Metering device fault early warning method, system and equipment based on industrial Internet of Things

Abstract

The invention discloses a metering device fault early warning method, system and device based on the industrial Internet of Things, and relates to the technical field of metering device fault early warning methods. The invention provides a metering instrument fault early warning method based on industrial Internet of Things, and the method comprises the steps: judging whether a target instrument has abnormal data or not according to a first data set of the target instrument in a preset time period; if the target instrument has the abnormal data, acquiring a second data set of the target instrument in a preset time period; according to the abnormal data and the second data set, evaluating the fault probability of the fault condition of the target instrument; and when the fault probability is greater than a preset threshold value, sending early warning information to a target user.

Description

Technical Field

[0001] This application relates to the field of metering instrument fault warning methods, and particularly to a metering instrument fault warning method, system and device based on the industrial Internet of Things. Background Art

[0002] With the rapid development of industrial Internet of Things (IIoT) technology, metering instruments, as the core monitoring devices in the industrial production process, real-time monitoring of their operating status and fault warning have become key requirements in the field of intelligent manufacturing.

[0003] In the prior art, alarms are usually triggered based on a single parameter threshold. However, in the industrial environment, instrument data is often interfered by complex working conditions, and such methods are prone to false alarms or missed alarms. Moreover, if the metering instrument is only detected after a fault occurs, it is easy to have a lag in maintenance, which is not conducive to early discovery of potential faults in the metering instrument and leads to waste of operation and maintenance resources. Summary of the Invention

[0004] The main purpose of this application is to provide a metering instrument fault warning method, system and device based on the industrial Internet of Things, aiming to solve the technical problem of the lag in maintenance that is prone to occur in the prior art by relying on a single parameter threshold to trigger an alarm.

[0005] To achieve the above object, in the first aspect, this application provides a metering instrument fault warning method based on the industrial Internet of Things, including:

[0006] Judging whether there is abnormal data in the target instrument according to the first data set of the target instrument in a preset time period;

[0007] If there is abnormal data in the target instrument, obtaining a second data set of the target instrument in a preset time period;

[0008] Evaluating the fault probability of the target instrument having a fault condition according to the abnormal data and the second data set;

[0009] When the fault probability is greater than a preset threshold, sending a warning message to the target user.

[0010] Optionally, the step of judging whether there is abnormal data in the target instrument according to the first data set of the target instrument in a preset time period includes:

[0011] Obtaining the first data set of the target instrument in a preset time period, and dividing the first data set into at least one data group according to the attributes of the data;

[0012] Traversing each data in the first data set, comparing each data with a first threshold, and judging whether there is first suspicious data in the first data set;

[0013] If there is first suspicious data in the first data set, mark the first suspicious data as abnormal data;

[0014] If there is no first suspicious data in the first data set, analyze the at least one data group to determine whether there is second suspicious data in the first data set;

[0015] If there is second suspicious data in the first data set, mark all the data in the data group where the second suspicious data is located as the abnormal data.

[0016] Optionally, the step of, if there is no first suspicious data in the first data set, analyzing the at least one data group to obtain whether there is second suspicious data in the first data set includes:

[0017] According to the attribute information of each data group, obtain the contribution degree of each data group to the fault situation when the target instrument has a historical fault;

[0018] According to the contribution degree of each data group to the fault of the target instrument, obtain the priority of each data group;

[0019] Sort all the data groups according to the priority to obtain a sorting result;

[0020] Analyze a preset proportion of data groups according to the sorting result to obtain the change trend of the data of each data group;

[0021] According to the change trend, determine whether there is second suspicious data in the first data set.

[0022] Optionally, the step of, if there is abnormal data in the target instrument, obtaining a second data set of the target instrument in a preset time period includes:

[0023] The second data set at least includes a vibration data packet, an audio data packet, a temperature data packet, and historical maintenance information of the target instrument.

[0024] Optionally, the step of evaluating the fault probability of the target instrument having a fault situation according to the abnormal data and the second data set includes:

[0025] Obtain a first evaluation factor according to the deviation degree and / or change trend of the abnormal data;

[0026] Obtain a second evaluation factor according to the vibration data packet of the second data set,

[0027] Obtain a third evaluation factor according to the audio data packet of the second data set,

[0028] Obtain a fourth evaluation factor according to the temperature data packet of the second data set.

[0029] Evaluate the failure probability of the target instrument having a failure according to the first evaluation factor, the second evaluation factor, the third evaluation factor, and the fourth evaluation factor.

[0030] Optionally, the step of evaluating the failure probability of the target instrument having a failure according to the abnormal data and the second data set further includes:

[0031] Predict the components of the target instrument having a failure according to the abnormal data and the second data set.

[0032] Obtain whether the predicted failed components belong to the components in the historical maintenance information according to the historical maintenance information of the target instrument.

[0033] If so, use the changing trend of the maintenance frequency of the component as the fifth evaluation factor.

[0034] Optionally, the step of evaluating the failure probability of the target instrument having a failure according to the abnormal data and the second data set includes:

[0035] The mathematical expression for evaluating the failure probability of the target instrument having a failure is:

[0036]

[0037] Among them, μ represents the mean value of the abnormal data, μ1 represents the normal operating condition reference value, σ represents the normal operating condition standard deviation, represents the slope of the data change trend, and α represents the weight coefficient of the first evaluation factor;

[0038] A max ,A min represents the safety amplitude range threshold, A represents the current vibration amplitude, B represents the total number of audio samples, b represents the number of abnormal frequency points, T0 represents the current temperature measurement value, T1 represents the temperature warning threshold, T2 represents the rated temperature, and β represents the weight coefficients of the second evaluation factor, the third evaluation factor, and the fourth evaluation factor;

[0039] N C (t) represents the number of repairs of the component within the time window [t - T, t], represents the frequency sensitivity coefficient, e -ρ(t-T) represents that the weight of the earlier maintenance record is lower, ρ represents the attenuation coefficient, and γ represents the weight coefficient of the fifth evaluation factor.

