Load current fault diagnosis and protection system

Abstract

The invention discloses a load current fault diagnosis and protection system, and relates to the technical field of power electronic equipment fault diagnosis, a current data acquisition module is used for acquiring a real-time current sampling value and a current sampling timestamp of load equipment, and synchronously acquiring operation condition parameters of the load equipment; and the current anomaly index calculation module is used for calculating a current fluctuation anomaly index and a working condition adaptation anomaly index according to the real-time current sampling value, the current sampling timestamp and the operation working condition parameters, and superposing the current fluctuation anomaly index and the working condition adaptation anomaly index to obtain a comprehensive current anomaly index. According to the method, the current data and the working condition parameters are synchronously collected, the multi-dimensional abnormal indexes (current fluctuation and working condition adaptation) are constructed, the problems of misjudgment and missed judgment of traditional single threshold value diagnosis are solved, the current abnormal degree can be accurately quantified by integrating the current abnormal indexes, and a scientific basis is provided for fault recognition.

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis technology for power electronic equipment, and in particular to a load current fault diagnosis and protection system. Background Technology

[0002] During the operation of industrial load equipment (such as motors, frequency converters, and high-power power supplies), abnormal current is one of the core causes of equipment failure, shortened service life, and even safety accidents. Currently, most mainstream current protection schemes rely on fixed threshold triggering (such as overcurrent relays and fuses), which can only protect against severe overcurrent or short-circuit faults exceeding preset thresholds, and cannot identify potential fault risks such as abnormal current fluctuations and operating condition mismatches.

[0003] Existing technologies suffer from two major problems: First, fault diagnosis relies on a single dimension, judging solely based on the absolute value of current, neglecting current change rate, timing correlation, and operating condition adaptability. This easily leads to misjudgments (e.g., a normal large current at startup being misjudged as a fault) or missed judgments (e.g., a slowly rising overload current not being identified in time). Second, protection actions lack a graded mechanism, employing power-off protection regardless of fault severity, resulting in equipment shutdowns for non-critical faults and affecting production continuity. Furthermore, traditional solutions cannot locate the fault area and related circuit nodes, leading to low fault diagnosis efficiency and increased equipment maintenance costs. Summary of the Invention

[0004] The purpose of this invention is to solve the problems of single-dimensional load current fault diagnosis, lack of flexibility in protection action, and difficulty in fault location in the prior art. It proposes a load current fault diagnosis and protection system, which realizes accurate diagnosis and efficient protection of load current faults through multi-dimensional anomaly index calculation, time-sequential fault location, and hierarchical protection decision-making.

[0005] To address the problems existing in the prior art, the present invention adopts the following technical solution:

[0006] Load current fault diagnosis and protection system, including:

[0007] The current data acquisition module is used to obtain the real-time current sampling value and current sampling timestamp of the load device, and to synchronously collect the operating condition parameters of the load device.

[0008] The current anomaly index calculation module is used to calculate the current fluctuation anomaly index and the operating condition adaptation anomaly index based on the real-time current sampling value, the current sampling timestamp, and the operating condition parameters, and then superimpose them to obtain the comprehensive current anomaly index.

[0009] The fault feature candidate point detection module is used to determine whether the load current is within the sensitive range of sudden change and the steady-state fluctuation range based on the comprehensive current anomaly index. It performs differential calculation on the real-time current sample value and the standard current threshold curve, searches for feature abrupt change points along the current change trend, and filters and records them to form a fault feature candidate point set. The fault area location module is used to filter out the effective fault point subset in the fault feature candidate point set based on the normal operating current stable range of the load equipment, performs time-series correlation analysis on the effective fault point subset, and generates a fault area location mask.

[0010] The protection action decision module is used to map and match the fault area location mask with the circuit topology partition of the load equipment, determine the correlation between the fault area and the key nodes of the circuit, obtain the correlation determination result, and determine the protection action level based on the correlation determination result.

