A rotating machine progressive fault diagnosis and multi-dimensional data tracing method based on effect space trajectory features

Abstract

The application discloses a kind of based on effect space trajectory feature Rotating machine progressive fault diagnosis and multidimensional data tracing method, belong to rotating equipment condition monitoring technical field, the rotating machine includes motor, generator and pure mechanical rotating equipment.This method relies on equilateral triangle effect coordinate system, with the help of motion state sensor acquisition multidimensional physical signal, extract multidimensional space feature vector and construct progressive fault evolution model, continuously capture the space topology shift law of device fault from sprout period, development period until critical state;Build local high-frequency snapshot and cloud end full life cycle archives collaborative storage architecture, output with fault positioning vector, evolution stage, intervention level Standardized signal, can directly interface bottom control system or independent / additional detection module.The application unifies the space analysis logic of three-phase electrical equipment and pure mechanical equipment, provides preposition data support for rotating equipment to realize predictive maintenance, diagnosis and control integration.

Description

Technical Field

[0001] This invention relates to the fields of electromechanical energy conversion equipment condition monitoring, rotating machine fault diagnosis and data traceability technology, and in particular to a method for progressive fault evolution analysis of a three-phase rotating motor (including motor and generator) based on multidimensional spatial feature vectors of an equilateral triangle effect coordinate system, local and cloud collaborative storage and control preparation signal output. Background Technology

[0002] Current fault monitoring systems for rotating equipment in the industrial sector generally rely on three traditional methods: time-domain threshold judgment, FFT Fourier spectrum analysis, and symmetrical component analysis. This mature technology, rooted in the time-reduction framework, has several inherent limitations when dealing with long-term degradation monitoring and complex operating conditions: First, it can only identify instantaneous faults such as short circuits and phase loss, failing to capture slow, drifting degradation in its early stages, such as bearing wear and insulation aging. Small imbalances in the equipment can accumulate over time, leading to sudden shutdowns and significant economic losses. Second, the output signals are mostly simple Boolean alarms, lacking structured data such as fault location and evolution stages. Downstream controllers can only shut down the equipment indiscriminately, lacking a smooth derating mechanism. Third, conventional terminals have limited local buffers, making it easy to lose original data when sudden faults are accompanied by instantaneous power outages, making accident reproduction difficult. Fourth, frequency domain analysis relies on signal steady-state conditions, which can easily lead to spectral leakage and distortion under variable frequency or load fluctuation conditions. Fifth, single voltage trajectory analysis is susceptible to transformer saturation or common-mode interference, resulting in frequent false alarms. Furthermore, electrical and purely mechanical rotating equipment often require two completely independent monitoring programs, leading to doubled software and hardware development and maintenance costs. However, existing general fault diagnosis architectures and mature algorithm models cannot directly adapt to the progressive evolution analysis of multidimensional spatial feature vectors in effects theory. Existing diagnostic systems (such as time-series-based expert systems or frequency-domain-based machine learning models) are designed for one-dimensional time series or spectral data. When the input variables are transformed into "multidimensional spatial feature vectors" (such as relative centroid offset vector, eccentricity, petal distortion rate, and spatial topological distortion degree) in an equilateral triangular effects theory coordinate system, traditional diagnostic models cannot analyze these feature data with rigorous spatial geometric meaning and physical adversarial logic. It is necessary to reconstruct the diagnostic logic from the bottom layer and establish an evolutionary quantification model specifically for progressive imbalance of spatial potential difference in order to transform the spatial geometric features collected by the underlying hardware into diagnostic conclusions with engineering guidance value. Summary of the Invention

[0003] The purpose of this invention is to provide a progressive fault diagnosis and multi-dimensional data tracing method for rotating machines based on the spatial trajectory characteristics of effects. With a unified spatial analysis logic, it accurately captures the progressive fault evolution process of various motors, meets the needs of accountability and maintenance through local and cloud collaborative storage, and provides accurate pre-data input for subsequent motor control.

[0004] To achieve the above objectives, the present invention adopts the following technical solution: S1: Simultaneous acquisition of multi-dimensional features: Extracting multi-dimensional spatial feature vectors based on the equilateral triangle effect coordinate system; S2: Progressive Fault Evolution Analysis: Construct an evolution model to determine the nascent stage, development stage, or critical stage; S3: Multi-dimensional data tracing through local and cloud collaboration: triggering differentiated storage strategies; S4: Control preparation signal output: Outputs a standardized control preparation signal.

[0005] Compared with the prior art, the present invention has the following significant technical effects: Unified underlying logic for accurate capture of progressive faults: This invention is not limited to a specific type of motor, but is based on the topological characteristics of the effect space. By calculating the feature degradation rate and spatial drift trajectory, it accurately removes transient noise and truly restores the progressive degradation process of the motor from its inception to its critical state.

[0006] It enables collaboration between local and cloud environments, meeting the dual needs of accountability and long-term maintenance: for high-security / low-fault-tolerance scenarios, the local "black box" locks high-frequency snapshots to ensure that data is not lost after catastrophic failures, providing solid evidence for post-event accountability; for high-value scenarios, the cloud builds a full lifecycle digital archive, providing data support for predictive maintenance and long-term health assessment.

