A method and system for diagnosing safety faults in the front and rear axle steering systems of forklifts
By constructing a twin mirror model, the physical structure and linkage of the forklift steering system are decomposed, and forward calculation and reverse identification are performed. This solves the problem of difficulty in identifying deep faults in the fault diagnosis of forklift steering systems, and achieves accurate fault location and reduces misjudgment.
Patent Information
- Application Number
- CN202610270054.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-03-06
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2046-03-06
AI Technical Summary
Existing fault diagnosis methods for forklift steering systems are insufficient for achieving system-level linkage analysis and accurate identification of deep faults, resulting in inadequate fault location accuracy and a tendency to miss or misdiagnose faults.
The physical structure and linkage relationship of the front and rear axle steering systems of forklifts are decomposed hierarchically, and a twin mirror model is constructed. High-confidence monitoring data is obtained through forward calculation and reverse identification, dynamic mapping relationship is established, calculation and decoupling identification are performed along the hierarchy, deviation information is identified and quantified, fault type is predicted and control decomposition feedback is performed.
It achieves system-level linkage analysis, accurately identifies deep-seated faults, improves fault location accuracy, and reduces missed and false diagnoses.
Smart Images

Figure CN122176820B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of fault diagnosis, and in particular to a method and system for diagnosing safety faults in the front and rear axle steering systems of forklifts. Background Technology
[0002] The operational reliability of the forklift's front and rear axle steering system directly determines the safety and efficiency of forklift operations, making it a core performance indicator that must be prioritized in the industrial vehicle sector. Current fault diagnosis for steering systems primarily employs threshold-based single-point monitoring methods and experience-based troubleshooting methods. The former collects key parameters such as steering pressure and angle and compares them with preset thresholds to identify anomalies, while the latter relies on the experience of maintenance personnel combined with partial disassembly and inspection to locate the fault point. Existing methods largely rely on single parameter threshold judgments, lacking a holistic model of the interconnected relationships among multiple system components. This results in insufficient fault location accuracy and makes it difficult to identify deep-seated faults caused by multi-physics coupling and progressive parameter degradation, easily leading to missed or false fault diagnoses.
[0003] Currently, in related technologies, fault diagnosis of forklift steering systems faces technical challenges in achieving system-level linkage analysis and accurate identification of deep-seated faults. Summary of the Invention
[0004] This application provides a method and system for diagnosing safety faults in the front and rear axle steering systems of forklifts. It employs a hierarchical decomposition of the physical structure and linkage relationships of the forklift steering system based on identification complexity, constructing a twin mirror model that includes forward calculation and reverse identification functions. The method aligns the forklift monitoring equipment with its state and assesses its quality, acquiring high-confidence monitoring data and establishing a dynamic mapping relationship between the data and the twin mirror model. The high-confidence monitoring data is input into the model, and forward calculations are performed from low to high levels to obtain the theoretical analytical state characteristics of each physical structure. Simultaneously, reverse decoupling and identification are performed from high to low levels to obtain the actual analytical state characteristics. By comparing the two, quantitative deviation information is identified. Based on the quantitative deviation information, fault types are matched from low to high levels to distinguish between surface faults and deep faults, forming diagnostic results. Based on the diagnostic results, control, decomposition, and feedback techniques are implemented. This method solves the technical problems of existing forklift steering system fault diagnosis, which struggle to achieve system-level linkage analysis and accurate identification of deep faults. It achieves the technical effects of realizing system-level linkage analysis, accurately identifying deep faults, improving fault location accuracy, and reducing missed and false fault diagnoses.
[0005] This application provides a method for diagnosing safety faults in the front and rear axle steering systems of a forklift, comprising: decomposing the physical structure and linkage relationships of the forklift's front and rear axle steering systems, constructing a twin mirror model, and dividing the physical structure into multiple levels from low to high according to the recognition complexity; performing state alignment and quality assessment on the forklift monitoring equipment to obtain high-confidence monitoring data, and establishing a dynamic mapping relationship between the high-confidence monitoring data and the twin mirror model; based on the dynamic mapping relationship, inputting the high-confidence monitoring data into the twin mirror model, and performing forward propagation calculations from low to high along the levels to obtain the theoretical analytical state characteristics of each physical structure; based on the high-confidence monitoring data, performing reverse decoupling identification from high to low along the levels to obtain the actual analytical state characteristics of each physical structure, comparing the actual analytical state characteristics with the corresponding theoretical analytical state characteristics to identify quantified deviation information; and, based on the quantified deviation information, performing fault type prediction matching sequentially from low to high along the levels to identify fault diagnosis results, wherein the fault diagnosis results include surface fault diagnosis results and deep fault diagnosis results, and performing control decomposition feedback based on the fault diagnosis results.
[0006] In a possible implementation, the physical structure and linkage relationships of the forklift's front and rear axle steering system are decomposed, a twin mirror model is constructed, and the following processing is performed: the forklift's front and rear axle steering system is decomposed hierarchically to obtain the target structure, geometric parameters, physical properties, and dynamic characteristics of each level. The levels are arranged in order of increasing recognition complexity as execution layer, transmission layer, power layer, control layer, and association layer. Based on the geometric constraint motion relationship, power linkage relationship, and rigid body mechanical coupling relationship of the hierarchical physical structure, lateral coupling relationship and longitudinal transmission relationship are established to obtain a linkage relationship network. The lateral coupling relationship is the parallel, series, or feedback connection relationship of the target structure within the same level; the longitudinal transmission relationship is the energy flow, signal flow, or material flow transmission relationship across levels. Based on the hierarchical physical structure and the linkage relationship network, the twin mirror model is constructed. The twin mirror model includes a forward calculation module and a reverse identification module. The forward calculation module is used to calculate the theoretical state of the upper level based on the lower level input, and the reverse identification module is used to invert the actual parameters of the lower level based on the upper level constraints.
[0007] In possible implementations, the following processes are performed: the target structure of the execution layer is the front steering axle, rear steering axle, and steering wheel assembly, characterized by directly observable geometric motion states; the target structure of the transmission layer is the steering cylinder, tie rod, and steering knuckle arm, characterized by mechanical displacement states of force transmission; the target structure of the power layer is the steering pump, fully hydraulic steering gear, and hydraulic lines, characterized by energy conversion states of fluid power; the target structure of the control layer is the steering valve core, feedback mechanism, and electronic control unit, characterized by dynamic response states of control logic; and the target structure of the association layer is the vehicle frame suspension, cargo center of gravity distribution, and ground contact coupling, characterized by multi-physics coupling states.
[0008] In a possible implementation, the forklift monitoring equipment undergoes state alignment and quality assessment, and the following processes are performed: Before operation, standard state data for the forklift is established, including the baseline response characteristics of each target structure in a healthy state; the standard state data is quantitatively mapped to the corresponding monitoring results of the forklift monitoring equipment to establish a monitoring mapping relationship; based on the standard state data, the monitoring equipment is state-corrected and aligned according to the monitoring mapping relationship to ensure that the monitoring data is consistent with the current actual state of the forklift; a quality assessment is performed based on the consistency verification results, and high-confidence monitoring data is marked.
[0009] In a possible implementation, the monitoring equipment is calibrated and aligned according to the monitoring mapping relationship, and the following processes are performed: establishing the calibration conditions between the forklift status and the monitoring equipment, wherein the calibration conditions include vehicle unloaded, stationary in a straight line, engine idling and hydraulic oil temperature in a stable operating range; triggering the calibration process of the monitoring equipment based on each calibration condition, and verifying the consistency between standard state data and monitoring results; calculating calibration compensation parameters based on the verification deviation, and calibrating and aligning the monitoring equipment with data.
