An abnormal intelligent identification system for a crawler excavator

Abstract

The application discloses a kind of abnormal intelligent identification system of crawler excavator, it is related to excavator abnormal identification technical field, including following module: multi-source sensing acquisition module, including pressure sensing unit, vibration sensing unit, temperature sensing unit and controller area network signal acquisition unit;Signal synchronization pre-processing module is connected with multi-source sensing acquisition module, for each sensor signal is time synchronization, anti-aliasing filtering and analog-digital conversion;In the scheme, not only will fault diagnosis from traditional single parameter threshold alarm to the qualitative change breakthrough of multi-feature fusion accurate positioning, can distinguish different fault modes of plunger wear, flow distribution plate wear, bearing damage and hydraulic oil intake fault and provide quantitative severity assessment, also by building the diagnosis mode of data-driven and knowledge-driven fusion, both use high-precision sensor to extract state characteristics, also through rule base embedding field experience, make fault identification more accurate and intelligent.

Description

Technical Field

[0001] This invention relates to the field of excavator anomaly recognition technology, specifically to an intelligent anomaly recognition system for tracked excavators. Background Technology

[0002] Tracked excavators are a type of heavy-duty construction machinery. Their core features are that they are driven by a hydraulic system, travel on a tracked chassis, and perform diverse earthmoving operations such as digging, loading, and leveling through the boom, stick, and bucket at the front end. Their powerful power system, excellent ground pressure ratio, and complex and precise hydraulic transmission and control technology make them an indispensable key construction equipment in various complex and harsh working conditions such as mining, construction, transportation, and water conservancy.

[0003] The hydraulic main pump is a critical component of excavators. Its malfunctions, such as piston wear, distributor plate damage, bearing damage, and abnormal air intake, can lead to a decline in overall machine performance or even shutdown. Existing technologies mainly rely on alarms for single parameters such as pressure values ​​exceeding thresholds. Single pressure or temperature alarms cannot distinguish between malfunctions and normal operating conditions such as impact load fluctuations, resulting in a very high false alarm rate. At the same time, the alarm only indicates pump malfunctions and cannot guide maintenance personnel to deal with piston or bearing problems, leading to blind maintenance. Therefore, existing excavators lack intelligence and precision in identifying and judging abnormalities in the hydraulic main pump. To address this, we propose an intelligent identification system for abnormalities in tracked excavators. Summary of the Invention

[0004] (a) Technical problems to be solved

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent anomaly identification system for tracked excavators, thereby solving the aforementioned problems in the prior art.

[0006] (II) Technical Solution

[0007] To achieve the above objectives, the present invention provides the following technical solution: an intelligent anomaly identification system for tracked excavators, comprising the following modules:

[0008] The multi-source sensing acquisition module includes a pressure sensing unit, a vibration sensing unit, a temperature sensing unit, and a controller area network signal acquisition unit.

[0009] The signal synchronization preprocessing module is connected to the multi-source sensor acquisition module and is used to perform time synchronization, anti-aliasing filtering and analog-to-digital conversion on the signals from each sensor.

[0010] The feature parameter quantization module is connected to the signal synchronization preprocessing module. It is used to acquire multiple predefined feature parameters, including pressure stability parameter A, vibration characteristic frequency energy ratio parameter B, and temperature-pressure coupling deviation parameter C, in real time based on the preprocessed signal.

[0011] The rule-based diagnostic decision engine module, connected to the feature parameter quantization module, stores an updatable fault diagnosis knowledge base. This engine module is configured to execute a multi-level diagnostic decision process, including abnormal parameter identification, fault mode association mapping, numerical logic verification, and diagnostic result generation.

[0012] Preferably, in the multi-source sensing acquisition module, the pressure sensing unit includes a high dynamic pressure sensor installed at the main pump outlet, the vibration sensing unit includes a triaxial acceleration sensor installed on the main pump housing in the axial and radial directions, the temperature sensing unit includes a temperature sensor immersed in the hydraulic oil tank, and the controller local area network signal acquisition unit acquires the real-time engine speed through a protocol interface.

[0013] Preferably, the feature parameter quantization module includes a working condition identification submodule, a feature calculation submodule, and an anomaly marking submodule:

[0014] The working condition identification submodule divides the excavator's working status into three benchmark working conditions in real time: no-load idling state, constant load running state, and impact load state, based on the engine load rate, multi-channel hydraulic cylinder pressure signal, and pilot pressure signal.

[0015] The feature calculation submodule employs different feature extraction algorithms for signals under different benchmark conditions;

[0016] Under constant load operation, the standard deviation of the main pump outlet pressure signal within a 2-second time window is obtained as the pressure stability parameter A.

[0017] After the hydraulic oil temperature reaches a stable equilibrium, the percentage deviation between the current main pump outlet pressure value and the historical health baseline curve at the same temperature is obtained as the temperature-pressure coupling deviation parameter C. The historical health baseline curve is obtained by recording the pressure values ​​corresponding to different temperatures under the initial health state of the equipment and fitting them.