[0011] The protection execution module is used to generate protection execution options, including emergency power outage, current limiting adjustment, and alarm prompts, based on the protection action level, and to trigger the corresponding protection operation.

[0012] Preferably, the real-time current sampling value and current sampling timestamp of the load device are obtained, and the operating condition parameters of the load device are collected synchronously, including:

[0013] The integrity of the current sampling data within the same load operation cycle is verified to confirm that the sampling frequency meets the preset 1kHz-10kHz range requirement and that there is no data loss or abnormal jump.

[0014] Extract the real-time current sampling value and the corresponding current sampling timestamp from each sampling point from the current sampling data, and convert the timestamp into a unified millisecond-level time base.

[0015] Through the load device's operating condition sensor interface, operating condition parameters are collected synchronously with current sampling, including load power, input voltage, ambient temperature, and device operating time.

[0016] Preferably, based on real-time current sampling values, current sampling timestamps, and operating condition parameters, a current fluctuation anomaly index and an operating condition adaptation anomaly index are calculated and superimposed to obtain a comprehensive current anomaly index, including:

[0017] The time interval between adjacent sampling points is calculated based on the current sampling timestamp. The difference between real-time current sampling values ​​is combined to obtain the current change rate. The current change rate is compared with the preset normal change rate threshold. The normalized value of the part exceeding the threshold is used as the current fluctuation abnormal index.

[0018] A condition-current adaptation model is constructed based on the operating condition parameters. The theoretical current range is obtained by inputting the current operating condition parameters. The percentage deviation between the real-time current sampling value and the midline of the theoretical current range is calculated as the condition adaptation anomaly index.

[0019] The current fluctuation anomaly index and the operating condition adaptation anomaly index are superimposed with a weighting of 6:4 to obtain the comprehensive current anomaly index. The index range is limited to 0-1, with a higher index indicating a more severe degree of current anomaly.

[0020] Preferably, based on the comprehensive current anomaly index, it is determined whether the load current is within the abrupt change sensitive range or the steady-state fluctuation range. A difference operation is performed between the real-time current sample value and the standard current threshold curve. Characteristic abrupt change points are searched along the current change trend direction, and these are screened and recorded to form a fault characteristic candidate point set, including:

[0021] When the comprehensive current anomaly index is less than 0.3, the current is determined to be within the normal range, and the fault feature candidate point set is set to an empty set.

[0022] When the comprehensive current anomaly index is greater than or equal to 0.3, the real-time current sampling value is smoothed and filtered to remove high-frequency noise, and the standard current threshold curve matching the current operating condition is retrieved. In the sudden change sensitive range of load current (absolute value of current change rate > 5A / ms) and the steady-state fluctuation range (absolute value of current change rate ≤ 5A / ms), the difference curve between the real-time current sampling value and the standard current threshold curve is calculated respectively, and the peak point and valley point of the difference curve are obtained.

[0023] Correlation analysis is performed on the extreme points of the differential curve along the direction of current change (rising or falling). Extreme points with more than 3 consecutive sampling points that satisfy the absolute value of the difference > the preset deviation threshold are selected as characteristic mutation points. The current value, timestamp and interval of each characteristic mutation point are recorded and merged to form a set of fault feature candidate points.

[0024] Preferably, based on the stable operating current range of the load equipment, a subset of valid fault points is selected from the candidate fault point set, including:

[0025] Retrieve historical normal operating current data of the load equipment under rated operating conditions, and use statistical analysis methods to determine the stable range of normal operating current. The upper and lower limits of the range are the mean of historical data ± 2 times the standard deviation.

[0026] Traverse each feature mutation point in the fault feature candidate point set and determine whether the corresponding real-time current sampling value exceeds the normal operating current stability range.

[0027] If the abrupt change exceeds the stable range, the abrupt change point is included in the subset of valid fault points; if it does not exceed the range, it is determined to be a normal fluctuation point and is removed.