[0010] In a possible implementation, forward propagation calculations are performed hierarchically from low to high to obtain the theoretical analytical state characteristics of each physical structure. The following processing is then performed: the directly measurable data from the high-confidence monitoring data is used as boundary input to drive the forward calculation module of the twin mirror model; calculations are performed layer by layer from low to high levels along the longitudinal propagation relationship, using the measured geometric quantities of the execution layer as input to calculate the theoretical force transmission state of the transmission layer; the calculation results of the transmission layer are used as input to calculate the theoretical energy state of the power layer; and so on, until the theoretical coupling state of the associated layer is obtained; the theoretical analytical state characteristics of each level are output, including theoretical geometric quantities, theoretical mechanical quantities, theoretical efficiency parameters, theoretical response characteristics, and theoretical coupling parameters.
[0011] In a possible implementation, the quantification deviation information is identified, and the following processing is performed: using the high-confidence monitoring data as an observation constraint, the inverse identification module of the twin mirror model is driven to invert layer by layer from high level to low level along the vertical transmission relationship. Using the macroscopic response of the correlation layer as a constraint, the actual dynamic parameters of the control layer are identified; using the identification results of the control layer as a constraint, the actual efficiency parameters of the power layer are identified; and so on, until the actual geometric state of the execution layer is obtained; the actual analytical state characteristics of each level are compared with the corresponding theoretical analytical state characteristics, and the deviation metric is calculated; the deviation metric is subjected to consistency verification and uncertainty assessment, and the quantification deviation information with confidence label is output.
[0012] In a possible implementation, based on the quantized deviation information, fault type prediction and matching are performed sequentially from low to high level to identify fault diagnosis results, and the following processing is performed: For the quantized deviation information of the execution layer and the transmission layer, known fault modes are directly matched. Known fault modes are surface faults, including abnormal tire pressure, insufficient steering fluid, excessive tie rod ball joint clearance, and poor lubrication. When the confidence level of the quantized deviation information meets a preset threshold, the surface fault diagnosis result is output. When the surface fault diagnosis cannot explain all quantized deviations, or abnormal deviations still exist after surface fault repair, or the quantized deviation information directly points to the power layer and above, deep fault diagnosis is triggered. For the quantized deviation information of the power layer, control layer, and related layers, fault propagation path reasoning is performed in conjunction with the linkage relationship network, and a fault hypothesis list sorted by probability is output. Deep faults include pump volumetric efficiency degradation, steering gear valve core wear, cylinder internal leakage, frame structure fatigue, and multi-system coupling failure. The surface fault diagnosis results and deep fault diagnosis results are combined to generate a fault diagnosis result, which includes fault level location and fault cause.
[0013] In a possible implementation, the following processing is performed based on the fault diagnosis results: Based on the surface fault diagnosis results, an immediate maintenance prompt is generated, including fault location guidance, on-site handling suggestions, and a list of required tools and spare parts; based on the deep fault diagnosis results, a predictive maintenance strategy is generated, including fault development trend prediction, suggested maintenance time windows, downtime impact assessment, and in-depth overhaul plan; wherein, according to the urgency of the fault diagnosis results, the forklift control strategy is dynamically adjusted, including speed-limited operation, steering angle restriction, power reduction mode, or forced safe shutdown.
[0014] This application also provides a safety fault diagnosis system for a forklift front and rear axle steering systems, comprising: a twin mirror model construction module, used to decompose the physical structure and linkage relationships of the forklift front and rear axle steering systems, construct a twin mirror model, and divide the physical structure into multiple levels from low to high according to recognition complexity; a high-confidence monitoring data acquisition module, used to perform state alignment and quality assessment on forklift monitoring equipment, obtain high-confidence monitoring data, and establish a dynamic mapping relationship between the high-confidence monitoring data and the twin mirror model; and a forward transfer calculation module, used to input the high-confidence monitoring data into the twin mirror model based on the dynamic mapping relationship, and transfer the data along the levels... A forward propagation calculation is performed from low to high to obtain the theoretical analytical state characteristics of each physical structure. A reverse decoupling identification module is used to perform reverse decoupling identification from high to low along the hierarchy based on the high-confidence monitoring data to obtain the actual analytical state characteristics of each physical structure. The actual analytical state characteristics are compared with the corresponding theoretical analytical state characteristics to identify quantification deviation information. A fault type prediction and matching module is used to perform fault type prediction and matching sequentially from low to high according to the quantification deviation information to identify fault diagnosis results. The fault diagnosis results include surface fault diagnosis results and deep fault diagnosis results, and control decomposition feedback is performed based on the fault diagnosis results.
[0015] This application proposes a method and system for diagnosing safety faults in the front and rear axle steering systems of forklifts. First, the physical structure and linkage relationships of the forklift's front and rear axle steering systems are decomposed, and a twin mirror model is constructed. The physical structure is then divided into multiple levels from low to high based on recognition complexity. Next, the forklift monitoring equipment undergoes state alignment and quality assessment to obtain high-confidence monitoring data. A dynamic mapping relationship is established between the high-confidence monitoring data and the twin mirror model. Finally, based on this dynamic mapping relationship, the high-confidence monitoring data is input into the twin mirror model, and positive correlation is performed from low to high levels. The method and system proposed in this application achieve the technical effects of system-level linkage analysis, accurate identification of deep faults, and improved fault location accuracy while reducing fault omissions and misjudgments. The actual analytical state characteristics of each physical structure are obtained through reverse decoupling and identification along the hierarchy from high to low based on the high-confidence monitoring data. These actual analytical state characteristics are then compared with the corresponding theoretical analytical state characteristics to identify quantified deviation information. Finally, based on the quantified deviation information, fault type prediction and matching are performed sequentially from low to high hierarchy to identify fault diagnosis results. These fault diagnosis results include surface fault diagnosis results and deep fault diagnosis results. Adjustment decomposition and feedback are then performed based on these fault diagnosis results. Attached Figure Description
[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.
[0017] Figure 1 This is a flowchart illustrating a method for diagnosing safety faults in the front and rear axle steering systems of a forklift, as provided in an embodiment of this application.
[0018] Figure 2 This is a structural schematic diagram of a safety fault diagnosis system for the front and rear axle steering systems of a forklift, provided in an embodiment of this application.
[0019] Figure labeling: Twin mirror model construction module 10, high confidence monitoring data acquisition module 20, forward propagation calculation module 30, reverse decoupling identification module 40, fault type prediction and matching module 50. Detailed Implementation
[0020] The above description is merely an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below.
[0021] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of this application will be provided in conjunction with the accompanying drawings. The described embodiments should not be considered as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0022] In the following description, references to "some embodiments" describe a subset of all possible embodiments; however, it is understood that "some embodiments" may be the same or different subsets of all possible embodiments and may be combined with each other without conflict. The terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only.
[0023] This application provides a method for diagnosing safety faults in the front and rear axle steering systems of a forklift, such as... Figure 1 As shown, the method includes:
[0024] Step S100: Decompose the physical structure and linkage relationship of the forklift's front and rear axle steering system, construct a twin mirror model, and divide the physical structure into multiple levels from low to high according to the recognition complexity.
[0025] Specifically, a hierarchical modeling method based on multibody dynamics and hydraulic system dynamics is adopted. The system is divided into execution layer, transmission layer, power layer, control layer, and correlation layer according to the complexity of identification from low to high. Geometric parameters, material properties, kinematic pair constraints, and dynamic characteristics of the target structure at each level are obtained through 3D scanning, component parameter manuals, and bench tests. Based on geometric constraint equations, force balance equations, flow continuity equations, and control logic equations, lateral coupling relationships within the same level and longitudinal transmission relationships across levels are established, forming a linkage network containing nodes and transmission paths. On this basis, a modular modeling approach is used to construct a twin mirror model. The forward calculation module adopts a bottom-up recursive solution structure, while the reverse identification module adopts a top-down hierarchical inversion structure. Data interaction between modules is achieved through standardized state variable interfaces. Model parameters are corrected through bench calibration and real vehicle tests to ensure dynamic consistency with the physical system.