[0040] Optionally, the step of sending a warning message to the target user when the failure probability is greater than the preset threshold includes:

[0041] Set at least two warning levels according to the preset threshold;

[0042] Obtain the warning level corresponding to the failure probability according to the failure probability;

[0043] Send a warning message to the target user according to the warning level;

[0044] If the target instrument triggers a warning three times in a row, the warning message sent to the target user is upgraded by one level on the basis of the original warning level.

[0045] In a second aspect, the present application provides a metering instrument failure warning system based on the industrial Internet of Things, including a management platform, a sensing network platform, and an object platform that are communicatively connected in sequence:

[0046] The sensing network platform is configured to:

[0047] Judge whether there is abnormal data in the target instrument according to the first data set of the target instrument in a preset time period;

[0048] If there is abnormal data in the target instrument, obtain the second data set of the target instrument in a preset time period;

[0049] Evaluate the failure probability of the target instrument in case of failure according to the abnormal data and the second data set;

[0050] The management platform is configured to:

[0051] When the failure probability is greater than the preset threshold, send a warning message to the target user.

[0052] In a third aspect, the present application provides a computer device, characterized in that the computer device includes a memory and a processor, a computer program is stored in the memory, and the processor executes the computer program to implement the method as described above.

[0053] The beneficial effects that the present application can achieve:

[0054] A method, system, and device for fault warning of a metering instrument based on the industrial Internet of Things proposed in the embodiments of this application include: judging whether there is abnormal data in the target instrument according to the first data set of the target instrument in a preset time period; if there is abnormal data in the target instrument, obtaining the second data set of the target instrument in the preset time period; evaluating the fault probability of the target instrument having a fault condition according to the abnormal data and the second data set; when the fault probability is greater than a preset threshold, sending a warning message to the target user. By monitoring and analyzing the data of the target instrument in real time, this method can identify abnormal data before a fault occurs, thereby discovering potential fault hazards in advance. This helps to avoid production interruptions or equipment damage caused by sudden faults, and reduces maintenance costs and production losses. Combining the first data set and the second data set of the target instrument in the preset time period, as well as the specific information of the abnormal data, to evaluate the fault probability. This multi-dimensional data analysis method improves the accuracy of fault warning and reduces the possibility of false alarms and missed alarms. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a schematic flowchart of the method for fault warning of a metering instrument according to the embodiments of this application;

[0056] Figure 2 It is a schematic framework diagram of the service platform involved in this application;

[0057] Figure 3 It is a schematic framework diagram of the management platform involved in this application;

[0058] Figure 4 It is a schematic framework diagram of the sensor network platform involved in this application.

[0059] The implementation, functional features, and advantages of the purpose of this application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0060] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts belong to the scope of protection of the present invention.

[0061] It should be noted that all directional indications (such as up, down, left, right, front, back...) in the embodiments of the present invention are only used to explain the relative position relationship and movement conditions between components in a specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indications will also change accordingly.

[0062] In the present invention, unless otherwise clearly specified or limited, terms such as "connection" and "fixation" shall be understood in a broad sense. For example, "fixation" can be a fixed connection, a detachable connection, or integrated; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and can be the communication inside two components or the interaction relationship between two components, unless otherwise clearly limited. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0063] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" can explicitly or implicitly include at least one of such features. In addition, the meaning of "and / or" appearing throughout the text includes three parallel scenarios. Taking "A and / or B" as an example, it includes Scenario A, or Scenario B, or the scenario where both A and B are satisfied simultaneously. In addition, the technical solutions between various embodiments can be combined with each other, but it must be based on the ability of those of ordinary skill in the art to implement. When the combination of technical solutions results in contradictions or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0064] Embodiment 1

[0065] Referring to Figure 1 , the first embodiment of the present application provides a method for fault warning of a measuring instrument based on the industrial Internet of Things, including the following operating steps:

[0066] S10. Determine whether there is abnormal data in the target instrument according to the first data set of the target instrument in a preset time period.

[0067] Optionally, the target meter can be an intelligent water meter in an industrial production process, or other fluid metering instruments, etc. The first data set of the target meter in a preset time period (such as the past 24 hours) is collected through an industrial Internet of Things gateway. The first data set at least includes: time series data of core parameters such as basic operating conditions data such as flow rate, pressure, temperature, voltage, etc. Operating status label: normal / abnormal mark. After collecting the first data set, missing values caused by communication interruptions are usually removed, and instantaneous noise is eliminated through moving window mean filtering. In the process of abnormal data detection, data analysis algorithms (such as statistical analysis, machine learning models, etc.) can be used to analyze the first data set to identify data points that deviate significantly from the normal operating mode, that is, abnormal data. A reasonable abnormal judgment criterion, such as data fluctuation range, change rate, etc., is set in the process of abnormal data detection to ensure accurate identification of abnormal data. After detecting abnormal data, an abnormal data report can be output, including the type of abnormality (deviation / trend), parameter list and time stamp.

[0068] S20. If there is abnormal data in the target meter, obtain the second data set of the target meter in a preset time period.