[0007] Seamless integration with control patent (Patent 6) to achieve "integrated diagnosis and control": The output of this invention is not a simple alarm code, but a standardized control preparation signal containing "spatial fault location vector, evolution stage, and suggested control mode", which is directly used as the feedforward compensation input for subsequent smooth derating and fault-tolerant control patents, thus completely opening up the technical closed loop from "perception diagnosis" to "execution control". Attached Figure Description

[0008] The present invention will now be described in further detail with reference to the accompanying drawings and specific implementation methods.

[0009] Figure 1 This is a schematic diagram of the progressive fault diagnosis and multidimensional data tracing process for motors in an embodiment of the present invention; Figure 2 This is a schematic diagram of the core module and data flow of the diagnostic device of the present invention. Detailed Implementation

[0010] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention.

[0011] 1. Prerequisites: 1.1 Positioning and Synergy of this Patent within the Entire Series of Patents This patent belongs to the Effects of Spatial Electricity patent family, which is divided into a principle family and an implementation family. In the principle family (family 1), patent 1 constructs an equilateral triangular geometric base and a dynamic overflow mechanism; patent 2 establishes hierarchical quantization rules to decouple spatiotemporal disturbances; patent 3 realizes composite trajectory anti-interference anchoring and outputs a four-dimensional standardized spatial feature vector; in the implementation family (family 2), hardware carrier patent 4 provides hardware-level synchronous acquisition, FPGA acceleration and power-down black box storage, and continuously outputs a distortion-free multi-dimensional spatial feature data stream.

[0012] This invention fully receives the standardized multidimensional spatial feature vector output by Patent 4, with underlying operations supported by the complete set of algorithmic principles from Patents 1, 2, and 3. Addressing the limitation of existing mature diagnostic models in handling spatial features, this invention, based on fundamental effects theories such as "slow drift in spatial potential difference represents continuous degradation of equipment static potential," specifically customizes a phased fault evolution quantification model, dividing it into three levels of judgment criteria: nascent, developing, and critical. It unifies the establishment of an equivalent spatial mapping analysis logic shared by three-phase electrical equipment and 120° distributed mechanical sensing equipment, breaking down electromechanical monitoring barriers. A dual-layer storage architecture of local fault snapshots and cloud-based full lifecycle archives is designed to ensure permanent retention of critical power outage trajectories. Finally, a standardized output carries a control preparation signal carrying the fault vector location, degradation stage, and suggested intervention level. This invention is only responsible for fault assessment and data tracing; control commands are executed by downstream Patent 6, forming a closed loop of decoupled diagnosis and control.

[0013] This invention belongs to the family of spatial electrical realizations of effects (family 2). Patents 1 to 6 are a series of inventions filed on the same day, independent yet closely related from the underlying principles to the application layer, and inseparable in terms of basic terminology and technical solutions. To clearly illustrate the position and synergistic relationship of this patent within the entire series of inventions, it is listed below:

[0014] 1.2 Unified Sampling and Progressive Analysis Logic This invention does not rely on the physical structure of a specific motor, but rather unifies the equilateral triangular effect coordinate system constructed in Patent 1 and the multidimensional spatial feature vector extracted in Patent 3 (| δ_rel|,ε,η,H _d In step S2, the construction logic of the progressive fault evolution model is as follows: Budding stage: When the magnitude of the relative centroid offset vector |δ _rel |or spatial topological distortion degree H _d When a slow, unidirectional drift is observed, and the drift rate exceeds the system noise floor threshold, it is determined to be either slight insulation aging or early bearing wear.

[0015] Development stage: When the eccentricity ε or petal distortion rate η exhibits periodic fluctuations and amplification, and the spatial topological drift trajectory significantly deviates from the reference point set (P) of the ideal standard circumcircle. _ideal When this occurs, it is determined to be either the initial stage of an inter-turn short circuit or rotor dynamic eccentricity.

[0016] Critical period: When the transient high-frequency fluctuation amplitude of the multidimensional feature vector exceeds the safety boundary, or the trajectory touches the dynamic overflow boundary of the equilateral triangle effect field, it is determined that the fault is about to lead to catastrophic failure.

[0017] 1.3 Joint Implementation Plan: Hardware and software co-engineering architecture for motion state sensor configuration and error compensation Motion Status Sensor Configuration Layer: Based on the physical type and monitoring requirements of the rotating machine, corresponding motion status sensors are configured to acquire multi-dimensional physical signals. For motors and generators with standard three-phase output or input, voltage transformers (PTs) and current transformers (CTs) are configured. These directly acquire three-phase electrical parameters, providing the most intuitive reflection of the spatial interaction state of the electromagnetic field inside the equipment, eliminating the need for additional mechanical sensors. For purely mechanical rotating equipment (such as centrifugal compressors and industrial fans), and special motors or generators that cannot directly extract standard three-phase electricity, a 120° spatially distributed group of mechanical motion status sensors is configured. This sensor group is configured differently according to the equipment type and fault characteristic frequency band, specifically including: eddy current displacement sensors (shaft vibration probes): used for non-contact acquisition of relative displacement of the rotor journal, suitable for monitoring the shaft trajectory of large, low-speed machinery; piezoelectric accelerometers (accelerometers): acquire high-frequency impacts on the housing, sensitive to bearing and gear fault distortion; and magnetoelectric velocity sensors: acquire the overall machine vibration intensity, suitable for monitoring conventional motors and fans.