[0026] In one possible implementation, the physical structure and linkage of the forklift's front and rear axle steering system are decomposed, and a twin mirror model is constructed. Step S100 further includes step S110, which involves hierarchically decomposing the forklift's front and rear axle steering system to obtain the target structure, geometric parameters, physical properties, and dynamic characteristics of each level. The levels, ranked from low to high in terms of recognition complexity, are: execution layer, transmission layer, power layer, control layer, and association layer. Specifically, the target structure of the execution layer is the front steering axle, rear steering axle, and steering wheel assembly, characterized by directly observable geometric motion states; the target structure of the transmission layer is the steering cylinder, tie rods, and steering knuckle, characterized by mechanical displacement states of force transmission; the target structure of the power layer is the steering pump, fully hydraulic steering gear, and hydraulic lines, characterized by energy conversion states of fluid power; the target structure of the control layer is the steering valve core, feedback mechanism, and electronic control unit, characterized by dynamic response states of control logic; and the target structure of the association layer is the chassis suspension, cargo center of gravity distribution, and ground contact coupling, characterized by multi-physics coupling states.
[0027] Specifically, a hierarchical decomposition is performed using 3D modeling software and a physical parameter library. The execution layer uses laser scanning and image measurement to obtain the geometric dimensions and installation positions of the front steering axle, rear steering axle, and steering wheel assembly. The transmission layer uses calipers, dial indicators, and force sensors to measure the length, cross-sectional dimensions, and stiffness of the steering cylinder, tie rods, and steering knuckle arms. The power layer uses hydraulic component manuals and flow and pressure test benches to obtain the displacement, rated pressure, pipe diameter, and flow resistance coefficient of the steering pump, fully hydraulic steering gear, and hydraulic lines. The control layer reads the steering valve core stroke, feedback mechanism potentiometer characteristics, and control algorithm parameters through the electronic control unit interface. The correlation layer uses chassis modal testing and load sensors to obtain the chassis suspension stiffness, cargo center of gravity distribution range, and tire ground contact stiffness and friction coefficient.
[0028] Step S120: Based on the geometric constraint motion relationships, dynamic linkage relationships, and rigid body mechanical coupling relationships of the hierarchical physical structure, establish lateral coupling relationships and longitudinal transmission relationships to obtain a linkage relationship network. The lateral coupling relationships are parallel, series, or feedback connection relationships of target structures within the same level; the longitudinal transmission relationships are energy flow, signal flow, or material flow transmission relationships across levels. Specifically, a multibody dynamics modeling method is used to construct lateral coupling relationships. Within the execution layer, the front steering axle, rear steering axle, and steering wheel assembly form series and parallel combinations with the steering trapezoidal mechanism through hinge constraints. Within the transmission layer, the steering cylinder, tie rods, and steering knuckle arms form a force transmission chain with the kinematic pairs through force constraints. Within the power layer, the steering pump, fully hydraulic steering gear, and hydraulic pipelines form a fluid network through flow continuity equations and pressure loss equations. Within the control layer, the steering valve core, feedback mechanism, and electronic control unit form a closed-loop control relationship through signal feedback loops. Within the association layer, the vehicle frame suspension, cargo center of gravity distribution, and ground contact form a coupling network through a multibody contact model and load transmission relationships. The longitudinal transmission relationship employs transmission equations for energy flow, signal flow, and material flow. The execution layer transmits steering wheel angle and ground reaction force to the transmission layer; the transmission layer transmits tie rod force and cylinder displacement to the power layer; the power layer transmits hydraulic pressure and flow signals to the control layer; and the control layer transmits control commands and system responses to the associated layers. For example, the steering wheel angle in the execution layer is converted into the lateral tie rod displacement in the transmission layer through the steering trapezoidal geometry. The displacement is equal to the steering knuckle arm length multiplied by the sine of the steering wheel angle. The tie rod force in the transmission layer is converted into hydraulic pressure in the power layer through the cylinder area. The pressure is equal to the tie rod force divided by the effective cylinder area. The hydraulic flow in the power layer is converted into the valve core position signal in the control layer through the steering gear opening. The signal value is proportional to the flow rate.
[0029] Step S130: Based on the hierarchical physical structure and the linkage network, construct the twin mirror model. The twin mirror model includes a forward calculation module and a reverse identification module. The forward calculation module is used to calculate the theoretical state of the upper layer based on the lower layer input, and the reverse identification module is used to invert the actual parameters of the lower layer based on the upper layer constraints. Specifically, a multiphysics coupling simulation platform is used to construct the twin mirror model. The forward calculation module adopts a recursive solution structure, using the measured geometric quantities of the execution layer as the initial input. It calculates the mechanical displacement of the transmission layer through geometric constraint equations, calculates the fluid state of the power layer through force balance equations, calculates the dynamic response of the control layer through control algorithms, and calculates the coupling state of the associated layer through multiphysics coupling equations. The calculation results of each layer are used as the input of the next layer. The reverse identification module adopts a hierarchical inversion structure, using the macroscopic response of the associated layer as the top-level constraint. It inverts the actual parameters of the control layer through a state observer, inverts the efficiency parameters of the power layer through parameter identification, inverts the force transmission characteristics of the transmission layer through force and displacement inversion, and inverts the actual geometric state of the execution layer through geometric inversion. The twin mirror model adopts a modular design, and state variables and parameters are passed between layers through standard interfaces.
[0030] Step S200: Perform status alignment and quality assessment on the forklift monitoring equipment to obtain high-confidence monitoring data, and establish a dynamic mapping relationship between the high-confidence monitoring data and the twin mirror model.
[0031] Specifically, the standard operating condition calibration method is adopted. Under conditions of no load, linear stationary state, engine idling speed, and hydraulic oil temperature within a stable operating range, the baseline response characteristics of the target structure at each level are collected to form a standard state database. The monitoring mapping relationship between the sensor output signal and the physical quantity is established by fitting the least squares method. Based on this mapping relationship, the sensor is subjected to zero-point offset correction, sensitivity correction, and nonlinear compensation. The monitoring data is evaluated for quality through triple verification of time consistency, spatial consistency, and physical rationality. Outliers and abrupt data are eliminated, and data that meets the verification conditions is marked as high-confidence data. The high-confidence data is converted into state variables that can be recognized by the twin mirror model using the lookup table method and function mapping method, so as to realize the dynamic binding of monitoring data and model variables.
[0032] In one possible implementation, the forklift monitoring equipment undergoes state alignment and quality assessment. Step S200 further includes step S210, which establishes standard state data for the forklift before operation. This standard state data includes the baseline response characteristics of each target structure under healthy conditions. Specifically, standard state data is established using a standard operating condition test method. Under no-load, straight-line stationary, engine idling, and hydraulic oil temperature within a stable range of 45 to 55 degrees Celsius, sensors collect data on: the steering wheel angle, kingpin inclination angle, and tire pressure of the actuator layer; the tie rod displacement, ball joint clearance, and cylinder stroke of the transmission layer; the pump outlet pressure, steering gear flow rate, and pipeline pressure drop of the power layer; the valve core position, feedback voltage, and control current of the control layer; and the frame attitude, suspension displacement, and ground reaction force of the correlation layer. Ten sets of data are collected for each parameter, and the average value is used as the baseline value. Simultaneously, ambient temperature, humidity, and hydraulic oil viscosity are recorded as correction factors.
