Intelligent detection and early warning method and system for copper sheath heater heating wire falling off

Abstract

The application provides a copper sleeve heater heating wire falling intelligent detection and early warning method and system, relates to the technical field of intelligent early warning, and acquires an initial benchmark data set and a real-time feature vector; a trend early warning mark is generated for the real-time feature vector; a multi-signal correlation decision-making process is automatically triggered for the trend early warning mark, and a judgment level is output; an abnormal area table decision bit is performed according to the judgment level, an abnormal area is generated, and an abnormal area level is acquired; a monitoring mechanical constraint state is discriminated according to the judgment level and the abnormal area, and a mechanical degradation confirmation mark is output; based on the judgment level, the trend early warning mark, the abnormal area, the abnormal area level and the mechanical degradation confirmation mark, the state of the copper sleeve heater is determined according to a certainty priority rule, and a decision is made; early and accurate detection and positioning of local loosening or falling of the heating wire of the copper sleeve heater can be performed, hierarchical early warning is realized, timely intervention is realized, serious falling accidents are avoided, and the reliability of equipment is improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent early warning technology, specifically to an intelligent detection and early warning method and system for the detachment of heating wires in copper sleeve heaters. Background Technology

[0002] In the industrial production field, especially in industries such as hot runner systems and injection molding, copper sleeve heaters are widely used. However, during the use of traditional copper sleeve heaters, problems such as the heating wire not being securely fixed often cause the heating wire to fall out of the groove in the copper sleeve. This falling out not only affects the heating efficiency and reduces the stability and reliability of the equipment, but may also cause safety hazards such as short circuits, leading to production interruptions and increasing maintenance and time costs. Therefore, intelligent detection and early warning are needed.

[0003] However, existing methods mainly rely on the delayed detection of single parameters (such as current and temperature), which cannot accurately detect early signs of heating wire detachment (such as poor contact or minute displacement). This leads to the accumulation of faults until they reach a critical stage before triggering an alarm, causing production interruptions. Sensors are mostly externally placed on the heater structure, making them susceptible to environmental interference and unable to monitor microscopic changes at the contact interface, resulting in low signal accuracy and a high false alarm rate. Early warning and response rely excessively on software algorithms and central control, lacking hardware-level real-time intervention and redundant protection mechanisms. In extreme cases, they cannot promptly block risks, easily leading to safety accidents such as short circuits and fires. As a result, traditional methods cannot achieve early warning, accurate intervention, and efficient maintenance, becoming a key bottleneck restricting equipment stability and production continuity. Summary of the Invention

[0004] To achieve the above objectives, the present invention provides the following technical solution: an intelligent detection and early warning method for heating wire detachment in a copper sleeve heater, the method comprising: Step S1: Obtain the initial benchmark dataset and real-time feature vectors; generate trend warning signs for the real-time feature vectors; automatically trigger the multi-signal association adjudication process based on the trend warning signs and output the judgment level. Step S2: Based on the judgment level, perform abnormal area voting and location, generate abnormal areas, and obtain the abnormal area level; Step S3: Based on the judgment level and abnormal area, determine the mechanical constraint status of the monitored machinery and output a mechanical degradation confirmation flag; Step S4: Based on the judgment level, trend warning sign, abnormal area, abnormal area level, and mechanical degradation confirmation sign, determine the status of the copper sleeve heater according to the deterministic priority rule and make a decision.

[0005] Furthermore, obtaining the initial benchmark dataset includes: The copper jacket heater is divided into N physical sections. When the equipment is powered on for the first time or after maintenance and reset, the initial insulation resistance, distributed capacitance, contact resistance, thermocouple temperature, shell resonant frequency, initial value of micro-gap capacitance and micro-vibration amplitude of each physical section are collected. Compare with the standard ranges in the standard database to generate a set of physical segments for manufacturing deviations; The initial baseline dataset is composed of the initial insulation resistance, distributed capacitance, contact resistance, thermocouple temperature, shell resonant frequency, initial value of micro-gap capacitance, micro-vibration amplitude, and manufacturing deviation of each physical segment.

[0006] Furthermore, the step of generating trend warning indicators for real-time feature vectors includes: During equipment operation, three hardware loops are invoked to collect high-frequency impedance signals, differential current signals, and high-frequency arc flash pulse signals, and multi-dimensional feature vectors are extracted to generate real-time feature vectors. For real-time feature vectors, the rate of change of the real-time feature vectors with a time interval of D from the current time is obtained to obtain a real-time feature rate of change vector at multiple time scales; and a preset rate of change threshold range is set. If the rate of change of real-time features in the multi-timescale real-time feature rate of change vector is not within the preset rate of change threshold for R consecutive times, a trend warning sign is generated.

[0007] Furthermore, the automatic triggering of the multi-signal association adjudication process, outputting the judgment level, includes: First-level judgment: When there is a trend warning sign for the rate of change of impedance offset, and there are no trend warning signs for the effective value of differential current, the number of over-limit events, and the rate of change of arc pulse count, it is determined that the contact is slowly degrading, triggering the first-level judgment. Second-level judgment: When there is a trend warning sign for the rate of change of impedance offset, and a trend warning sign for the effective value of differential current or the count of over-limit events, it is determined to be asymmetrical contact, triggering the second-level judgment. The third layer of judgment: When the rate of change of the arc flash event count exceeds the preset rate of change threshold, and the arc flash count shows an increasing trend in the most recent Y statistical periods, and there is already a first or second layer warning, it is determined to be intermittent dropout, triggering the third layer of judgment. For trend warning flag combinations of features beyond the three-layer judgment, only logs are recorded, and no judgment is triggered.

[0008] Furthermore, the step of performing abnormal region voting and locating to generate abnormal regions includes: Real-time reading of micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature of N physical segments, and retrieval of corresponding data from the initial benchmark dataset; If the real-time readings of micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature exceed the sensor's range; if the deviation between the real-time readings of micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature and the corresponding data in the retrieved initial reference dataset exceeds a preset reasonable deviation threshold; or if the sensor reports an error, then the sensor will be marked as "untrustworthy"; otherwise, it will be marked as "trustworthy". For N physical segments, if the number of sensors marked as "trusted" in a physical segment is greater than or equal to M, then the corresponding physical segment is marked as a "data-sufficient segment". If there are fewer than M sensors marked as "trusted" in a physical segment, then mark the physical segment as "data insufficient segment", do not perform any further operations, and record it in the log; The voting rules are set according to the "data-sufficient segment" to obtain the abnormal area.