[0018] Error Compensation and Preprocessing Layer: A built-in sensor error compensation matrix addresses PT / CT angular differences and mechanical sensor installation deviations. Simultaneously, the static centroid calibration subtraction method described in Patent 3 is employed: the static centroid O is calculated and stored under no-load or baseline healthy operating conditions. _static During runtime, all dynamic centroid offsets are differentially calculated to eliminate fixed common-mode errors.

[0019] Mechanical-Electrical Equivalent Mapping Rule: The three-phase electricity of the motor relies on the time phase difference to form three signals with a 120° phase difference. The mechanical rotor relies on sensors arranged in a 120° circumferential space to form three synchronous vibration signals. The two are mathematically isomorphic. The 120° spatial physical quantities MA, MB, and MC collected by the mechanical sensors are equivalent to the three-phase electrical amplitudes, which are mapped to the OA, OB, and OC effect axes through the effector orthogonal projection matrix. The circumferential force imbalance caused by the mechanical rotor's mass eccentricity, local fouling, and bearing wear will cause an imbalance in the amplitudes of the three mechanical signals. This imbalance can be equivalent to the imbalance of the motor's three-phase electromagnetic field, which is ultimately unified into the centroid offset in the effector space. The electrical and mechanical scenarios share the same set of geometric measurement models to carry out quantitative calculations such as centroid offset, trajectory area, and distortion rate.

[0020] Effects computation and feature extraction layer: Main control FPGA / DSP operation composite trajectory generation, spatial anchor point positioning, low-computing-power topology measurement module, unified output of four-dimensional feature vector (|δ _rel |、ε、η、H _d Patent 4's hardware supports two feature output modes: low-computational-power approximation and high-precision. This method is compatible with both types of feature vectors to complete fault classification and determination.

[0021] Data output link: The extracted multidimensional feature vector is sent to the diagnostic logic of Patent 5 via the industrial bus, or directly pushed to the control terminal of Patent 6.

[0022] Based on this general acquisition scheme, after generating multi-dimensional feature vectors and completing fault classification and judgment, the system archives and stores fault data in layers, and encapsulates and generates standardized control preparation signals to be sent to the downstream control end (Patent 6). The storage rules are shown in the third subsection below, and the standardized signal encapsulation and transmission rules are shown in the fourth subsection below.

[0023] 1.4 Differentiated Storage Strategy for Local and Cloud Collaboration In step S3, the system performs differentiated storage based on the fault stage: Local storage: a circular buffer continuously overwrites normal data. When it is determined that the system has entered the development or critical period, the "black box" mechanism is triggered, storing high-frequency feature snapshots and the original sampling point set (P) within a preset time window before and after the fault (e.g., 5 seconds before and after). _actual Encrypted data is written to local non-volatile memory to ensure data recovery after power outages or physical damage. Cloud storage: Dimensionally reduced feature trends and health scores are uploaded to the cloud via lightweight protocols such as MQTT to build timestamp-based digital twin profiles for long-term trend analysis and AI model iteration.

[0024] 1.5 Standardized output of control preparation signals In step S4, the control preparation signal output by the present invention includes: a fault spatial location vector (directly pointing to the effect axis where the overloaded phase or faulty phase is located), a fault evolution stage code (such as Code_Development), and a suggested control mode (such as Mode_Smooth_Derating) (Note: Code_Development is the development stage code, and Mode_Smooth_Derating is the smooth derating control mode). This signal is directly used as the feedforward input for subsequent control patent families.

[0025] Based on the above unified logic, the engineering applicability of the present invention will be explained in detail below in combination with the characteristics of different scenarios.

[0026] 2. Implementation Examples 2.1 Example 1: Progressive Degradation Analysis of High-Value and Long-Lead-Cycle Equipment Long-distance pipelines involve extremely high-cost equipment such as compressor drive motors, resulting in substantial downtime losses. The core value of data recording lies in "predictive maintenance" and "long-term health reference." Analysis focus and implementation strategy: (1) Long-term trend prediction and fault type mapping: The system focuses on monitoring the relative centroid offset vector magnitude |δ _rel The long-term drift trajectory of the motor. Because the operating conditions of the electric motor and the generator are different, this invention achieves differentiated fault mapping based on the same set of spatial topology drift logic: For generators (such as large hydro / thermal generators): If |δ is found _rel | Over several months, it exhibits a slow, unidirectional drift pointing towards a specific effect axis, accompanied by periodic small fluctuations in eccentricity ε. The system identifies this as a generator-specific progressive fault, such as stator winding thermal aging, rotor eccentricity, or excitation system asymmetry. For electric motors (such as main propulsion motors for ships and drive motors for large wind turbines): If |δ is found _rel The drift and petal distortion rate η worsen with increasing load, which the system identifies as a typical progressive motor fault such as bearing mechanical wear, rotor bar breakage, or stator inter-turn insulation deterioration.

