A fault modeling diagnosis method based on multi-value function function

Abstract

The application discloses a kind of fault modeling diagnosis methods based on multi-value function function, comprising the following steps: step 1: radar function analysis and signal classification;Step 2: element definition of fault diagnosis model;Step 3: the construction of system level multi-task long chain model;Step 4: modeling of unit level explicit function function;Step 5: modeling of unit level implicit function function;Step 6: diagnosis analysis and decision suggestion.The present application is through the carding decomposition of radar operation mode, the explicit mechanism model is constructed to the function function of clear signal relationship, the implicit data model is constructed to the function function of fuzzy signal relationship, based on the comprehensive diagnosis of the transmission characteristics of multi-task long chain, the testability design is strengthened, the output result of this method is beneficial to user to improve fault detection and positioning efficiency, while helping research and development department to scientifically analyze weak link of reliability, optimize maintenance support decision, and as the basis for driving design end to improve new equipment.

Description

Technical Field

[0001] This invention relates to the field of integrated support engineering technology for radar equipment, and specifically to a fault modeling and diagnosis method based on multi-valued function. Background Technology

[0002] Radar's fault diagnosis capabilities impact operational effectiveness. The breadth of diagnostic coverage and positioning accuracy are closely related to the system's operating mode, the signal transmission paths of each module, and their conversion principles. To help users accurately locate and quickly troubleshoot faults, it is necessary to extract valuable information from increasingly complex fault phenomena, continuously improve radar testing and diagnostic design capabilities, and thereby enhance on-site operational support capabilities.

[0003] In the design process of existing radar systems, the following model is generally adopted: 1) Defining equipment states using a binary approach of fault and normal conditions; 2) Modeling according to the system-subsystem-LRU (module) responsibility hierarchy; 3) Enumerating module fault modes based on experience and directly summing upwards. For a long time, insufficient integration with telecommunications design and inadequate utilization of a large amount of physical knowledge and statistical data have led to radar fault diagnosis exhibiting characteristics of incomplete detection, incorrect isolation, and deviation from established principles. Summary of the Invention

[0004] To address the above problems, this invention proposes a fault modeling and diagnosis method based on multi-valued function, comprising the following steps:

[0005] Step 1: Radar Function Analysis and Signal Classification: Analyze the complete working profile of the radar system from deployment to relocation, sort out the sub-functions of each unit related to the system functions, and classify them according to the logical characteristics of the input and output signals of each function to ensure that the subsequent fault model accurately reflects the radar operating status.

[0006] Step 2: Element definition of the fault diagnosis model: Redefine the elements of the fault diagnosis model based on the functions and signal categories determined in Step 1;

[0007] Step 3: Construction of the system-level multi-task long link model: Referencing the signal transmission path, build a system fault model; based on the time-sharing independence of tasks, build single-task long link models one by one; combining the reusability of unit sub-functions, splice the single-task long link models into a fault model of the entire system's multi-task long link; each single-task long link includes several sequentially connected nodes, and each node includes working input signals, working functions, working output signals, operating condition input signals, operating condition functions, operating condition output signals, and internally generated operating condition signals;

[0008] Step 4: Modeling of explicit functional functions at the unit level: For nodes in the link of Step 3 where the working condition output signal only reports parameters and the state is unclear, if the functional function corresponding to the function has an analytical form, the input-output relationship is clarified through the functional function, and then the state corresponding to different ranges of the working output signal is obtained. The corresponding three-valued logic is agreed upon as follows: 0 represents the normal state; 1 represents the failure state; 2 represents the defective state.

[0009] Step 5: Modeling of unit-level implicit function: For nodes in the link where the function corresponding to the function in Step 3 has no analytical form, effective statistical data is filtered through preprocessing measures. Combined with response surface methodology and Bayesian principle modeling, the states corresponding to different ranges of the working output signal are obtained. The corresponding three-valued logic is defined as follows: 0 represents normal state; 1 represents failure state; 2 represents defective state.

[0010] Step 6: Diagnostic Analysis and Decision Recommendations: For nodes with complete input and output information, perform module self-diagnosis based on multi-valued logic operations; for nodes with insufficient input and output, refer to the D matrix and diagnostic tree algorithm, perform comprehensive diagnosis of the link-related node information in order from near to far, and propose assurance decision recommendations based on the impact of the fault.

[0011] Furthermore, the steps for building the single-task long-link model in step 3 are broken down as follows:

[0012] Step 3.1: Define the operation rules for the combinational logic values ​​of the working input signals, and perform preprocessing before the function is implemented;

[0013] Step 3.2: Based on the results of Step 3.1, define the logical operation rules of the working function when the combination of working input signals generates the working output signal;

[0014] Step 3.3: Based on the results of Step 3.2, define the logical operation rules for the function of the operating condition when the operating condition input signal, the operating condition output signal and the internally generated operating condition signal are spliced ​​together to form the operating condition output signal.