[0018] The anomaly marking submodule stores a dynamic threshold table for each feature parameter under different operating conditions. When the real-time feature parameter value exceeds the corresponding threshold range, a corresponding anomaly mark is generated.

[0019] Preferably, the specific method for the feature calculation submodule to obtain the vibration characteristic frequency energy ratio parameter B under constant load operation is as follows:

[0020] Step 1: Calculate the main pump frequency E based on the real-time engine speed D: E = D / 60;

[0021] Step 2: Calculate the piston passing frequency F based on the inherent structural parameter Z of the hydraulic main pump: F = E × Z;

[0022] Step 3: Perform a 1024-point Fast Fourier Transform on the vibration acceleration signal, apply the Hanning window function to reduce spectral leakage and obtain the power spectral density;

[0023] Step 4: Calculate the low-frequency band energy G: Perform numerical integration on the power spectral density in the frequency range [0.5E, 3E];

[0024] Step 5: Calculate the high-frequency band energy H: Perform numerical integration on the power spectral density in the frequency ranges [F-8Hz, F+8Hz] and [2F-8Hz, 2F+8Hz] respectively, and then sum the results.

[0025] Step 6: Calculate the vibration characteristic frequency energy ratio parameter B: B = H / G.

[0026] Preferably, the rule-based diagnostic decision engine module includes a fault association mapping submodule, a multi-level criterion verification submodule, and a diagnostic report generation submodule:

[0027] The fault association mapping submodule stores a fault parameter association matrix table, which defines the correspondence between each fault mode and its required set of abnormal parameters. The fault association mapping submodule receives an anomaly tag set from the feature parameter quantization module and filters out all candidate fault modes by querying the fault parameter association matrix table. The fault parameter association matrix table includes:

[0028] The necessary abnormal parameter set associated with the plunger wear failure mode is {C,B}; the necessary abnormal parameter set associated with the distributor plate wear failure mode is {C,A}; the core abnormal parameter associated with the bearing damage failure mode is {B}; and the necessary abnormal parameter set associated with the hydraulic oil intake failure mode is {A,B}.

[0029] Preferably, the multi-level criterion verification submodule applies its corresponding multi-level numerical logic criteria to accurately verify each candidate fault mode. The multi-level criterion verification submodule only determines a candidate fault mode as a valid diagnostic result when the candidate fault mode satisfies all of its first-level necessity criteria.

[0030] The verification criteria for plunger wear failure include: Level 1 necessity criterion: C < -12% and B > 3.5; Level 2 severity criterion: if A > twice the benchmark value is met at the same time, then the diagnosis result should indicate severe wear accompanied by severe pressure pulsation.

[0031] The verification criteria for bearing damage include: core feature criterion: B<0.6; auxiliary exclusion criterion: C>-8%, used to exclude pressure drop caused by severe wear.

[0032] Preferably, the multi-level criterion verification submodule also includes a diagnostic conflict arbitration unit. When multiple candidate failure modes simultaneously meet their respective first-level necessity criteria, the arbitration unit makes a decision in the following priority order:

[0033] Step 1: Prioritize identifying fault modes with larger deviations from the normal baseline for abnormal parameters, and compare them by obtaining the sum of the standardized offsets of each abnormal parameter;

[0034] Step 2: When the sum of the standardized offsets is the same, prioritize identifying the fault modes that have a higher degree of matching between the associated set of necessary abnormal parameters and the current set of actual abnormal parameters.

[0035] Step 3: If a decision still cannot be made, mark it as requiring confirmation through manual disassembly and inspection.

[0036] Preferably, the rule-based diagnostic decision engine module is followed by a diagnostic knowledge optimization module, which specifically includes:

[0037] Step 1: Establish a diagnostic feedback loop by receiving repair work order information uploaded from the repair terminal. The repair work order information includes the actual faulty component, the type of damage, and the severity.

[0038] Step 2: Link the repair work order information with historical data from the diagnostic process to create a tagged feedback sample library;

[0039] Step 3: Use statistical process control methods to perform distribution analysis on the feature parameter values ​​in the feedback sample library and dynamically adjust the anomaly judgment threshold in the feature parameter quantification module;

[0040] Step 4: When the accumulated feedback samples of the same type of fault exceed the preset number, the automatic optimization process of the fault diagnosis knowledge base in the rule-based diagnostic decision engine module is triggered to adjust the logical judgment threshold of the relevant fault modes.

[0041] Preferably, after the diagnostic knowledge optimization module, a multi-level early warning output module is also included. Specifically, the multi-level early warning output module generates early warning signals of different levels based on the severity of the diagnostic results output by the rule-based diagnostic decision engine module.

[0042] Primary warning: When a single characteristic parameter is abnormal but not associated with a specific fault mode, a suggested inspection prompt will be displayed on the excavator's on-board display unit;

[0043] Intermediate warning: When a specific fault mode is confirmed but its severity is low, an audible and visual alarm is generated and a warning message is sent to the remote monitoring center;

[0044] Advanced warning: When it is confirmed that the fault is serious and may cause secondary damage, in addition to audible and visual alarms, a command is sent to the hydraulic control system to limit the maximum output power of the main pump, and an emergency repair work order is pushed to the management platform.