[0028] Preferably, a temporal correlation analysis is performed on a subset of valid fault points to generate a fault area location mask, including:

[0029] A time series clustering algorithm is used to divide the effective fault point subset into continuous fault time periods according to timestamps, with the time interval between fault points in each fault time period not exceeding 100ms;

[0030] Based on the current sampling timing of the load device, a time axis pixel domain synchronized with the sampling period is constructed, with each pixel corresponding to a 1ms sampling time. For each fault period, the corresponding time interval is marked in the time axis pixel domain, and the gray value of the marked pixel is determined according to the maximum current deviation value in that period (the larger the deviation, the higher the gray value), thus forming the initial fault area.

[0031] The initial fault region is morphologically dilated to fill the gaps in the time interval. Then, the fault region contour is extracted by the edge detection algorithm to generate the fault region location mask.

[0032] Preferably, the fault area location mask is mapped and matched with the circuit topology partition of the load equipment, the correlation between the fault area and the key nodes of the circuit is determined, and the correlation determination result is obtained. Based on the correlation determination result, the protection action level is determined, including:

[0033] Obtain the circuit topology diagram of the load device, divide the circuit topology into sections such as power input area, power conversion area, and load output area, and mark the key circuit nodes (such as rectifier bridge, IGBT, and filter capacitor) in each section.

[0034] Establish the time axis of the fault area location mask and the current path timing mapping relationship of the circuit topology partition to determine the circuit operation stage and associated topology partition corresponding to each fault period.

[0035] Calculate the current influence coefficient between the fault area and each key node of the circuit. A coefficient greater than 0.7 is considered high correlation, 0.3-0.7 is considered medium correlation, and less than 0.3 is considered low correlation.

[0036] When a highly correlated critical node exists, the protection action level is set to Level 1 (emergency power failure); when only a medium-correlated critical node exists, the level is set to Level 2 (current limiting adjustment); when only a low-correlated critical node exists, the level is set to Level 3 (alarm indication).

[0037] Preferably, based on the protection action level, protection execution options including emergency power outage, current limiting adjustment, and alarm prompts are generated, and corresponding protection operations are triggered, including:

[0038] Construct a three-element protection execution set that includes emergency power outage, current limiting adjustment, and alarm prompts, and clarify the triggering conditions and execution logic of each option;

[0039] When the protection action level is Level 1, the emergency power-off operation is triggered first, and an alarm signal is sent to the monitoring terminal at the same time; when the level is Level 2, the current limiting adjustment is triggered to limit the current to within 80% of the rated value, and the current change is continuously monitored; when the level is Level 3, only the alarm prompt is triggered, and the fault location and related node information are displayed.

[0040] After the protection operation is executed, the current recovery data of the load device is collected in real time. If the current does not return to normal within 10 seconds, the protection action level is automatically upgraded (level 3 to level 2, level 2 to level 1); if it returns to normal, the fault information is recorded and the protection mode is exited.

[0041] Compared with the prior art, the beneficial effects of the present invention are:

[0042] In this invention, by synchronously collecting current data and operating parameters, a multi-dimensional anomaly index (current fluctuation + operating condition adaptation) is constructed, solving the problem of misjudgment and missed judgment in traditional single-threshold diagnosis. The comprehensive current anomaly index can accurately quantify the degree of current anomaly, providing a scientific basis for fault identification. Based on time-series correlation analysis, a fault area location mask is generated, realizing the time dimension location of the fault. Combined with circuit topology mapping and matching, the key nodes of fault association can be clearly identified, greatly improving the efficiency of fault diagnosis and reducing maintenance costs. A three-level protection action level is established, which flexibly selects emergency power cut-off, current limiting adjustment or alarm prompt according to the severity of the fault, avoiding equipment downtime under non-critical faults, ensuring production continuity, and ensuring equipment safety under severe faults through an automatic upgrade mechanism. Attached Figure Description

[0043] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0044] Figure 1 This is a functional block diagram of the load current fault diagnosis and protection system of the present invention. Detailed Implementation