[0027] Cloud-based full lifecycle record: The system constructs a full lifecycle health record for the motor from manufacturing to retirement in the cloud. Based on the evolution model, the system outputs "predictive maintenance" control preparation signals several months in advance, guiding maintenance personnel to replace components within the planned downtime window.

[0028] (3) Smooth derating recommendation: During the maintenance waiting period, the system outputs a "smooth derating mode" control preparation signal, recommending that the control system reduce the output power, force the centroid of the composite trajectory to return to the origin, achieve smooth operation with defects, and avoid huge economic losses caused by unplanned downtime.

[0029] 2.2 Example 2: Synergistic Mechanism of Multidimensional Signal Output Based on "Diagnosis and Control Decoupling" and Subsequent Control Patent Family As the core of the analysis and diagnosis implementation family, this invention's physical boundaries are strictly limited to "sensing, computing, storage, and signal output." It only sends critical intervention control preparation signals to Patent 6, and the downstream control device executes power cut-off and brake operation. This method does not directly output hardware drive commands. This "diagnosis-control decoupling" design ensures the independence of the diagnostic system and avoids the risk of equipment malfunctions caused by diagnostic algorithm misjudgments.

[0030] Based on the automation level, safety standards, and regulatory requirements of the monitored equipment, this invention provides two distinct signal output options in step S4: (1) Option 1: Pure alarm and trace signal output (output only, no control) For special equipment (such as elevator traction machines and passenger ropeways), highly regulated scenarios (such as motors for large amusement equipment and emergency generators for nuclear power plants), or old equipment without closed-loop control capabilities, the system only outputs standardized alarm and traceability signals without interfering with the original operating logic of the equipment.

[0031] Action executed: The system only outputs audible and visual alarm codes and text prompts (such as "Stator insulation deterioration - development stage, planned maintenance recommended") to the central control room or maintenance terminal, and forcibly triggers the local black box lock, saving high-frequency original waveform slices before and after the fault.

[0032] Project value: It meets the safety and compliance requirements of special equipment that "faults must be confirmed and intervened manually", while providing irrefutable evidence for post-event accountability and eliminating secondary safety risks caused by automatic miscontrol of the system.

[0033] (2) Option 2: Structured control preparation signal output (diagnosis and control collaboration) For modern equipment with advanced closed-loop control capabilities (such as drive motors for new energy vehicles and intelligent wind turbines). This invention encapsulates the diagnostic results into a structured data protocol (such as JSON / ModbusTCP / IEC61850GOOSE) and uses it as a "control preparation signal" to directly interface with Patent 6 in the implementation family.

[0034] Example of signal elements: When this invention determines that the motor is in the "development stage" and the spatial positioning vector points to phase A, the output control preparation signal encapsulates at least the following core data fields: fault phase identifier (phase A), evolution stage code (development stage), suggested control mode (smooth derating), and traceability ID (e.g., 0x9A). This signal can be encapsulated and sent via JSON, XML, or industry standard protocols (e.g., IEC61850GOOSE).

[0035] Collaborative Logic: After receiving the signal, Patent 6 does not need to perform complex spatial topology diagnosis again. Instead, it directly calls its internal "reconstruction of the remaining phase space counterbalance point" algorithm to perform smooth derating or fault-tolerant control.

[0036] (3) Architectural advantages This "decoupled diagnosis and control, signal coordination" architecture allows the diagnostic layer (Patent 5) to focus on "accurate calculation and complete storage," while the control layer (Patent 6) focuses on "stable control." The two layers communicate via standardized signals, ensuring no interference. When the control algorithm of Patent 6 is upgraded, there is no need to modify the diagnostic logic of this patent; conversely, when this patent introduces a new AI diagnostic model or adjusts the evolution threshold, there is no need to modify the underlying control code of Patent 6. This design significantly improves the real-time performance, reliability, and scalability of the entire implementation family (Family 2) system, ensuring a seamless closed loop from perception and diagnosis to execution and control.

[0037] 2.3 Example 3: Dual-mode deployment and traceability scenario based on independent detection equipment and additional analysis module When implemented in engineering projects, this invention supports two physical forms: "attached analysis modules" and "standalone testing equipment," to adapt to the needs of scenarios with different security levels and asset values, thus completely establishing a closed loop from "data perception" to "asset protection." (1) Form 1: An add-on analysis module (embedded or bypassed) is added to the existing frequency converter, BMS, or industrial control computer as a software algorithm library or hardware coprocessor. This form directly shares the sampling data and computing power of the original system, achieving progressive fault diagnosis at extremely low modification costs. It is suitable for cost-sensitive industrial continuous production equipment (such as chemical pump sets and metallurgical rolling mills). When the fault is determined to be in the development stage, a "smooth derating" control preparation signal is sent directly to the original controller through the internal bus to achieve operation with the fault.