[0045] Preferably, the system adopts a distributed computing architecture: the multi-source sensor acquisition module, the signal synchronization preprocessing module, and the feature parameter quantization module are all deployed on the excavator's onboard edge computing unit; the rule-based diagnostic decision engine module and the diagnostic knowledge optimization module are both deployed on the cloud server; the onboard edge computing unit and the cloud server are connected via a wireless communication network, the onboard edge computing unit is responsible for real-time feature extraction and primary anomaly detection, and only uploads the feature parameter set and operating condition information under abnormal conditions to the cloud server; the cloud server is responsible for running complex rule reasoning, long-term trend analysis, and the maintenance and updating of the diagnostic knowledge base, and periodically distributes the optimized diagnostic rules to the onboard edge computing unit.

[0046] (III) Beneficial Effects

[0047] This invention provides an intelligent anomaly identification system for tracked excavators, which has the following beneficial effects:

[0048] This solution not only represents a qualitative breakthrough in fault diagnosis, moving from traditional single-parameter threshold alarms to precise localization through multi-feature fusion, enabling clear differentiation of different fault modes such as plunger wear, distributor plate wear, bearing damage, and hydraulic oil intake failure, and providing quantitative severity assessments, but also constructs a diagnostic model that integrates data-driven and knowledge-driven approaches. This model utilizes high-precision sensors to extract state features and embeds domain experience into a rule base, making fault identification more accurate and intelligent. Attached Figure Description

[0049] Figure 1 This is a flowchart of an intelligent anomaly identification system for a tracked excavator according to the present invention. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] Please see Figure 1 This invention provides an intelligent anomaly identification system for tracked excavators, comprising the following modules:

[0052] The multi-source sensing acquisition module includes a pressure sensing unit, a vibration sensing unit, a temperature sensing unit, and a controller area network signal acquisition unit.

[0053] The signal synchronization preprocessing module is connected to the multi-source sensor acquisition module and is used to perform time synchronization, anti-aliasing filtering and analog-to-digital conversion on the signals from each sensor.

[0054] The feature parameter quantization module is connected to the signal synchronization preprocessing module. It is used to acquire multiple predefined feature parameters, including pressure stability parameter A, vibration characteristic frequency energy ratio parameter B, and temperature-pressure coupling deviation parameter C, in real time based on the preprocessed signal.

[0055] The rule-based diagnostic decision engine module, connected to the feature parameter quantization module, stores an updatable fault diagnosis knowledge base. This engine module is configured to execute a multi-level diagnostic decision process, including abnormal parameter identification, fault mode association mapping, numerical logic verification, and diagnostic result generation.

[0056] In this embodiment, the solution uses a multi-source sensor acquisition module as the sensing front end of the system. It consists of a standardized sensor array composed of a high dynamic pressure sensor, a triaxial vibration acceleration sensor, an oil temperature sensor, and a bus interface. The module synchronously acquires raw state signals from four dimensions: hydraulic pipeline, pump body structure, oil thermodynamics, and vehicle control network, ensuring the comprehensiveness and accuracy of the data source.

[0057] In this scheme, the signal synchronization preprocessing module serves as the data channel. Through the coordinated operation of the hardware synchronization sampling circuit, the anti-aliasing filtering and conditioning circuit, and the high-precision analog-to-digital converter, it transforms various analog signals into time-aligned and denoised standardized digital signal sequences, providing a precise timing input basis for subsequent analysis.

[0058] In this solution, the feature parameter quantization module functions as a signal analyzer, with a built-in working condition identification algorithm and feature calculation engine. It can automatically determine whether the equipment is under different working conditions such as no-load, constant load, or impact based on engine load and hydraulic action, and specifically calculate pressure stability parameter A, vibration characteristic frequency energy ratio parameter B, and temperature-pressure coupling deviation parameter C, which are quantitative indicators with clear physical meaning, realizing the intelligent conversion from the original waveform to healthy feature values.

[0059] In this solution, the rule-based diagnostic decision engine module serves as the intelligent hub of the system. It has built a structured fault diagnosis knowledge base and simulates human diagnostic thinking through a three-step reasoning process of abnormal parameter identification, fault mode association mapping, and multi-layer numerical logic verification. The final output includes specific fault types, confidence levels, and judgment criteria.

[0060] This solution achieves full automation from multi-source heterogeneous data acquisition to transparent intelligent diagnosis through the precise coordination of four major modules. The beneficial effects are mainly reflected in two aspects: First, it realizes a qualitative breakthrough in fault diagnosis from traditional single-parameter threshold alarms to precise positioning through multi-feature fusion, which can clearly distinguish different fault modes such as plunger wear, distributor plate wear, bearing damage, and hydraulic oil intake failure, and provide quantitative severity assessment. Second, it constructs a diagnostic model that integrates data-driven and knowledge-driven approaches. It utilizes high-precision sensors to extract state features and embeds domain experience through rule bases, making fault identification more accurate and intelligent.