[0045] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0046] Example: This example provides a load current fault diagnosis and protection system, including: a current data acquisition module, used to acquire real-time current sampling values ​​and current sampling timestamps of the load device, and simultaneously acquire the operating condition parameters of the load device; a current anomaly index calculation module, used to calculate the current fluctuation anomaly index and the operating condition adaptation anomaly index based on the real-time current sampling values, current sampling timestamps, and operating condition parameters, and superimpose them to obtain a comprehensive current anomaly index; and a fault feature candidate point detection module, used to determine whether the fault is within the sensitive range of sudden changes and the steady-state fluctuation range of the load current based on the comprehensive current anomaly index, perform differential calculations on the real-time current sampling values ​​and the standard current threshold curve, and search for feature sudden changes along the current change trend direction. The system consists of four modules: a fault feature candidate point set, a fault area location module, and a protection action decision module. The fault area location module filters out a subset of valid fault points from the candidate point set based on the stable operating current range of the load equipment. It then performs time-series correlation analysis on the valid fault point subset to generate a fault area location mask. The protection action decision module maps the fault area location mask to the circuit topology partitions of the load equipment, determines the correlation between the fault area and key circuit nodes, obtains the correlation determination result, and determines the protection action level based on the correlation determination result. Finally, the protection execution module generates protection execution options, including emergency power cut-off, current limiting adjustment, and alarm prompts, based on the protection action level, and triggers the corresponding protection operation.

[0047] Acquire real-time current sampling values ​​and current sampling timestamps from the load device, and simultaneously collect the operating condition parameters of the load device, including:

[0048] The integrity of current sampling data within the same load operation cycle is verified to ensure that the sampling frequency meets the preset requirement of 1kHz-10kHz and that there is no data loss or abnormal jumps. Specifically, the integrity of the data is verified using a cyclic redundancy check (CRC) algorithm. If the verification fails, resampling is triggered until valid data is obtained. The real-time current sampling value and the corresponding current sampling timestamp of each sampling point are extracted from the current sampling data. The timestamps are uniformly converted into millisecond-level time bases to ensure the time synchronization of different sampling devices. Through the load device's operating condition sensor interface (such as RS485, Modbus protocol), operating condition parameters synchronized with the current sampling are collected, including load power (accuracy ±2%), input voltage (accuracy ±1%), ambient temperature (range -20℃-85℃, accuracy ±0.5℃), and device operating time. The sampling interval is consistent with the current sampling (1ms-10ms).

[0049] Based on real-time current sampling values, current sampling timestamps, and operating condition parameters, the current fluctuation anomaly index and the operating condition adaptation anomaly index are calculated and superimposed to obtain a comprehensive current anomaly index, including:

[0050] The time interval Δt between adjacent sampling points is calculated based on the current sampling timestamp. Combined with the difference ΔI between real-time current sampling values, the current change rate k = ΔI / Δt is obtained. k is compared with a preset normal change rate threshold k0 (set according to load type, e.g., k0 = 3A / ms for motor load, k0 = 5A / ms for power supply load). If k > k0, the current fluctuation anomaly index = (k - k0) / k0; if k ≤ k0, the index = 0. The final index is normalized to the range of 0-0.6. A condition-current adaptation model is constructed based on the operating parameters, using a multiple linear regression algorithm with load power P, input voltage U, and ambient temperature T as independent variables and current I as the dependent variable. The model expression is: ( These are the fitting coefficients. (For error terms); input the current operating parameters to obtain the theoretical current range. Calculate the real-time current sampling value With respect to the theoretical current range centerline The percentage deviation, i.e., the operating condition adaptation anomaly index. The index is normalized to the range of 0-0.4. The current fluctuation abnormality index and the operating condition adaptation abnormality index are superimposed with a weight of 6:4 to obtain the comprehensive current abnormality index, that is, comprehensive index = 0.6 × fluctuation index + 0.4 × adaptation index. The index range is limited to 0-1. An index ≥ 0.8 is a serious abnormality, 0.5-0.8 is a moderate abnormality, 0.3-0.5 is a slight abnormality, and < 0.3 is normal.