[0069] Optionally, after detecting that there is abnormal data, in order to more comprehensively understand the operating status of the meter, the second data set is further obtained, and the factors used to evaluate the failure probability of the target meter are expanded through the second data set, so as to realize multi-dimensional evaluation of the failure probability of the target meter and reduce the situation where the probability of misjudgment is relatively high when only a single parameter threshold is used for failure probability prediction in the prior art.

[0070] S30. Evaluate the failure probability of the target meter in case of failure according to the abnormal data and the second data set.

[0071] Optionally, analyze the abnormal data and the second data set to identify possible failure modes or causes. Use a failure prediction model (such as a prediction algorithm based on machine learning) to combine historical failure data and current data to calculate the probability of the target meter failing. Regularly verify and update the failure prediction model to ensure its accuracy and reliability.

[0072] S40. When the failure probability is greater than a preset threshold, send a warning message to the target user.

[0073] Optionally, according to the importance of the meter and the severity of the consequences of the failure, set a reasonable failure probability threshold. When the failure probability exceeds the threshold, automatically generate a warning message, including the failure probability, possible failure causes, recommended maintenance measures, etc. Through the industrial Internet of Things system, the warning message is sent to the target user (such as maintenance personnel, management personnel, etc.) in a timely manner so that they can take corresponding actions.

[0074] Embodiment 2

[0075] Based on Embodiment 1, this embodiment provides a method for fault warning of metering instruments based on the industrial Internet of Things, including the following operating steps:

[0076] S10. According to the first data set of the target instrument in a preset time period, determine whether there is abnormal data in the target instrument.

[0077] Optionally, the step of determining whether there is abnormal data in the target instrument according to the first data set of the target instrument in a preset time period includes:

[0078] S101. Obtain the first data set of the target instrument in a preset time period, and divide the first data set into at least one data group according to the attributes of the data;

[0079] Specifically, based on the data set of the first data set, the grouping basis of the data set can be time-series data groups, and the time-series data groups can also be further divided into flow, pressure, temperature, etc.; it can also be grouped according to sensor types, such as flow sensor groups, temperature sensor groups, and vibration sensor groups, etc.; grouped according to environmental data groups, such as environmental temperature and humidity, vibration level; grouped according to physical locations, such as instrument housing groups, installation base groups; grouped according to status data groups, which can include digital signals, such as power status, valve opening and closing; grouped according to device function modules, such as power module groups, control module groups. The division rules for data include: homologous grouping: merging multi-dimensional data of the same sensor (such as the X / Y / Z axis data of a three-axis vibration sensor are grouped into the same group). Time-series alignment: performing timestamp interpolation on asynchronously collected data to ensure time synchronization of data within the group (error < 10ms).

[0080] S102. Traverse each data in the first data set, compare each data with a first threshold, and determine whether there is first suspicious data in the first data set;

[0081] Specifically, check each data point in the first data set one by one. Compare each data point with a preset first threshold. The first threshold can be set according to the normal operating range of the instrument, historical data, or industry standards. If a certain data point exceeds the range of the first threshold, it is marked as first suspicious data. The first threshold can also be a dynamic threshold, which can be generated according to historical data statistics (such as mean ± 3ε, ε is the rolling standard deviation). It can also detect outliers through a pre-trained Isolation Forest model, and when the confidence level > 90%, it is determined as suspicious. And output a list of suspicious data, which can include parameter names, timestamps, deviation values (such as "temperature sensor T1, 2025-02-21 14:05:00, measured value 92°C vs threshold 85°C").

[0082] S103: If there is first questionable data in the first data set, mark the first questionable data as abnormal data;

[0083] Specifically, if a single parameter exceeds the threshold for three consecutive sampling periods, it is marked as a deterministic anomaly. If two or more parameters in the same data group exceed the limit simultaneously (for example, flow and pressure drop at the same time), it is marked as a correlation anomaly. The first suspicious data point is also marked as an anomaly. This means that these data points have significantly deviated from the normal operating range, which may indicate a fault or anomaly in the target instrument, or indicate a potential fault.

[0084] S104: If the first data set does not contain the first questionable data, analyzing the at least one data set to determine whether the first data set contains the second questionable data;

[0085] Specifically, if no data point in the first data set is marked as the first suspicious data, the next step of analysis is performed. This may include statistical analysis, trend analysis, pattern recognition, etc. to identify abnormal patterns or trends in the data group. Through analysis, if it is found that the data points in a certain data group show abnormal patterns or trends, the data points in the data group are marked as the second suspicious data.

[0086] Cluster analysis: Perform K-means clustering (k=2) on the data set. If the distance between the centers of normal and abnormal clusters exceeds 20% of the historical baseline, the data is considered suspicious. Alternatively, a Fourier transform can be performed to detect spectral energy anomalies (e.g., a 50% drop in the fundamental frequency amplitude) of periodic parameters (e.g., flow pulses). Alternatively, a sliding window entropy value can be performed to calculate the information entropy of the data within the window. If the entropy value suddenly increases (e.g., from 1.2 to 2.5), it is considered random noise interference.

[0087] Optionally, if the first data set does not contain the first questionable data, the step of analyzing the at least one data set to determine whether the first data set contains the second questionable data includes:

[0088] S1041. Obtain, based on the attribute information of each data group, the contribution of each data group to the fault condition when the target instrument has a historical fault;

[0089] Specifically, the target instrument's fault history from the past three years is extracted and categorized by fault type (e.g., mechanical wear, circuit failure, sensor drift). For each fault event, the parameter anomaly records for each data group (e.g., temperature group, vibration group) at the time of the fault are correlated. A random forest model is trained on this historical fault data, outputting a feature importance score for each data group and its contribution to the fault.