[0009] Furthermore, the setting of voting rules includes: For the "data-sufficient segment", the deviations of the micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature collected by the sensors in the physical segment are obtained in real time. If a “sufficient data segment” contains micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature; and the deviation of the micro-gap capacitance is greater than or equal to a preset reasonable deviation threshold, while the deviation of the micro-vibration amplitude or the thermocouple temperature is also greater than or equal to a preset reasonable deviation threshold, then the “sufficient data segment” is an abnormal segment. If there is no micro-gap capacitance, micro-vibration amplitude, or thermocouple temperature in the "data-sufficient segment", then two existing deviations are identified. If both existing deviations are greater than or equal to the preset reasonable deviation threshold, then the segment is identified as abnormal. Adjacent physical segments are merged to obtain an abnormal region. If two abnormal segments are not adjacent, the abnormal segment is directly defined as an abnormal region.

[0010] Furthermore, obtaining the abnormal region level includes: For each physical segment in the abnormal region, calculate the correction score for the "trustworthy" sensor in that physical segment. The base score for the physics section is obtained by averaging the corrected scores. If the physical segment is in the set of physical segments with manufacturing deviations, or if it is marked as "untrustworthy", then the base score is adjusted to obtain the fusion score; The confidence level of the abnormal region is obtained by taking the weighted average of the fusion scores of all physical segments within the abnormal region. When the confidence level of an outlier region is greater than or equal to the upper limit of the baseline confidence level, it is considered a high-level outlier region. When the confidence level of an outlier region is between the upper and lower limits of the baseline confidence level, it is considered a medium-level outlier region. When the confidence level of an outlier region is less than or equal to the lower limit of the baseline confidence level, it is considered a low-level outlier region.

[0011] Furthermore, the step of determining the monitoring mechanical constraint status based on the judgment level and abnormal area, and outputting a mechanical degradation confirmation flag, includes: Based on the judgment level and abnormal area, a dynamic decision is adopted to apply an excitation signal to the shell. Under the application of the excitation signal, the shell generates forced vibration and produces a real-time vibration response signal. The real-time shell resonant frequency is extracted by the peak search algorithm and then compared with the initial shell resonant frequency in the initial benchmark dataset to obtain the shell resonant frequency offset. The mechanical constraint state is monitored by a dual-condition discrimination method based on the shell resonant frequency offset. If either condition is met, it is considered mechanical degradation, and the mechanical degradation confirmation flag W(mech) = 1 is output; otherwise, it is considered mechanically normal, and the mechanical degradation confirmation flag W(mech) = 0 is output.

[0012] Furthermore, determining the state of the copper bushing heater based on the deterministic priority rule includes: Highest priority: If any condition is met, the current status is determined to be a Level 3 warning. Condition 1: Mechanical degradation, i.e., mechanical degradation confirmation criterion W(mech) = 1, judgment level is two or three, and there is an abnormal area, the abnormal area level is high; Condition 2: The judgment level is three, and there is an abnormal area, and the level of the abnormal area is medium or high; Second highest priority: If the judgment level is greater than or equal to level two (i.e., level two or level three judgment), and there is an abnormal area, and the level of the abnormal area is high or medium; if these conditions are met, the current status is determined to be a level two warning. Medium priority: The judgment level is level 1, or there is a trend warning indicator; if either condition is met, the current state is judged as level 1 warning. Lowest priority: If a state does not meet the highest, second-highest, or medium priority, then the current state is a safe state.

[0013] A smart detection and early warning system for heating wire detachment in a copper bushing heater, the system comprising: Situation analysis engine: acquires initial benchmark dataset and real-time feature vectors; generates trend warning signs based on real-time feature vectors; automatically triggers multi-signal association adjudication process based on trend warning signs and outputs judgment level; Anomaly domain identification unit: Based on the judgment level, it performs anomaly region voting and location, generates anomaly regions, and obtains the anomaly region level; Mechanical deterioration verification unit: Based on the judgment level and abnormal area, it determines the constraint status of the monitored machinery and outputs a mechanical degradation confirmation mark; Overall machine status decision center: Based on the judgment level, trend warning signs, abnormal areas, abnormal area levels, and mechanical degradation confirmation signs, the copper sleeve heater status is determined and a decision is made according to the deterministic priority rules.

[0014] This invention provides an intelligent detection and early warning method and system for the detachment of heating wires in copper sleeve heaters. It has the following beneficial effects: By axially segmenting the copper sleeve heater and integrating multiple types of sensors, combined with three signals—global high-frequency impedance, differential current, and arc flash pulse—it has for the first time achieved full-dimensional online monitoring from local contact degradation to overall mechanical condition. It can accurately locate the axial position of heating wire loosening and trigger graded early warnings at the micron-level gap change and milliohm-level contact resistance increase stages. This transforms the traditional passive mode of "fault repair and overall replacement" into proactive predictive maintenance of "early warning, precise location, and on-demand maintenance," significantly shortening fault diagnosis time, avoiding unplanned downtime, extending heater lifespan, and reducing maintenance costs. Employing a multi-physical quantity voting and confidence assessment mechanism effectively eliminates manufacturing deviations and environmental interference, ensuring high reliability of positioning results. Based on deterministic rules, a tiered early warning system (Level 1 maintenance prompt, Level 2 planned shutdown, Level 3 emergency power failure) and MCU / CPLD hardware-software collaborative control balances production continuity with microsecond-level safety protection under extreme conditions. Simultaneously, active frequency sweep diagnosis of the shell resonant frequency offset serves as macroscopic mechanical verification, further enhancing system robustness. The overall design does not rely on complex algorithm models, making it easy to implement in engineering and industrial certification, providing a reusable intelligent detection framework for the health management of copper-jacketed heaters and similar electrothermal equipment. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the steps of the intelligent detection and early warning method for the detachment of heating wire in a copper sleeve heater according to the present invention. Figure 2 This is a flowchart of the overall system status decision unit of the present invention; Figure 3 This is a diagram illustrating the architecture of the intelligent detection and early warning system for the detachment of the heating wire in the copper sleeve heater of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] like Figures 1 to 2 As shown, the intelligent detection and early warning method for heating wire detachment in a copper bushing heater includes the following steps: Acquire the initial benchmark dataset and real-time feature vectors; generate trend warning indicators for the real-time feature vectors; automatically trigger the multi-signal association adjudication process based on the trend warning indicators and output the judgment level; Based on the judgment level, anomaly area voting is performed to locate the abnormal area, generate the abnormal area, and obtain the abnormal area level; Based on the judgment level and abnormal area, the mechanical constraint status of the monitored machinery is determined, and a mechanical degradation confirmation mark is output; Based on the judgment level, trend warning signs, abnormal areas, abnormal area levels, and mechanical degradation confirmation signs, the status of the copper bushing heater is determined and a decision is made according to the deterministic priority rule.