[0051] Based on the comprehensive current anomaly index, it is determined whether the load current is within the sensitive range of sudden changes or the steady-state fluctuation range. A difference operation is performed between the real-time current sample value and the standard current threshold curve. Characteristic abrupt change points are searched along the current change trend, and these are screened and recorded to form a set of candidate fault feature points, including:

[0052] When the comprehensive current anomaly index is less than 0.3, the current is determined to be within the normal range, and the fault characteristic candidate point set is set to an empty set. When the comprehensive current anomaly index is greater than or equal to 0.3, the real-time current sample value is processed by a 5-point moving average filter to remove high-frequency noise (filter cutoff frequency 100Hz), and a standard current threshold curve matching the current operating condition is retrieved (generated by fitting historical normal operating condition data, including upper and lower thresholds). In the abrupt change sensitive range of the load current (absolute value of current change rate > 5A / ms) and the steady-state fluctuation range (absolute value of current change rate ≤ 5A / ms), the difference curve between the real-time current sample value and the upper limit value of the standard current threshold curve is calculated respectively. The difference curve between the lower limit and the lower limit value Extract the peak and valley points with absolute values ​​> 0 from the two difference curves; perform correlation analysis on the extreme points of the difference curves along the direction of current change (rising or falling); if the absolute value of the difference of three consecutive sampling points is greater than the preset deviation threshold (e.g., 5% of the rated current), then the intermediate sampling point is taken as the characteristic mutation point; record the current value, timestamp, and interval (mutation sensitive area / steady-state fluctuation area) of each characteristic mutation point, and merge them to form a fault feature candidate point set; if the number of candidate points is < 3, it is determined to be an occasional fluctuation, and the candidate point set is cleared.

[0053] Preferably, based on the stable operating current range of the load equipment, a subset of valid fault points is selected from the candidate fault point set, including:

[0054] Retrieve historical normal operating current data of the load equipment under rated operating conditions for the past 30 days (sample size ≥ 1000 sets), and use... The rule determines the stable operating current range, i.e., the lower limit of the range equals the historical data average. Upper limit of the interval ( (Based on the standard deviation of historical data), ensuring that 95.45% of normal currents fall within the range; traversing each characteristic abrupt change point in the fault characteristic candidate point set, determining its corresponding real-time current sampling value. Does it exceed the stable range? lower limit or If the upper limit is reached, the abrupt change point is included in the subset of valid fault points; if it is not exceeded, it is judged as a normal fluctuation point and is removed; the subset of valid fault points is screened a second time, and the time interval between adjacent fault points is calculated. If the interval is >500ms, it is judged as an independent fluctuation point and is removed. Finally, continuous and dense fault points are retained.

[0055] Preferably, a temporal correlation analysis is performed on a subset of valid fault points to generate a fault area location mask, including:

[0056] The K-means clustering algorithm is used to divide the effective fault point subset into consecutive fault time periods according to timestamps. The number of clusters K is automatically determined based on the time interval (adjacent points with an interval ≤100ms are grouped into the same cluster), and each fault time period corresponds to one cluster. Based on the current sampling sequence of the load equipment (sampling period T = 1ms-10ms), a time axis pixel domain is constructed, with a pixel size of 1ms × 1px and a pixel domain length equal to the current operating cycle duration (e.g., 10s corresponds to 10000px). For each fault time period, the corresponding time interval (e.g., 500ms-800ms) is marked within the time axis pixel domain, and the maximum current deviation value within that time period is used as the basis for the clustering. Determine the grayscale value of the marked pixel: The grayscale value is 255 (pure white). Gray value = 180 The initial fault region is formed by setting the grayscale value to 120. The initial fault region is then subjected to 3×3 pixel morphological dilation to fill the gaps in the time interval (such as blank pixels caused by single-point removal). The fault region contour is then extracted using the Canny edge detection algorithm (threshold 100-200) to generate a binarized fault region location mask (fault region pixel value = 255, non-fault region = 0).