[0061] Specifically, the multi-source sensing acquisition module includes: a pressure sensing unit comprising a high dynamic pressure sensor installed at the main pump outlet; a vibration sensing unit comprising a triaxial acceleration sensor installed on the main pump housing in both axial and radial directions; a temperature sensing unit comprising a temperature sensor immersed in the hydraulic oil tank; and a controller area network signal acquisition unit acquiring the real-time engine speed via a protocol interface.

[0062] In this embodiment, the pressure sensing unit uses a high-frequency response, high-precision piezoelectric sensor, which is directly and rigidly installed on the high-pressure pipeline at the main pump outlet. Its core function is to capture the most direct pressure pulsation characteristics and steady-state pressure value of the hydraulic system in real time. The precise selection of the installation position ensures that the pressure signal can truly reflect the output characteristics of the main pump rather than the signal after pipeline attenuation.

[0063] In this scheme, the vibration sensing unit uses an industrial-grade triaxial IEPE accelerometer, which is installed at the axial and radial measurement points of the main pump housing. The axial mounting point is sensitive to the axial movement of the bearing and the swashplate swing, while the radial mounting point is sensitive to the lateral impact of the plunger and the vibration of the distribution plate. This multi-directional arrangement can collect the vibration generated by the mechanical interaction inside the pump body from all directions, providing complete spatial vibration information for subsequent spectrum analysis.

[0064] In this scheme, the temperature sensing unit uses a PT100 platinum resistance sensor immersed in the oil circulation area of ​​the hydraulic oil tank. Its measurement point avoids local heat sources and dead zones, ensuring that the average oil temperature that represents the overall thermal balance of the system is obtained. This parameter is not only a key indicator for evaluating the thermal load of the hydraulic system, but also a basic variable for conducting temperature-pressure coupling analysis to eliminate the influence of temperature drift.

[0065] In this solution, the controller area network signal acquisition unit is a communication interface that combines hardware and software. Through the physical layer and data link layer design that conforms to the J1939 protocol standard, it reads the information of the engine control unit from the excavator's CAN bus network in real time in a non-intrusive manner, and obtains accurate speed information directly related to the main pump drive shaft without the need to install an additional speed sensor.

[0066] The feature parameter quantization module includes a working condition identification submodule, a feature calculation submodule, and an anomaly marking submodule:

[0067] The working condition identification submodule divides the excavator's working status into three benchmark working conditions in real time: no-load idling state, constant load running state, and impact load state, based on the engine load rate, multi-channel hydraulic cylinder pressure signal, and pilot pressure signal.

[0068] The feature calculation submodule employs different feature extraction algorithms for signals under different benchmark conditions;

[0069] Under constant load operation, the standard deviation of the main pump outlet pressure signal within a 2-second time window is obtained as the pressure stability parameter A.

[0070] After the hydraulic oil temperature reaches a stable equilibrium, the percentage deviation between the current main pump outlet pressure value and the historical health baseline curve at the same temperature is obtained as the temperature-pressure coupling deviation parameter C. The historical health baseline curve is obtained by recording the pressure values ​​corresponding to different temperatures under the initial health state of the equipment and fitting them.

[0071] The anomaly marking submodule stores a dynamic threshold table for each feature parameter under different operating conditions. When the real-time feature parameter value exceeds the corresponding threshold range, a corresponding anomaly mark is generated.

[0072] In this embodiment, the original data is transformed into a decisionable health status value through the progressive collaboration of three functional sub-modules. The working condition identification sub-module is used to analyze the combined characteristics of load rate signals from the engine, multi-channel hydraulic cylinder pressure signals, and pilot pressure signals in real time, and intelligently identify the complex excavator operation process into three working conditions: no-load idling, constant load operation, and impact load. This identification logic not only considers a single parameter, but also focuses on the coordinated changes of multiple parameters in the time series. For example, the constant load working condition is confirmed by monitoring the coordinated stability of the boom and stick pressures over a specific time period.

[0073] In this solution, the feature calculation submodule dynamically switches the corresponding feature extraction method for different operating conditions to eliminate operating condition interference. Within a 2-second time window identified as constant load operation, this submodule calculates the standard deviation of the main pump outlet pressure signal as the pressure stability parameter A. In the steady-state stage where the hydraulic system oil temperature reaches thermal equilibrium, this submodule initiates temperature-pressure coupling analysis. By querying the pre-stored pressure-temperature reference curve, it calculates the percentage deviation between the current measured pressure and the reference value at the same temperature as the temperature-pressure coupling deviation parameter B. The pressure-temperature reference curve is obtained by calibration tests conducted on the equipment under initial health conditions within different ambient and operating temperature ranges. The anomaly marking submodule maintains a multi-dimensional dynamic threshold table. This threshold table sets different boundaries for each feature parameter under different operating conditions. When the real-time calculated feature value exceeds the threshold boundary of its corresponding operating condition, this submodule generates an anomaly mark.