[0090] S1042. Obtain the priority of each data group according to the contribution degree of each data group to the failure of the target instrument;

[0091] Specifically, map the contribution degree to the priority level. Further, if a data group frequently triggers an anomaly in the near future (such as within 30 days), temporarily increase the priority (for example, +1 level).

[0092] S1043. Sort all data groups according to the priority to obtain a sorting result;

[0093] Specifically, sort from high to low according to the priority, and refine the sorting of groups with the same level according to the contribution degree value. For failure types involving safety risks (such as overvoltage, leakage), their associated data groups are forced to the top.

[0094] S1044. Analyze a preset proportion of data groups according to the sorting result to obtain the change trend of the data of each data group;

[0095] Specifically, by default, select the top 20% of the data groups in the sorting (such as selecting the top 2 groups from 10 groups), which can be configured to be 10% - 30%. If the high-priority groups contain safety-critical parameters (such as pressure, voltage), force all to be analyzed. The specific analysis of the change trend is as follows: calculate the mean change rate within a 5-minute window size, use the CUSUM algorithm to identify the moment of trend mutation, perform FFT on the vibration data, and detect the change in the energy of the characteristic frequency (such as the energy of the bearing fault characteristic frequency increases by 30%).

[0096] S1045. Judge whether there is a second suspicious data in the first data set according to the change trend;

[0097] Specifically, for the determination of the second suspicious data, the trend anomaly criteria are as follows: the slope continuously exceeds 2 times the standard deviation of the historical baseline for 10 minutes; the CUSUM detects that 3 consecutive sampling points exceed the control line; the change rate of the characteristic frequency energy > 50%, etc. If a data group meets any of the above trend anomaly criteria, it is determined as the second suspicious data. If multiple data groups are abnormal at the same time and there is a physical correlation (such as the temperature group and the vibration group belong to the same motor module), it is marked as a composite anomaly.

[0098] S105. If there is a second suspicious data in the first data set, mark all the data in the data group where the second suspicious data is located as the abnormal data.

[0099] Specifically, if ≥30% of the parameters within a group are detected as latent anomalies, or the timing pattern of the key parameters within the group (such as the main flowmeter) differs from the historical baseline by more than a threshold (DTW distance > 5), all the data within that group are marked as abnormal data. Further, if the group-level anomaly persists for ≥5 minutes, all the data of the entire group during that time period are marked as abnormal. If adjacent data groups (such as the temperature group and the vibration group) are abnormal simultaneously, it is marked as a systematic failure. This avoids misjudgment caused by local interference (such as transient noise from a single sensor) and improves the detection sensitivity for complex faults (such as multi-parameter drift caused by mechanical looseness).

[0100] S20. If there is abnormal data in the target instrument, obtain the second data set of the target instrument within a preset time period.

[0101] Optionally, the second data set at least includes a vibration data packet, an audio data packet, a temperature data packet, and the historical maintenance information of the target instrument.

[0102] Specifically, for the vibration data packet, its data acquisition can be carried out through a triaxial acceleration sensor. The key parameters of the vibration data packet mainly include:

[0103] Time-domain features: vibration amplitude (RMS), peak value, waveform factor.

[0104] Frequency-domain features: energy ratio in the 1 - 5 kHz frequency band, bearing fault characteristic frequencies (such as BPFI / BPFO).

[0105] Acquisition rule: Continuously acquire at least 5 minutes of data after the anomaly is triggered, covering 2 minutes before and after the anomaly period.

[0106] For the audio data packet, its data acquisition can be carried out through an industrial-grade noise-canceling microphone array (frequency response range 20 Hz - 20 kHz, sampling rate 44.1 kHz); it mainly includes:

[0107] Sound pressure level (dB): Real-time monitoring of the noise intensity.

[0108] Mel-frequency cepstral coefficients (MFCC): Extract 13-dimensional features to identify abnormal sound patterns (such as friction, impact sounds).

[0109] Acquisition rule: Synchronously record the audio waveform during the anomaly period and perform frame division processing (each frame is 30 ms).

[0110] For the temperature data packet, its data acquisition can use an infrared thermometer (accuracy ±0.5°C) and an embedded thermal resistor (PT100).

[0111] The temperature monitoring points mainly include:

[0112] Mechanical components: motor windings, bearing housings, gearbox surfaces.

[0113] Electronic components: circuit board hot spots, heat sinks for power modules.

[0114] The main key parameters collected include:

[0115] Temperature gradient: Calculate the temperature rise rate per minute (℃ / min).

[0116] Thermal field distribution: Generate the standard deviation of the temperature field through multi-point temperature measurement.

[0117] Regarding the historical maintenance information, its data source can be the enterprise maintenance management system, covering the records of the past 3 years; the historical maintenance information includes key fields: the name and code of the faulty component (such as bearing B-2039). Maintenance type: replacement, calibration, lubrication. Maintenance cycle statistics: mean time between failures (MTBF), recent maintenance frequency trend.

[0118] S30. According to the abnormal data and the second data set, evaluate the failure probability of the target instrument in case of failure.

[0119] Optionally, the step of evaluating the failure probability of the target instrument in case of failure according to the abnormal data and the second data set includes:

[0120] S301. Obtain a first evaluation factor according to the deviation degree and / or change trend of the abnormal data;

[0121] Specifically, clean and preprocess the collected abnormal data, remove noise and outliers to ensure data accuracy. Calculate the deviation degree of the abnormal data from the normal data range (such as the range defined by statistical quantities such as mean and standard deviation). This can be achieved by calculating the difference between the abnormal data and the mean, and then dividing by the standard deviation to obtain a standardized deviation degree value. Analyze the change trend of the abnormal data over time, such as whether it shows a continuous upward or downward trend, or whether it has the characteristics of periodic fluctuations. According to the analysis results of the deviation degree and change trend, comprehensively give a first evaluation factor to reflect the abnormality degree and potential failure risk of the abnormal data.