[0057] Preferably, the fault area location mask is mapped and matched with the circuit topology partition of the load equipment, the correlation between the fault area and the key nodes of the circuit is determined, and the correlation determination result is obtained. Based on the correlation determination result, the protection action level is determined, including:

[0058] Obtain the circuit topology diagram (CAD format) of the load device, and use image recognition algorithms to divide the circuit topology into sections such as the power input area (including rectifier circuit and filter capacitor), the power conversion area (including IGBT and transformer), and the load output area (including contactor and terminal block). Each section is marked with a unique ID. Simultaneously, key circuit nodes within each section are marked, such as the rectifier bridge output, IGBT collector, and both ends of the filter capacitor. The current path and timing characteristics of these key nodes are recorded (e.g., the power input area corresponds to the startup phase, and the power conversion area corresponds to the operation phase). Establish a time axis mapping relationship between the fault location mask and the current path timing of the circuit topology sections: using the device control timing table, determine the circuit operating section corresponding to each time stage (e.g., 0-100ms is the startup phase, corresponding to the power input area; after 100ms is the operation phase, corresponding to the power conversion area). The fault period is mapped to a specific topology partition; the current influence coefficient between the fault area and each key node of the circuit is calculated: based on the circuit simulation model (such as PSpice), the degree of influence of the abnormal current in the fault area on the current of the key nodes is simulated. The influence coefficient = the change in node current after the fault / the normal node current. A coefficient > 0.7 is considered high correlation, 0.3-0.7 is considered medium correlation, and less than 0.3 is considered low correlation. When there are high correlation key nodes (such as IGBTs and rectifier bridges), the protection action level is set to Level 1 (emergency power failure), which requires cutting off the main power supply within 10ms; when there are only medium correlation key nodes (such as filter capacitors), the level is set to Level 2 (current limiting regulation), which limits the current to within 80% of the rated value through PWM control; when there are only low correlation key nodes (such as output terminals), the level is set to Level 3 (alarm prompt), which only triggers audible and visual alarms.

[0059] Preferably, based on the protection action level, protection execution options including emergency power outage, current limiting adjustment, and alarm prompts are generated, and corresponding protection operations are triggered, including:

[0060] The system constructs a three-element protection execution set including emergency power failure, current limiting regulation, and alarm indication: Emergency power failure triggers a relay to cut off the main circuit, with a response time ≤10ms; current limiting regulation controls the duty cycle of power devices (such as IGBTs) through PWM signals output by the MCU, with an adjustment step size of 5% of the rated current; alarm indication drives a buzzer (frequency 2kHz) and an LED indicator (red flashing, period 1s), and simultaneously sends alarm information to the monitoring terminal via Ethernet; when the protection action level is Level 1, the emergency power failure operation is triggered first, and the fault time, fault area, and associated node information are recorded and stored to a local SD card (storage capacity ≥16GB); when the level is Level 2, the current limiting regulation is triggered. The system collects current data every 50ms. If the current is ≤80% of the rated value for three consecutive times, the current limiting state is maintained. If the current still exceeds the limit, it is upgraded to Level 1 protection after 10s. When the level is Level 3, only an alarm prompt is triggered, displaying the fault location (such as "Power Conversion Zone IGBT Related Fault") and related node information. If no manual confirmation is made within 1 minute, it is upgraded to Level 2 protection. After the protection operation is executed, the current recovery data of the load equipment is collected in real time (sampling frequency = 1kHz). If the current recovers to the normal stable range (μ±σ) within 10s, the fault recovery time is recorded and the protection mode is exited. If it does not recover, the protection action level is automatically upgraded (Level 3 to Level 2, Level 2 to Level 1) until the fault is cleared or the equipment is powered off.