[0074] The specific method used by the feature calculation submodule to obtain the vibration characteristic frequency energy ratio parameter B under constant load operation is as follows:

[0075] Step 1: Calculate the main pump frequency E based on the real-time engine speed D: E = D / 60;

[0076] Step 2: Calculate the piston passing frequency F based on the inherent structural parameter Z of the hydraulic main pump: F = E × Z;

[0077] Step 3: Perform a 1024-point Fast Fourier Transform on the vibration acceleration signal, apply the Hanning window function to reduce spectral leakage and obtain the power spectral density;

[0078] Step 4: Calculate the low-frequency band energy G: Perform numerical integration on the power spectral density in the frequency range [0.5E, 3E];

[0079] Step 5: Calculate the high-frequency band energy H: Perform numerical integration on the power spectral density in the frequency ranges [F-8Hz, F+8Hz] and [2F-8Hz, 2F+8Hz] respectively, and then sum the results.

[0080] Step 6: Calculate the vibration characteristic frequency energy ratio parameter B: B = H / G.

[0081] In this embodiment, the main pump frequency E is obtained by dividing the real-time engine speed D by 60, and the plunger passing frequency F is obtained by the number of plungers Z of the hydraulic main pump. This frequency corresponds to the impact frequency generated by the periodic oil discharge of the plunger and is a characteristic frequency representing the working state of the plunger. A 1024-point fast Fourier transform is performed on the vibration acceleration signal, and a Hanning window function is applied to suppress the spectral leakage effect to obtain the power spectral density. By defining the low-frequency band energy integration interval [0.5E, 3E], the low-frequency vibration energy caused by bearing failure and rotor imbalance is captured. The low-frequency band energy G is obtained by numerically integrating the power spectral density within the band; similarly, the high-frequency band energy H is calculated; finally, the vibration characteristic frequency energy ratio parameter B is obtained by dividing the high-frequency band energy H by the low-frequency band energy G. By normalizing and comparing the high-frequency impact energy and the low-frequency vibration energy, a dimensionless sensitive index is formed. In this scheme, characteristic frequency bands directly related to specific fault mechanisms are defined, with the low-frequency band corresponding to bearing faults and the high-frequency band corresponding to plunger faults. This enables the accurate extraction of fault characteristic signals from wide-spectrum noise, which is beneficial to improving the accuracy of fault identification.

[0082] The rule-based diagnostic decision engine module includes a fault association mapping submodule, a multi-level criterion verification submodule, and a diagnostic report generation submodule.

[0083] The fault association mapping submodule stores a fault parameter association matrix table, which defines the correspondence between each fault mode and its required set of abnormal parameters. The fault association mapping submodule receives an anomaly tag set from the feature parameter quantization module and filters out all candidate fault modes by querying the fault parameter association matrix table. The fault parameter association matrix table includes:

[0084] The necessary abnormal parameter set associated with the plunger wear failure mode is {C,B}; the necessary abnormal parameter set associated with the distributor plate wear failure mode is {C,A}; the core abnormal parameter associated with the bearing damage failure mode is {B}; and the necessary abnormal parameter set associated with the hydraulic oil intake failure mode is {A,B}.

[0085] In this embodiment, the diagnostic experience is transformed into diagnostic rules through a predefined fault parameter correlation matrix table. Precise correspondences are established for the four most common typical fault modes of the hydraulic main pump in tracked excavators: For a composite fault of piston wear reflecting decreased volumetric efficiency and enhanced periodic impact, the necessary abnormal parameter set is defined as {temperature-pressure coupling deviation parameter C, vibration characteristic frequency energy ratio parameter B}. This combination requires evidence of both pressure capacity degradation and increased high-frequency impact, consistent with the physical mechanism of the fault. For a fault of distributor plate wear mainly manifesting as abnormal pressure pulsation and decreased volumetric efficiency, the necessary abnormal parameter set is defined as {temperature-pressure coupling deviation parameter C, pressure stability parameter A}. The core characteristic of this fault is deteriorated pressure stability. For purely mechanical faults such as bearing damage that mainly induce low-frequency vibration, the core abnormal parameter is defined as {vibration characteristic frequency energy ratio parameter B}. For hydrodynamic faults such as hydraulic oil air ingress that cause wideband random vibration and severe pressure fluctuations, the necessary abnormal parameter set is defined as {pressure stability parameter A, vibration characteristic frequency energy ratio parameter B}, simultaneously capturing pressure instability and vibration anomalies.

[0086] The multi-level criterion verification submodule applies its corresponding multi-level numerical logic criteria to accurately verify each candidate fault mode. The multi-level criterion verification submodule only determines a candidate fault mode as a valid diagnostic result when it meets all of its first-level necessity criteria.

[0087] The verification criteria for plunger wear failure include: Level 1 necessity criterion: C < -12% and B > 3.5; Level 2 severity criterion: if A > twice the benchmark value is met at the same time, then the diagnosis result should indicate severe wear accompanied by severe pressure pulsation.

[0088] The verification criteria for bearing damage include: core feature criterion: B<0.6; auxiliary exclusion criterion: C>-8%, used to exclude pressure drop caused by severe wear.

