A device state monitoring method and system based on automatic working condition calibration and multi-dimensional signal analysis

Abstract

The present application relates to the technical field of industrial equipment state monitoring, and particularly relates to a kind of equipment state monitoring method and system based on automatic working condition calibration and multidimensional signal analysis, comprising: the present application is combined by automatic working condition calibration, multidimensional signal accurate analysis, dynamic anomaly diagnosis and intelligent interactive feedback technology, realizes the full-link automation closed-loop monitoring of equipment operation state from calibration to diagnosis to abnormal response, significantly improves the real-time, accuracy and intelligent level of equipment state monitoring process, solves the problems of inaccurate working condition recognition, single feature analysis dimension and insensitive abnormal state recognition in the prior art, and effectively improves the equipment operation reliability, reduces the fault risk and maintenance cost.

Description

Technical Field

[0001] This invention relates to the technical field of industrial equipment condition monitoring, and in particular to a method and system for equipment condition monitoring based on automatic operating condition calibration and multidimensional signal analysis. Background Technology

[0002] With the continuous improvement of the intelligence level of industrial equipment, the traditional equipment condition monitoring mode that relies on manual inspection and experience judgment is gradually showing its shortcomings and limitations because it is difficult to meet the comprehensive needs of the current industrial field for real-time, accuracy and intelligence of equipment condition monitoring. Especially in the operation scenarios of rotating machinery equipment such as pumps, compressors and fans, the equipment operating status is varied and complex, and the required monitoring feature dimensions are diversified, which puts forward higher requirements for the analysis and diagnosis of equipment health status.

[0003] However, current equipment condition monitoring technologies still largely rely on single-type signals (such as vibration signals) for condition determination in practical applications. They lack dynamic identification and real-time calibration mechanisms for equipment operating conditions, making it difficult to comprehensively reflect the actual operating status of equipment from multiple dimensions. This easily leads to lag and misjudgment in equipment condition identification. Therefore, there is an urgent need to develop an intelligent monitoring technology that integrates automatic operating condition calibration, standardized signal acquisition, and multi-dimensional data analysis algorithms to achieve a more accurate and intelligent comprehensive assessment of the operating status of industrial equipment.

[0004] In the prior art, CN112613646A discloses a device status prediction method based on multi-dimensional data fusion, which performs noise reduction, wavelet packet analysis and feature extraction on the status signals during the device's life cycle, and completes model training and device status prediction in a cloud-edge collaborative manner; while CN119249345A discloses an online monitoring system, which establishes an online monitoring platform to achieve real-time diagnosis of device status by collecting and analyzing temperature, vibration and pressure signals.

[0005] However, the existing technical solutions mentioned above have the following shortcomings: First, the lack of automatic real-time calibration of equipment operating conditions results in a lack of unified reference standards for operating condition identification, reducing the ability to accurately identify different operating states; second, data analysis is mostly macroscopic overall analysis, without refined modeling and fusion of multi-dimensional signal features, which limits the sensitivity and accuracy of abnormal state identification; third, the abnormal judgment process mostly relies on preset fixed thresholds for evaluation, ignoring the dynamic feature drift caused by real-time changes in operating conditions, which reduces the accuracy and stability of condition monitoring. Summary of the Invention

[0006] The purpose of this section is to outline some aspects of the embodiments of the present invention and to briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section, as well as in the abstract and title of the present application, to avoid obscuring the purpose of this section, the abstract and title of the invention. Such simplifications or omissions shall not be used to limit the scope of the present invention.

[0007] In view of the aforementioned existing problems, the present invention is proposed.

[0008] Therefore, the technical problem solved by the present invention is that the existing technology lacks an automatic calibration and dynamic identification mechanism for equipment operating conditions and has low accuracy in multi-dimensional feature analysis.

[0009] To solve the above-mentioned technical problems, the present invention provides the following technical solution: collecting operating parameters during the equipment installation phase, and establishing a working condition benchmark vector based on outlier removal and mean calculation;

[0010] During the equipment operation phase, real-time operating parameters are acquired and compared with the operating condition benchmark vector to generate a comprehensive deviation metric, and the current operating condition label is determined through a continuity judgment strategy.

[0011] Based on the operating condition label, the corresponding signal analysis model is invoked to extract spectral features, shaft center trajectory features, and operating parameter numerical features.

[0012] The acquired multidimensional features are compared with preset thresholds to determine whether the device status is abnormal and generate alarm information.

[0013] The operating condition labels, feature curves, anomaly categories, and alarm records are output in a visual manner, and question-and-answer interaction based on a large language model is supported.

[0014] As a preferred embodiment of the equipment condition monitoring system based on automatic operating condition calibration and multidimensional signal analysis described in this invention, the monitoring system includes a multi-source sensor module, a data acquisition unit, an analysis and storage unit, and a display and interaction unit, wherein:

[0015] The multi-source sensor module is used to collect vibration, sound, temperature and electrical parameter signals;

[0016] The data acquisition unit is used to condition and digitize the acquired signals;

[0017] The analysis and storage unit is used to perform working condition calibration, operation status identification and multi-dimensional feature analysis, and to determine anomalies, save operation data and support trend analysis, playback and edge computing;

[0018] The display and interaction unit is used to visualize the monitoring results and, in conjunction with a large language model, to achieve question-and-answer interaction and remote data communication.