[0015] Furthermore, based on the logical characteristics of each function's input and output signals, the signals are categorized into three types: power supply signals, control / RF / data signals, and motion / medium signals.

[0016] Furthermore, the model elements include signal elements, functional elements, path elements, and frame elements; signal elements include working signals and operating condition signals, and working signals need to be "copied" into operating condition signals for integration into BIT message transmission; it is agreed that operating condition signals have four logical values: operating condition signal 0 represents normal state; operating condition signal 1 represents failure state; operating condition signal 2 represents bad state; operating condition signal 3 represents rest state, corresponding to invalid or null values ​​in BIT message content; functional elements include working functions and operating condition functions; it is agreed that the signal elements corresponding to functional elements are also defined as having four logical values: 0 represents normal state; 1 represents failure state; 2 represents bad state; 3 represents ambiguous state, indicating that the current information is insufficient to determine whether the function is in state 0, 1, or 2; path elements include working paths and operating condition paths.

[0017] Furthermore, the diagnostic steps in step 6 are broken down as follows:

[0018] Step 6.1: Determine the power supply signal;

[0019] Step 6.2: Based on the information from the front-end nodes, back-end nodes, and multi-task status, determine the control / RF / data signals;

[0020] Step 6.3: Determine the motion / medium signal and complete the non-working output state;

[0021] Step 6.4: Determine the degree of impact of the fault and propose decision-making recommendations.

[0022] Furthermore, the construction of a single-task long-link model includes the following steps:

[0023] For a simplified unit-level model, the basic rules for combining the logic values ​​of the working input signals are defined as shown in Table 2.

[0024] Table 7. Logic Operation Rules for Combinations of Working Input Signals

[0025]

[0026]

[0027] The working input signals are combined and generated into working output signals by the working function. Considering that the actual logic value of the working function has only three cases: 0, 1, and 2, the logic rules shown in Table 3 are defined.

[0028] Table 8 Logical Operation Rules for Functional Roles

[0029]

[0030] In Table 3, a message-based input combinational logic value of 3 indicates a pause, meaning the module is not yet powered or the control command information requires no response. When a response is triggered, the rule of taking 2 from 2 applies.

[0031] The operating condition input signal, the working output signal, and the internally generated operating condition signal are spliced ​​together to form the operating condition output signal. The operating condition function only collects and splices signals, does not rely on control signal information, and does not affect radio frequency / data signals. The only input signals that affect the operating condition function are power supply and communication, and their rules are defined as shown in Table 4.

[0032] Table 9 Logical operation rules for the function of operating conditions

[0033]

[0034] Furthermore, when concatenating the single-task long-link model into a fault model of the entire system's multi-task long-link model, the single-task long-link model is first built; then, the reused functions are further concatenated and merged as common nodes, and the identifier Case i is added to facilitate differentiation and help with information complementarity in subsequent diagnosis; where case i indicates that the reused functions are respectively in the i-th task, and i is a positive integer.

[0035] Furthermore, step 6.1 specifically includes:

[0036] Based on the power supply signal, determine whether the link is in working condition. If it is not in working condition, both the input and output operating condition signals are 3, and the corresponding function diagnosis is 3.

[0037] Furthermore, step 6.2 specifically involves:

[0038] Based on the control, radio frequency, and data signals, confirm the current operating mode, trigger the corresponding fault link model and unit-level explicit or implicit functional function model, and classify them according to the known input and output information as follows:

[0039] When the direct inputs and outputs of the current link node are known, the logical operation rules shown in Table 6 are followed. In Table 6, × indicates that this situation does not exist.

[0040] Table 10. Diagnostic rules with complete input and output

[0041]

[0042] When the direct output of a node is known but the input combination is unknown, the signal of the link front-end node is meaningless for diagnosis and follows the logic operation rules shown in Table 7.

[0043] Table 11 Diagnostic rules for outputting unknown values

[0044]

[0045] When the direct input combination of a node is known but the output is unknown, the diagnostic output signal of the back-end node of the dependent link follows the logical operation rules shown in Table 8.

[0046] Table 12 Diagnostic rules for unknown inputs

[0047]

[0048] If the input combination or output of the current Case i is unknown, but the information in other tasks is complete and the diagnosis is not ambiguous, the diagnostic results of other Cases shall replace the input combination or output of the current task.

[0049] Furthermore, step 6.3 specifically involves:

[0050] Based on the current working mode, the range of motion and medium signals input by non-working functions is determined, and they are classified according to the interval of their direct output values.