[0089] In this embodiment, for progressive complex faults such as plunger wear, the judgment criteria are set in two layers: the first layer of necessity criterion is set as C < -12% and B > 3.5, and these two thresholds can be selected according to the specific application; the second layer of severity criterion is defined as if the pressure stability parameter A > twice the benchmark value is met simultaneously. This additional condition is not used to determine whether the fault exists, but to assess the development stage of the fault. When the pressure stability parameter A also deteriorates significantly, it indicates that the wear has developed to the point of causing severe pressure fluctuations. At this time, the system will automatically increase the fault confidence in the diagnostic results, providing a quantitative basis for maintenance priority decision-making; for bearing damage, etc. For mechanical faults characterized primarily by vibration, the verification criteria combine feature criteria and exclusion criteria. The core feature criterion is set as the vibration characteristic frequency energy ratio parameter B < 0.6. This setting, which is much lower than the plunger wear threshold, accurately captures the reversal of the spectrum characteristics caused by bearing failure, where low-frequency vibration energy dominates. The auxiliary exclusion criterion is defined as the temperature-pressure coupling deviation parameter C > -8%. Pure bearing damage does not directly affect the volumetric efficiency of the hydraulic pump in the early stages. Therefore, the pressure coupling deviation parameter must be kept within a relatively normal range to effectively eliminate the confusion caused by severe plunger wear, which may also reduce the vibration characteristic frequency energy ratio parameter B, ensuring the accuracy of the diagnosis.

[0090] The multi-level criterion verification submodule also includes a diagnostic conflict arbitration unit. When multiple candidate failure modes simultaneously meet their respective first-level necessity criteria, the arbitration unit makes a decision in the following priority order:

[0091] Step 1: Prioritize identifying fault modes with larger deviations from the normal baseline for abnormal parameters, and compare them by obtaining the sum of the standardized offsets of each abnormal parameter;

[0092] Step 2: When the sum of the standardized offsets is the same, prioritize identifying the fault modes that have a higher degree of matching between the associated set of necessary abnormal parameters and the current set of actual abnormal parameters.

[0093] Step 3: If a decision still cannot be made, mark it as requiring confirmation through manual disassembly and inspection.

[0094] In this embodiment, when multiple candidate faults meet the conditions at the same time, the system automatically makes a decision using three levels of rules: First, the sum of the standardized offsets of the associated parameters of each fault is compared, and the pattern with the largest deviation is confirmed first; if they are the same, the matching degree between the fault parameter set and the actual anomaly set is calculated, and the pattern with the higher matching degree is confirmed first; if a decision still cannot be made, manual confirmation is prompted.

[0095] Following the rule-based diagnostic decision engine module is a diagnostic knowledge optimization module, which specifically includes:

[0096] Step 1: Establish a diagnostic feedback loop by receiving repair work order information uploaded from the repair terminal. The repair work order information includes the actual faulty component, the type of damage, and the severity.

[0097] Step 2: Link the repair work order information with historical data from the diagnostic process to create a tagged feedback sample library;

[0098] Step 3: Use statistical process control methods to perform distribution analysis on the feature parameter values ​​in the feedback sample library and dynamically adjust the anomaly judgment threshold in the feature parameter quantification module;

[0099] Step 4: When the accumulated feedback samples of the same type of fault exceed the preset number, the automatic optimization process of the fault diagnosis knowledge base in the rule-based diagnostic decision engine module is triggered to adjust the logical judgment threshold of the relevant fault modes.

[0100] In this embodiment, by receiving work order information containing actual faulty components and types uploaded by the maintenance terminal, this information is precisely associated and bound with historical feature parameters and intermediate judgment records in the system diagnosis process to form a feedback sample library with truth value labels. Subsequently, statistical process control methods are used to analyze the sample distribution and dynamically adjust the anomaly judgment threshold in the feature parameter quantification module. When the accumulation of similar fault samples reaches a preset scale, the optimization process of the fault diagnosis knowledge base in the rule-based diagnosis decision engine is automatically triggered to calibrate the logical judgment threshold.

[0101] Following the diagnostic knowledge optimization module is a multi-level early warning output module, which specifically generates early warning signals of different levels based on the severity of the diagnostic results output by the rule-based diagnostic decision engine module.

[0102] Primary warning: When a single characteristic parameter is abnormal but not associated with a specific fault mode, a suggested inspection prompt will be displayed on the excavator's on-board display unit;

[0103] Intermediate warning: When a specific fault mode is confirmed but its severity is low, an audible and visual alarm is generated and a warning message is sent to the remote monitoring center;

[0104] Advanced warning: When it is confirmed that the fault is serious and may cause secondary damage, in addition to audible and visual alarms, a command is sent to the hydraulic control system to limit the maximum output power of the main pump, and an emergency repair work order is pushed to the management platform.