[0019] As a preferred embodiment of the equipment condition monitoring system based on automatic operating condition calibration and multidimensional signal analysis described in this invention, it further includes one or more processors;

[0020] The memory stores operable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations, including the flow of the equipment condition monitoring method based on automatic operating condition calibration and multidimensional signal analysis as described above.

[0021] As a preferred embodiment of the computer-readable medium for storing software according to the present invention, the software includes instructions executable by one or more computers, the instructions causing the one or more computers to perform operations, the operations including the process of the equipment condition monitoring method based on automatic operating condition calibration and multidimensional signal analysis as described above.

[0022] The beneficial effects of this invention are as follows: By combining automatic operating condition calibration, real-time dynamic identification of operating conditions, precise analysis of multi-dimensional state characteristics, and intelligent health assessment, this invention comprehensively improves the real-time performance, accuracy, and intelligence level of equipment monitoring. It effectively compensates for the shortcomings of existing technologies, such as insufficient single-dimensional analysis, lack of a unified operating condition calibration mechanism, and insufficient dynamic identification capability of abnormal states. It can be widely applied to rotating machinery equipment such as pumps, compressors, and fans, ensuring stable and reliable operation of equipment and reducing operation and maintenance costs. Attached Figure Description

[0023] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0024] Figure 1 This is a flowchart illustrating the equipment condition monitoring method based on automatic operating condition calibration and multidimensional signal analysis as shown in this invention.

[0025] Figure 2 This is a schematic diagram of the modular unit structure of the equipment condition monitoring system based on automatic operating condition calibration and multidimensional signal analysis as shown in this invention.

[0026] Figure 3 This is a schematic diagram of the signal processing flow module of the equipment condition monitoring system based on automatic operating condition calibration and multidimensional signal analysis, as shown in this invention. Detailed Implementation

[0027] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0028] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort should fall within the scope of protection of this invention.

[0029] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0030] According to an embodiment of the present invention, in combination Figure 1 The flowchart shown illustrates a method for equipment condition monitoring based on automatic operating condition calibration and multidimensional signal analysis, which specifically includes the following steps:

[0031] S1. Collect operating parameters during the equipment installation phase, and establish a working condition baseline vector based on outlier removal and mean calculation.

[0032] S2. During the equipment operation phase, real-time operating parameters are acquired and compared with the operating condition baseline vector to generate a comprehensive deviation metric, and the current operating condition label is determined through a continuity judgment strategy.

[0033] S3. Based on the operating condition label, call the corresponding signal analysis model to extract spectral features, shaft center trajectory features and operating parameter numerical features;

[0034] S4. Compare the acquired multidimensional features with preset thresholds to determine whether the device status is abnormal and generate alarm information.

[0035] S5 outputs operating condition labels, characteristic curves, anomaly categories, and alarm records in a visual manner, and supports question-and-answer interaction based on a large language model.

[0036] By obtaining a stable parameter sample sequence through step S1, the mean value of each operating parameter is accurately calculated, and a reliable operating condition reference vector is finally formed. This step ensures the accuracy and stability of the equipment's personalized reference model, providing a unified and accurate reference benchmark for subsequent operating condition identification and condition analysis, and fundamentally improving the accuracy of the entire condition monitoring process.

[0037] Step S2 enables high-precision real-time monitoring of equipment operating status, quickly and accurately identifying operating condition switching, greatly improving the timeliness and accuracy of equipment status monitoring, and avoiding the risks of false or delayed reporting of operating condition switching under traditional methods.

[0038] Step S3 enables the extraction of multi-dimensional features of equipment status, ensuring the refinement and targeting of signal processing under different operating conditions. This effectively avoids the risk of missed diagnoses or misjudgments that are easily caused by single-feature dimension analysis, and significantly enhances the comprehensiveness and accuracy of equipment status diagnosis.

[0039] Step S4 enables rapid location and accurate diagnosis of abnormal equipment conditions, significantly improving the early warning capability of equipment failures, providing maintenance personnel with timely and clear maintenance guidance, and avoiding the risk of false alarms or missed alarms caused by the single threshold setting in traditional condition monitoring solutions.

[0040] Step S5 enables efficient interaction between equipment status information and maintenance personnel, clearly and intuitively displaying equipment status and trends, greatly improving the on-site response efficiency and decision-making ability of maintenance personnel, and ultimately significantly reducing the complexity of equipment maintenance and the difficulty of personnel operation.

[0041] In summary, this invention combines automatic operating condition calibration, multi-dimensional signal precision analysis, dynamic anomaly diagnosis, and intelligent interactive feedback technologies to achieve fully automated closed-loop monitoring of equipment operating status from calibration to diagnosis to anomaly response. This significantly improves the real-time performance, accuracy, and intelligence of the equipment status monitoring process, and solves the problems of inaccurate operating condition identification, single feature analysis dimensions, and insensitive anomaly identification in existing technologies. It effectively improves equipment operating reliability and reduces failure risks and maintenance costs.