[0033] Step S220: Quantitative mapping is performed using the standard state data and the corresponding monitoring results of the forklift monitoring equipment to establish a monitoring mapping relationship. Specifically, piecewise linear fitting and calibration experiments are used to establish the monitoring mapping relationship. For example, for angle sensors, standard angles and sensor output voltages are collected, and the linear relationship between voltage and angle is obtained through least squares fitting; for pressure sensors, standard pressures and output currents are collected, and the relationship between current and pressure is obtained through fitting; for displacement sensors, the standard displacement of a grating ruler is fitted with the sensor pulse count. The monitoring mapping relationship is stored as a lookup table and fitting equations, and sensor signals are converted into physical quantities through equations.
[0034] Step S230: Using the standard state data as a benchmark, the monitoring equipment is calibrated and aligned according to the monitoring mapping relationship to ensure that the monitoring data matches the actual state of the forklift. Specifically, a closed-loop calibration method is used to convert the standard state data into theoretical sensor output values through the mapping relationship. The deviation is then compared with the actual monitored values. The sensor calibration parameters are adjusted based on the deviation; for example, the zero-point offset and gain coefficient are adjusted for the angle sensor, the zero-point current and sensitivity are adjusted for the pressure sensor, and the pulse equivalent and initial position are adjusted for the displacement sensor. After calibration, standard state data is collected again for verification until the deviation is less than the allowable range.
[0035] Step S240: Based on the consistency verification results, a quality assessment is performed, and high-confidence monitoring data is marked. Specifically, a triple consistency verification method is adopted: the first level is time consistency, where the rate of change of five consecutive sampling data is less than the physical limit; the second level is spatial consistency, where the difference in steering wheel angle between the left and right steering wheels is less than a preset degree, conforming to the steering trapezoidal constraint; and the third level is physical rationality, where pressure, flow rate, displacement, etc., satisfy the basic equations. Data that passes the triple verification is marked as high-confidence, and data that fails is marked as low-confidence and discarded.
[0036] In one possible implementation, the monitoring equipment is calibrated and aligned according to the monitoring mapping relationship. Step S230 further includes step S231, establishing calibration conditions for the forklift status and the monitoring equipment. These calibration conditions include vehicle unloaded, stationary in a straight line, engine idling, and hydraulic oil temperature within a stable operating range. Specifically, an automatic condition identification and triggering mechanism is adopted. The load sensor detects a load weight of less than 50 kg, determining it as unloaded. The gyroscope and tilt sensor detect the frame being level and the steering wheels being in a straight line, determining it as stationary in a straight line. The engine speed sensor detects an engine speed between 750 and 850 rpm, determining it as idling. The temperature sensor detects the hydraulic oil temperature between 45 and 55 degrees Celsius, determining it as a stable range. When all conditions are met simultaneously, the calibration conditions are triggered, and the calibration process begins.
[0037] Step S232: Based on each calibration condition, the monitoring equipment calibration process is triggered, and the consistency between the standard state data and the monitoring results is verified. Specifically, an automatic calibration process is adopted, all sensor data are collected under the calibration condition, converted into physical quantities through mapping relationships, compared point by point with the standard state data, and the absolute and relative deviations are calculated. Sensors with deviations greater than a preset threshold are determined to require calibration.
[0038] Step S233: Calculate calibration compensation parameters based on the verification deviation and perform data calibration and alignment on the monitoring equipment. Specifically, a parameter compensation method is used: for zero-point offset deviations, the sensor zero-point parameters are directly adjusted; for gain deviations, the sensitivity coefficient is adjusted; and for nonlinear deviations, a piecewise compensation table is used. The compensation parameters for angle sensors include zero-point offset and gain coefficient; for pressure sensors, they include zero-point current and slope; and for displacement sensors, they include initial pulse count and pulse equivalent. After compensation, data is re-acquired for verification until the deviation is within the allowable range.
[0039] Step S300: Based on the dynamic mapping relationship, the high-confidence monitoring data is input into the twin mirror model, and forward propagation calculation is performed from low to high along the hierarchy to obtain the theoretical analytical state characteristics of each physical structure.
[0040] Specifically, an explicit recursive solution method is adopted. The directly measurable geometric quantities and force parameters of the execution layer are used as initial inputs. The theoretical displacement and force transmission state of the transmission layer are calculated through the steering trapezoidal geometric equation and rigid body mechanics equation. The theoretical energy conversion state of the power layer is calculated through the hydraulic system flow and pressure equation. The theoretical logic state of the control layer is calculated through the control algorithm and dynamic response equation. The theoretical coupling state of the related layers is calculated through the multi-physics coupling equation. The solution process adopts fixed time step iteration, and the calculation results of each layer are used as the input conditions of the next layer. Finally, the theoretical geometric quantities, theoretical mechanical quantities, theoretical efficiency parameters, theoretical response characteristics and theoretical coupling parameters of each layer are output, forming a complete set of theoretical analytical states.
[0041] In one possible implementation, forward propagation calculations are performed hierarchically from low to high to obtain the theoretical analytical state characteristics of each physical structure. Step S300 further includes step S310, which uses the directly measurable data from the high-confidence monitoring data as boundary inputs to drive the forward calculation module of the twin mirror model. Specifically, a data interface module reads the high-confidence monitoring data, filters out directly measurable data at the execution layer, including steering wheel left and right steering angles, kingpin inclination angles, tire pressure, and ground reaction forces. These data are formatted as model input variables, converted into internal model units through a data mapping table, and the input variables are passed to the forward calculation module through the interface as initial boundary conditions.
[0042] Step S320: Calculate layer by layer along the longitudinal transmission relationship from lower to higher levels. Using the measured geometric quantities of the execution layer as input, calculate the theoretical force transmission state of the transmission layer; using the calculation results of the transmission layer as input, calculate the theoretical energy state of the power layer; and so on, until the theoretical coupling state of the associated layer is obtained. Specifically, a recursive solution method is adopted. For example, the execution layer inputs the steering wheel angle, calculates the transmission layer tie rod displacement through the steering trapezoidal geometric equation, calculates the tie rod force through the force balance equation, and uses the transmission layer's output displacement and force as the power layer input. The hydraulic pressure is calculated through the cylinder force equation, and the flow rate is calculated through the flow equation. The power layer's output pressure and flow rate are used as the control layer input. The valve core stroke is calculated through the valve core flow characteristic equation, and the control current is calculated through the control algorithm. The control layer outputs the valve core position and current as the associated layer input. The suspension displacement is calculated through the frame dynamics equation, and the tire ground force is calculated through the grounding model. Each layer's calculation uses explicit solution with a set time step to ensure calculation stability.
[0043] Step S330 outputs the theoretical analytical state characteristics of each level, including theoretical geometric quantities, theoretical mechanical quantities, theoretical efficiency parameters, theoretical response characteristics, and theoretical coupling parameters. Specifically, after the forward calculation is completed, the result output module outputs the theoretical analytical state characteristics of each level. For example, the execution layer outputs the theoretical steering wheel angle, theoretical kingpin inclination angle, and theoretical tire contact area; the transmission layer outputs the theoretical tie rod displacement, theoretical tie rod force, and theoretical ball joint clearance; the power layer outputs the theoretical hydraulic pressure, theoretical flow rate, and theoretical pump volumetric efficiency; the control layer outputs the theoretical valve core stroke, theoretical response time, and theoretical control error; and the correlation layer outputs the theoretical frame attitude angle, theoretical suspension stiffness, and theoretical coupling coefficient. All parameters are output in text format.
[0044] Step S400: Based on the high-confidence monitoring data, reverse decoupling and identification are performed from high to low along the hierarchy to obtain the actual analytical state characteristics of each physical structure. The actual analytical state characteristics are compared with the corresponding theoretical analytical state characteristics to identify quantification deviation information.