[0038] (2) Form Two: Independent Testing Equipment (External Black Box or Portable Diagnostic Instrument) As a completely independent physical entity, it possesses an independent signal acquisition interface, local non-volatile memory, and a cloud communication module. This form plays an irreplaceable role in high-security / low-fault-tolerance and high-value scenarios. In special equipment and high-safety scenarios (such as elevator traction machines and motors for large amusement rides), independent monitoring equipment is forcibly mounted using a "bypass monitoring" method. Even if the original motor control system malfunctions, experiences logical errors, or is tampered with, the independent equipment can still objectively record the spatial topology drift trajectory based on effects spatial analysis of the underlying logic. Its locally locked high-frequency snapshots and cloud-encrypted files constitute an undeniable legal chain of evidence, providing support for fault cause analysis and subsequent accountability.

[0039] In large-scale power equipment scenarios (offshore wind turbines, ship main propulsion motors): independent testing equipment resides long-term as a "full life cycle health steward." It does not occupy the real-time control computing power of the original control system, but is dedicated to local caching and cloud synchronization of heavy data, providing an independent data foundation for predictive maintenance, spare parts scheduling, and life extension assessment.

[0040] 2.4 Example 4: Dynamic Online Diagnosis of Traction Motors in High-Speed ​​Railways / EMUs The pain point of online health monitoring of traction motors in high-speed trains / EMUs lies in the fact that traction motors are constantly exposed to high speed, heavy load, and strong electromagnetic interference generated by the high-frequency PWM switching of the traction inverter. Traditional frequency domain analysis (such as FFT) suffers from severe spectrum ambiguity under variable frequency conditions of train acceleration / deceleration and is highly susceptible to false alarms due to inverter noise interference; while static detection during shutdown cannot capture progressive degradation characteristics such as rotor bar breakage or insulation aging. The advantage of this patent is that, relying on the "current anchor point composite trajectory anti-interference mechanism" of patent 3, it effectively filters out measurement signal distortion caused by high-frequency switching of the traction inverter by using the system inertia of the current trajectory as a spatial reference during high-speed dynamic operation of the train; at the same time, based on the strict decoupling characteristics of effects in space and time, it overcomes the defect of traditional frequency domain analysis relying on the assumption of signal stability, and extracts the small spatial topological drift caused by slight insulation degradation of the stator or early breakage of the rotor in real time and accurately under variable frequency conditions, providing highly robust pre-data for subsequent multi-motor collaborative fault-tolerant control.

[0041] 2.5 Example 5: A closed-loop system for mandatory maintenance, regulatory tracing, and diagnostic control of special equipment (elevators as an example) based on effects spatial feature tracing. This embodiment takes the elevator traction machine (motor) as an example to illustrate in detail how this patent can completely solve the pain points of existing elevator management and control schemes through breakthroughs in underlying spatial analysis technology in the scenario of "high safety, low fault tolerance, and strong supervision", and build a complete safety closed loop in collaboration with subsequent control patents (patent 6).

[0042] Analysis of Pain Points and Technical Root Causes of Existing Management Solutions: The current elevator industry heavily relies on periodic mandatory maintenance and simple threshold controls, but in actual operation, numerous management pain points have been exposed. The root cause lies in the lack and outdated underlying technical solutions: First, periodic maintenance is often merely a formality, and supervision lacks effective means. The management pain point lies in the frequent occurrence of "over-maintenance" or "inadequate maintenance (going through the motions)." The technical root cause is: First, existing electrical monitoring technologies lack objective and continuous quantitative monitoring methods for the progressive deterioration of traction machines, bearings, and steel cables. Maintenance can only rely on fixed cycles and subjective human judgment, making it impossible for regulatory authorities to obtain objective data on the true health status of equipment, thus hindering digital supervision. Second, accident liability is difficult to determine, and there is a lack of conclusive evidence for tracing the source. A management pain point is that when accidents such as overshooting, bottoming out, or entrapment occur, disputes easily arise between the manufacturer, maintenance provider, and user. The technical root cause is that existing elevator operation recorders (black boxes) only record floor, speed, and simple overcurrent / overvoltage fault codes, lacking the original high-frequency electromagnetic and mechanical waveforms at the moment of the fault. This makes it impossible to reconstruct the essence of the accident from a physical perspective, resulting in a lack of legally conclusive evidence for determining liability. Third, the control logic is crude, easily leading to passenger entrapment. A management pain point is that once the elevator reports a fault, it stops operating and entraps passengers, triggering public complaints. The technical root cause is that existing elevator control strategies rely only on simple threshold judgments based on one-dimensional time series (Boolean logic), lacking a quantitative assessment of the "degree of spatial imbalance" of the equipment. Unable to make a smooth transition within the safety boundaries, the only option is to take a brutal, "black and white" approach to stopping.