[0042] The following describes in more detail the implementation process and / or effects of certain embodiments of the present invention, with reference to some preferred or optional examples.

[0043] Operating condition calibration:

[0044] To achieve accurate identification of the operating status of industrial equipment, the operating condition baseline values ​​are calibrated first during the equipment installation process. Depending on the installation environment, the operating condition calibration is divided into two types:

[0045] Work Mode Calibration: Data is collected during equipment online operation;

[0046] Idle Mode Calibration: Data is collected when the equipment is stopped or idle.

[0047] The calibration results are used to construct a personalized reference state model for the equipment and serve as a benchmark for subsequent operating condition identification and anomaly analysis; the specific process is as follows:

[0048] ① Selection of operating parameters

[0049] A working condition vector is formed by selecting representative equipment operating parameters, denoted as:

[0050]

[0051] in, Let be the current at time t. Let be the rotational speed at time t. The power at time t;

[0052] This vector is used to describe the load characteristics of the device's operating status;

[0053] ② Calibration mode distinction

[0054] Set calibration mode parameters according to the installation scenario. There are two installation modes to choose from, and the possible values ​​are:

[0055] idle: Installation calibration during shutdown. In this mode, the calibration equipment is in an unloaded state, and the parameter baseline value is low.

[0056] work: Run installation calibration. In this mode, the equipment operating status is calibrated. It is a load-bearing operating condition with high parameter baseline values.

[0057] ③ Preprocessing of operating parameters

[0058] Within the set calibration time window Within this period, the calibration process takes place, during which key operating condition parameter sequences are collected, including current. Rotation speed ,power Construct the working condition vector:

[0059]

[0060] To improve the robustness of the baseline operating conditions, an outlier removal mechanism based on interquartile range (IQR) is introduced, with the following processing steps:

[0061] For any operating condition parameter Calculate the upper and lower quartiles of one of its samples:

[0062] First quartile (25%): ;

[0063] Third quartile (75%): ;

[0064] IQR is defined as: ;

[0065] Set the upper and lower limits as follows:

[0066]

[0067] in, This is an adjustable parameter with a value range of 1.5 to 3; the default setting is... ;

[0068] All unsatisfied Samples that are considered outliers will be removed.

[0069] This method is applied to respectively , , Three dimensions are used to independently filter out outliers for each working condition parameter. The outlier data is then used for subsequent baseline value calculations to ensure that the reference model has higher stability and noise resistance.

[0070] ④ Calculation of operating condition reference values

[0071] For the pre-processed operating parameters, calculate their average value over a specified time interval, which will serve as a reference benchmark for this installation mode:

[0072]

[0073] This forms a complete operating condition reference vector:

[0074]

[0075] in, It is the first Current values ​​at each sampling point It is the first Rotational speed values ​​at each sampling point It is the first Power values ​​at each sampling point This is the total number of sampling points. , , These are the average values ​​of current, speed, and power, used to construct the operating condition reference vector. ;

[0076] For different installation and calibration processes, save the corresponding baseline value vectors for either the idle or work operating conditions. This is used for subsequent working condition identification.

[0077] Operating condition identification:

[0078] To achieve intelligent sensing and trend analysis of equipment operating status, this embodiment of the invention refines the operating condition identification process into two steps:

[0079] Deviation identification from the benchmark condition: Determine whether the current state deviates from the historical calibration condition;

[0080] Identifying the continuity of operating condition changes: avoiding misjudgment of instantaneous fluctuations and capturing trend changes;

[0081] It should be noted that the identification process is based on a unique reference model and uses a fixed operating condition mode (idle or work only) to ensure the consistency and rigor of the identification.

[0082] ① Deviation identification from the benchmark condition

[0083] Load the baseline model of the specified working condition at startup, such as idle or work:

[0084]

[0085] The calibration mode remains unchanged during device operation to prevent judgment errors caused by mixing different modes.

[0086] Collect the actual operating parameters of the equipment at the current moment:

[0087]

[0088] Calculate the relative deviation for each dimension:

[0089]

[0090] Define a weighted overall deviation metric:

[0091]

[0092] in, , , These represent the relative deviations of the current, speed, and power from their reference values, respectively. , , For the corresponding weights (satisfying) The weights can be set empirically based on parameter stability, and the overall deviation measurement is used. Used to assess the overall deviation between the current operating condition and the reference operating condition;

[0093] like If the value is 0.3, it indicates that the current state has deviated significantly from the reference condition and is marked as another condition (e.g., if mode=idle, the current point is identified as the running condition; if mode=work, the current point is identified as the no-load condition).

[0094] ② Continuity of Operation Identification and Verification

[0095] Considering the instantaneous fluctuations in industrial systems (such as load jumps and speed regulation disturbances), a sliding window continuous judgment method is further adopted to improve the stability and accuracy of operating condition identification.