[0018] Traditional copper sleeve heaters use an integral structure and can only output a single total current, total voltage and average temperature signal. They cannot distinguish between local contact degradation and overall uniform aging. When an abnormality is detected, it has often progressed to a stage of severe detachment. Furthermore, it is impossible to locate the axial position of the fault. Maintenance personnel need to disassemble the entire heater for troubleshooting, resulting in low maintenance efficiency and long downtime. Obtain the initial benchmark dataset, including: Along the axial direction of the copper sleeve heater, the copper sleeve heater is divided into N physical segments (this division does not actually cut the copper sleeve heater into N segments, but rather divides it into equal length segments according to the length of the copper sleeve heater), and these segments are numbered; the materials, thicknesses and processes of the copper sleeve, insulation layer and heating wire contained in each segment are consistent, but each segment is electrically isolated from the others (for example, through annular insulation partitions or reserved gaps). With the segmented design, each physical segment can collect multiple physical quantities (such as insulation resistance, distributed capacitance, contact resistance, temperature, and micro-vibration). This not only enables early identification and precise location of local anomalies (outputting the faulty segment) through multi-physical quantity voting, but also eliminates manufacturing deviations and environmental interference by comparing signals from adjacent segments. This triggers graded early warnings when the heating wire becomes slightly loose or the gap increases, guiding maintenance personnel to make precise repairs. Ultimately, this achieves a shift from "passive emergency repair" to "proactive predictive maintenance," significantly improving the reliability and service life of the heater. The outer copper sleeve heater and the inner heating wire are considered as a pair of coaxial cylindrical electrodes. In each physical segment, one wire is led out from the copper sleeve and the heating wire (two wires in total) to form a two-wire detection interface. This interface is used to apply a low-voltage DC signal to measure the insulation resistance of N physical segments and to apply a low-voltage AC signal to measure the distributed capacitance. Two wires are led out from the copper sleeve and the heating wire (four wires in total) to form a four-wire detection interface. This interface is used to apply a constant DC current to measure the contact resistance of N physical segments. Insulation resistance and distributed capacitance can reflect the overall integrity of the insulation layer of each physical segment. A decrease in insulation resistance indicates that the insulation material is damp or carbonized. Changes in distributed capacitance reflect changes between the heating wire and the copper sleeve heater. An increase in contact resistance indicates a decrease in local contact pressure or a decrease in contact area, which is the most direct electrical evidence of a loose heating wire. Each physical segment is pre-embedded with a flexible thin-film sensing unit (including a micro-gap capacitance sensor and a micro-vibration sensor), which applies an excitation signal and records the micro-gap capacitance and micro-vibration amplitude. Among them, the high-frequency low-amplitude AC signal (typical frequency 1MHz~10MHz, voltage AC 1V~5V) applied to the micro-gap capacitance sensor is used as the excitation; the micro-vibration sensor is applied with a high-frequency carrier signal (such as a 100kHz sine wave) as the excitation. Thin-film thermocouples are pre-embedded in N physical segments. These thermocouples have the characteristics of fast response and small size. They are used to accurately measure the temperature of thermocouples in different segments. They reflect whether the heat generated by the heating wire in this segment can be effectively conducted to the copper sleeve heater. When the contact is poor, the thermal resistance of this segment increases, and the temperature will rise abnormally under the same power. In addition, a piezoelectric ceramic plate (PZT) is installed on the outer shell of the copper-shelter heater. This piezoelectric ceramic plate has a reserved frequency sweep signal interface. By applying a small-amplitude frequency sweep excitation signal through the interface (such as a frequency range of 200Hz to 20kHz, a step of 10Hz, a voltage of AC 5V to 30Vpp, and a sweep duration of 10 to 60 seconds), the vibration response signal of the shell is collected. Using a peak search algorithm, the resonant frequency of the shell can be obtained. The shedding of the heating wire will change the system constraints, causing the resonant frequency of the shell to shift. This is a macroscopic and holistic mechanical diagnostic indicator. The peak search algorithm, a current technology, takes the collected shell vibration response signal as input and performs bandpass filtering, amplitude extraction, and smoothing on the signal to obtain the stable response amplitude corresponding to the frequency, thereby constructing a frequency amplitude response curve. On this curve, candidate peak points are identified by local maxima determination, and noise interference is eliminated by combining an amplitude threshold (which can be set empirically). Among the screened candidate peaks, the main peak with the largest amplitude located in the neighborhood of the initial reference resonant frequency is preferentially selected as the target resonant peak. Further interpolation fitting is performed using this peak point and its adjacent frequency points to improve frequency resolution, ultimately obtaining the shell resonant frequency. After the equipment (copper jacket heater itself) is powered on for the first time or after maintenance reset, initial data acquisition is performed. Within the initial preset time window, insulation resistance, distributed capacitance, contact resistance, thermocouple temperature, micro-gap capacitance, and micro-vibration amplitude, as well as the shell resonant frequency, are collected for each physical segment. Preprocessing and average value calculations are then performed to obtain the initial insulation resistance, distributed capacitance, contact resistance, thermocouple temperature, micro-gap capacitance, micro-vibration amplitude, and shell resonant frequency. By statistically analyzing the insulation resistance, distributed capacitance, contact resistance, thermocouple temperature, micro-gap capacitance, micro-vibration amplitude, and shell resonant frequency within the initial preset time window, and averaging these values, the average insulation resistance, distributed capacitance, contact resistance, thermocouple temperature, micro-gap capacitance, micro-vibration amplitude, and shell resonant frequency are obtained. These average insulation resistance, distributed capacitance, contact resistance, thermocouple temperature, micro-gap capacitance, micro-vibration amplitude, and shell resonant frequency are then used as the initial insulation resistance, distributed capacitance, contact resistance, thermocouple temperature, micro-gap capacitance, micro-vibration amplitude, and shell resonant frequency. For the standard database (a built-in database of the control system that stores parameter ranges of a large number of devices under normal manufacturing processes, such as insulation resistance ranging from 100MΩ to 5000MΩ), the collected data of N physical segments (initial insulation resistance, distributed capacitance, contact resistance, thermocouple temperature, micro-gap capacitance, micro-vibration amplitude, and shell resonant frequency) are compared with the standard ranges in the standard database. If the collected data of N physical segments are not within the standard range, it is determined that there is a manufacturing deviation in that physical segment, and the physical segment number of the manufacturing deviation is recorded to generate a set of physical segments with manufacturing deviations, such as {3,7}, indicating that the 3rd and 7th physical segments have manufacturing deviations. If the collected data of N physical segments are within the standard range, no special marking is made, and it is regarded as a standard healthy area. The initial insulation resistance, distributed capacitance, contact resistance, thermocouple temperature, micro-gap capacitance, micro-vibration amplitude, shell resonant frequency, and manufacturing deviation physical segments collected within the initial preset time window are integrated into an initial benchmark dataset and stored in non-volatile memory. If the copper bushing heater remains stable for an extended period without any subsequent warnings, the initial baseline dataset will be updated once a day. For example, if the copper bushing heater is stable on the first day, real-time insulation resistance, distributed capacitance, contact resistance, thermocouple temperature, micro-gap capacitance, micro-vibration amplitude, and shell resonant frequency will be collected at some point on the second day as a new initial baseline dataset.