[0105] In this embodiment, differentiated responses are generated based on the severity of the diagnostic results: a primary warning suggests checking when a single parameter is abnormal but not associated with a specific fault; a secondary warning triggers an audible and visual alarm and sends information to the remote monitoring center when a specific fault is confirmed and its severity is low; and a high-level warning, when it is determined that the fault is serious and may cause secondary damage, sends a direct instruction to the hydraulic control system to limit the maximum output power of the main pump, in addition to audible and visual alarms and pushing emergency work orders.

[0106] The system adopts a distributed computing architecture: the multi-source sensor acquisition module, signal synchronization preprocessing module, and feature parameter quantization module are all deployed on the excavator's onboard edge computing unit; the rule-based diagnostic decision engine module and diagnostic knowledge optimization module are both deployed on the cloud server; the onboard edge computing unit and the cloud server are connected via a wireless communication network. The onboard edge computing unit is responsible for real-time feature extraction and primary anomaly detection, and only uploads the feature parameter set and operating condition information under abnormal conditions to the cloud server; the cloud server is responsible for running complex rule reasoning, long-term trend analysis, and the maintenance and updating of the diagnostic knowledge base, and periodically distributes the optimized diagnostic rules to the onboard edge computing unit.

[0107] In this embodiment, the multi-source sensor acquisition, signal synchronous preprocessing, and feature parameter quantization modules are deployed on the vehicle edge computing unit to ensure data real-time performance and reduce the bandwidth requirements of raw data; the rule-based diagnostic decision engine and diagnostic knowledge optimization module are deployed on the cloud server, thereby significantly saving communication costs while ensuring real-time diagnostic performance.

[0108] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0109] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0110] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. An intelligent anomaly identification system for a tracked excavator, characterized in that, Includes the following modules: The multi-source sensing acquisition module includes a pressure sensing unit, a vibration sensing unit, a temperature sensing unit, and a controller area network signal acquisition unit; The signal synchronization preprocessing module is connected to the multi-source sensor acquisition module and is used to perform time synchronization, anti-aliasing filtering and analog-to-digital conversion on the signals from each sensor. The feature parameter quantization module is connected to the signal synchronization preprocessing module. It is used to acquire multiple predefined feature parameters, including pressure stability parameter A, vibration characteristic frequency energy ratio parameter B, and temperature-pressure coupling deviation parameter C, in real time based on the preprocessed signal. The rule-based diagnostic decision engine module, connected to the feature parameter quantization module, stores an updatable fault diagnosis knowledge base. This engine module is configured to execute a multi-level diagnostic decision process, including abnormal parameter identification, fault mode association mapping, numerical logic verification, and diagnostic result generation.

2. The intelligent anomaly identification system for a tracked excavator according to claim 1, characterized in that: Specifically, the multi-source sensing acquisition module includes: a pressure sensing unit comprising a high dynamic pressure sensor installed at the main pump outlet; a vibration sensing unit comprising a triaxial acceleration sensor installed on the main pump housing in both axial and radial directions; a temperature sensing unit comprising a temperature sensor immersed in the hydraulic oil tank; and a controller area network signal acquisition unit acquiring the real-time engine speed via a protocol interface.

3. The intelligent anomaly identification system for a tracked excavator according to claim 1, characterized in that: The feature parameter quantization module includes a working condition identification submodule, a feature calculation submodule, and an anomaly marking submodule: The working condition identification submodule divides the excavator's working status into three benchmark working conditions in real time: no-load idling state, constant load running state, and impact load state, based on the engine load rate, multi-channel hydraulic cylinder pressure signal, and pilot pressure signal. The feature calculation submodule employs different feature extraction algorithms for signals under different benchmark conditions; Under constant load operation, the standard deviation of the main pump outlet pressure signal within a 2-second time window is obtained as the pressure stability parameter A. After the hydraulic oil temperature reaches a stable equilibrium, the percentage deviation between the current main pump outlet pressure value and the historical health baseline curve at the same temperature is obtained as the temperature-pressure coupling deviation parameter C. The historical health baseline curve is obtained by recording the pressure values ​​corresponding to different temperatures under the initial health state of the equipment and fitting them. The anomaly marking submodule stores a dynamic threshold table for each feature parameter under different operating conditions. When the real-time feature parameter value exceeds the corresponding threshold range, a corresponding anomaly mark is generated.

4. The intelligent anomaly identification system for a tracked excavator according to claim 3, characterized in that: The specific method used by the feature calculation submodule to obtain the vibration characteristic frequency energy ratio parameter B under constant load operation is as follows: Step 1: Calculate the main pump frequency E based on the real-time engine speed D: E = D / 60; Step 2: Calculate the piston passing frequency F based on the inherent structural parameter Z of the hydraulic main pump: F = E × Z; Step 3: Perform a 1024-point Fast Fourier Transform on the vibration acceleration signal, apply the Hanning window function to reduce spectral leakage and obtain the power spectral density; Step 4: Calculate the low-frequency band energy G: Perform numerical integration on the power spectral density in the frequency range [0.5E, 3E]; Step 5: Calculate the high-frequency band energy H: Perform numerical integration on the power spectral density in the frequency ranges [F-8Hz, F+8Hz] and [2F-8Hz, 2F+8Hz] respectively, and then sum the results. Step 6: Calculate the vibration characteristic frequency energy ratio parameter B: B = H / G.

