely new, it is systematically recorded, and correlations with other operating parameters are sought to create a signature for this new fault. This signature forms the basis for classifying and analyzing the previously unknown fault as "known" in the future.

[0010] The decisive advantage of the invention lies in its self-learning nature. Every new fault that occurs in the fleet expands the system's knowledge base. Over time, the device becomes increasingly robust and can accurately predict an ever-wider range of faults. This enables early warning of the operator, even for new fault patterns, and facilitates well-informed maintenance planning, leading to a significant reduction in downtime and maximized machine availability. BRIEF DESCRIPTION OF THE DRAWING

[0011] The invention is explained in more detail below with reference to the accompanying drawings, which show an advantageous embodiment of the invention. Fig. Figure 1 shows a schematic block diagram illustrating the functional units of a device according to the invention and their logical interaction for online diagnosis of known and unknown faults. DETAILED DESCRIPTION

[0012] The present invention is described below with reference to certain embodiments. However, the present invention is not limited to the specific embodiments described in the following detailed description; rather, the described embodiments merely illustrate some aspects of the present invention, the scope of which is defined by the claims.

[0013] Further modifications and variations of the present invention are obvious to a person skilled in the art. The present description therefore encompasses all modifications and / or variations of the present invention whose scope of protection is defined by the claims.

[0014] Fig. Figure 1 shows a schematic block diagram illustrating the functional units of a device according to the invention and their logical interaction for the online prediction of unknown faults in drivetrain components of a mobile work machine. The functional units shown are preferably implemented as software modules that run on a central, cloud-based computing unit and communicate with one or more vehicle-side units in the work machine.

[0015] The in Fig.The diagnostic process shown in Figure 1 begins at a starting point 100, for example, through the continuous monitoring of the machine during operation. First, a vehicle-side anomaly detection unit 101 records one or more operating parameters BG, such as temperatures, pressures, currents, voltages, or vibrations, supplied by sensors on the powertrain components. The anomaly detection unit 101 is designed to recognize an atypical behavior, deviating from normal behavior, in at least one of these operating parameters and to signal this as a potential fault.

[0016] This error information is forwarded to a first inspection unit 102, which is part of the cloud-based unit. The first inspection unit 102 accesses an initial database containing patterns and signatures of previously analyzed errors. It checks whether the detected atypical behavior can be attributed to one of these known error patterns.

[0017] If this check is positive, a processing unit for known errors (103) is activated. This unit performs a predictive diagnostic procedure in which the detected error is assigned to a specific hybrid error model in the cloud. This model analyzes the transmitted operational parameters, determines the impact of the error on the component's health (e.g., the state of health, SoH), and calculates a specific error severity. By comparing the current, error-related state with a predicted aging model, an accurate remaining operating time (ROT) is calculated and provided to the operator before the process is terminated (113).

[0018] If, however, the fault behavior is classified as unknown by the first test unit 102 because no match was found in the database, the process is transferred to a second test unit 104. This second test unit 104 checks a second, separate fault database to see if data already exists for precisely this specific, but not yet definitively diagnosed, unknown fault. Such data could originate from other machines in the fleet where the same unknown fault occurred at an earlier time.

[0019] If data is already available, an analysis module 105 is activated. This analysis module 105 uses the collected historical fleet data to determine a statistically sound estimate of the current severity of the fault and the remaining service life (ROT) for the affected component, even without a complete understanding of the fault. This valuable information is transmitted to the machine to prevent an unforeseen failure, whereupon the process for this cycle ends (113).

[0020] However, if the second test unit 104 determines that no data exists for the unknown fault, it is a completely new fault, detected for the first time. In this case, a recording and storage unit 106 is activated, which records the progression of the anomalous operating parameter(s) and stores it in the fault database for unknown faults. This lays the foundation for a new, self-learning, and growing database from which all fleet vehicles can benefit in the future.

[0021] Downstream of the recording unit 106 is a correlation unit 107, which systematically checks whether other operating parameters of the machine also exhibit anomalous behavior, either simultaneously or at a defined time interval after the occurrence of the first anomaly. If further anomalous operating parameters are detected, a map generation unit 108 is activated. This unit 108 is configured to generate and store a multidimensional correlation map from the multivariate combination of the various anomalous operating parameters over time. This map represents a complex signature of the new fault and enables a significantly more precise diagnosis and characterization at a later time, for example, in the workshop.