[0096] Define the sliding window size (Default setting is 60 sampling points), count the number of time points where the deviation exceeds the limit within each window. :

[0097]

[0098] in, For indicator functions;

[0099] Set a continuity determination threshold (Default setting is 15), the judgment logic is as follows:

[0100] like If so, then a change in operating conditions has been confirmed;

[0101] like If so, it is considered a normal fluctuation and no response is required;

[0102] The final working condition identification result is output as follows:

[0103]

[0104] in, Indicates the working condition type that is the opposite of the current mode:

[0105]

[0106]

[0107] Final output labels This is the result of the working condition identification at the current time point.

[0108] State analysis:

[0109] It should be further explained that, under the identified operating conditions, a state analysis process is performed. This process takes high-frequency vibration signals and key operating parameters as inputs, calculates multiple types of state characteristics, and constructs a benchmark model based on the characteristic distribution of historical normal operating stages for subsequent anomaly identification and health assessment.

[0110] The benchmark model includes at least three sub-models: vibration spectrum analysis, shaft center trajectory calculation, and parameter trend analysis. The output of each model is divided into numerical features and shape features, and benchmark thresholds or reference spectra are established for each.

[0111] ① Vibration spectrum analysis (shape-related features)

[0112] During the operating period identified as the running condition, the vibration acceleration signal... Perform a frequency domain transformation and calculate the spectrum using the Fast Fourier Transform (FFT):

[0113]

[0114] in, The extracted spectrum is the sampling frequency. Characterizes the vibration excitation features of the equipment during operation;

[0115] A reference spectrum is generated by averaging the spectra from multiple runtime segments. As a standard shape feature under normal conditions:

[0116]

[0117] ② Axial trajectory analysis (shape-based features)

[0118] The trajectory of the shaft center is synchronously acquired using two axial displacement sensors (e.g., in the X and Y directions) installed on the rotating equipment. To obtain the dynamic behavior characteristics of the rotor in the plane;

[0119] When the operating condition is identified as the running period, trajectory data from multiple cycles is extracted. The time series coordinates are directly normalized and aligned, and the average trajectory is calculated as a standard reference trajectory:

[0120]

[0121] Each of them The axis trajectory curve within one period;

[0122] ③ Parameter trend analysis (numerical features)

[0123] For time series parameters (such as vibration amplitude, temperature, etc.), denoted as Extract its statistical trend characteristics over time, including:

[0124] Parameter mean: ;

[0125] Parameter volatility: ;

[0126] Trend slope, using least squares method Fitting a first-order straight line slope That is, the trend slope. ;

[0127] in, It refers to the sample size, that is, the total number of observation points in the time series. It is the first The time points corresponding to each observation value It is a point in time. Parameter observations on;

[0128] For data under normal operating conditions, a statistical threshold interval is constructed for each numerical feature. For any statistical trend feature, a reference mean is set as... The reference standard deviation is Define the decision interval:

[0129]

[0130] in, This is the confidence factor (default is 3), used to set the alarm threshold.

[0131] Health assessment:

[0132] To achieve real-time determination of equipment status and anomaly alarm, after the status analysis is completed, the health assessment stage is entered. In this stage, the corresponding anomaly identification logic is executed according to the different types of status characteristics, and specific alarm information is output when a deviation is detected.

[0133] The health assessment process is divided into numerical feature and shape feature similarity recognition based on feature type. Each type of feature is evaluated separately. If it is determined to be abnormal, the alarm result for that type is output, and the current time, deviation degree, feature value, and reference value information are recorded.

[0134] ① Numerical feature anomaly identification

[0135] For numerical features output by state analysis (such as parameter trend slope, mean, and maximum value), set historical reference threshold intervals under operating conditions:

[0136]

[0137] The exception detection logic is as follows:

[0138]

[0139] ② Shape-based feature anomaly identification

[0140] For shape-related features such as spectrograms and axis trajectories, similarity evaluation is performed based on normalized multi-frame vector sequences;

[0141] The real-time feature shape is a vector sequence. The standard reference shape is: Each of them , for dimension vectors (such as) The time represents a planar trajectory point; both the spectrum diagram and the axisymmetric trajectory diagram are planar trajectory points.

[0142] Calculate the root mean square value of the inter-frame vector distance:

[0143]

[0144] Set an abnormal shape threshold (Default value is 0.3), the exception handling logic is defined as follows:

[0145]

[0146] For various features, if the anomaly identification logic result of a certain feature value is 1, then the indicator is judged to be abnormal and an alarm of that type is issued.

[0147] Question and answer interaction:

[0148] The method provided in this embodiment supports natural language question-and-answer interaction. By combining a built-in large language model with data interfaces and prompt word templates, it automatically answers user questions and realizes intelligent explanation of device status and decision assistance.