[0019] For real-time feature vectors, generate trend warning indicators, including: During the operation of the equipment (i.e., the copper bushing heater), three hardware circuits are invoked (the three hardware circuits include a high-frequency impedance signal circuit, a differential current signal circuit, and a high-frequency arc flash pulse signal circuit, all of which are existing hardware; installed in the junction box or control panel of the copper bushing heater itself), to collect high-frequency impedance signals, differential current signals, and high-frequency arc flash pulse signals, and to extract multi-dimensional feature vectors to generate real-time feature vectors; Specifically, impedance offset and impedance change rate are extracted using high-frequency impedance signals; impedance offset is the difference between the current impedance value and the initial impedance value; a fixed time window is set, and the impedance values ​​within the window are differentially calculated to obtain the impedance change rate; The effective value of the differential current, the peak current, and the count of out-of-limit events are extracted from the differential current signal. The instantaneous value of the differential current is collected at a fixed sampling rate (e.g., 10kHz), and the root mean square is calculated as the effective value of the differential current within a time window. The largest instantaneous value of the differential current within the same time window is the peak current. When the peak current exceeds the preset peak current (e.g., 5% of the rated current) for c consecutive time windows, an out-of-limit event is considered to have occurred, and the count is incremented for each event. Here, c is an integer greater than or equal to 3, preferably 3, 4, or 5. The flash pulse event count, pulse width, and pulse interval are extracted from the high-frequency flash pulse signal. Bandpass filtering is applied to the high-frequency flash pulse signal to extract the flash pulses and their corresponding time points. A dynamic flash threshold (obtained from relevant literature) is set. When a flash pulse exceeds the flash threshold within c consecutive time windows, it is considered a valid flash event and counted to obtain the flash pulse event count. The rising and falling edges of the flash pulse events are captured using a timer, and the difference between the rising and falling edges is calculated to obtain the pulse width. The difference between the rising edges of two adjacent flash pulse times is continuously recorded as the pulse interval. The rising edge is the instant when the flash pulse signal jumps from low to high, marking the start of the pulse; the falling edge is the instant when the pulse signal jumps back from high to low, marking the end of the pulse. The impedance offset, impedance change rate, differential current RMS value, peak current, out-of-range event count, arc pulse event count, pulse width, and pulse interval are time-aligned at the same time granularity (e.g., 1 second) to generate a real-time feature vector, which is then stored. The specific alignment method is as follows: For example, a time series index from start to end is generated at 1-second intervals; for continuous features such as impedance offset, impedance change rate, effective value of differential current, and peak current, the forward filling method is used to fill the current second point with the feature value corresponding to the most recent second point with an effective feature value on the time axis before each second point; for event counting features such as over-limit event count and arc pulse event count, the count value of each second point is summed and statistically analyzed, and the count value of the second point without events is automatically set to 0; for pulse width and pulse interval, the average value of all pulse widths and pulse intervals within each second point is calculated, and for the second point without pulses, the forward filling method is used to fill the current second point without pulses with the average value of pulse width and average value of pulse interval corresponding to the most recent second point with a pulse on the time axis before the current second point; After the above processing, each second point has a complete feature value, which is combined row by row in chronological order to form the real-time feature vector; High-frequency impedance signals, differential current signals, and high-frequency flash pulse signals are collected because they reflect the contact state between the heating wire and the copper sleeve from three independent physical dimensions: electrical contact interface (impedance), circuit asymmetry (differential current), and intermittent discharge (flash pulse), respectively. By extracting specific features, the signals are converted into quantized, time-aligned real-time feature vectors, providing structured input for subsequent trend warning and fault identification.

[0020] For the real-time feature vector, obtain the rate of change of the real-time feature vector at the current time interval D (e.g., 60 seconds) to obtain a multi-time-scale real-time feature rate of change vector; for the rate of change of each feature in the multi-time-scale real-time feature rate of change vector, set a preset rate of change threshold range (obtained based on expert experience); the rate of change of the real-time feature vector is the difference between the real-time feature vector and the real-time feature vector at the current time interval D, and the ratio of the difference to the real-time feature vector at the current time interval D. If the rate of change of real-time features in the multi-timescale real-time feature rate of change vector is not within the preset rate of change threshold for R consecutive times, a trend warning sign is generated; where the time interval D ranges from 30 seconds to 120 seconds, with a typical value of 60 seconds, and the time interval can be increased or shortened according to specific circumstances; R is usually between 3 and 10 times. For example: The impedance offset at the current moment is 1Ω, and the impedance offset 60 seconds after the current time point is 2Ω, so the change rate is 1Ω. The preset change rate threshold range for the impedance offset change rate is 0.5Ω to 0.7Ω, so the result is out of limit. Calculate the change rate of the impedance offset at the current moment and the impedance offset 60 seconds after the current time point. If it is not within the preset change rate threshold range, it is marked as out of limit. Monitor 10 times continuously. If it is out of limit every time, it indicates that the feature has a continuous deterioration trend. If the change rates of other features are normal, a trend warning sign W[1] is generated. If the change rate of the real-time feature is within the preset change rate threshold range during the 5th monitoring (i.e., the monitoring from the 1st to the 4th time is not within the preset change rate threshold range), the continuity is interrupted and the monitoring is restarted. If the change rate of all real-time features is not within the corresponding preset change rate threshold range in 10 consecutive monitoring, a trend warning sign W[1,2,3,4,5,6,7,8] is generated.