[0043] Addressing the aforementioned technical challenges, this patent introduces an equilateral triangular electromechanical coordinate system, elevating elevator traction machine monitoring from a one-dimensional time domain to a two-dimensional spatial topological domain, recording the gradual degradation process. By extracting objective geometric quantitative indicators, it provides an immutable data foundation for government digital supervision and precise maintenance; by locking in the complete transient waveform through a "local black box," it provides physical evidence for accident liability determination; and by utilizing the coordinate system's "dynamic overflow mechanism" and "inscribed circular basis," it provides a precise "spatial safety envelope" for the control system. This electromechanical spatial technology solution fundamentally fills the gaps in underlying data and quantitative assessment, thereby supporting a comprehensive upgrade of elevator management solutions.

[0044] Based on the aforementioned technological breakthroughs, this patent, in conjunction with the subsequent control patent (Patent 6), is implemented as follows: 1. Maintenance and Repair Assistance and Digital Government Supervision (Technology-Enabled Precision Management): The system monitors the multi-dimensional spatial characteristic vector of the traction machine in real time. When stator insulation degradation is detected, leading to a relative centroid shift vector |δ... _rel When a device exhibits slow, unidirectional drift (emergence stage) or high-frequency petal distortion (development stage), the system automatically generates a maintenance work order containing the degradation rate and predicted remaining lifespan. Simultaneously, this multi-dimensional spatial characteristic data is directly connected to the government's special equipment supervision platform via an encrypted protocol, providing regulatory authorities with objective and tamper-proof geometric quantitative data. This achieves a shift from "regular, blind maintenance" to "predictive, precise maintenance based on spatial topological drift trends," completely eliminating fraudulent maintenance.

[0045] 2. Ironclad Evidence System for Accident Liability Determination and Tracing (Technology-Enabled Accident Liability Determination): In the event of an elevator safety accident, this patent provides absolute data evidence through a "local black box" and a "cloud-based encrypted traceability chain." The system utilizes the dynamic overflow mechanism of the underlying coordinate system to ensure that spatial coordinates are not truncated when physical quantities severely exceed limits at the moment of the accident, fully recording all high-frequency original waveform slices of the accident precursors and transient changes. Combined with the spatial trajectory archive throughout the entire lifecycle, regulatory authorities and appraisal institutions can accurately trace the source: if the pre-accident characteristics have drifted for a long time but have not been addressed, it is determined as maintenance deficiency responsibility; if the pre-accident spatial trajectory is normal but a sudden spatial topological distortion degree H occurs... _d A sudden, abrupt change in response accompanied by a delay in brake response is defined as a failure of the mechanical actuator or a manufacturing defect. This mechanism provides an undeniable legal chain of evidence for determining liability in an accident.

[0046] 3. Control Assistance and Physical Safety Backup (Technology-Enabled Safety and Trapping Prevention): The standardized control preparation signal output by this patent, which includes "spatial fault location vector and evolution stage," is directly used as the feedforward input for subsequent control patents (Patent 6) or the original elevator safety circuit. During normal elevator operation, if this patent determines that the traction machine is in the "development stage," the subsequent control system, upon receiving the signal, executes "smooth derating" control through the frequency converter. Without affecting passenger comfort, it dynamically limits the maximum load or operating speed of the elevator, forcing the spatial trajectory to return to the safety envelope. This ensures that the elevator can still safely operate to the nearest floor and open the door even when it is malfunctioning, reducing the frequency of entrapment caused by machine stoppage. If this patent determines that it is in the "critical period" (such as a sudden phase loss causing the trajectory area to be smaller than the area of ​​the inscribed circle), the system immediately triggers the quantization cutoff mechanism, cutting off the traction machine power supply within milliseconds and linking the elevator brake and safety clamp to perform emergency braking, safely locking the car on the guide rails and increasing the physical safety backup under extreme conditions.

[0047] 2.6 Example 6: Dynamic Balancing Detection Equipment Based on Effects Spatial Trajectory Analysis This embodiment can be implemented in a dynamic balancing testing device based on effector spatial trajectory. This device, as an independent physical entity or additional module, directly calls the feature extraction algorithm in the aforementioned common implementation scheme, and is specifically designed for on-site dynamic balancing and long-term degradation monitoring of purely mechanical rotating equipment (such as centrifugal compressors, fan rotors, and steam turbines).

[0048] 1. Hardware architecture of dynamic balancing testing equipment This device physically comprises the following core hardware modules: Multi-channel synchronous acquisition interface: used to connect mechanical motion state sensors (such as eddy current displacement sensors) that are spatially distributed in a 120° space on site to acquire multi-dimensional physical signals.

[0049] Effects feature calculation unit: Built-in FPGA or DSP chip, it uses the aforementioned "static centroid calibration subtraction method" to eliminate installation errors and runs a low-computing-power topology metric algorithm to output multi-dimensional spatial feature vectors (|δ) in real time. _rel |,ε,η,H _d ).