[0045] Specifically, a hierarchical Kalman filter and least squares parameter identification method are employed. The macroscopic response of the correlation layer is used as the top-level constraint. The actual dynamic parameters of the control layer are inverted through a state observer, and the control layer identification results are used as constraints. The actual efficiency parameters of the power layer are identified through flow-pressure inversion equations, and the power layer identification results are used as constraints. The actual force transmission characteristics of the transmission layer are identified through force-displacement back-calculation, and the transmission layer identification results are used as constraints. Finally, the actual geometric state of the execution layer is identified through geometric back-calculation. The actual analytical state characteristics of each level are compared parameter-by-parameter with the theoretical analytical state characteristics, and the absolute and relative deviations are calculated. The deviation uncertainty is evaluated through statistical analysis of multiple inversion results. A deviation confidence level is assigned based on the standard deviation, and the final output is quantitative deviation information containing the deviation value, deviation type, and confidence level.
[0046] In one possible implementation, identifying quantification deviation information, step S400 further includes step S410, using the high-confidence monitoring data as an observation constraint to drive the inverse identification module of the twin mirror model to invert layer by layer from high level to low level along the longitudinal transmission relationship, using the macroscopic response of the correlation layer as a constraint to identify the actual dynamic parameters of the control layer; using the identification result of the control layer as a constraint to identify the actual efficiency parameters of the power layer; and so on, until the actual geometric state of the execution layer is obtained. Specifically, a hierarchical recursive parameter identification mechanism is adopted. The macroscopic response monitoring data of the correlation layer is used as the top-level constraint. The actual dynamic parameters of the steering valve core, feedback mechanism and electronic control unit in the control layer are inverted through state observation and parameter estimation methods. The actual parameters identified in the control layer are used as intermediate constraints. The actual efficiency and leakage parameters of the steering pump, full hydraulic steering gear and hydraulic pipeline in the power layer are identified through flow and pressure characteristic inversion and energy conservation analysis. The identification results of the power layer are used as constraints. The actual force transmission and clearance parameters of the steering cylinder, tie rod and steering knuckle in the transmission layer are identified through force balance and displacement back calculation methods. The identification results of the transmission layer are used as constraints. The actual geometric and kinematic parameters of the front steering axle, rear steering axle and steering wheel set in the execution layer are identified through geometric constraint back calculation methods. The identification process of each level adopts an iterative optimization algorithm with the goal of minimizing the residual between the monitoring data and the model output. After convergence, the actual analytical state parameters of each level are output.
[0047] Step S420 involves comparing the actual analytical state characteristics of each level with the corresponding theoretical analytical state characteristics to calculate the deviation metric. Specifically, a one-to-one correspondence is established between the parameters at each level, and for each set of actual and theoretical parameters, the absolute deviation and relative deviation are calculated. The absolute deviation is the difference between the actual and theoretical parameter values, and the relative deviation is the ratio of the absolute deviation to the theoretical parameter value. A unified deviation calculation rule is adopted for different types of parameters to ensure the comparability and consistency of the deviation metrics, ultimately forming a set of deviation metrics for each level.
[0048] Step S430: Perform consistency verification and uncertainty assessment on the deviation metric, and output quantified deviation information with confidence level labels. Specifically, perform time consistency verification on the deviation metric to verify whether the deviation obtained from multiple consecutive identifications is within a stable range. Perform physical rationality verification on the deviation metric to determine whether the deviation exceeds the physical limit range. Calculate the standard deviation of the deviation based on the statistical distribution of multiple inversion results to assess the uncertainty. Divide the confidence level according to the uncertainty magnitude, bind the deviation metric with the corresponding confidence level, form quantified deviation information with confidence level labels, and output it.
[0049] Step S500: Based on the quantified deviation information, perform fault type prediction and matching in order of hierarchy from low to high, identify fault diagnosis results, the fault diagnosis results include surface fault diagnosis results and deep fault diagnosis results, and perform regulation decomposition feedback based on the fault diagnosis results.
[0050] Specifically, a fault mode and impact analysis rule base is used to directly match the quantified deviation information between the execution layer and the transmission layer. When the deviation confidence level meets a preset threshold, it is judged as a surface fault, and location and handling information is output. When the surface fault cannot explain all deviations or the deviation points to the power layer or higher, a fault propagation directed graph and Bayesian inference method are used based on the linkage network to trace the fault path of deviations in the power layer, control layer, and related layers, generating a deep fault hypothesis list sorted by probability. Combining the surface fault and deep fault results, a diagnostic report is generated, including the fault level, faulty component, and fault cause. Based on the diagnostic results, the maintenance knowledge base is called to generate real-time maintenance prompts and predictive maintenance strategies. According to the severity of the fault, the electronic control unit dynamically adjusts the forklift operating parameters to achieve control feedback such as speed limiting, steering angle limiting, power reduction, or safe stopping.
[0051] In one possible implementation, based on the quantized deviation information, fault type prediction and matching are performed sequentially from low to high level to identify fault diagnosis results. Step S500 further includes step S510, whereby the quantized deviation information between the execution layer and the transmission layer is directly matched with known fault modes. The known fault modes are surface faults, including abnormal tire pressure, insufficient steering fluid, excessive tie rod ball joint clearance, and poor lubrication. When the confidence level of the quantized deviation information meets a preset threshold, the surface fault diagnosis result is output. Specifically, a rule base corresponding to surface fault modes and deviation features is established. The quantized deviation information between the execution layer and the transmission layer is classified according to parameter type and deviation direction. It is matched item by item with the typical deviation features of surface faults such as abnormal tire pressure, insufficient steering fluid, excessive tie rod ball joint clearance, and poor lubrication in the rule base. For successfully matched fault types, it is verified whether the confidence level of their corresponding deviations reaches a preset threshold. If it does, the fault type, corresponding deviation parameters, and confidence level are integrated into a surface fault diagnosis result and output.
[0052] Step S520: When surface fault diagnosis cannot explain all quantization deviations, or abnormal deviations still exist after surface fault repair, or quantization deviation information directly points to the power layer or higher, deep fault diagnosis is triggered. Specifically, the residual value between the deviation explained by surface fault diagnosis and the total deviation is calculated, and it is determined whether the residual value exceeds a preset range. Simultaneously, the source level of the quantization deviation information is detected to determine whether it directly originates from the power layer, control layer, or related layer. If the residual value exceeds the preset range, or abnormal deviations exceeding a threshold still exist after simulated surface fault repair, or deviation information directly points to the power layer or higher, the deep fault diagnosis process is triggered, and fault propagation path reasoning is initiated.
[0053] Step S530 involves using the quantified deviation information of the power layer, control layer, and related layers, combined with the aforementioned linkage network, to perform fault propagation path reasoning, and outputting a fault hypothesis list sorted by probability. Deep-level faults include pump volumetric efficiency degradation, steering gear valve core wear, cylinder internal leakage, frame structure fatigue, and multi-system coupling failure. Specifically, a directed fault propagation graph is constructed based on the linkage network, clarifying the source level, transmission path, and associated influence parameters corresponding to each deep-level fault. The quantified deviation information of the power layer, control layer, and related layers is used as fault feature input. Based on the path relationships in the directed fault propagation graph, the possible sources and propagation processes of the deviations are traced. For deep-level fault types such as pump volumetric efficiency degradation, steering gear valve core wear, cylinder internal leakage, frame structure fatigue, and multi-system coupling failure, the probability of occurrence of each fault is calculated by combining the deviation magnitude, confidence level, and propagation path matching degree. The fault hypothesis list is then sorted from high to low probability and output.