[0122] S302. Obtain a second evaluation factor according to the vibration data packet of the second data set;

[0123] Specifically, extract the vibration data packet from the second data set, including information such as vibration frequency, amplitude, and vibration mode. Conduct feature analysis on the vibration data, such as spectral analysis and time-domain analysis, to extract characteristic parameters that can reflect the operating state of the equipment. According to the results of the vibration feature analysis, determine a second evaluation factor to reflect the abnormality degree and potential failure risk of the equipment vibration state.

[0124] S303. Obtain a third evaluation factor based on the audio data packets of the second data set;

[0125] Extract audio data packets from the second data set, including information such as the frequency, amplitude, and timbre of the audio signal. Conduct feature analysis on the audio data, such as voice recognition and spectrum analysis, to extract voice feature parameters that can reflect the operating state of the device. Based on the results of the audio feature analysis, determine a third evaluation factor to reflect the degree of abnormality in the device's voice state and the potential failure risk.

[0126] S304. Obtain a fourth evaluation factor based on the temperature data packets of the second data set;

[0127] Extract temperature data packets from the second data set, including information such as the temperature values and temperature change rates of various parts of the device. Conduct feature analysis on the temperature data, such as temperature distribution analysis and temperature trend prediction, to extract feature parameters that can reflect the thermal state of the device. Based on the results of the temperature feature analysis, determine a fourth evaluation factor to reflect the degree of abnormality in the device's thermal state and the potential failure risk.

[0128] S305. Evaluate the failure probability of the target instrument in the event of a failure based on the first evaluation factor, the second evaluation factor, the third evaluation factor, and the fourth evaluation factor.

[0129] Optionally, the step of evaluating the failure probability of the target instrument in the event of a failure based on the abnormal data and the second data set further includes:

[0130] S310. Predict the components of the target instrument in the event of a failure based on the abnormal data and the second data set;

[0131] Specifically, construct a feature comparison library, such as establishing a typical fault vibration spectrum library (e.g., the BPFI frequency corresponding to the inner ring fault of the bearing, the sideband corresponding to the broken gear teeth); store standard fault voiceprints (e.g., the high-frequency whistling of the motor winding short circuit, the low-frequency rumbling of the pump cavitation); record the overheating characteristics of components (e.g., the bearing overheating temperature gradient > 2 °C / min, the circuit board hot spot temperature difference > 10 °C). During the prediction process, calculate the similarity between the real-time vibration spectrum and the feature library (a cosine similarity > 0.8 is determined to be a match), and match the abnormal audio with the fault voiceprint library through MFCC dynamic time warping (DTW). If the temperature data is consistent with the typical overheating mode of a certain component (e.g., the motor winding temperature field shows a gradient distribution), then mark that component. Output the prediction result, a list of predicted faulty components (e.g., bearing B-2039, circuit board C-205), along with the confidence level (e.g., the bearing failure probability is 78%).

[0132] For ease of understanding, an example is given as follows:

[0133] An intelligent water meter (device ID: WT-203) has collected abnormal data.

[0134] The abnormal data is as follows: the average flow rate has decreased to 6 m 3 / day (normal benchmark is 10 m 3 / day), and the trend slope is -1.5 m 3 / day 2 .

[0135] The second dataset collected is as follows:

[0136] The vibration spectrum shows abnormal high-frequency energy (the energy ratio in the range of 1 - 3 kHz is 80%).

[0137] Periodic "click" sounds are detected in the audio (MFCC matches the tooth missing sound pattern of the gear).

[0138] The temperature monitoring is normal (25 °C).

[0139] Historical records: The gear set has been replaced 2 times in the past year, with an average interval of 180 days, and the last repair was 60 days ago.

[0140] The vibration spectrum analysis is as follows:

[0141] Feature extraction: A significant peak appears in the vibration spectrum at 2.5 kHz (the energy ratio in this frequency band under normal operating conditions is < 30%).

[0142] Matching the fault library: This frequency corresponds to tooth missing or wear of the gear set (the feature frequency matching degree is 92%).

[0143] Preliminary conclusion: The probability of gear set failure is high (vibration confidence level is 85%).

[0144] The audio feature recognition analysis is as follows:

[0145] MFCC comparison: The cosine similarity between the Mel cepstral coefficients of the real-time audio and the tooth missing sound pattern library of the gear is 0.82 (the threshold > 0.7).

[0146] Zero-crossing rate detection: 12 high-frequency transient pulses are detected per second (normal < 5 times), which conforms to the characteristics of gear jamming.

[0147] Supporting conclusion: Further confirmation of gear set abnormality (audio confidence level is 78%).

[0148] Temperature data exclusion method:

[0149] The current temperature is 25 °C (rated value is 25 °C), and the temperature gradient \(\Delta T / \Delta t = 0 °C / min\).

[0150] Eliminate overheating faults: There are no abnormalities in heating components such as magnetic couplers and bearings, reducing the suspicion of related components.

[0151] Historical maintenance association:

[0152] Maintenance records: The gear set has been replaced twice in the past year, and the maintenance interval has recently been shortened to 60 days (historical average 180 days).

[0153] Trend analysis: Maintenance interval attenuation coefficient \lambda = 0.15λ = 0.15 (>0.1 indicates accelerated deterioration).