[0051] Furthermore, step 6.4 specifically includes:

[0052] Any malfunctions affecting current safety, including leaks, sparking, poor cooling, and mechanical failures, should be repaired immediately.

[0053] Faults affecting the current task, including those in signal processing such as the receiving and transmitting boards being inoperable, performance falling below the threshold, or other thresholds of redundant equipment, should be repaired immediately.

[0054] For failures that affect future missions, including mechanical equipment with fixed aging / damage cycles, single-path failures in dual-backup systems, or multi-redundant equipment whose performance is predicted to degrade to an unacceptable level, maintenance should be carried out regularly or as appropriate.

[0055] For faults that affect future lifespan, including equipment parameters such as temperature and pressure that are continuously out of tolerance but not exceeding safe values, maintenance should be carried out regularly or as appropriate.

[0056] For faults with insignificant impact, including individual non-safety faults in multi-redundant equipment and faults in monitoring / simulation equipment, repairs should be carried out as appropriate.

[0057] Compared with the prior art, the advantages of this invention include:

[0058] 1) Scientific measurement points and accurate positioning

[0059] By analyzing the relationship between signal transmission and conversion, and closely integrating with the scientific selection of test points in telecommunications design, potential redundancy points are reduced compared to experience-based design, and key functional points are highlighted to supplement omissions and refine isolation.

[0060] 2) The model is scientific and the information is comprehensive.

[0061] Compared to the "pyramid" structure model, this approach fully utilizes knowledge and statistical information to build a system-level model based on multi-task long-linkage and a unit-level model based on explicit mechanisms and implicit data, thus avoiding the decoupling of logical operations from working modes.

[0062] 3) Scientific diagnosis and sound decision-making

[0063] Based on multi-valued logic rules, module-independent diagnosis and link-related diagnosis are performed. Scientific assurance suggestions are proposed according to the degree of fault impact, reducing the probability of under-maintenance and over-maintenance, and driving the continuous optimization of the six-character design of new equipment. Attached Figure Description

[0064] Figure 1 This is a flowchart illustrating the implementation of the present invention.

[0065] Figure 2 This is a schematic diagram of model elements in an embodiment of the present invention.

[0066] Figure 3 This is a schematic diagram of a single-task long-link model in an embodiment of the present invention.

[0067] Figure 4 This is a schematic diagram of a multi-task long-link model in an embodiment of the present invention.

[0068] Figure 5 This is the array pattern of the sinusoidal space in an embodiment of the present invention.

[0069] Figure 6 This is a schematic diagram of the transmission mechanism in an embodiment of the present invention. Detailed Implementation

[0070] The purpose of this invention is to provide a fault modeling and diagnosis method based on multi-valued function, which makes full use of signal relationships and statistical data information to explore the analytical logic and potential correlation between telecommunications functions and equipment status, and to achieve scientific modeling and accurate diagnosis.

[0071] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0072] like Figure 1 As shown, the technical solution of this embodiment specifically includes the following steps:

[0073] I. Radar Functional Analysis and Signal Classification

[0074] Analyze the complete operational profile of the radar system from deployment to relocation, sort out the sub-functions of each unit with interconnected system functions, and classify them according to the logical characteristics of the input and output signals of each function to ensure that the subsequent fault model accurately reflects the radar's operating status.

[0075] First, following the principle of actual transfer and conversion between devices, a functional analysis of the flattened structure is conducted:

[0076] 1) Organize all system functions and their temporal relationships according to the workflow triggered by the operation, such as:

[0077] Power station power-on → Servo power-on;

[0078] Servo power-on → Leveling → Unlocking → Lifting / lowering;

[0079] 2) Analyze the sub-functions of each module. Each system function has several relatively independent paths, which need further refinement. Taking leveling as an example, under the premise that the device is powered on normally, pressing the button triggers three paths simultaneously:

[0080] Control valves are used to open valve groups, primarily for control signal conversion.

[0081] Starting the pump aims to operate the oil pump in two ways: firstly, by converting control signals, and secondly, by converting electrical energy into mechanical energy.

[0082] Leveling, the purpose of which is to extend and retract the outriggers, involves only the passive transmission of hydraulic oil.

[0083] 3) Describe the complete function's operation process according to the transmission path;

[0084] Connection format: Module|Function|Signal → Module|Function|Signal →…;

[0085] Encoding format: Modules M1, M2…Mn; Functions G1, G2…Gn; Signals S1, S2…Sn;

[0086] Taking a typical working profile as an example, the functional analysis after symbol simplification is shown in the table below.