[0021] Automatically triggers a multi-signal correlation adjudication process and outputs the judgment level, including: First-level judgment: When there is a trend warning sign for the rate of change of impedance offset, and there are no trend warning signs for the effective value of differential current, the number of over-limit events, and the rate of change of arc pulse count, it is determined that the contact is slowly degrading, triggering the first-level judgment. Second-level judgment: When there is a trend warning sign for the rate of change of impedance offset, and a trend warning sign for the effective value of differential current or the count of over-limit events, it is determined to be asymmetrical contact, triggering the second-level judgment. The third layer of judgment: When the rate of change of the arc flash pulse event count exceeds the preset rate of change threshold, and the arc flash pulse count shows an increasing trend in the most recent Y statistical periods (taking the 3 to 5 times before the current period), and there is already a first or second layer of judgment, it is determined to be intermittent dropout, triggering the third layer of judgment. For trend warning flag combinations of features beyond the above three-level judgment, only logs are recorded and no judgment is triggered; The rate of change of impedance offset directly reflects the continuous degradation of the electrical contact interface between the heating wire and the copper sleeve heater, and is the most sensitive indicator of the deterioration of the contact condition; the effective value of differential current and the number of over-limit events directly characterize the circuit imbalance caused by contact asymmetry, and are important evidence of the middle stage of detachment; the number of flash pulse events directly corresponds to intermittent discharge, and is a strong sign that detachment is imminent. Peak current is greatly affected by power supply fluctuations and load changes, and instantaneous peaks may occur even if the contact is normal; pulse width and pulse interval are affected by the characteristics of the flash arc and the parameters of the detection circuit, and have a weak linear relationship with the severity of the detachment, and are easily affected by electromagnetic noise interference; if the rate of change of these auxiliary characteristics is also included in the early warning conditions, the false alarm rate will be significantly increased and the system reliability will be reduced. Using core physical quantities as criteria makes early warning rules clear, easy to verify and maintain; abnormalities in auxiliary characteristics (such as peak current, pulse width, and pulse interval) often occur simultaneously with the above core characteristics. Even if they do not participate in early warning, their change information is still recorded in the log for verification and detailed diagnosis during post-fault analysis.

[0022] When a judgment level is triggered (when a first-level warning or a higher-level second or third-level warning is triggered), an abnormal area is located by voting and an abnormal area is generated; if no judgment level is triggered, it enters standby mode. Perform anomaly region voting to locate and generate anomaly regions, including: Real-time reading of micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature of N physical segments, and retrieval of corresponding data from the initial benchmark dataset; If the real-time readings of micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature exceed the sensor's range (i.e., the real-time readings of micro-gap capacitance, micro-vibration amplitude, or thermocouple temperature exceed the upper or lower limits specified in the sensor design); if the deviation (i.e., the difference) between the real-time readings of micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature and the corresponding data in the retrieved initial reference dataset exceeds the preset reasonable deviation threshold (set according to empirical methods); or if the sensor reports an error, then the sensor will be marked as "untrustworthy," otherwise it will be marked as "trustworthy." For N physical segments, if the number of sensors marked as "trusted" in a physical segment is greater than or equal to M, then the corresponding physical segment is marked as a "data-sufficient segment". If there are fewer than M sensors marked as "trustworthy" in a physical segment, then the physical segment is marked as "data insufficient segment", no further operations are performed, and this is recorded in the log; M is set to 2, because each physical segment has at least two types of reliable sensors, which can ensure that the data in the segment has enough information for anomaly detection; The voting rules are set according to the "data-sufficient segment" to obtain the abnormal area.

[0023] The specific steps for setting voting rules are as follows: For the "data-sufficient segment", the deviations of the micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature collected by the sensors in the physical segment are obtained in real time. If the "data-sufficient segment" contains micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature; and the micro-gap capacitance deviation (i.e., the difference between the currently acquired micro-gap capacitance and the initial micro-gap capacitance) is greater than or equal to the preset reasonable deviation threshold (set according to empirical method), and the micro-vibration amplitude deviation or thermocouple temperature deviation is also greater than or equal to the preset reasonable deviation threshold, then the "data-sufficient segment" is an abnormal segment; all three sensors are present. If there are no micro-gap capacitance, micro-vibration amplitude, or thermocouple temperature in the "data sufficient segment" (i.e., the sensor is marked as "unreliable"), then two existing deviations are identified. If both existing deviations are greater than or equal to the preset reasonable deviation threshold, then it is determined to be an abnormal segment; only two sensors exist. Adjacent physical segments are merged to obtain an abnormal region. If two abnormal segments are not adjacent, the abnormal segment is directly defined as an abnormal region. By employing a multi-sensor consensus voting and segment merging mechanism, the system can accurately locate and suppress misjudgments of abnormal areas while ensuring anti-interference capabilities, thereby improving the reliability and accuracy of fault detection.