[0154] Statistical inference: The gear set enters a period of accelerated wear due to material fatigue, and the risk of failure increases significantly.

[0155] Final prediction: Gear set (G-205 model) failure, 92% confidence.

[0156] Vibration and audio data together point to the gear train, forming a closed chain of evidence. Normal temperatures eliminate interference from other heat-generating components, improving prediction accuracy. Shortened maintenance intervals (60 days vs. 180 days) reveal accelerated degradation patterns, supporting the recommendation to replace rather than repair.

[0157] S320, determining, based on the historical maintenance information of the target instrument, whether the component predicted to have a fault belongs to the component in the historical maintenance information;

[0158] Specifically, based on the equipment's unique code (e.g., serial number SN-2039), maintenance work orders from the past three years are extracted from the MRO system, screening for records involving the predicted component. Historical maintenance information primarily includes: repair date, component code, fault description, and maintenance action (replacement / calibration / lubrication). When matching prediction results, precise matches can be performed: where the predicted component code is identical to the component code in the historical maintenance record (e.g., B-2039); or fuzzy matches can be performed: for uncoded components (e.g., "motor bearings"), natural language processing (NLP) is used to extract keywords and match them to historical descriptions.

[0159] S330: If the component predicted to fail is a component with historical maintenance information, the maintenance frequency change trend of the component is used as a fifth evaluation factor.

[0160] Optionally, the step of evaluating the failure probability of the target instrument failing based on the abnormal data and the second data set includes:

[0161] The mathematical expression for evaluating the failure probability of the target instrument in the event of a failure is:

[0162]

[0163] Among them, μ represents the mean value of abnormal data, which is used to reflect the overall deviation level of abnormal data; μ1 is the reference value under normal working conditions, which is obtained based on historical statistical data; σ represents the standard deviation under normal working conditions, which is used to standardize the deviation degree; represents the slope of the data change trend, which is used to reflect the deterioration rate of data change; ω represents the weight coefficient of the deviation degree and the trend, which ranges from 0 to 1; α represents the weight coefficient of the first evaluation factor; is used to measure the standardization degree of data deviation from the normal state. For example, if μ = μ1 + 3σ, the deviation degree is 3, indicating a significant abnormality; is used to capture the severity of parameter changes. For example, a 5% decrease in flow per minute (a large absolute value of the slope) may indicate a blockage; ω adjusts the contribution ratio of the two (for example, steady-state parameters focus on the deviation degree, and dynamic parameters focus on the trend). It is used for the evaluation of abnormal data.

[0164] A max ,A min represents the safety amplitude range threshold, A represents the current vibration amplitude (such as the RMS value), B represents the total number of audio samples, b represents the number of abnormal frequency points, T0 represents the current temperature measurement value, T1 represents the temperature alarm threshold, T2 represents the rated temperature, and β represents the weight coefficients of the second, third, and fourth evaluation factors. represents the normalized vibration energy, and its value ranges from 0 to 1; represents the proportion of abnormal audio, and its value ranges from 0 to 1; represents the relative temperature threshold, and its value ranges from 0 to 1. Normalization eliminates the dimension difference, enabling the integration of vibration, audio, and temperature data on the same scale (0 - 1). For example. It is used for the evaluation of multi-sensor data. N C (t) represents the number of repairs of the component within the time window [t - T, t], and N C (t) The higher it is, the more frequent the component failures are recently (aging acceleration); represents the exponentially decaying weighted integral of the historical repair times; represents the frequency sensitivity coefficient, which is used to adjust the influence amplitude of the repair times on the evaluation; e -ρ(t-T) represents that the weight of the earlier repair record is lower, ρ represents the decay coefficient, and γ represents the weight coefficient of the fifth evaluation factor. tanh(·) represents the hyperbolic tangent function, which compresses the input to [-1, 1], and here the positive value (0 - 1) is taken. It is used for historical maintenance correction. α + 3β + γ ≤ 1 to ensure that F ∈ [0, 1].

[0165] For easy understanding, the following is an example:

[0166] An intelligent water meter (device ID: WT-203) has recently shown abnormal fluctuations in flow readings, and its failure probability needs to be evaluated. The following data is known:

[0167] Normal operating condition benchmark: average daily flow μ1 = 10 m 3 / day, standard deviation σ = 2 m 3 / day.

[0168] Sensor configuration:

[0169] Vibration sensor: safe vibration range A min = 0.1 g, A max = 1.0 g, current vibration A = 0.7 g.

[0170] Audio sensor: total number of sampling points B = 1000, detected abnormal frequency points b = 20.

[0171] Temperature sensor: current temperature T0 = 35 °C, rated temperature T2 = 25 °C, warning threshold T1 = 40 °C.

[0172] Historical maintenance records: the gear set has been maintained once in the past six months (the last time was 60 days ago), the average annual maintenance frequency is 2 times, and the attenuation coefficient ρ = 0.01.

[0173] ① Normal data evaluation items:

[0174] Parameter values:

[0175] Abnormal flow mean μ = 6 m 3 / day, trend slope

[0176] Weight coefficients: ω = 0.6, α = 0.4

[0177] ② Sensor data evaluation items:

[0178] Parameter values: vibration amplitude A = 0.7 g, abnormal frequency points b = 20, temperature T0 = 35 °C.

[0179] Weight coefficient: β = 0.1

[0180] ③ Historical maintenance trend evaluation items:

[0181] Parameter values: number of maintenance times N in the past six months C (t) = 1, sensitivity coefficient Weight γ = 0.1.

[0182] Historical maintenance records: once six months ago, and the earlier one was 240 days ago.