[0054] Step S540: Integrate the surface fault diagnosis results and the deep fault diagnosis results to generate a fault diagnosis result, which includes fault level location and fault cause. Specifically, the surface fault diagnosis results are merged with the deep fault hypothesis list, duplicate or contradictory fault items are eliminated, and the execution layer, transmission layer, power layer, control layer or related layer to which the retained fault items belong is identified. The causes of each fault are analyzed by combining deviation characteristics and fault mode characteristics. The fault level location, faulty component, fault cause and corresponding confidence level are integrated into a unified fault diagnosis result and output.
[0055] In one possible implementation, the control decomposition feedback is performed based on the fault diagnosis results. Step S500 further includes step S550, generating an instant maintenance prompt based on the surface fault diagnosis results, including fault location guidance, on-site handling suggestions, and a list of required tools and spare parts. Specifically, a knowledge base mapping surface faults and maintenance information is established. Based on the fault type and location information in the surface fault diagnosis results, the corresponding fault location guidance, on-site handling steps, required tool types and specifications, and applicable spare parts models and quantities are matched from the knowledge base. The above information is integrated into an instant maintenance prompt, ensuring that the prompt content can be directly used for rapid on-site handling.
[0056] Step S560: Based on the deep fault diagnosis results, a predictive maintenance strategy is generated, including fault development trend prediction, suggested maintenance time window, downtime impact assessment, and in-depth overhaul plan. Specifically, based on the type of deep fault, current deviation level, and confidence level, combined with the fault decay model and life prediction model, the development trend and degradation rate of the fault are deduced, and the suggested maintenance time window is determined according to the degradation rate and system safety threshold. Based on the fault level and impact range, the impact of fault development on forklift operating efficiency, operational safety, and downtime is assessed. The corresponding fault disassembly process, detection method, repair process, and replacement component requirements are matched from the in-depth overhaul knowledge base to form an in-depth overhaul plan. The trend prediction, time window, impact assessment, and overhaul plan are integrated into a predictive maintenance strategy.
[0057] Step S570: Based on the urgency of the fault diagnosis results, dynamically adjust the forklift control strategy, including speed-limited operation, steering angle limitation, power reduction mode, or forced safety stop. Specifically, establish fault urgency classification rules, dividing faults into three levels: low urgency, medium urgency, and high urgency, based on the fault's level, deviation magnitude, confidence level, and impact range. For low-urgency faults, the electronic control unit (ECU) executes speed-limited operation and steering angle limitation control; for medium-urgency faults, power reduction operation control is superimposed on speed-limited and steering angle-limited operation; for high-urgency faults, a forced safety stop control command is directly triggered. By outputting corresponding control signals through the ECU, the engine power, steering actuator travel, and driving parameters are adjusted to achieve dynamic adjustment of the control strategy.
[0058] This application's embodiments employ a hierarchical decomposition of the physical structure and linkage relationships of the forklift's front and rear axle steering systems based on recognition complexity. A twin mirror model containing forward calculation and reverse identification functions is constructed. State alignment and quality assessment of the forklift monitoring equipment are performed, high-confidence monitoring data is acquired, and a dynamic mapping relationship is established between this data and the twin mirror model. The high-confidence monitoring data is input into the model, and forward calculation is performed hierarchically from low to high to obtain the theoretical analytical state characteristics of each physical structure. Simultaneously, reverse decoupling identification is performed hierarchically from high to low to obtain the actual analytical state characteristics. By comparing the two, quantitative deviation information is identified. Based on the quantitative deviation information, fault types are matched hierarchically from low to high to distinguish between surface faults and deep faults, forming diagnostic results. Based on the diagnostic results, control, decomposition, and feedback techniques are implemented. This solves the technical problem of existing forklift steering system fault diagnosis, which struggles to achieve system-level linkage analysis and accurate identification of deep faults. It achieves the technical effects of realizing system-level linkage analysis, accurately identifying deep faults, improving fault location accuracy, and reducing missed and false fault diagnoses.
[0059] In the above text, refer to Figure 1 A method for diagnosing safety faults in the front and rear axle steering systems of a forklift according to an embodiment of the present invention is described in detail. Next, reference will be made to... Figure 2 A safety fault diagnosis system for the front and rear axle steering systems of a forklift, according to an embodiment of the present invention, is described.
[0060] A safety fault diagnosis system for a forklift front and rear axle steering systems according to an embodiment of the present invention addresses the technical problem of difficulty in achieving system-level linkage analysis and accurate identification of deep faults in existing forklift steering system fault diagnosis methods. The system achieves system-level linkage analysis, accurate identification of deep faults, and improves fault location accuracy while reducing missed and false fault diagnoses. The safety fault diagnosis system for a forklift front and rear axle steering systems includes: a twin mirror model construction module 10, a high-confidence monitoring data acquisition module 20, a forward transmission calculation module 30, a reverse decoupling identification module 40, and a fault type prediction and matching module 50.
[0061] The twin mirror model construction module 10 is used to decompose the physical structure and linkage relationship of the forklift's front and rear axle steering system, construct a twin mirror model, and divide the physical structure into multiple levels from low to high according to recognition complexity; the high-confidence monitoring data acquisition module 20 is used to perform state alignment and quality assessment on the forklift monitoring equipment, obtain high-confidence monitoring data, and establish a dynamic mapping relationship between the high-confidence monitoring data and the twin mirror model; the forward transfer calculation module 30 is used to input the high-confidence monitoring data into the twin mirror model based on the dynamic mapping relationship, and perform forward transfer calculation along the levels from low to high to obtain each Theoretical analytical state characteristics of physical structures; reverse decoupling identification module 40, used to perform reverse decoupling identification from high to low level based on the high confidence monitoring data, to obtain the actual analytical state characteristics of each physical structure, and compare the actual analytical state characteristics with the corresponding theoretical analytical state characteristics to identify quantification deviation information; fault type prediction and matching module 50, used to perform fault type prediction and matching from low to high level according to the quantification deviation information, to identify fault diagnosis results, the fault diagnosis results including surface fault diagnosis results and deep fault diagnosis results, and to perform control decomposition feedback according to the fault diagnosis results.
[0062] The detailed description of the specific configuration of the twin mirror model construction module 10 is explained as follows: As mentioned above, the physical structure and linkage relationship of the forklift's front and rear axle steering system are decomposed to construct a twin mirror model. The twin mirror model construction module 10 may further include: a hierarchical decomposition unit for performing hierarchical decomposition of the forklift's front and rear axle steering system to obtain the target structure, geometric parameters, physical properties, and dynamic characteristics of each level. The levels are arranged in order of increasing recognition complexity as execution layer, transmission layer, power layer, control layer, and association layer; and a linkage relationship network establishment unit for establishing the motion relationship and power linkage relationship based on the geometric constraints of the hierarchical physical structure. The rigid body mechanics coupling relationship is established by creating lateral coupling relationships and longitudinal transmission relationships to obtain a linkage network. The lateral coupling relationship refers to the parallel, series, or feedback connection relationships of target structures within the same level; the longitudinal transmission relationship refers to the energy flow, signal flow, or material flow transmission relationship across levels. A twin mirror model construction unit is used to construct the twin mirror model based on the hierarchical physical structure and the linkage network. The twin mirror model includes a forward calculation module and a reverse identification module. The forward calculation module is used to calculate the theoretical state of the upper layer based on the lower layer input, and the reverse identification module is used to invert the actual parameters of the lower layer based on the upper layer constraints.
[0063] The hierarchical decomposition unit may further include: the target structure of the execution layer is the front steering axle, rear steering axle, and steering wheel assembly, characterized by directly observable geometric motion states; the target structure of the transmission layer is the steering cylinder, tie rod, and steering knuckle arm, characterized by mechanical displacement states of force transmission; the target structure of the power layer is the steering pump, fully hydraulic steering gear, and hydraulic lines, characterized by energy conversion states of fluid power; the target structure of the control layer is the steering valve core, feedback mechanism, and electronic control unit, characterized by dynamic response states of control logic; and the target structure of the association layer is the vehicle frame suspension, cargo center of gravity distribution, and ground contact coupling, characterized by multi-physics coupling states.