[0087] Table 13: Functional Analysis and Signal Classification Cases (Numbers omitted)

[0088]

[0089]

[0090] Based on the table above, and considering the characteristics of the bonding operation and testing, the signals are divided into:

[0091] ●Power supply signal

[0092] The power supply signal is generated by the power supply equipment, converted and rectified by the power distribution equipment, and consumed by the power-consuming equipment as mechanical energy or heat energy. It does not involve the utilization of the signal information.

[0093] ●Control\RF\Data Signals

[0094] Control signals are generated by operations and converted into timing instructions by control equipment to trigger mode switching and parameter loading. Radio frequency and data signals, on the other hand, utilize the information contained in the control signals to synthesize electromagnetic waves of the expected waveform by radio frequency equipment. The reflected echoes from the target are received and converted into data for display on the interface by processing equipment, which are the main operation signals of the system.

[0095] ●Motion / Media Signal

[0096] Motion signals are generated by motor-driven or hydraulically assisted motion, involving the conversion of electrical energy into mechanical energy; medium signals include temperature, flow rate, and pressure. These two types of signals do not have direct information in the operational chain but are indirectly transmitted based on sensor parameterization.

[0097] II. Element Definitions of the Fault Diagnosis Model

[0098] Based on the functions and signal categories determined in step one, the element categories, forms, and functions of the fault diagnosis model are redefined according to the principle of easy identification. This is conducive to preserving the independence of functions for node logic conversion, while standardizing the link logic operation for homogeneous factors.

[0099] Considering the requirements of logical operations, we extract common causes from the cases and define four types of model elements, such as... Figure 2 As shown.

[0100] 1) Signal elements

[0101] ① Operating signals: Signals generated by the radar system during fault diagnosis modeling.

[0102] The square ■ symbol Wi represents the working input signal, which is the working signal input from the outside.

[0103] The square symbol Wo (■) represents the working output signal, which is the working signal output to the outside.

[0104] ② Operating condition signals, signals generated during fault diagnosis by the fault diagnosis model.

[0105] The square ■ in the symbol Bi represents the working condition input signal, which is the working condition signal input from the outside;

[0106] The square ■ in the symbol Bo represents the operating condition output signal, which is the operating condition signal output to the outside;

[0107] The square symbol ■ of Ei represents the internally generated operating condition signal.

[0108] Operating signals are not directly "visible" values; they need to be "copied" into operating condition signals and integrated into BIT messages for transmission. Unlike the traditional two states of fault and normal operation, operating condition signals are defined to have four logical values:

[0109] A status signal of 0 indicates a normal operating condition.

[0110] A condition signal of 1 indicates a failure state;

[0111] A working condition signal of 2 indicates a poor condition, meaning that the working performance has decreased but the device has not yet failed.

[0112] The operating condition signal is 3, which represents a resting state. For example, the BIT will report an invalid value of 255 or a null value.

[0113] 2) Functional elements

[0114] First, the functions are subdivided according to their input-output relationships.

[0115] ①N collection 1

[0116] Note: The splicing process transmits the signal to the backend node, but the signal content remains unchanged;

[0117] Applicable to: Operating condition signals.

[0118] ②Take K from N

[0119] Note: When several identical signals meet the threshold number, output the result, while keeping the signal content unchanged;

[0120] Applicable to: situations where there are 2 backups for 1 or 3 backups for 2 in the working signal.

[0121] ③N terms count 1

[0122] Note: All mathematical operations and physical conversions have altered the signal content;

[0123] Applicable to: general functions in working signals.

[0124] Then, consider the mapping of signal elements, using two types of elements to describe the function.

[0125] ①Working functions

[0126] The circular ● symbol Wg covers the functions of power supply, control, radio frequency, data signal generation and processing.

[0127] ②BIT (Balance of Intake) Function

[0128] The circular ● symbol Bg represents the function of measuring motion / medium signal and monitoring the "coupling" of Wo.

[0129] The functional status is not "visible" and needs to be determined in reverse based on the known operating condition signals. Its logical value and signal element are also defined as four categories: 0, 1, 2, and 3. The difference is that: 3 for the operating condition signal represents rest, corresponding to the invalid value / empty value of the BIT message content; while 3 for the function represents ambiguity, that is, the current information is insufficient to determine whether the function is in a 0, 1, or 2 state.

[0130] 3) Path elements

[0131] ①Work path

[0132] Solid line with arrowhead →; Figure 2 In the diagram, the working process of the radar system is from left to right, which is consistent with the modeling process of the fault diagnosis model. That is, the working input signal Wi of the node is processed by the working function Wg to obtain the working output signal Wo.

[0133] ②Working path

[0134] Dashed line with dots -·; Figure 2 In the fault diagnosis model, the diagnostic process is to deduce from right to left along the dotted line direction, that is, to deduce the working function Wg from the node's working condition output signal Bo, working condition function Bg, internal self-generated working condition signal Ei, and working condition input signal Bi.