[0183] F = 0.72 + 0.13 + 0.02 = 0.87

[0184] Result analysis and handling:

[0185] Fault probability: 0.87 (exceeding the preset threshold of 0.7, an alarm message is sent);

[0186] Predicted faulty component: Gear set (vibration spectrum matches wear characteristics, historical maintenance interval has shortened).

[0187] Analysis of key contributions:

[0188] Abnormal data dominant (contribution degree 72%): Sudden drop in flow rate and trend deterioration.

[0189] Sensor verification (contribution degree 13%): Abnormal vibration and temperature support mechanical faults.

[0190] Historical trend assistance (contribution degree 2%): Increased recent maintenance frequency.

[0191] Disposal instructions:

[0192] Immediately cut off the water supply to prevent complete damage to the gears.

[0193] Replace the gear set (spare part G - 205) and synchronously calibrate the flow sensor.

[0194] Shorten the next maintenance cycle to 30 days.

[0195] S40. When the fault probability is greater than the preset threshold, send a warning message to the target user.

[0196] Optionally, the step of sending a warning message to the target user when the fault probability is greater than the preset threshold includes:

[0197] S401. Set at least two warning levels according to the preset threshold;

[0198] Specifically, the specific classification of the warning levels is shown in the following table:

[0199]

[0200] S402. Obtain the warning level corresponding to the fault probability according to the fault probability;

[0201] Specifically, after calculating F in real - time, match the level according to the preset interval (for example, if F = 0.85, an orange warning is sent).

[0202] S403. Send a warning message to the target user according to the warning level;

[0203] Specifically, push visualization data such as vibration spectrogram and temperature trend curve. Associate with the maintenance manual link (such as the operation guide for gear set replacement).

[0204] S404. If the target instrument triggers a warning three times in a row, the warning information sent to the target user is upgraded by one level based on the original warning level.

[0205] Specifically, query the warning history database and count the number of warnings for the device within a 30-day time window. If the same-level warning is triggered three times in a row (such as three level I warnings), the upgrade condition is triggered. Mark the upgrade record in the device file for subsequent root cause analysis of faults. After the upgrade is triggered, reset the continuous count and start counting again.

[0206] For easy understanding, an example is given as follows:

[0207] Scenario: Upgrade of warning for wear of gear set of intelligent water meter

[0208] Initial warning (February 1, 2025, F = 0.72):

[0209] Level I warning, notify the operation and maintenance personnel by email and suggest an inspection within 72 hours.

[0210] Second warning (February 8, 2025, F = 0.75):

[0211] Still level I, the text message reminder has not been processed, and the system marks "to be followed up".

[0212] Third warning (February 15, 2025, F = 0.78):

[0213] The condition for continuous warning is triggered, and the system automatically upgrades to level II.

[0214] Push the information to the operation and maintenance supervisor and require a response within 4 hours.

[0215] Disposal after upgrade: Remotely diagnose and confirm gear wear, dispatch spare part G-205, and complete the replacement within 48 hours.

[0216] Embodiment 3

[0217] Based on Embodiment 1, this embodiment provides a method for warning of faults of a metering instrument based on the industrial Internet of Things. The step of sending a warning message to a target user when the fault probability is greater than a preset threshold includes:

[0218] Set at least two warning levels according to the preset threshold;

[0219] Obtain the warning level corresponding to the fault probability according to the fault probability;

[0220] Send a warning message to the target user according to the warning level;

[0221] If the target instrument triggers a warning three times in a row, the warning message sent to the target user will be upgraded by one level based on the original warning level.

[0222] The user platform is configured to provide functions for users and front-end services; users obtain the required perception service information through the user platform, process the perception service information, and convert it into user perception information; users analyze the user perception information and make corresponding decisions based on their own wishes, and convert the user perception information into user control information through the corresponding information system and send it to the service platform, thereby showing the user's corresponding service demand intention.

[0223] The physical entities of the user platform include various user terminals, such as mobile phones, computers, dedicated terminals, etc., and realize the services at the user end through the combination with the user information system software.

[0224] The service platform is configured as an API server or other servers used to establish communication between the management platform and the user platform to implement corresponding functions; the physical entities of the service platform include various servers.

[0225] The management platform is configured to perform at least one of device operation status monitoring and management, data monitoring and management, device parameter management, and life cycle management; the management platform is the operation and coordination platform of the Internet of Things, and may include various management sub-platforms, and different management sub-platforms perform different management services; the physical entities of the management platform include various servers.

[0226] The sensor network platform is configured to perform at least one of network management, instruction management, device status management, data protocol management, data parsing, data classification, data transmission monitoring, and data transmission security management. The sensor network platform provides functions such as communication transmission, parsing, identification, and classification of data, avoiding the direct aggregation of data from various object platforms in the management platform, resulting in redundant data in the management platform and low data processing efficiency; the physical entities of the object platform include various gateways, edge computing devices, etc.

[0227] The object platform is configured to perform specific production control, detection, measurement and other production work; the physical entities in the object platform include various production equipment, sensors, etc.

[0228] Optionally, the sensor network platform includes a general database communicating with the management platform and at least two sensor network sub-platforms communicating with the general database;

[0229] Optionally, each sensor network sub-platform corresponds to an API function or an API server.

[0230] Embodiment 4

[0231] This embodiment provides a computer device, which includes a memory and a processor. A computer program is stored in the memory, and the processor executes the computer program to implement any of the above methods.

[0232] The above are only the preferred embodiments of the present application, and do not limit the patent scope of the present application accordingly. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present application, or directly or indirectly applied to other related technical fields, shall be included in the patent protection scope of the present application by the same token.