[0149] ①Workflow

[0150] User input: Questions can be entered via a web interface or touch terminal;

[0151] Intent recognition: Identify the intent of the question and match the corresponding template;

[0152] Status query class → Call template Q1;

[0153] Alarm explanation class → Call template Q2;

[0154] Trend Analysis Class → Call Template Q3;

[0155] Health assessment category → Call template Q4;

[0156] Historical comparison class → Call template Q5;

[0157] Other free classes → Call template Q6;

[0158] Data acquisition: Call the data interface to obtain real-time or historical feature values;

[0159] Prompt word construction: Populate the data into the preset prompt word template;

[0160] Model invocation: Input the constructed Prompt into the LLM to generate a response;

[0161] Results display: Results are returned to the user in natural language, and also support display modules such as linked alarms and trend charts;

[0162] ② Prompt word template

[0163] Q1: Current Device Status Inquiry

[0164] Prompt word structure:

[0165] Based on the following data, please determine the current operating status of the device and answer the user in concise and natural language:

[0166] Operating mode: {Operating mode}

[0167] Current running parameters:

[0168] Current: {current} A

[0169] Rotational speed: {rotational speed} rpm

[0170] Power: {Power} kW

[0171] State analysis characteristics:

[0172] Spectral similarity: {spectral similarity}

[0173] Axis trajectory similarity: {Axis trajectory similarity}

[0174] Parameter trend deviation: {trend deviation}

[0175] Based on the above information, please briefly describe whether the equipment is in normal working order and point out its key characteristics;

[0176] Input variables: {Operating mode}, {Current}, {Speed}, {Power}, {Spectrum similarity}, {Shaft center trajectory similarity}, {Trend deviation};

[0177] Applicable scenarios: When users ask, "Is the device working properly now?", "What is the current status of the device?", or "Are there any problems with its operation?"

[0178] Q2: Explanation of Abnormal Alarms

[0179] Prompt word structure:

[0180] If a user requests details of an alarm, please explain based on the following alarm information:

[0181] Alarm type: {alarm type}

[0182] Trigger time: {alarm time}

[0183] Feature type: {feature type}

[0184] Current value: {current value}

[0185] Normal reference value: {reference value}

[0186] Degree of deviation: {Degree of deviation}

[0187] Please explain the cause of this alarm, the affected areas, and recommended actions in a professional yet easy-to-understand way;

[0188] Input variables: {alarm type}, {alarm time}, {feature type}, {current value}, {reference value}, {deviation level};

[0189] Applicable scenarios: When users ask questions such as "Why did an alarm go off?", "What type of anomaly is this?", or "Is the alarm serious?"

[0190] Q3: Trend Change Analysis

[0191] Prompt word structure:

[0192] Please analyze the changing trends of the following key equipment parameters over the past {time interval}, and briefly explain whether their changing characteristics are abnormal:

[0193] Parameter trend data:

[0194] Vibration fluctuation: {Vibration fluctuation}

[0195] Average temperature: {average temperature}

[0196] Trend slope: {trend slope}

[0197] Please indicate whether there is a clear upward or fluctuating trend, and explain the possible reasons;

[0198] Input variables: {time interval}, {oscillation fluctuation}, {mean temperature}, {trend slope};

[0199] Applicable scenarios: When users ask questions such as "Has the status changed recently?", "Is the device operating normally?", or "Is there a deteriorating trend?"

[0200] Q4: Health Assessment Inquiry

[0201] Prompt word structure:

[0202] Please summarize the health status of the device based on the following health score data:

[0203] Overall health score: {Overall score} / 100

[0204] Key feature scores:

[0205] Vibration characteristics: {Vibration score}

[0206] Spectrum shape: {Spectrum score}

[0207] Parameter trend: {trend score}

[0208] Please evaluate whether the current status is good and suggest whether preventative maintenance is needed;

[0209] Input variables: {Overall score}, {Vibration score}, {Spectrum score}, {Trend score}

[0210] Applicable scenarios: When users ask questions such as "Is the device healthy?", "What is its health score?", or "Does it need maintenance now?"

[0211] Q5: Historical Comparative Analysis

[0212] Prompt word structure:

[0213] Please compare the current device status with the historical reference status:

[0214] Current state ({current time}):

[0215] Vibration amplitude: {current vibration}

[0216] Spectral similarity: {Current spectral similarity}

[0217] Operating mode: {Current operating condition}

[0218] Historical reference status ({historical time period}):

[0219] Vibration amplitude: {historical vibration}

[0220] Spectral similarity: {Historical Spectral Similarity}

[0221] Operating mode: {Historical operating conditions}

[0222] Please explain whether the current state shows a trend of deterioration compared to the past, and point out the possible sources of the change;

[0223] Input variables: {current time}, {current vibration}, {current spectrum similarity}, {current operating condition}, {historical time period}, {historical vibration}, {historical spectrum similarity}, {historical operating condition};

[0224] Applicable scenarios: When users ask questions like "Is it better or worse now than last month?", "Has the condition deteriorated recently?", or "Is the device showing signs of aging?"