[0135] 4) Frame element

[0136] An unfilled square □ represents a node to which the function or signal within the box belongs.

[0137] III. Construction of a System-Level Multi-Task Long-Link Model

[0138] Based on the signal transmission path, a system fault model is constructed. Given the time-sharing independence of tasks, a single-task long-link model is first built line by line; then, considering the reusability of unit sub-functions, a fault model of the entire system's multi-task long-link is constructed.

[0139] 1) Single-task long-link model

[0140] For universality, the boxed symbols [i], [o], and [g] represent the input / output signals and function sets corresponding to each node, such as... Figure 3 As shown, Figure 3 In this context, L, C, and R represent the front-end node, the current node, and the back-end node, respectively. More specifically, [i] includes the working input signal [Wi], the operating condition input signal [Bi], and the internally generated operating condition signal [Ei]; [o] includes the working output signal [Wo] and the operating condition input signal [Bo]; and [g] includes the working function [Wg] and the operating condition function [Bg]. The steps for transferring and converting functions are broken down as follows.

[0141] ① Working input signal combination

[0142] For the simplified unit-level model, the basic rules for combining the logic values ​​of the working input signals are defined, as shown in Table 2. For cases where these combination rules do not apply, subsequent modeling is performed using unit-level function functions (including unit-level explicit function functions and unit-level implicit function functions).

[0143] Table 14 Logic Operation Rules for Combinations of Working Input Signals

[0144]

[0145] ② The working input signals are combined and processed by the working function to generate the working output signal.

[0146] Considering that the actual logical value of the function has only three cases: 0, 1, and 2, the logical rules shown in the table below are defined.

[0147] Table 15 Logical Operation Rules for Functional Roles

[0148]

[0149]

[0150] In the table, when the input combinational logic value based on the message is 3 (pause), it may be because the module has not yet been powered on, or the control command information requires no response at this time. When a response is triggered, the rule of taking 2 out of 2 is followed.

[0151] ③ The operating condition input signal, the operating output signal, and the internally generated operating condition signal are spliced ​​together to form the operating condition output signal.

[0152] The operating condition function only involves signal acquisition and splicing, does not rely on control signal information, and does not affect RF / data signals; it is an N-to-1 type function. The only input signals affecting the operating condition function are power supply and communication signals, and their definitions are shown in the table below.

[0153] Table 16 Logical Operation Rules for Operating Conditions and Functions

[0154]

[0155] 2) Fault Model for Multi-Task Long Link

[0156] Pay attention to the common functions triggered under different work mode settings.

[0157] by Figure 4 For example, when building a single-task long-link model, two paths are generated. The transmission and conversion process of the signal through any function is still divided into three steps. Furthermore, the reused functions are used as common nodes to splice and merge them, and the identifiers Case1 and Case2 are added to facilitate differentiation and help with information complementarity in subsequent diagnosis.

[0158] IV. Modeling of Entity-Level Explicit Functions

[0159] For nodes in the third step of the link where the working condition output signal only reports parameters (only reports parameter values ​​instead of logical values) and the status is unclear (the logical value is unclear), if the function has an analytical form, the input-output relationship can be clearly defined through the function, and then the status corresponding to different ranges of the working output signal can be obtained. The corresponding three-valued logic is agreed upon as follows: 0 represents the normal state; 1 represents the failure state; 2 represents the defective state.

[0160] Some unit signals only report parameters without logic values. The reasons include: the parameter range corresponding to four-valued logic is different under different tasks; whether the signal is normal or not does not depend on its own value, but is affected by coupling with other parameters.

[0161] Taking the external radiation function of a certain product as an example, it is not advisable to use the rules in step three to model the antenna elements separately. Instead, a spatial virtual node should be set up to construct a functional function model to describe the actual composite function of the element output in the spatial domain.

[0162] The analytical expression for the far-field radiation pattern of the array is:

[0163]

[0164] r i It is the position vector of the i-th oscillator relative to the origin of the coordinate system.

[0165] Unit vector along the observation direction

[0166] This is the vector pattern of the oscillator;

[0167] R is the distance from the origin of the coordinate system to the observation point P((x,y,z));

[0168] c i It is the weighted (voltage or current) coefficient of the oscillator.

[0169] in, These are the variables of each antenna element (which will degrade and deviate from the wave control), r i , R,c i It is a theoretically geometrically definite quantity or a controllable definite quantity. The effect of E(r) on the system's power, accuracy, and resolution varies depending on the task; the impact of concentrated or discrete deviation regions on the system also differs, requiring separate modeling.

[0170] 1) Focus on missions that emphasize power.