[0024] Obtain the anomaly zone level, including: For abnormal regions, the confidence level of the abnormal region is obtained based on the number of sensors in the abnormal region, the reference ratio of the real-time measured data of the sensors (i.e., the ratio of the real-time read micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature to the corresponding data in the initial reference data), manufacturing deviation markings, and "unreliable" markings; the real-time measured data of the sensors are the real-time read micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature. The specific steps are as follows: For each physical segment in the abnormal region, calculate the individual score of the "trustworthy" sensor in each segment. The individual score is obtained based on the ratio of the real-time measured data of the sensor to the corresponding data in the initial benchmark dataset (i.e., the ratio of the real-time read micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature). For example, when the ratio is 0 to 1, the individual score is 0.6; when the ratio is 1 to 2, the individual score is 0.8; and when the ratio exceeds 2, the individual score is 1. The real-time measured data of the sensor refers to the real-time read micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature. If the baseline ratio of the real-time measured data by the sensor is lower than the preset limit (set according to the empirical method), then the individual score is adjusted (the individual score is reduced by 0.2 times) to obtain the corrected score. If it is higher than or equal to the preset limit, then no correction is performed, and the individual score is the corrected score. The base score of a physical segment is obtained by averaging the correction scores of the sensors (for example, if the abnormal area is a physical segment with three sensors, the base score of the physical segment is obtained by averaging the correction scores of the real-time data measured by the three sensors). If the physical segment is in the set of manufacturing deviation physical segments, or if it is marked as "unreliable", the base score is adjusted to obtain the fusion score (e.g., if the abnormal area is a physical segment in which one sensor is marked as "unreliable" or marked as a manufacturing deviation after initial power-on and maintenance reset, the base scores of other "reliable" sensors are adjusted (e.g., decreased by 0.2 times) to obtain the fusion score). The confidence level of the abnormal region is obtained by taking a weighted average of the fusion scores of all physical segments within the abnormal region; all weights in the above weighted average are equal. Map the confidence level of outlier regions to three levels: high, medium, and low. When the confidence level of an anomaly region is greater than or equal to the upper limit of the baseline confidence level, it is a high-level anomaly region, indicating that the localization results are very reliable and can be directly used for decision-making. When the confidence level of the outlier region is between the upper and lower limits of the baseline confidence level, it is a medium-level outlier region, indicating that the localization result is basically reliable. When the confidence level of an anomaly region is less than or equal to the lower limit of the baseline confidence level, it is considered a low-level anomaly region, indicating that the reliability of the localization result is insufficient; the upper and lower limits of the baseline confidence level are set according to empirical methods. By integrating sensor data consistency, benchmark ratio, manufacturing deviation, and "unreliable" marking, the confidence level of abnormal areas is quantified and classified, enabling reliability assessment and decision support for abnormal location results, thereby improving the accuracy and operability of fault determination.

[0025] Based on the judgment level and abnormal area, the mechanical constraint status of the monitored machinery is determined, and mechanical degradation confirmation indicators are output, including: Based on the judgment level and abnormal area, a dynamic decision is adopted to apply an excitation signal to the shell. Under the application of the excitation signal, the shell generates forced vibration and produces a real-time vibration response signal. The real-time shell resonant frequency is extracted by the peak search algorithm and then compared with the corresponding shell resonant frequency benchmark value in the initial benchmark dataset to obtain the shell resonant frequency offset. The mechanical constraint state is monitored by a dual-condition discrimination method based on the shell resonant frequency offset. If either condition is met, it is considered mechanical degradation, and a mechanical degradation confirmation flag W(mech) = 1 is output; otherwise, it is considered mechanically normal, and a mechanical degradation confirmation flag W(mech) = 0 is output.

[0026] The specific steps for applying excitation signals to the shell using dynamic decision-making are as follows: If there are no abnormal areas and no judgment level, apply a fast frequency scan excitation to the entire shell once a day. If there is no abnormal area but there is one or more layers of judgment, then apply a fast frequency scan excitation to the entire shell once per hour; If an abnormal area exists, a rapid frequency scanning excitation is applied to the local shell area corresponding to the abnormal area once per hour. Among them, "full shell" refers to the entire outer shell of the copper bushing heater (i.e., the copper bushing body); "rapid frequency sweep excitation" refers to the piezoelectric ceramic plate sweeping across a preset wide frequency range (e.g., 200Hz~20kHz) in a short period of time (e.g., 10 seconds) with fewer frequency points or larger frequency steps. Mechanical constraint refers to the physical fixation and contact condition between the heating wire and the copper sleeve heater. Specifically, it includes whether the heating wire is tightly attached to the inner wall of the copper sleeve heater, whether it is loose or detached, and changes in contact pressure. When the mechanical constraint between the heating wire and the copper sleeve heater is good, they form a rigid connection, the overall stiffness of the shell is high, and the shell resonant frequency is stable. When the heating wire is loose or detached, the mass distribution and constraint conditions of the system change, resulting in a decrease in the mechanical stiffness of the shell and a shift in the shell resonant frequency. Therefore, by monitoring the trend of the shell resonant frequency change, it is possible to indirectly determine whether the mechanical constraint between the heating wire and the copper sleeve has deteriorated.

[0027] The specific steps of the two-condition discrimination method include: Condition a: The shell resonant frequency offset changes in the same direction for A consecutive frequency sweeps (either a continuous positive offset or a continuous negative offset); where, the reasonable empirical range for A is usually 3 to 5 times, because taking 3 to 5 times can balance the suppression of random fluctuations and the rapid response to the real offset. That is, when the same offset occurs for 3 to 5 consecutive times and exceeds the threshold, the anomaly can be reliably determined, and there will be no misjudgment due to a single or two random fluctuations. Condition b: The absolute value of the change in the shell resonant frequency offset during each frequency sweep exceeds a preset absolute value (e.g., 0.5% of the initial shell resonant frequency).

[0028] The status of the copper bushing heater is determined according to the deterministic priority rule, including: Highest priority: If any condition is met, the current status is determined to be a Level 3 warning. Condition 1: Mechanical degradation, i.e., mechanical degradation confirmation criterion W(mech) = 1, judgment level is two or three, and there is an abnormal area, the abnormal area level is high; Condition 2: The judgment level is three, and there is an abnormal area, and the level of the abnormal area is medium or high; Second highest priority: If the judgment level is greater than or equal to level two (i.e., level two or level three judgment), and there is an abnormal area, and the level of the abnormal area is high or medium; if these conditions are met, the current status is determined to be a level two warning. Medium priority: The judgment level is level 1, or there is a trend warning indicator; if either condition is met, the current state is judged as level 1 warning. By using hierarchical priority rules, the status of the copper bushing heater is associated with the level of abnormal area and mechanical degradation indicators, thereby enabling early warning classification and judgment, which enhances the pertinence and timeliness of fault response.