[0225] Q6: Diagnosis and Recommendation Q&A (Enhanced Free-for-All Questions)

[0226] Prompt word structure:

[0227] User Question: {User Issue}

[0228] Current data supports:

[0229] Operating condition identification result: {Operating condition label}

[0230] List of abnormal characteristics: {abnormal list}

[0231] Alarm Log: {Alarm Information}

[0232] Parameter trend summary: {trend summary}

[0233] Please combine data and fault knowledge to generate a natural language answer, including the judgment result, the basis, and suggested actions;

[0234] Input variables: {User Issues}, {Operating Condition Labels}, {Abnormal List}, {Alarm Information}, {Trend Summary};

[0235] Applicable scenarios: When users ask questions like "What do you think of this status?", "Is there a possibility of problems later?", or "Should we shut down the service immediately?"

[0236] The scoring calculation method for the aforementioned health assessment can be carried out using existing technologies and methods, and will not be elaborated further in this example.

[0237] Reference Figure 2 and Figure 3 This invention provides an equipment condition monitoring system based on automatic operating condition calibration and multidimensional signal analysis. The monitoring system includes a multi-source sensor module, a data acquisition unit, an analysis and storage unit, a display and interaction unit, a communication interface module, and a power supply module, wherein:

[0238] The multi-source sensor module is installed on the monitored equipment to synchronously collect signals of vibration, temperature, acoustics, current, speed, and power, and outputs analog signals through BNC, RS-485 and 4~20mA standard physical interfaces to obtain full operating information covering high and low frequencies.

[0239] The data acquisition unit includes a sensor interface, power isolation, analog conditioning and analog-to-digital conversion circuits, which are used to filter, amplify and sample analog signals from multi-source sensor modules to form a digital signal stream, and send the digital signal stream to the analysis and storage unit in real time via SPI, CAN and Ethernet buses;

[0240] The analysis storage unit, which integrates an embedded processor and a local hot database, is used to perform hierarchical processing on digital signals, specifically including:

[0241] The automatic operating condition calibration module generates an operating condition reference vector based on the shutdown or operating mode during the installation phase.

[0242] The operating condition identification module outputs operating condition labels in real time based on comprehensive deviation measurement and sliding window algorithm during the operation phase;

[0243] The condition analysis model module can call spectrum analysis, shaft center trajectory analysis and parameter trend analysis models according to the working conditions to extract multi-dimensional features of shape and numerical values.

[0244] The health assessment module performs threshold determination on numerical features and similarity determination on shape features, and outputs health scores and alarm records.

[0245] The data interface service module provides the evaluation results to the display unit and communication interface module through a unified data interface, enabling local and remote access.

[0246] The display interaction unit is a touch screen or an external HDMI display. It communicates with the analysis and storage unit through a serial port or HDMI interface. It is used to display operating condition labels, characteristic curves, alarm information and health scores in real time in the form of curves, spectrum graphs, axis trajectory graphs, etc., and supports local query and confirmation by users.

[0247] The communication interface module consists of USB, RS-485, Ethernet, HDMI and audio interfaces built into the analysis and storage unit. It is used to send alarm codes and control commands to the host computer and PLC (RS-485), synchronize monitoring data, configuration parameters and logs to the remote monitoring platform (Ethernet), provide firmware upgrade, data export or debugging (USB), and output audible and visual alarm prompts (audio).

[0248] The power module provides DC12V / 24V regulated power and is equipped with overvoltage and undervoltage protection circuits to provide safe and reliable power for the data acquisition unit, analysis and storage unit and display and interaction unit.

[0249] The multi-source sensor module is connected to the data acquisition unit via a standard cable. The data acquisition unit is connected to the analysis and storage unit via a bus interface. The analysis and storage unit is connected to the display unit via a serial port and HDMI. All modules are installed in an integrated chassis and are internally connected via a PCB backplane and ribbon cables.

[0250] It should be noted that this invention collects equipment operation data in real time through multi-source sensors such as vibration, temperature, and sound. After numerical conversion and preprocessing by the data acquisition unit, the data is sent to the analysis and storage unit. The system automatically identifies the current operating condition, calls the corresponding professional models such as spectrum analysis, shaft center trajectory, and health factors to extract state features, and completes the equipment status assessment. The analysis results are transmitted to the display and interaction unit for chart display via the data interface. At the same time, it supports intelligent agent workflow linkage, realizing closed-loop monitoring of the entire process from signal acquisition and status recognition to result display and intelligent feedback.

[0251] In an optional implementation, the analysis storage unit is used to perform operating condition calibration and identification, status analysis, and health assessment on the collected multidimensional raw data, including the following functions:

[0252] (1) Automatic calibration module for operating conditions

[0253] Operating condition calibration: By analyzing operating condition parameters (current, speed, power), the operating condition reference calibration is performed during the installation process. If the installation is performed offline and the machine is stopped, the parameter reference values ​​for the no-load operating condition are calibrated. If the installation is performed online and running, the parameter reference values ​​for the operating condition are calibrated.

[0254] Operating condition identification: During operation, if the change threshold of the baseline value based on the operating condition calibration parameters exceeds 20%, it is considered that the operating condition has changed, and the operating status of the equipment (such as no-load or running) is automatically identified.