[0171] Gain G in the power equation t G r The shape parameter Z of the array pattern m Impact, Z m Related to the main lobe peak value and noise floor, different θ values ​​are calculated based on array monitoring. Given E(r), further find the maximum value D and the noise floor N of E(r), and calculate the difference:

[0172]

[0173] 2) Tasks that focus on accuracy

[0174] Azimuth and elevation angles σ in the accuracy equation θ , The shape parameter Z of the array pattern d Impact, Z d It is related to the angle where the maximum value of the main lobe is located. When modeling, locate the angular coordinates corresponding to the peak value of E(r), and then calculate the difference with the set angle:

[0175]

[0176] 3) Focus on tasks requiring high resolution

[0177] θ in the resolution equation 3dB , The shape parameter Z of the array pattern r Impact, Z r Related to the main lobe width, during modeling, locate the angular coordinates 3dB below the peak value of E(r), and then calculate the difference:

[0178]

[0179] 4) Tasks that simultaneously focus on power, accuracy, and resolution

[0180] In other cases, the shape parameter Z needs to be considered. m Z d Z r Three types of factors are modeled with certain weights:

[0181]

[0182] Finally, based on the task metrics and the resulting shape parameter range, a three-valued logic is defined:

[0183]

[0184] The shape of the array pattern in sinusoidal space is as follows: Figure 5 As shown.

[0185] V. Modeling of Unit-Level Implicit Functions

[0186] For nodes in the third step of the link where the functional function has no analytical form, effective statistical data is filtered through preprocessing measures such as cleaning and merging distributed sources. Combined with response surface methodology and Bayesian principle modeling, the states corresponding to different ranges of the working output signal are obtained, and corresponding three-valued logic is agreed upon: 0 represents the normal state; 1 represents the failure state; and 2 represents the defective state.

[0187] For nodes in the message that have no direct state and no analytical expression for their principle, implicit function descriptions of their operation process need to be established. For example... Figure 6 The servo device shown has a mechanism that, after assembly, is constrained by space and its motion state cannot be directly observed by sensors. During the degradation process, the impact of the abrupt changes in the clearance of the kinematic pairs and the complex particle wear intensifies, and theoretical formulas cannot cover these factors.

[0188] First, define the motion limit state function:

[0189] S=g(x=g(x1,x2,…,x) n )

[0190] In the formula, S can be displacement, velocity, acceleration, or other motion parameters of interest to the system, and x = [x1, x2, ..., x...]. n ] T This represents n observable and statistically significant uncertain variables related to S, such as the vibration amplitude / frequency of the outer shell in several directions, noise amplitude / frequency, etc. These are not necessarily direct influencing factors of the mechanism's motion.

[0191] Mapping product performance metrics to functional levels, we define the state where the function just reaches a threshold as the limit state. For traditional binary logic, mathematically, S>0 indicates a reliable state where the intended function is achieved, and S<0 indicates a fault state; therefore, S=0 is the limit state. Similarly, for functions conforming to [0,1,2] ternary logic, there are two transition points: from 0 (normal) to 2 (defective) and from 2 (defective) to 1 (failure). Therefore, we define two limit state functions:

[0192] S0=g(x=g(x1,x2,…,x) n ),for 0→2

[0193] S1=h(x=h(x1,x2,…,x) n ), for 2→1

[0194] The representation quantities do not have a real physical formula; they can only be approximated as closely as possible to the real S0 and S1 through data fitting.

[0195] 1) Data Filtering

[0196] Data screening was performed, independent factors were compared, and far-field vibrations were excluded to avoid variable coupling and overfitting.

[0197] 2) Response surface model

[0198] Taking both accuracy and efficiency into consideration, a quadratic polynomial response surface is used to fit S0 and S1:

[0199]

[0200]

[0201] Where A0 = [a, b1, ..., c i A1 = [d,e]; i ,…,f i [,…] represents two sets of 2n+1 undetermined coefficients each. When the number of collected motion states and corresponding variable samples [S(X)] exceeds the coefficient dimension, the least squares method is used to solve the system of equations, and the vector of undetermined coefficients can be fitted.

[0202]

[0203]

[0204] 3) Bayes model

[0205] Considering the limitations of response surface models, namely when variables have states but no numerical values ​​or when the current number of samples is less than the number of undetermined coefficients and cannot be solved, the corrective power of Bayes' principle is used to supplement the implicit function model.

[0206] The following table shows the results of recording the operational status and variables of a certain organization.