[0050] Explanation of cross-patent citations and implementation logic: The "static centroid calibration deduction method" is an equivalent extension and isomorphic application of the "current trajectory as spatial anchor point" anti-interference mechanism in the aforementioned "A System State Feature Extraction Method Based on Multiphase Composite Spatial Trajectory Geometric Moments and Low Computational Power Topology Measurement" (Patent 3), in the field of pure mechanical sensing. The specific implementation logic is as follows: Under the initial healthy state of the equipment (or cold-state calibration after shutdown), multi-channel mechanical displacement signals are collected to construct an initial static trajectory, and its geometric center is defined as the "static spatial anchor point" (i.e., the physical rotation center). In subsequent dynamic monitoring, the offset of the dynamic trajectory relative to this "static spatial anchor point" is calculated in real time, thereby completely eliminating common-mode interference caused by sensor installation angle deviation and initial mechanical deflection, ensuring the extracted relative centroid offset vector magnitude |δ _rel It truly reflects the dynamic imbalance of the rotor.

[0051] Dynamic balancing mapping engine: Based on the isomorphism of the effector coordinate system, it shifts the relative centroid offset vector δ _rel Length of the module It is directly mapped to the magnitude of the imbalance, and its direction angle is directly mapped to the phase angle of the physical counterweight on the rotor circumference.

[0052] Human-Machine Interface (HMI): Equipped with a high-brightness touchscreen, physical operation buttons and communication interface, it is used to render the equilateral triangle effect trajectory diagram, display the counterweight guidance, and support data export or control command issuance.

[0053] 2. Significant advantages compared to existing technologies: Accuracy and Efficiency (Minimized Trial Weights and Decoupling of Complex Faults): Compared to the traditional "trial weight method," which requires multiple start-ups and shutdowns to add trial weights, is time-consuming, and prone to causing mechanical fatigue, this equipment, based on the direct mapping of efficiencies spatial positioning vectors, significantly reduces trial weight balancing, compressing debugging time from hours to minutes. Simultaneously, the equipment utilizes spatial topology to decouple complex faults: if the trajectory only shows centroidal shift with smooth edges, the equipment determines it as a simple mass imbalance; if accompanied by significant high-frequency petal distortion, the equipment automatically indicates "imbalance accompanied by shaft misalignment or mechanical loosening," providing highly accurate guidance for precise on-site troubleshooting.

[0054] Economic Efficiency and Convenience (Lowered Barriers and Predictive Maintenance): Compared to the shortcomings of traditional "spectrum analysis methods" that rely on complex Bode plots and require highly specialized personnel, this equipment uses an HMI unit to transform abstract spectrum data into intuitive "centroid offset direction" and "petal distortion morphology." Ordinary field maintenance personnel no longer need to interpret complex spectrum data; they can complete high-precision operations simply by following the intuitive on-screen guidance, significantly reducing the company's costs associated with manual training and reliance on experts. Combined with cloud-based archives, the equipment can also track centroid drift trends, enabling predictive maintenance and preventing unplanned downtime.

[0055] 3. Examples of Deployment of Multi-Form Physical Devices To adapt to different scenarios, this device offers the following two physical forms during engineering deployment: Form 1: Portable intelligent dynamic balancing tester (with minimalist on-site interaction) This device is a handheld or portable physical unit that integrates a data acquisition interface, a calculation unit, and a touchscreen. During on-site debugging, the screen directly renders the effect trajectory diagram and uses dynamic arrows to intuitively indicate the "suggested counterweight position and mass." After the maintenance personnel add the counterweight, the device refreshes the trajectory in real time. The personnel can visually observe whether the trajectory centroid returns to the origin O and whether the "residual imbalance" value meets the standard, achieving a simplified "what you see is what you get" on-site interaction.

[0056] Form 2: Online dynamic balancing monitoring and automatic balancing cabinet (central control closed-loop interactive form) This form factor is a standard rack-mount physical entity deployed in the central control room or next to the equipment. Its HMI screen displays the "Effective Space Health Matrix" of all equipment in the plant. In addition to monitoring and alarm functions, its control output interface is directly connected to the "Automatic Oiling / Throwing Counterweight Actuator" or "Active Magnetic Bearing Controller" via an industrial bus. When the equipment detects that the centroid offset exceeds the threshold, it automatically issues control commands, and the actuator dynamically adjusts the counterweight while the equipment is running at full speed. The system monitors the centroid return process in real time, forming an unmanned closed loop of "diagnosis-execution-feedback" to achieve proactive fault-tolerant operation without stopping the machine.

[0057] The embodiments of the present invention have been illustrated above through specific examples. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention.