5. The intelligent anomaly identification system for a tracked excavator according to claim 1, characterized in that: The rule-based diagnostic decision engine module includes a fault association mapping submodule, a multi-level criterion verification submodule, and a diagnostic report generation submodule. The fault association mapping submodule stores a fault parameter association matrix table, which defines the correspondence between each fault mode and its required set of abnormal parameters. The fault association mapping submodule receives an anomaly tag set from the feature parameter quantization module and filters out all candidate fault modes by querying the fault parameter association matrix table. The fault parameter association matrix table includes: The necessary abnormal parameter set associated with the plunger wear failure mode is {C,B}; the necessary abnormal parameter set associated with the distributor plate wear failure mode is {C,A}; the core abnormal parameter associated with the bearing damage failure mode is {B}; and the necessary abnormal parameter set associated with the hydraulic oil intake failure mode is {A,B}.

6. The intelligent anomaly identification system for a tracked excavator according to claim 5, characterized in that: The multi-level criterion verification submodule applies its corresponding multi-level numerical logic criteria to accurately verify each candidate fault mode. The multi-level criterion verification submodule only determines a candidate fault mode as a valid diagnostic result when it meets all of its first-level necessity criteria. The verification criteria for plunger wear failure include: Level 1 necessity criterion: C < -12% and B > 3.5; Level 2 severity criterion: if A > twice the benchmark value is met at the same time, then the diagnosis result should indicate severe wear accompanied by severe pressure pulsation. The verification criteria for bearing damage include: core feature criterion: B<0.6; auxiliary exclusion criterion: C>-8%, used to exclude pressure drop caused by severe wear.

7. The intelligent anomaly identification system for a tracked excavator according to claim 6, characterized in that: The multi-level criterion verification submodule also includes a diagnostic conflict arbitration unit. When multiple candidate failure modes simultaneously meet their respective first-level necessity criteria, the arbitration unit makes a decision in the following priority order: Step 1: Prioritize identifying fault modes with larger deviations from the normal baseline for abnormal parameters, and compare them by obtaining the sum of the standardized offsets of each abnormal parameter; Step 2: When the sum of the standardized offsets is the same, prioritize identifying the fault modes that have a higher degree of matching between the associated set of necessary abnormal parameters and the current set of actual abnormal parameters. Step 3: If a decision still cannot be made, mark it as requiring confirmation through manual disassembly and inspection.

8. The intelligent anomaly identification system for a tracked excavator according to claim 1, characterized in that: Following the rule-based diagnostic decision engine module is a diagnostic knowledge optimization module, which specifically includes: Step 1: Establish a diagnostic feedback loop by receiving repair work order information uploaded from the repair terminal. The repair work order information includes the actual faulty component, the type of damage, and the severity. Step 2: Link the repair work order information with historical data from the diagnostic process to create a tagged feedback sample library; Step 3: Use statistical process control methods to perform distribution analysis on the feature parameter values ​​in the feedback sample library and dynamically adjust the anomaly judgment threshold in the feature parameter quantification module; Step 4: When the accumulated feedback samples of the same type of fault exceed the preset number, the automatic optimization process of the fault diagnosis knowledge base in the rule-based diagnostic decision engine module is triggered to adjust the logical judgment threshold of the relevant fault modes.

9. An intelligent anomaly identification system for a tracked excavator according to claim 8, characterized in that: Following the diagnostic knowledge optimization module is a multi-level early warning output module, which specifically generates early warning signals of different levels based on the severity of the diagnostic results output by the rule-based diagnostic decision engine module. Primary warning: When a single characteristic parameter is abnormal but not associated with a specific fault mode, a suggested inspection prompt will be displayed on the excavator's on-board display unit; Intermediate warning: When a specific fault mode is confirmed but its severity is low, an audible and visual alarm is generated and a warning message is sent to the remote monitoring center; Advanced warning: When it is confirmed that the fault is serious and may cause secondary damage, in addition to audible and visual alarms, a command is sent to the hydraulic control system to limit the maximum output power of the main pump, and an emergency repair work order is pushed to the management platform.

10. An intelligent anomaly identification system for a tracked excavator according to claim 1, characterized in that: The system adopts a distributed computing architecture: the multi-source sensor acquisition module, the signal synchronization preprocessing module, and the feature parameter quantization module are all deployed on the excavator's onboard edge computing unit; the rule-based diagnostic decision engine module and the diagnostic knowledge optimization module are all deployed on the cloud server. The vehicle-mounted edge computing unit is connected to the cloud server via a wireless communication network. The vehicle-mounted edge computing unit is responsible for real-time feature extraction and primary anomaly detection, and only uploads the feature parameter set and operating condition information under abnormal conditions to the cloud server. The cloud server is responsible for running complex rule reasoning, long-term trend analysis, and maintaining and updating the diagnostic knowledge base, and regularly distributes the optimized diagnostic rules to the vehicle edge computing unit.

Images