[0207] Table 17 Record of Mechanism Motion State and Variables

[0208]

[0209] ① Define a random variable and obtain its likelihood function

[0210] Let A i (i = 1, 2, 3) represents the motion state of the mechanism, and θ represents its uncertain response; B is the observable event related to θ, such as X-axis vibration and Y-axis vibration, and its independent random variable form is x = [x1, x2]. T The joint distribution P(x; θ) is equivalent to the conditional probability distribution P(x|θ) of x with respect to θ, which is also equal to P(x1|θ)P(x2|θ). The likelihood function can be obtained as follows:

[0211] P(B|A1)=P(x1|A1)P(x2|A1)=0.05×0.05=0.0025

[0212] P(B|A2)=0.06

[0213] P(B|A3)=0.275

[0214] ② Determine the prior distribution

[0215] The prior distribution can be fitted using motion data or test results from similar products under approximate working conditions:

[0216] P(A1)=0.6; P(A2)=0.3; P(A3)=0.1

[0217] ③ Solve for the posterior distribution using Bayes' principle

[0218] Based on Bayes' formula for maximum likelihood estimation, we can obtain:

[0219]

[0220] P(A2|B)=0.38

[0221] P(A3|B)=0.59

[0222] The results show that when vibrations occur simultaneously in the X and Y directions, the probability of the mechanism's motion state being 1 (failure) is the highest.

[0223] Define event C as vibration in only the X or Y direction, and event D as no vibration. Further, we can obtain:

[0224]

[0225] The results show that when only unidirectional vibration occurs, the probability of the mechanism's motion state being 2 (poor) is the highest; when there is no vibration, the probability of the mechanism's motion state being 0 (normal) is the highest.

[0226] VI. Diagnostic Analysis and Decision Recommendations

[0227] For nodes with complete input and output bit information, module self-diagnosis is performed based on multi-valued logic operations; for nodes with insufficient input and output, the information of the linked nodes is comprehensively diagnosed in order from near to far, referring to the D matrix and diagnostic tree algorithm, and protection decision suggestions are proposed according to the impact of the fault.

[0228] Diagnostic analysis is the inverse process of modeling, and the specific steps are as follows:

[0229] 1) Determine the power supply signal

[0230] Based on the power supply signal, first determine whether the link is in working condition. If it is not working, the logic values ​​of both the working condition input signal and the working condition output signal are 3, and the corresponding working condition function diagnosis is 3: fuzzy.

[0231] 2) Determine control / RF / data signals

[0232] Based on the control, radio frequency, and data signals, confirm the current operating mode, trigger the corresponding fault link model and unit-level functional function model (including unit-level explicit functional function model and unit-level implicit functional function model), and classify them into the following categories according to the known input and output information.

[0233] ●When the direct inputs and outputs of the current link node are known, the logical operation rules shown in Table 6 shall be followed.

[0234] Table 18 Diagnostic Rules for Complete Input and Output

[0235]

[0236] In the table, × indicates that this situation does not exist.

[0237] ● When the direct output of a node is known, but the combination of inputs is unknown.

[0238] At this point, the signals from the front-end nodes of the link are meaningless for diagnosis and follow the logical operation rules shown in Table 7.

[0239] Table 19 Diagnostic rules for outputting unknown information

[0240]

[0241] ● When the direct input combination of a node is known, but the output is unknown.

[0242] At this time, the output signals of the diagnostic dependency link backend nodes follow the logical operation rules shown in Table 8.

[0243] Table 20 Diagnostic rules for unknown inputs

[0244]

[0245] ●Multi-task information joint diagnosis

[0246] If the input combination or output is unknown in the current Case i task, but the information is complete and the diagnosis is clear in other tasks, the diagnosis results of other cases are temporarily stored in the system. When switching back to Case i, the clear results will still be displayed.

[0247] 3) Determine the motion / medium signal

[0248] Based on the current working mode, determine the range of motion and medium signals in non-working function inputs, and classify them according to the interval of their direct output values.

[0249] 4) Determine the degree of impact of the failure and propose decision-making recommendations:

[0250] ●Affecting current security

[0251] If there are issues such as leakage, sparking, poor cooling, or mechanical failure, repairs should be carried out immediately.

[0252] ● Affects the current task

[0253] If the signal processing receiving and transmitting boards are not working; the power, accuracy, or other performance characteristics have fallen below the threshold (multiple non-safety-related faults in multi-redundant equipment); or other thresholds of multi-redundant equipment, repair should be carried out immediately.

[0254] ●Impact on future missions

[0255] Mechanical equipment with a fixed aging / damage cycle; dual-backup single-path failure; or multi-redundant equipment whose performance is predicted to degrade to an unacceptable level; should be maintained regularly or as needed.

[0256] ●Affects future lifespan

[0257] If equipment parameters such as temperature and pressure consistently exceed tolerances but do not exceed safe limits, maintenance should be performed regularly or as needed.

[0258] ●The impact is not obvious

[0259] Individual non-safety faults in multi-redundant equipment and faults in monitoring / simulation equipment should be repaired as appropriate.