[0255] (2) State Analysis Model Module

[0256] Select the appropriate model for signal processing based on the equipment's operating conditions, including:

[0257] Spectrum calculation: Fast Fourier Transform (FFT) is used to analyze spectral characteristics and obtain the spectrum.

[0258] Axis trajectory analysis: Obtains the axis position, axis trajectory shape, inner equivalent circle, and outer equivalent circle;

[0259] Parameter analysis and calculation: parameter trends, parameter exceedances, parameter volatility;

[0260] After calculating the features, the state analysis model module takes the operating condition data based on the working condition identification results, calibrates the normal baseline values ​​for the output features of the analysis model, sets thresholds for numerical features (such as parameter trends, parameter over-limits, parameter fluctuations, shaft center position, inner equivalent circle of the shaft center, and outer equivalent circle radius), and sets standard shape references for shape features (such as spectrum and shaft center trajectory shape).

[0261] (3) Health assessment module

[0262] The output features of the status analysis model are input into the evaluation logic to perform qualitative (normal / abnormal) and quantitative (scoring) judgments on the current equipment status, generating evaluation results and alarm records, which are then saved to the analysis database. The evaluation logic is divided into numerical and shape types according to the feature type: numerical features calculate the deviation between the real-time feature and the benchmark value set threshold, and if it exceeds the set level, it is considered abnormal; shape features compare the similarity between the real-time shape and the benchmark value set standard shape reference, and if the similarity is lower than the set level, it is considered abnormal.

[0263] For each type of state analysis model feature, if an anomaly is identified, an alarm for that type will be output.

[0264] In a preferred embodiment, the display interaction unit receives the analysis results in real time and presents them visually through a web interface or touch display terminal. The displayed content includes sensor data curves, spectrum diagrams, axis trajectory diagrams, equipment operating status, alarm information, and operating condition identification results. It also provides a unified data access interface to support operations such as data reading, alarm status acquisition, and equipment status query, which facilitates the integration of the intelligent agent workflow.

[0265] Furthermore, the display interaction unit can trigger rule-based processes (such as fault Q&A, knowledge retrieval, and alarm linkage) based on changes in device operating status or user requests. After the locally deployed large language model calls the data interface to query data, it performs prompt word enhancement (user question + query result) to generate a natural language answer and output it to the user interaction window.

[0266] In the application of the above embodiments, the equipment status monitoring system based on automatic operating condition calibration and multidimensional signal analysis disclosed in this embodiment of the invention further includes one or more processors and a memory.

[0267] The memory is used to store operable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations, including the flow of the equipment condition monitoring method based on automatic operating condition calibration and multidimensional signal analysis in the foregoing embodiments, especially... Figure 1 The flowchart of the method is shown.

[0268] Other aspects disclosed in the embodiments of the present invention also propose a computer-readable medium for storing software including instructions executable by one or more computers, which, upon execution, cause the one or more computers to perform operations including the flow of the equipment condition monitoring method based on automatic operating condition calibration and multidimensional signal analysis of the foregoing embodiments, particularly... Figure 1 The flowchart of the method is shown.

[0269] It should be recognized that embodiments of the present invention may be implemented or carried out by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer-readable storage medium.

[0270] The method can be implemented using standard programming techniques, including a non-transitory computer-readable storage medium configured with a computer program in the computer program, wherein the storage medium is configured such that the computer operates in a specific and predefined manner.

[0271] Each program can be implemented in a high-level procedural or object-oriented programming language to communicate with the computer system; however, if required, the program can be implemented in assembly or machine language.

[0272] In any case, the language can be either compiled or interpreted.

[0273] Furthermore, for this purpose, the program can run on programmed application-specific integrated circuits.

[0274] The processes described herein (or variations and / or combinations thereof) can be executed under the control of one or more computer systems configured with executable instructions, and can be implemented by hardware or a combination thereof as code (e.g., executable instructions, one or more computer programs, or one or more applications) that commonly executes on one or more processors. The computer program includes a plurality of instructions executable by one or more processors.

[0275] Furthermore, the method can be implemented in any suitable computing platform, including but not limited to personal computers, minicomputers, mainframes, workstations, networked or distributed computing environments, standalone or integrated computer platforms, or in communication with charged particle tools or other imaging devices.

[0276] Various aspects of the present invention can be implemented in machine-readable code stored on a non-transitory storage medium or device, whether portable or integrated into a computing platform, such as a hard disk, optical read and / or write storage medium, RAM, ROM, etc., such that it can be read by a programmable computer, and when the storage medium or device is read by the computer, it can be used to configure and operate the computer to perform the processes described herein.

[0277] Furthermore, machine-readable code, or parts thereof, can be transmitted via wired or wireless networks.

[0278] When such media includes instructions or programs that combine with a microprocessor or other data processor to implement the steps described above, the invention described herein includes these and other different types of non-transitory computer-readable storage media.