Claims

1. A progressive fault diagnosis and multidimensional data tracing method for rotating machines based on the spatial trajectory characteristics of effector science, characterized in that, The rotating machine includes an electric motor, a generator, or a purely mechanical rotating device; the method is based on an equilateral triangular effect coordinate system containing a centroid O, three mutually disparate effect axes of 120°, and a dynamic overflow mapping mechanism, and includes the following steps: S1: Real-time synchronous acquisition of multi-dimensional physical raw signals via motion state sensors installed on the rotating machine; complete error compensation preprocessing of the raw signals; calculation and storage of the trajectory static centroid O under reference steady-state conditions. _static All subsequent dynamic centroid offsets are deducted from O. _static Simultaneously, the sensor compensation matrix is ​​invoked to correct amplitude, phase, temperature drift, and channel imbalance errors; the compensated signal is mapped to an equilateral triangle effect coordinate system, and a multidimensional spatial feature vector is extracted based on the composite spatial trajectory geometric moment algorithm. This multidimensional spatial feature vector at least includes the relative centroid offset vector |δ _rel | eccentricity ε, petal distortion rate η, spatial topological distortion H _d When adapted to electric motors or generators, the motion state sensor uses voltage transformers and current transformers to collect three-phase electrical parameters; when adapted to purely mechanical rotating equipment or special motors that cannot collect standard three-phase electrical parameters, the motion state sensor uses a group of mechanical motion sensors arranged at 120° along space. S2: Based on multidimensional spatial feature vector time series data, calculate feature degradation rate and spatial topology drift trajectory and construct progressive fault evolution model to distinguish three types of fault stages: equipment initiation stage, development stage and critical stage; synchronously monitor the instantaneous jump of spatial topology distortion degree and relative centroid offset vector to identify sudden change faults; S3: Execute a tiered storage strategy based on the determined fault stage and transient sudden events: continuously store high-frequency feature snapshots for progressively deteriorating faults, and lock the full original waveform slices of the fault window for transient sudden events; at the same time, establish a full life cycle health record and encrypted data traceability chain for the device in the cloud; S4: Converts fault type, evolution stage, spatial topology drift orientation, and recommended intervention level into standardized control preparation signals, and outputs them to the underlying closed-loop control system or independent additional control module.

2. The method according to claim 1, characterized in that, The sensor compensation matrix mentioned in step S1 includes PT / CT amplitude and phase correction coefficients, mechanical sensor temperature drift compensation coefficients, and multi-channel gain equalization coefficients; after equipment overhaul, sensor replacement, or long-term shutdown restart, it is automatically recalibrated under the reference steady-state conditions. _static With all parameters of the compensation matrix.

3. The method according to claim 1, characterized in that, The progressive fault evolution model judgment rule in step S2 is as follows: if the relative centroid offset vector magnitude or spatial topological distortion degree undergoes a slow unidirectional drift and the drift rate is greater than a preset noise threshold, it is judged as the nascent stage; if the eccentricity and petal distortion rate periodically fluctuate and amplify and the spatial drift trajectory deviates from the ideal standard circumscribed circle reference point set, it is judged as the development stage; if the instantaneous high-frequency fluctuation of multidimensional features exceeds the safety threshold, or the trajectory reaches the dynamic overflow boundary of the coordinate system, or the trajectory area is less than the area of ​​the inscribed circle / judgment circle, it is judged as the critical stage; the dynamic overflow boundary and the geometric reference of the inscribed circle / judgment circle are the built-in geometric judgment reference of the equilateral triangle effect coordinate system.

4. The method according to claim 1, characterized in that, The tiered storage strategy in step S3 includes two parts: local storage and cloud storage. Local storage: Local edge devices are configured with a circular buffer, which normally cyclically overwrites the dimensionality reduction feature data. When it is determined that the equipment has entered the development period or critical period, the hardware black box latching mechanism is triggered to encrypt and store the high-frequency snapshots and original sampling point sets within the preset time window before and after the fault into the local non-volatile memory. Cloud storage: Upload dimensionality reduction feature trends, device health scores, and key fault event slices through a general lightweight communication protocol to build a digital archive of the entire device lifecycle.

5. The method according to claim 1, characterized in that, The standardized control preparation signal output in step S4 includes the fault space location vector, the fault evolution stage code, and the recommended control mode. For motors and generators, the fault space location vector points to the corresponding effect axis of the fault / heavy load. For purely mechanical rotating equipment, the fault space location vector maps to the equivalent effect axis of the coordinate system corresponding to the rotor unbalance mass.

6. A progressive fault diagnosis and multidimensional data tracing device for rotating machines based on the spatial trajectory characteristics of effector science, characterized in that, The device is an embedded (additional) analysis module of the original control system of the rotating equipment or an independent external detection device; the device includes a multi-dimensional feature extraction module, an error compensation module, an evolution analysis module, a collaborative storage module, and a signal output module; the multi-dimensional feature extraction module receives the original physical signals from the sensor to generate original trajectory data, and transmits the data bidirectionally with the error compensation module, and the original trajectory data is first sent to the error compensation module to complete multi-level offset correction; The corrected feature vector is input into the evolution analysis module; the evolution analysis module interacts with the collaborative storage module and the signal output module respectively. The collaborative storage module and the signal output module read and write snapshots and equipment files bidirectionally. The signal output module generates a standardized control preparation signal and outputs it outward. The signal output module can also send the control preparation signal back to the evolution analysis module to complete the closed-loop analysis.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements all the steps of the method according to any one of claims 1 to 5.

Images