[0064] The detailed description of the specific configuration of the high-confidence monitoring data acquisition module 20 is explained as follows: As mentioned above, to perform state alignment and quality assessment on the forklift monitoring equipment, the high-confidence monitoring data acquisition module 20 may further include: a standard state data establishment unit for establishing standard state data of the forklift before operation, wherein the standard state data includes the baseline response characteristics of each target structure in a healthy state; a monitoring mapping relationship establishment unit for quantitatively mapping the standard state data with the corresponding monitoring results of the forklift monitoring equipment to establish a monitoring mapping relationship; a state correction alignment unit for performing state correction alignment of the monitoring equipment according to the monitoring mapping relationship based on the standard state data, so that the monitoring data is consistent with the current actual state of the forklift; and a quality assessment unit for performing quality assessment based on the consistency verification results and marking high-confidence monitoring data.
[0065] The monitoring equipment is calibrated and aligned according to the monitoring mapping relationship. The state calibration and alignment unit may further include: a calibration condition establishment subunit for establishing the calibration conditions of the forklift state and the monitoring equipment, wherein the calibration conditions include vehicle unloaded, stationary in a straight line, engine idling and hydraulic oil temperature in a stable operating range; a consistency verification subunit for triggering the calibration process of the monitoring equipment based on each calibration condition and performing consistency verification using standard state data and monitoring results; and a data calibration and alignment subunit for calculating calibration compensation parameters based on verification deviation and performing data calibration and alignment of the monitoring equipment.
[0066] The specific configuration of the forward transfer calculation module 30 is described in detail below: As mentioned above, forward transfer calculations are performed from low to high levels to obtain the theoretical analytical state characteristics of each physical structure. The forward transfer calculation module 30 may further include: a forward calculation module driving unit used to drive the forward calculation module of the twin mirror model by taking the directly measurable data in the high confidence monitoring data as boundary input; a related layer theoretical coupling state acquisition unit used to calculate layer by layer from low to high levels along the longitudinal transfer relationship, using the measured geometric quantities of the execution layer as input to calculate the theoretical force transfer state of the transmission layer; using the calculation results of the transmission layer as input to calculate the theoretical energy state of the power layer; and so on, until the theoretical coupling state of the related layer is obtained; and a theoretical analytical state characteristic output unit used to output the theoretical analytical state characteristics of each level, including theoretical geometric quantities, theoretical mechanical quantities, theoretical efficiency parameters, theoretical response characteristics, and theoretical coupling parameters.
[0067] The detailed description of the specific configuration of the reverse decoupling identification module 40 is explained as follows: As mentioned above, to identify quantization deviation information, the reverse decoupling identification module 40 may further include: an execution layer actual geometric state acquisition unit used to use the high-confidence monitoring data as an observation constraint to drive the reverse identification module of the twin mirror model to invert layer by layer from high level to low level along the longitudinal transmission relationship, using the macroscopic response of the correlation layer as a constraint to identify the actual dynamic parameters of the control layer; using the identification result of the control layer as a constraint to identify the actual efficiency parameters of the power layer; and so on, until the actual geometric state of the execution layer is obtained; a feature comparison unit used to compare the actual analytical state features of each level with the corresponding theoretical analytical state features to calculate the deviation metric; and a quantization deviation information output unit used to perform consistency verification and uncertainty assessment on the deviation metric and output quantization deviation information with confidence markers.
[0068] The detailed description of the specific configuration of the fault type prediction and matching module 50 is explained as follows: As mentioned above, based on the quantified deviation information, fault type prediction and matching are performed sequentially from low to high level to identify fault diagnosis results. The fault type prediction and matching module 50 may further include: a surface fault diagnosis result output unit for directly matching the quantified deviation information between the execution layer and the transmission layer with known fault modes. The known fault modes are surface faults, including abnormal tire pressure, insufficient steering fluid, excessive tie rod ball joint clearance, and poor lubrication. When the confidence level of the quantified deviation information meets a preset threshold, the surface fault diagnosis result is output; a deep fault diagnosis trigger unit is used when the surface fault diagnosis is not possible. When the method explains all quantification deviations, or when abnormal deviations still exist after surface fault repair, or when quantification deviation information directly points to the power layer or higher, deep fault diagnosis is triggered. The fault propagation path reasoning unit is used to perform fault propagation path reasoning on the quantification deviation information of the power layer, control layer, and related layers, combined with the linkage relationship network, and outputs a fault hypothesis list sorted by probability. Deep faults include pump volumetric efficiency degradation, steering gear valve core wear, cylinder internal leakage, frame structure fatigue, and multi-system coupling failure. The fault diagnosis result generation unit is used to integrate the surface fault diagnosis results and deep fault diagnosis results to generate fault diagnosis results, which include fault level location and fault cause.
[0069] The fault type prediction and matching module 50, which performs control decomposition and feedback based on the fault diagnosis results, may further include: an immediate maintenance prompt generation unit for generating immediate maintenance prompts based on surface fault diagnosis results, including fault location guidance, on-site handling suggestions, and a list of required tools and spare parts; a predictive maintenance strategy generation unit for generating predictive maintenance strategies based on deep fault diagnosis results, including fault development trend prediction, suggested maintenance time windows, downtime impact assessment, and in-depth repair plans; and a forklift control strategy dynamic adjustment unit for dynamically adjusting the forklift control strategy according to the urgency of the fault diagnosis results, including speed-limited operation, steering angle limitation, power reduction mode, or forced safe shutdown.
[0070] The safety fault diagnosis system for the front and rear axle steering system of a forklift provided in this embodiment of the invention can execute the safety fault diagnosis method for the front and rear axle steering system of a forklift provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method.
[0071] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.
[0072] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application. In some cases, the actions or steps described in this application can be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
Claims
1. A safety failure diagnosis method for a forklift front and rear axle steering system, characterized by, include: Decompose the physical structure and linkage relationship of the front and rear axle steering system of the forklift, construct a twin mirror model, and divide the physical structure into execution layer, transmission layer, power layer, control layer and association layer according to the recognition complexity from low to high. Perform status alignment and quality assessment on forklift monitoring equipment to obtain high-confidence monitoring data, and establish a dynamic mapping relationship between the high-confidence monitoring data and the twin mirror model; Based on the dynamic mapping relationship, the high-confidence monitoring data is input into the twin mirror model, and forward propagation calculation is performed from low to high along the hierarchy to obtain the theoretical analytical state characteristics of each physical structure; Based on the high-confidence monitoring data, reverse decoupling and identification are performed from high to low along the hierarchy to obtain the actual analytical state characteristics of each physical structure. The actual analytical state characteristics are compared with the corresponding theoretical analytical state characteristics to identify quantification deviation information. Based on the quantified deviation information, fault type prediction and matching are performed sequentially from low to high level to identify fault diagnosis results. The fault diagnosis results include surface fault diagnosis results and deep fault diagnosis results. Adjustment decomposition feedback is performed based on the fault diagnosis results. This includes decomposing the physical structure and linkage relationships of the forklift's front and rear axle steering systems, and constructing a twin mirror model, including: The front and rear axle steering systems of the forklift are decomposed hierarchically to obtain the target structure, geometric parameters, physical properties and dynamic characteristics of each level. The levels are arranged in order of increasing recognition complexity as execution layer, transmission layer, power layer, control layer and association layer. Based on the geometric constraint motion relationships, dynamic linkage relationships, and rigid body mechanical coupling relationships of the hierarchical physical structure, lateral coupling relationships and longitudinal transmission relationships are established to obtain a linkage relationship network. The lateral coupling relationship is the parallel, series, or feedback connection relationship of the target structure within the same level; the longitudinal transmission relationship is the energy flow, signal flow, or material flow transmission relationship across levels. Based on the hierarchical physical structure and the interconnected network, the twin mirror model is constructed. The twin mirror model includes a forward calculation module and a reverse identification module. The forward calculation module is used to calculate the theoretical state of the upper layer based on the lower layer input, and the reverse identification module is used to invert the actual parameters of the lower layer based on the upper layer constraints.