Claims

1. A fault warning method for measuring instruments based on the industrial Internet of Things, characterized in that, Including: Based on the first data set of the target instrument in a preset time period, determine whether there is abnormal data in the target instrument; If there is abnormal data in the target instrument, obtain the second data set of the target instrument in the preset time period; Based on the abnormal data and the second data set, evaluate the failure probability of the target instrument in case of a failure; When the failure probability is greater than a preset threshold, send a warning message to the target user.

2. The method for early warning of meter failure based on industrial Internet of Things according to claim 1, characterized in that: The step of determining whether there is abnormal data in the target instrument based on the first data set of the target instrument in the preset time period includes: Obtain the first data set of the target instrument in the preset time period, and divide the first data set into at least one data group according to the attributes of the data; Traverse each data in the first data set, compare each data with a first threshold, and determine whether there is first suspicious data in the first data set; If there is first suspicious data in the first data set, mark the first suspicious data as abnormal data; If there is no first suspicious data in the first data set, analyze the at least one data group to determine whether there is second suspicious data in the first data set; If there is second suspicious data in the first data set, mark all the data in the data group where the second suspicious data is located as the abnormal data.

3. The fault warning method for metering instruments based on industrial Internet of Things according to claim 2, characterized in that, The step of, if there is no first suspicious data in the first data set, analyzing the at least one data group to obtain whether there is second suspicious data in the first data set includes: According to the attribute information of each data group, obtain the contribution degree of each data group to the failure situation when the target instrument has a historical failure; According to the contribution degree of each data group to the failure of the target instrument, obtain the priority of each data group; According to the priority, sort all the data groups to obtain a sorting result; Analyze a preset proportion of the data groups according to the sorting result to obtain the change trend of the data in each data group; According to the change trend, determine whether there is second suspicious data in the first data set.

4. The fault warning method for a metering instrument based on the industrial Internet of Things according to claim 1, wherein, The step of, if there is abnormal data in the target instrument, obtaining the second data set of the target instrument in the preset time period includes: The second data set at least includes a vibration data packet, an audio data packet, a temperature data packet, and the historical maintenance information of the target instrument.

5. The fault warning method of the metering instrument based on the industrial Internet of Things according to claim 1, wherein, The step of, based on the abnormal data and the second data set, evaluating the failure probability of the target instrument in case of a failure includes: Obtain a first evaluation factor according to the deviation degree and / or change trend of the abnormal data; Obtain a second evaluation factor according to the vibration data packet of the second data set; Obtain a third evaluation factor according to the audio data packet of the second data set; Obtain a fourth evaluation factor according to the temperature data packet of the second data set; Based on the first evaluation factor, the second evaluation factor, the third evaluation factor, and the fourth evaluation factor, evaluate the failure probability of the target instrument in case of a failure.

6. The fault warning method for a metering instrument based on the industrial Internet of Things according to claim 5, characterized in that, The step of, based on the abnormal data and the second data set, evaluating the failure probability of the target instrument in case of a failure further includes: Predict the components of the target instrument that have failed according to the abnormal data and the second data set; According to the historical maintenance information of the target instrument, obtain whether the components predicted to have failed belong to the components in the historical maintenance information; If so, use the changing trend of the maintenance frequency of the component as the fifth evaluation factor.

7. The fault warning method for a metering instrument based on the industrial Internet of Things according to claim 6, wherein The step of evaluating the failure probability of the target instrument in case of failure according to the abnormal data and the second data set includes: The mathematical expression for evaluating the failure probability of the target instrument in case of failure is: Among them, μ represents the mean value of abnormal data, μ1 represents the reference value under normal working conditions, σ represents the standard deviation under normal working conditions, represents the slope of the data change trend, ω represents the weight coefficient of the deviation degree and the trend, and α represents the weight coefficient of the first evaluation factor; A max ,A min represents the safe amplitude range threshold, A represents the current vibration amplitude, B represents the total number of audio samples, b represents the number of abnormal frequency points, T0 represents the current temperature measurement value, T1 represents the temperature alarm threshold, T2 represents the rated temperature, and γ represents the weight coefficients of the second, third, and fourth evaluation factors; N C (t) represents the number of repairs of the component within the time window [t - T, t]. represents the frequency sensitivity coefficient, e -ρ(t-T) indicates that the weight of the earlier repair records is lower, ρ represents the attenuation coefficient, and γ represents the weight coefficient of the fifth evaluation factor.

8. The fault warning method for a metering instrument based on the industrial Internet of Things according to claim 1, characterized in that, The step of sending a warning message to the target user when the failure probability is greater than the preset threshold includes: Set at least two warning levels according to the preset threshold; Obtain the warning level corresponding to the failure probability according to the failure probability; Send a warning message to the target user according to the warning level; If the target instrument triggers a warning three times in a row, the warning message sent to the target user is upgraded by one level on the basis of the original warning level.

9. A metering instrument fault warning system based on the industrial Internet of Things, characterized in that, Including a management platform, a sensor network platform, and an object platform that establish communications in sequence: The sensor network platform is configured to: Judge whether there is abnormal data of the target instrument according to the first data set of the target instrument in a preset time period; If there is abnormal data of the target instrument, obtain the second data set of the target instrument in a preset time period; Evaluate the failure probability of the target instrument in case of failure according to the abnormal data and the second data set; The management platform is configured to: Send a warning message to the target user when the failure probability is greater than the preset threshold.

10. A computer device, characterized in that, The computer device includes a memory and a processor. A computer program is stored in the memory, and the processor executes the computer program to implement the method according to any one of claims 1-8.

Images