[0061] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A load current fault diagnosis and protection system, characterized in that, include: The current data acquisition module is used to obtain the real-time current sampling value and current sampling timestamp of the load device, and to synchronously collect the operating condition parameters of the load device. The current anomaly index calculation module is used to calculate the current fluctuation anomaly index and the operating condition adaptation anomaly index based on the real-time current sampling value, the current sampling timestamp, and the operating condition parameters, and then superimpose them to obtain the comprehensive current anomaly index. The fault feature candidate point detection module is used to determine whether the load current is in the sensitive range of sudden change and the steady-state fluctuation range based on the comprehensive current anomaly index. It performs differential calculation on the real-time current sample value and the standard current threshold curve, searches for feature sudden change points along the current change trend, and filters and records them to form a fault feature candidate point set. The fault area location module is used to filter out a subset of valid fault points from the set of candidate fault feature points based on the stable range of the normal operating current of the load equipment, perform time-series correlation analysis on the subset of valid fault points, and generate a fault area location mask. The protection action decision module is used to map and match the fault area location mask with the circuit topology partition of the load equipment, determine the correlation between the fault area and the key nodes of the circuit, obtain the correlation determination result, and determine the protection action level based on the correlation determination result. The protection execution module is used to generate protection execution options, including emergency power outage, current limiting adjustment, and alarm prompts, based on the protection action level, and to trigger the corresponding protection operation.

2. The load current fault diagnosis and protection system according to claim 1, characterized in that, Acquire real-time current sampling values ​​and current sampling timestamps from the load device, and simultaneously collect the operating condition parameters of the load device, including: The integrity of the current sampling data within the same load operation cycle is verified to confirm that the sampling frequency meets the preset 1kHz-10kHz range requirement and that there is no data loss or abnormal jump. Extract the real-time current sampling value and the corresponding current sampling timestamp from each sampling point from the current sampling data, and convert the timestamp into a unified millisecond-level time base. Through the load device's operating condition sensor interface, operating condition parameters are collected synchronously with current sampling, including load power, input voltage, ambient temperature, and device operating time.

3. The load current fault diagnosis and protection system according to claim 1, characterized in that, Based on real-time current sampling values, current sampling timestamps, and operating condition parameters, the current fluctuation anomaly index and the operating condition adaptation anomaly index are calculated and superimposed to obtain a comprehensive current anomaly index, including: The time interval between adjacent sampling points is calculated based on the current sampling timestamp. The difference between real-time current sampling values ​​is combined to obtain the current change rate. The current change rate is compared with the preset normal change rate threshold. The normalized value of the part exceeding the threshold is used as the current fluctuation abnormal index. A condition-current adaptation model is constructed based on the operating condition parameters. The theoretical current range is obtained by inputting the current operating condition parameters. The percentage deviation between the real-time current sampling value and the midline of the theoretical current range is calculated as the condition adaptation anomaly index. The current fluctuation abnormality index and the operating condition adaptation abnormality index are superimposed with a weight of 6:4 to obtain the comprehensive current abnormality index. The index range is limited to 0-1, and the higher the index, the more severe the current abnormality.

4. The load current fault diagnosis and protection system according to claim 1, characterized in that, Based on the comprehensive current anomaly index, it is determined whether the load current is within the sensitive range of sudden changes or the steady-state fluctuation range. A difference operation is performed between the real-time current sample value and the standard current threshold curve. Characteristic abrupt change points are searched along the current change trend, and these are screened and recorded to form a set of candidate fault feature points, including: When the comprehensive current anomaly index is less than 0.3, the current is determined to be within the normal range, and the fault feature candidate point set is set to an empty set. When the comprehensive current anomaly index is greater than or equal to 0.3, the real-time current sampling value is smoothed and filtered to remove high-frequency noise, and the standard current threshold curve matching the current operating condition is retrieved. In the sudden change sensitive range and steady-state fluctuation range of the load current, the difference curve between the real-time current sampling value and the standard current threshold curve is calculated respectively to obtain the peak point and valley point of the difference curve. Correlation analysis is performed on the extreme points of the differential curve along the direction of current change. Extreme points with more than three consecutive sampling points that satisfy the absolute value of the difference > the preset deviation threshold are selected as characteristic mutation points. The current value, timestamp and interval of each characteristic mutation point are recorded and merged to form a set of fault feature candidate points.