[0029] The specific steps for making a decision are as follows: Based on the judgment of the copper sleeve heater status, physical control actions are performed through the division of labor and cooperation between MCU and CPLD; When a Level 3 warning is detected, the highest level of emergency response will be triggered. The MCU (Microcontroller Unit) will immediately send a shutdown command to the CPLD (Complex Programmable Logic Device). Upon receiving the signal, the CPLD will directly cut off the power circuit of the heater without any software delay, ensuring that the power is cut off within microseconds to prevent the accident from escalating. At the same time, an audible and visual alarm will be activated, with the red light remaining constantly on and the buzzer continuously sounding to warn on-site personnel to evacuate immediately or take emergency measures. When a Level 2 warning is issued, it indicates that the equipment has a serious potential for failure and needs to be shut down for maintenance. At this time, the MCU takes over control and drives the indicator light to flash rapidly to remind the operator. At the same time, the MCU will output a planned shutdown request signal, which is usually manifested as the closure of relay contacts, to notify the host computer or main control system to arrange an appropriate time to stop the equipment operation and avoid the impact of sudden power failure on production. When a Level 1 warning is issued, it indicates an initial risk warning, and the system as a whole remains under control. The MCU will control the indicator light to flash slowly as a visual reminder. At the same time, it will send a maintenance reminder message to the host computer through the communication interface, informing the device that there may be a minor abnormality or that it is about to enter a deterioration stage. It is recommended to check and maintain the device during the subsequent downtime window to prevent failure. The MCU and CPLD are the core control components integrated into the copper sleeve heater.

[0030] like Figure 3 As shown, the intelligent detection and early warning system for heating wire detachment in a copper bushing heater includes: Situation analysis engine unit: acquires initial benchmark dataset and real-time feature vectors; generates trend warning signs based on real-time feature vectors; automatically triggers multi-signal association adjudication process based on trend warning signs and outputs judgment level; Anomaly domain identification unit: Based on the judgment level, it performs anomaly region voting and location, generates anomaly regions, and obtains the anomaly region level; Mechanical deterioration verification unit: Based on the judgment level and abnormal area, it determines the constraint status of the monitored machinery and outputs a mechanical degradation confirmation mark; Overall machine status decision unit: Based on the judgment level, trend warning signs, abnormal areas, abnormal area levels, and mechanical degradation confirmation signs, the copper bushing heater status is determined and a decision is made according to the deterministic priority rules.

[0031] In this embodiment, by axially segmenting the copper sleeve heater and integrating multiple types of sensors, and combining global high-frequency impedance, differential current, and arc flash pulse signals, a full-dimensional online monitoring of the entire mechanical state from local contact degradation to overall mechanical condition is achieved for the first time. It can accurately locate the axial position of the heating wire loosening and trigger graded early warnings at the stage of micron-level gap changes and milliohm-level contact resistance increases. This transforms the traditional passive mode of "fault repair and overall replacement" into proactive predictive maintenance of "early warning, precise location, and on-demand maintenance", significantly shortening fault diagnosis time, avoiding unplanned downtime, extending heater life, and reducing maintenance costs. Employing a multi-physical quantity voting and confidence assessment mechanism effectively eliminates manufacturing deviations and environmental interference, ensuring high reliability of positioning results. Based on deterministic rules, a tiered early warning system (Level 1 maintenance prompt, Level 2 planned shutdown, Level 3 emergency power failure) and MCU / CPLD hardware-software collaborative control balances production continuity with microsecond-level safety protection under extreme conditions. Simultaneously, active frequency sweep diagnosis of the shell resonant frequency offset serves as macroscopic mechanical verification, further enhancing system robustness. The overall design does not rely on complex algorithm models, making it easy to implement in engineering and industrial certification, providing a reusable intelligent detection framework for the health management of copper-jacketed heaters and similar electrothermal equipment.

[0032] This application also provides an electronic device. The electronic device may include one or more processors and one or more memories. The memory stores computer-readable code, which, when executed by the one or more processors, can perform the intelligent detection and early warning method and system for the detachment of the heating wire in the copper sleeve heater as described above.

[0033] The methods and systems according to the embodiments of this application can also be implemented using the architecture of the electronic device shown in this application. The electronic device may include a bus, one or more CPUs, ROM, RAM, a communication port connected to a network, input / output, a hard disk, etc. The storage device in the electronic device, such as a ROM or hard disk, may store the intelligent detection and early warning method and system for the detachment of the heating wire of the copper sleeve heater provided in this application. Furthermore, the electronic device may also include a user interface. Of course, the architecture shown in this application is merely exemplary; when implementing different devices, one or more components in the electronic device shown in this application may be omitted according to actual needs.

[0034] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising a reference structure" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0035] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent detection and early warning of heating wire detachment in a copper bushing heater, characterized in that, The method includes: Step S1: Obtain the initial benchmark dataset and real-time feature vectors; generate trend warning signs for the real-time feature vectors; automatically trigger the multi-signal association adjudication process based on the trend warning signs and output the judgment level. Step S2: Based on the judgment level, perform abnormal area voting and location, generate abnormal areas, and obtain the abnormal area level; Step S3: Based on the judgment level and abnormal area, determine the mechanical constraint status of the monitored machinery and output a mechanical degradation confirmation flag; Step S4: Based on the judgment level, trend warning sign, abnormal area, abnormal area level, and mechanical degradation confirmation sign, determine the status of the copper sleeve heater according to the deterministic priority rule and make a decision.

2. The intelligent detection and early warning method for heating wire detachment in a copper sleeve heater according to claim 1, characterized in that, The process of obtaining the initial benchmark dataset includes: The copper jacket heater is divided into N physical segments. When it is first powered on or after maintenance reset, the initial insulation resistance, distributed capacitance, contact resistance, thermocouple temperature, shell resonant frequency, initial value of micro gap capacitance and micro vibration amplitude of each physical segment are collected. Compare with the standard ranges in the standard database to generate a set of physical segments for manufacturing deviations; The initial baseline dataset is composed of the initial insulation resistance, distributed capacitance, contact resistance, thermocouple temperature, shell resonant frequency, initial value of micro-gap capacitance, micro-vibration amplitude, and manufacturing deviation of each physical segment.

3. The intelligent detection and early warning method for heating wire detachment in a copper sleeve heater according to claim 2, characterized in that, The process of generating trend warning indicators for real-time feature vectors includes: During equipment operation, three hardware loops are invoked to collect high-frequency impedance signals, differential current signals, and high-frequency arc flash pulse signals, and multi-dimensional feature vectors are extracted to generate real-time feature vectors, including impedance offset, impedance change rate, effective value of differential current, peak current, over-limit event count, arc flash pulse event count, pulse width, and pulse interval. For real-time feature vectors, the rate of change of the real-time feature vectors with a time interval of D from the current time is obtained to obtain a real-time feature rate of change vector at multiple time scales; and a preset rate of change threshold range is set. If the rate of change of real-time features in the multi-timescale real-time feature rate of change vector is not within the preset rate of change threshold for R consecutive times, a trend warning sign is generated.