[0279] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for monitoring equipment condition based on automatic operating condition calibration and multidimensional signal analysis, characterized in that, include: During the equipment installation phase, operating parameters are collected, and a working condition baseline vector is established based on outlier removal and mean calculation. The method for constructing the operating condition reference vector includes: Raw sample sequences of current, rotational speed, and power were collected within the calibration time window. For each type of sample, an interquartile range outlier removal mechanism is applied, using an adjustable coefficient of 1.5 to 3 to filter out outliers; The mean of the removed samples is calculated to obtain the parameter baseline values ​​for the shutdown condition and the operating condition, which are the operating condition baseline vectors. The mathematical expression of the working condition reference vector is as follows: in, , , These are the average values ​​of current, speed, and power, used to construct the operating condition reference vector. ; During equipment operation, real-time operating parameters are acquired and compared with the operating condition baseline vector to generate a comprehensive deviation metric. The current operating condition label is then determined using a continuity determination strategy. The continuity determination strategy includes: Establish a first-in-first-out buffer with a length of 60 sampling points and update the deviation judgment results in real time; Count the number of samples whose deviation exceeds the limit within the buffer; When the cumulative number of out-of-limit samples reaches 15, confirm that the operating condition has been switched and update the operating condition label immediately. When the number of out-of-limit samples is less than 15, it is considered a normal fluctuation, and the original operating condition label remains unchanged. The weighted overall deviation metric is defined as follows: in, , , These represent the relative deviations of the current, speed, and power from their reference values, respectively. , , For the corresponding weights; Based on the operating condition label, the corresponding signal analysis model is invoked to extract spectral features, shaft center trajectory features, and operating parameter numerical features; the method for obtaining the shaft center trajectory features includes: The rotor's X and Y displacement data are simultaneously collected by two orthogonal displacement sensors to form a planar trajectory; The trajectory sequences of multiple complete rotor cycles are time-aligned and coordinate-normalized to eliminate the influence of speed fluctuations. The average curve of each normalized trajectory is calculated as the standard reference trajectory, which is the shape feature of the axis trajectory. Numerical features of mean, volatility, and trend slope are extracted from real-time operating parameters, including: Establish a configurable sliding observation window for each real-time operating parameter to collect continuous sampled values ​​within the window; The average value of the sampled values ​​is calculated within the observation window to obtain the instantaneous mean of the parameter in the current window, which is used to characterize the steady-state level. The standard deviation of the sampled values ​​is calculated within the same window, and this standard deviation is used as a parameter volatility index to measure short-term dispersion. The least squares method is used to perform a linear fit on the curve of the sampled values ​​within the window over time. The slope of the fitted line is the trend slope, which reflects the trend of parameter rise and fall. The mean, volatility, and trend slope obtained above are combined to form the numerical feature vector at the current moment; the acquired multidimensional features are compared with preset thresholds to determine whether the equipment status is abnormal and generate alarm information; the operating condition labels, feature curves, abnormal categories, and alarm records are output in a visual manner and support question-and-answer interaction based on a large language model.

2. The equipment condition monitoring method based on automatic operating condition calibration and multidimensional signal analysis according to claim 1, characterized in that, The method for generating the comprehensive deviation metric includes: At startup, the variance of each operating parameter during the historical stable period is read, and weights are allocated according to the principle that the smaller the variance, the higher the weight. Calculate the relative deviations between the real-time operating condition parameters and the operating condition reference vector, multiply the deviations by their corresponding weights, and then normalize them to form a comprehensive deviation metric.

3. The equipment condition monitoring method based on automatic operating condition calibration and multidimensional signal analysis according to claim 2, characterized in that, The method for obtaining the spectral features includes: The original vibration signal is first subjected to window function weighting and bandpass filtering to suppress end effect and eliminate irrelevant frequency bands; The vibration amplitude spectrum is calculated using Fast Fourier Transform, and the amplitude spectra of multiple time periods are averaged under operating conditions to generate a reference spectrum. The similarity between the real-time amplitude spectrum and the reference spectrum is used as a shape feature.

4. A monitoring system based on the equipment condition monitoring method of automatic operating condition calibration and multidimensional signal analysis as described in claim 1, characterized in that, The monitoring system includes a multi-source sensor module, a data acquisition unit, an analysis and storage unit, and a display and interaction unit, wherein: The multi-source sensor module is used to collect vibration, sound, temperature and electrical parameter signals; The data acquisition unit is used to condition and digitize the acquired signals; The analysis and storage unit is used to perform working condition calibration, operation status identification and multi-dimensional feature analysis, and to determine anomalies, save operation data and support trend analysis, playback and edge computing; The display and interaction unit is used to visualize the monitoring results and, in conjunction with a large language model, to achieve question-and-answer interaction and remote data communication.

5. The equipment condition monitoring system based on automatic operating condition calibration and multidimensional signal analysis according to claim 4, characterized in that, It also includes one or more processors; The memory stores operable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations, including the flow of the equipment condition monitoring method based on automatic operating condition calibration and multidimensional signal analysis as described in any one of claims 1 to 3.

6. A computer-readable medium for storing software, characterized in that: The software includes instructions executable by one or more computers, which cause the one or more computers to perform operations, including the flow of the equipment condition monitoring method based on automatic operating condition calibration and multidimensional signal analysis as described in any one of claims 1 to 3.

Images