2. The method for diagnosing safety faults in the front and rear axle steering systems of a forklift according to claim 1, characterized in that, The target structure of the execution layer is the front steering axle, the rear steering axle, and the steering wheel assembly, which is characterized by directly observable geometric motion states. The target structure of the transmission layer is the steering cylinder, tie rod, and steering knuckle arm, characterized by the mechanical displacement state transmitted by force. The target structure of the power layer is a steering pump, a fully hydraulic steering gear, and hydraulic pipelines, which are characterized by the energy conversion state of fluid dynamics. The target structure of the control layer is a steering valve core, a feedback mechanism, and an electronic control unit, which is characterized by the dynamic response state of the control logic. The target structure of the associated layer is the vehicle frame suspension, cargo center of gravity distribution, and ground contact coupling, which is characterized by multi-physics coupling state.
3. The method for diagnosing safety faults in the front and rear axle steering systems of a forklift according to claim 1, characterized in that, Perform status alignment and quality assessment on forklift monitoring equipment, including: Before operation, standard state data of the forklift is established, which includes the baseline response characteristics of each target structure in a healthy state. The standard state data is quantitatively mapped to the corresponding monitoring results of the forklift monitoring equipment to establish a monitoring mapping relationship. Based on the standard state data, the monitoring equipment is corrected and aligned according to the monitoring mapping relationship so that the monitoring data is consistent with the current actual state of the forklift. Quality assessment is conducted based on the consistency verification results, and high-confidence monitoring data is marked.
4. The method for diagnosing safety faults in the front and rear axle steering systems of a forklift according to claim 3, characterized in that, According to the aforementioned monitoring mapping relationship, the monitoring equipment is calibrated and aligned to reflect its status, including: Establish calibration conditions for forklift status and monitoring equipment. These calibration conditions include vehicle unloaded, stationary in a straight line, engine idling, and hydraulic oil temperature within a stable operating range. The calibration process of the monitoring equipment is triggered based on each calibration condition, and the consistency is verified by using standard state data and monitoring results. Based on the verification deviation, the calibration compensation parameters are calculated, and the monitoring equipment is calibrated and aligned.
5. A method for diagnosing safety faults in the front and rear axle steering systems of a forklift according to claim 2, characterized in that, A forward propagation calculation is performed from low to high levels to obtain the theoretical analytical state characteristics of each physical structure, including: The directly measurable data from the high-confidence monitoring data is used as boundary input to drive the forward calculation module of the twin mirror model; The calculation proceeds layer by layer from the lower to the higher levels along the longitudinal transmission relationship. The measured geometric quantities of the execution layer are used as input to calculate the theoretical force transmission state of the transmission layer; the calculation results of the transmission layer are used as input to calculate the theoretical energy state of the power layer; and so on, until the theoretical coupling state of the associated layer is obtained. Output the theoretical analytical state characteristics of each level, including theoretical geometric quantities, theoretical mechanical quantities, theoretical efficiency parameters, theoretical response characteristics, and theoretical coupling parameters.
6. A method for diagnosing safety faults in the front and rear axle steering systems of a forklift according to claim 5, characterized in that, Identify quantification bias information, including: Using the high-confidence monitoring data as observation constraints, the inverse identification module of the twin mirror model is driven to invert layer by layer from high level to low level along the vertical transmission relationship. The actual dynamic parameters of the control layer are identified by using the macroscopic response of the correlation layer as constraints; the actual efficiency parameters of the power layer are identified by using the identification results of the control layer as constraints; and so on, until the actual geometric state of the execution layer is obtained. The actual analytical state characteristics at each level are compared with the corresponding theoretical analytical state characteristics, and the deviation metric is calculated. The deviation metric is subjected to consistency verification and uncertainty assessment, and the quantified deviation information with confidence level is output.
7. A method for diagnosing safety faults in the front and rear axle steering systems of a forklift according to claim 6, characterized in that, Based on the quantified deviation information, fault type prediction and matching are performed sequentially from low to high level to identify fault diagnosis results, including: The quantitative deviation information between the execution layer and the transmission layer is directly matched with known fault modes. The known fault modes are surface faults, including abnormal tire pressure, insufficient steering fluid, excessive tie rod ball joint clearance, and poor lubrication. When the confidence level of the quantitative deviation information meets the preset threshold, the surface fault diagnosis result is output. When surface fault diagnosis cannot explain all quantization deviations, or when abnormal deviations still exist after surface fault repair, or when quantization deviation information directly points to the power layer and above, deep fault diagnosis is triggered. The quantitative deviation information of the power layer, control layer and related layer is combined with the linkage relationship network to perform fault propagation path reasoning, and output a fault hypothesis list sorted by probability. Deep faults include pump volumetric efficiency degradation, steering valve core wear, cylinder internal leakage, frame structure fatigue and multi-system coupling failure. By combining the surface fault diagnosis results and the deep fault diagnosis results, a fault diagnosis result is generated, which includes the fault level location and the fault cause.
8. A method for diagnosing safety faults in the front and rear axle steering systems of a forklift according to claim 7, characterized in that, Based on the fault diagnosis results, control decomposition feedback is performed, including: Based on the surface fault diagnosis results, generate real-time maintenance prompts, including fault location guidance, on-site handling suggestions, and a list of required tools and spare parts; Based on the results of in-depth fault diagnosis, predictive maintenance strategies are generated, including fault development trend prediction, recommended maintenance time windows, downtime impact assessment, and in-depth overhaul plans. Based on the urgency of the fault diagnosis results, the forklift control strategy is dynamically adjusted, including speed-limited operation, steering angle restriction, power reduction mode, or forced safety stop.
9. A safety fault diagnosis system for the front and rear axle steering systems of a forklift, characterized in that, The system is used to implement a safety fault diagnosis method for a forklift front and rear axle steering system according to any one of claims 1-8, the system comprising: The twin mirror model construction module is used to decompose the physical structure and linkage relationship of the front and rear axle steering system of a forklift, construct a twin mirror model, and divide the physical structure into execution layer, transmission layer, power layer, control layer and association layer according to the recognition complexity from low to high. The high-confidence monitoring data acquisition module is used to perform state alignment and quality assessment on forklift monitoring equipment, obtain high-confidence monitoring data, and establish a dynamic mapping relationship between the high-confidence monitoring data and the twin mirror model. The forward propagation calculation module is used to input the high-confidence monitoring data into the twin mirror model based on the dynamic mapping relationship, and perform forward propagation calculation from low to high along the hierarchy to obtain the theoretical analytical state characteristics of each physical structure. The reverse decoupling identification module is used to perform reverse decoupling identification from high to low along the hierarchy based on the high confidence monitoring data, obtain the actual analytical state characteristics of each physical structure, compare the actual analytical state characteristics with the corresponding theoretical analytical state characteristics, and identify the quantification deviation information. The fault type prediction and matching module is used to predict and match fault types in order of increasing level according to the quantified deviation information, identify fault diagnosis results, including surface fault diagnosis results and deep fault diagnosis results, and perform regulation, decomposition and feedback based on the fault diagnosis results.