5. The load current fault diagnosis and protection system according to claim 4, characterized in that, Based on the stable operating current range of the load equipment, a subset of valid fault points is selected from the candidate fault point set, including: Retrieve historical normal operating current data of the load equipment under rated operating conditions, and use statistical analysis methods to determine the stable range of normal operating current. The upper and lower limits of the range are the mean of historical data ± 2 times the standard deviation. Traverse each feature mutation point in the fault feature candidate point set and determine whether the corresponding real-time current sampling value exceeds the normal operating current stability range. If the abrupt change exceeds the stable range, the abrupt change point is included in the subset of valid fault points; if it does not exceed the range, it is determined to be a normal fluctuation point and is removed.

6. The load current fault diagnosis and protection system according to claim 5, characterized in that, Perform temporal correlation analysis on a subset of valid fault points to generate a fault area location mask, including: A time series clustering algorithm is used to divide the effective fault point subset into continuous fault time periods according to timestamps, with the time interval between fault points in each fault time period not exceeding 100ms; Based on the current sampling timing of the load device, a time axis pixel domain synchronized with the sampling period is constructed, with each pixel corresponding to a 1ms sampling time; for each fault period, the corresponding time interval is marked in the time axis pixel domain, and the gray value of the marked pixel is determined according to the maximum current deviation value in that period to form the initial fault area; The initial fault region is morphologically dilated to fill the gaps in the time interval. Then, the fault region contour is extracted by the edge detection algorithm to generate the fault region location mask.

7. The load current fault diagnosis and protection system according to claim 1, characterized in that, The fault area location mask is mapped and matched with the circuit topology partitions of the load equipment. The correlation between the fault area and the key nodes of the circuit is determined to obtain the correlation determination result. Based on the correlation determination result, the protection action level is determined, including: Obtain the circuit topology diagram of the load device, divide the circuit topology into sections such as power input area, power conversion area, and load output area, and mark the key circuit nodes in each section; Establish the time axis of the fault area location mask and the current path timing mapping relationship of the circuit topology partition to determine the circuit operation stage and associated topology partition corresponding to each fault period. Calculate the current influence coefficient between the fault area and each key node of the circuit. A coefficient greater than 0.7 is considered high correlation, 0.3-0.7 is considered medium correlation, and less than 0.3 is considered low correlation. When a highly correlated critical node exists, the protection action level is set to Level 1; when only a moderately correlated critical node exists, the level is set to Level 2; and when only a lowly correlated critical node exists, the level is set to Level 3.

8. The load current fault diagnosis and protection system according to claim 7, characterized in that, Based on the protection action level, protection execution options including emergency power outage, current limiting adjustment, and alarm prompts are generated, and corresponding protection operations are triggered, including: Construct a three-element protection execution set that includes emergency power outage, current limiting adjustment, and alarm prompts, and clarify the triggering conditions and execution logic of each option; When the protection action level is Level 1, the emergency power-off operation is triggered first, and an alarm signal is sent to the monitoring terminal at the same time; when the level is Level 2, the current limiting adjustment is triggered to limit the current to within 80% of the rated value, and the current change is continuously monitored; when the level is Level 3, only the alarm prompt is triggered, and the fault location and related node information are displayed. After the protection operation is executed, the current recovery data of the load device is collected in real time. If the current does not return to normal within 10 seconds, the protection action level is automatically upgraded; if it returns to normal, the fault information is recorded and the protection mode is exited.

Images