[0022] If, however, after testing by the correlation unit 107, only a single anomalous operating parameter is found, the system enters a first, monitored operating state 109, in which the machine is allowed to continue operating under continuous observation of this single parameter until it possibly reaches a limit value.

[0023] The characteristic map generated in unit 108 and the state from block 109 are processed by a central limit value monitoring unit 110. This unit 110 continuously compares whether at least one of the operating parameters identified as anomalous touches, i.e., exceeds, a critical limit value (SG limit) stored in a vehicle control unit for the respective component. If such a limit value is touched, a stop unit 111 is activated. This unit immediately issues a signal to stop further operation, which typically results in an urgent warning message to the operator to prevent consequential damage or hazardous situations. The process then proceeds to end 113.

[0024] If, however, the limit value monitoring unit 110 does not detect any incursion of a critical limit value, a second operating state unit 112 is activated. This signals that continued operation is considered acceptable despite the detected, but not yet critical, unknown fault. The machine can thus continue its task while data continues to be collected in the background. This state also leads to the end of the process 113 for the current test cycle.

[0025] If, however, the limit value monitoring unit 110 does not detect any incursion of a critical limit value, a second operating state unit 112 becomes active. This signals that continued operation is considered acceptable despite the detected, but not yet critical, unknown fault. The machine can thus continue its task while data continues to be collected in the background. This state also leads to the end 113 of the process for the current test cycle. While the present invention has been described with reference to the embodiments described above, it is clear to those skilled in the art that it is possible to implement various modifications, variations, and improvements of the present invention in light of the teaching described above and within the scope of the appended claims without deviating from the scope of protection of the invention.

[0026] Furthermore, the areas in which experts are likely to be knowledgeable have not been described here in order to avoid unnecessarily obscuring the described invention.

[0027] Accordingly, the invention should not be limited by the specific illustrative embodiments, but only by the scope of protection of the attached claims.

Claims

[1] Device for online diagnosis of faults in drive train components of a mobile working machine, comprising: o an anomaly detection unit (101) designed to detect an atypical behavior of at least one operating parameter (OP) of the powertrain components; o a first test unit (102) connected to the anomaly detection unit (101), which is designed to check whether the detected atypical behavior can be attributed to a known fault behavior; characterized by through a second test unit (104) connected to the first test unit (102), which, if the fault behavior is classified as unknown, is configured to check in a fault database whether data on this unknown fault exists; and o a recording and storage unit (106) connected to the second test unit (104), which is designed to store the course of the anomalous operating size (BG) in the fault database if no data is yet available for the unknown fault. [2] Device according to claim 1, characterized by , that it further comprises a known fault processing unit (103) which is connected to the first inspection unit (102) and is designed to determine a fault severity and a remaining operating time (ROT) on the basis of a hybrid fault model in the case of a fault classified as known. [3] Device according to claim 1 or 2, characterized by, that it further comprises an analysis module (105) which is connected to the second test unit (104) and is designed to determine, in the presence of data on the unknown fault, a fault severity and / or a remaining operating time (ROT) from this data, in particular from fleet data. [4] Device according to any one of the preceding claims, characterized by , that it further exhibits: a correlation unit (107) downstream of the recording and storage unit (106), which is designed to check whether other operating parameters exhibit anomalous trends; and o a characteristic map formation unit (108) connected to the correlation unit (107), which is designed to form and store a correlation characteristic map linking several anomalous operating variables when several anomalous operating variables are present. [5] Device according to claim 4, characterized by , that it further exhibits: a limit value monitoring unit (110) designed to compare the anomalous operating parameters with stored limit values; and a stop unit (111) connected to the limit value monitoring unit (110), which is designed to output a signal to stop further work when a limit value is exceeded or fallen below. [6] Device according to claim 5, characterized by , that it further comprises a second operating state unit (112) connected to the limit value monitoring unit (110), which, if a limit value is not touched, is designed to send a signal that allows the working machine to continue working. [7] Device according to any one of the preceding claims, characterized by, that the units (102, 103, 104, 105, 106, 107, 108, 110, 111, 112) are implemented as software modules in a cloud-based unit which is in communication with a vehicle-side unit which includes the anomaly detection unit (101).

Images