4. The intelligent detection and early warning method for heating wire detachment in a copper sleeve heater according to claim 3, characterized in that, The automatic triggering of the multi-signal correlation adjudication process outputs the judgment level, including: First-level judgment: When there is a trend warning sign for the rate of change of impedance offset, and there are no trend warning signs for the effective value of differential current, the number of over-limit events, and the rate of change of arc pulse count, it is determined that the contact is slowly degrading, triggering the first-level judgment. Second-level judgment: When there is a trend warning sign for the rate of change of impedance offset, and a trend warning sign for the effective value of differential current or the count of over-limit events, it is determined to be asymmetrical contact, triggering the second-level judgment. The third layer of judgment: When the rate of change of the arc flash event count exceeds the preset rate of change threshold, and the arc flash count shows an increasing trend in the most recent Y statistical periods, and there is already a first or second layer warning, it is determined to be intermittent dropout, triggering the third layer of judgment. For trend warning flag combinations of features beyond the three-layer judgment, only logs are recorded, and no judgment is triggered.

5. The intelligent detection and early warning method for heating wire detachment in a copper sleeve heater according to claim 4, characterized in that, The process of performing abnormal region voting and locating, and generating abnormal regions, includes: Real-time reading of micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature of N physical segments, and retrieval of corresponding data from the initial benchmark dataset; If the real-time readings of micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature exceed the sensor's range; if the deviation between the real-time readings of micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature and the corresponding data in the retrieved initial reference dataset exceeds a preset reasonable deviation threshold; or if the sensor reports an error, then the sensor will be marked as "untrustworthy"; otherwise, it will be marked as "trustworthy". For N physical segments, if the number of sensors marked as "trusted" in a physical segment is greater than or equal to M, then the corresponding physical segment is marked as "data-sufficient segment". If there are fewer than M sensors marked as "trusted" in a physical segment, then mark the physical segment as "data insufficient segment", do not perform any further operations, and record it in the log; The voting rules are set according to the "data-sufficient segment" to obtain the abnormal area.

6. The intelligent detection and early warning method for heating wire detachment in a copper sleeve heater according to claim 5, characterized in that, The setting of voting rules includes: For the "data-sufficient segment", the deviations of the micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature collected by the sensors in the physical segment are obtained in real time. If a "sufficient data segment" contains micro-gap capacitance, micro-vibration amplitude, and thermocouple temperature; and the deviation of the micro-gap capacitance is greater than or equal to a preset reasonable deviation threshold, while the deviation of the micro-vibration amplitude or the thermocouple temperature is also greater than or equal to a preset reasonable deviation threshold, then the "sufficient data segment" is an abnormal segment. If there is no micro-gap capacitance, vibration amplitude or thermocouple temperature in the "data sufficient segment", then two existing deviations are identified. If both existing deviations are greater than or equal to the preset reasonable deviation threshold, then it is determined to be an abnormal segment. Adjacent physical segments are merged to obtain an abnormal region. If two abnormal segments are not adjacent, the abnormal segment is directly defined as an abnormal region.

7. The intelligent detection and early warning method for heating wire detachment in a copper sleeve heater according to claim 6, characterized in that, The acquisition of the abnormal region level includes: For each physical segment in the abnormal region, calculate the correction score for the "trustworthy" sensor in that physical segment. The base score for the physics section is obtained by averaging the corrected scores. If the physical segment is in the set of physical segments with manufacturing deviations, or if it is marked as "untrustworthy", then the base score is adjusted to obtain the fusion score; The confidence level of the abnormal region is obtained by taking the weighted average of the fusion scores of all physical segments within the abnormal region. When the confidence level of an outlier region is greater than or equal to the upper limit of the baseline confidence level, it is considered a high-level outlier region. When the confidence level of an outlier region is between the upper and lower limits of the baseline confidence level, it is considered a medium-level outlier region. When the confidence level of an outlier region is less than or equal to the lower limit of the baseline confidence level, it is considered a low-level outlier region.

8. The intelligent detection and early warning method for heating wire detachment in a copper sleeve heater according to claim 7, characterized in that, The process of determining the mechanical constraint status based on the judgment level and abnormal area, and outputting a mechanical degradation confirmation flag, includes: Based on the judgment level and abnormal area, a dynamic decision is adopted to apply an excitation signal to the shell. Under the application of the excitation signal, the shell generates forced vibration and produces a real-time vibration response signal. The real-time shell resonant frequency is extracted by the peak search algorithm and then compared with the initial shell resonant frequency in the initial benchmark dataset to obtain the shell resonant frequency offset. The mechanical constraint state is monitored by a dual-condition discrimination method based on the shell resonant frequency offset. If either condition is met, it is considered mechanical degradation; otherwise, it is considered normal mechanical operation, and a mechanical degradation confirmation flag is output.

9. The intelligent detection and early warning method for heating wire detachment in a copper sleeve heater according to claim 8, characterized in that, The determination of the state of the copper bushing heater based on the deterministic priority rule includes: Highest priority: If any condition is met, the current status is determined to be a Level 3 warning. Condition 1: Mechanical degradation, with a judgment level of two or three, and the existence of an abnormal area, the abnormal area being of a high level; Condition 2: The judgment level is three, and there is an abnormal area, and the level of the abnormal area is medium or high; Second highest priority: If the judgment level is greater than or equal to level two, and there is an abnormal area, and the level of the abnormal area is high or medium; if these conditions are met, the current status is judged as a level two warning. Medium priority: The judgment level is level 1, or there is a trend warning indicator; if either condition is met, the current state is judged as level 1 warning. Lowest priority: If a state does not meet the highest, second-highest, or medium priority, then the current state is a safe state.

10. A smart detection and early warning system for heating wire detachment in a copper sleeve heater, characterized in that, The system includes: Situation analysis engine: acquires initial benchmark dataset and real-time feature vectors; generates trend warning signs based on real-time feature vectors; automatically triggers multi-signal association adjudication process based on trend warning signs and outputs judgment level; Anomaly domain identification unit: Based on the judgment level, it performs anomaly region voting and location, generates anomaly regions, and obtains the anomaly region level; Mechanical deterioration verification unit: Based on the judgment level and abnormal area, it determines the constraint status of the monitored machinery and outputs a mechanical degradation confirmation mark; Overall machine status decision center: Based on the judgment level, trend warning signs, abnormal areas, abnormal area levels, and mechanical degradation confirmation signs, the copper sleeve heater status is determined and a decision is made according to the deterministic priority rules.

Images