[0260] This concludes the fault modeling and analysis process for the entire system.

[0261] This invention decomposes radar operating modes, constructs explicit mechanistic models for functional functions with clear signal relationships, and implicit data models for functional functions with ambiguous signal relationships. Based on the transmission characteristics of multi-task long-link systems, it performs comprehensive diagnosis and enhances testability design. The output of this method helps users improve fault detection and location efficiency, while also assisting research and development departments in scientifically analyzing reliability weaknesses, optimizing maintenance and support decisions, and serving as a basis for driving design improvements to new equipment.

[0262] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A fault modeling and diagnosis method based on multivalued function, characterized in that, Includes the following steps: Step 1: Radar Function Analysis and Signal Classification: Analyze the complete working profile of the radar system from deployment to relocation, sort out the sub-functions of each unit related to the system functions, and classify them according to the logical characteristics of the input and output signals of each function to ensure that the subsequent fault model accurately reflects the radar operating status. Step 2: Element definition of the fault diagnosis model: Redefine the elements of the fault diagnosis model based on the functions and signal categories determined in Step 1; Step 3: Construction of the system-level multi-task long link model: Referencing the signal transmission path, build a system fault model; based on the time-sharing independence of tasks, build single-task long link models one by one; combining the reusability of unit sub-functions, splice the single-task long link models into a fault model of the entire system's multi-task long link; each single-task long link includes several sequentially connected nodes, and each node includes working input signals, working functions, working output signals, operating condition input signals, operating condition functions, operating condition output signals, and internally generated operating condition signals; Step 4: Modeling of explicit functional functions at the unit level: For nodes in the link of Step 3 where the working condition output signal only reports parameters and the state is unclear, the functional function corresponding to the function has an analytical form. The input-output relationship is clarified through the functional function, and then the state corresponding to different ranges of the working output signal is obtained. The corresponding three-valued logic is agreed upon, namely: 0, representing the normal state; 1 represents a failure state; 2 represents an unfavorable condition; Step 5: Modeling of unit-level implicit function: For nodes in the link where the function corresponding to the function in Step 3 has no analytical form, effective statistical data is filtered through preprocessing measures. Combined with response surface and Bayes principle modeling, the state corresponding to different ranges of the working output signal is obtained. The corresponding three-valued logic is defined as follows: 0 represents the normal state. 1 represents a failure state; 2 represents an unfavorable condition; Step 6: Diagnostic analysis and decision-making suggestions: For nodes with complete input and output information, perform module self-diagnosis based on multi-valued logic operations; For nodes with limited input and output, referencing the D matrix and diagnostic tree algorithm, comprehensively diagnose the information of the linked nodes in order from near to far, and propose safeguard decision suggestions based on the impact of the fault. The steps for building the single-task long-link model in step 3 are broken down as follows: Step 3.1: Define the operation rules for the combinational logic values ​​of the working input signals, and perform preprocessing before the function is implemented; Step 3.2: Based on the results of Step 3.1, define the logical operation rules of the working function when the combination of working input signals generates the working output signal; Step 3.3: Based on the results of Step 3.2, define the logical operation rules for the function of the operating condition when the operating condition input signal, the operating condition output signal and the internally generated operating condition signal are spliced ​​together to form the operating condition output signal; Based on the logical characteristics of each function's input and output signals, the signals are classified into three categories: power supply signals, control / RF / data signals, and motion / medium signals. The model includes signal elements, functional elements, path elements, and frame elements; signal elements include working signals and operating condition signals, and working signals are copied into operating condition signals and integrated into BIT message transmission; it is agreed that operating condition signals have 4 logical values: the operating condition signal is 0, which represents the normal state; A condition signal of 1 indicates a failure state; A working condition signal of 2 represents a faulty state; a working condition signal of 3 represents a dormant state, corresponding to an invalid / empty value in the BIT message content; functional elements include working functions and working condition functions; the signal elements corresponding to the functional elements are also defined with 4 logical values: 0 represents a normal state; 1 represents a failure state; 2 represents an unfavorable condition; 3 represents a vague state, indicating that the current information is insufficient to determine whether the function is in state 0 / 1 / 2; path elements include working path and operating condition path.

2. The fault modeling and diagnosis method based on multi-valued function according to claim 1, characterized in that, The diagnostic steps in step 6 are broken down as follows: Step 6.1: Determine the power supply signal; Step 6.2: Based on the information from the front-end nodes, back-end nodes, and multi-task status, determine the control / RF / data signals; Step 6.3: Determine the motion / medium signal and complete the non-working output state; Step 6.4: Determine the degree of impact of the fault and propose decision-making recommendations.

Images