Low-voltage load fault identification method, device, equipment and medium

Abstract

The invention belongs to the technical field of distribution line fault identification, and particularly discloses a low-voltage load fault identification method, device and equipment and a medium, and the method comprises the steps: obtaining multi-dimensional data of a low-voltage load loop; calculating a real-time characteristic quantity representing a load state based on the multi-dimensional data; generating a self-adaptive threshold value through load operation baseline self-learning based on the historical characteristic quantity representing the load state of the low-voltage load loop within the specified time length; the real-time characteristic quantity representing the load state is compared with a corresponding self-adaptive threshold value, multi-level logic judgment of time sequence correlation analysis is carried out based on a comparison result, and a fault priority is determined; and performing fault early warning and fault protection based on the fault priority. According to the method, early and accurate identification and grading protection of various fault types can be realized on the premise of ensuring low calculation complexity.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power distribution line fault identification, and particularly relates to a low-voltage load fault identification method, device, equipment and medium. BACKGROUND

[0002] In industrial, commercial and civil buildings, the low-voltage power distribution system is the final link of power transmission and distribution, and the stable and reliable operation of its load equipment (such as motors, frequency converters, lighting devices, etc.) is directly related to production efficiency, equipment safety and personal safety. Traditionally, the fault protection in this field mainly relies on mechanical and electrical devices such as circuit breakers and leakage protectors. These devices can quickly cut off the circuit when serious faults such as overcurrent, short circuit, and ground fault occur, effectively preventing the expansion of accidents, and belong to a "post-protection" mechanism. However, with the increasing demand for continuous production and intelligent operation and maintenance in modern industry, this protection method that only acts after the fault has occurred and caused an impact cannot meet the urgent need for early warning, identification and preventive maintenance of potential faults.

[0003] Therefore, in the field of low-voltage load fault detection and protection, several fault detection methods based on current, voltage and other signals have been formed, but all have certain limitations.

[0004] For example, threshold-based overcurrent protection sets a fixed action threshold for the current, and when the sampled current exceeds the threshold, the protection device trips. This method is simple and reliable, but it cannot detect early signs of faults far below the trip threshold, such as slight inter-turn short circuits in motor windings, gradual aging of cable insulation, etc. For example, the diagnostic method based on harmonic analysis diagnoses faults such as motor rotor broken bars and bearing damage by analyzing the harmonic components in the load current. However, there are a large number of nonlinear loads in low-voltage power distribution systems, which produce rich background harmonics and cause serious interference to fault characteristic harmonics. In addition, some research has proposed models based on complex algorithms and artificial intelligence for low-voltage load fault detection, such as intelligent diagnostic models based on deep learning, support vector machines and other complex algorithms. These solutions have high identification accuracy in theory, but their implementation relies on a large amount of high-quality training data and powerful computing platforms. Moreover, their huge computational load and storage requirements are in sharp conflict with the low-cost, low-power embedded hardware resources widely used in the field. This "resource conflict" greatly limits the practical application and promotion of such advanced algorithms in industrial sites, especially in distributed and large-scale node deployment scenarios.

[0005] It can be seen that the existing low-voltage load fault protection technology has the defects of hysteresis, insufficient reliability, difficulty in balancing high precision and low cost, and lack of fault identification capability, which limits the further development and improvement of low-voltage power distribution systems. SUMMARY

[0006] The application aims to provide a low-voltage load fault identification method, device, equipment and medium, which can realize rapid positioning and efficient management of line loss anomalies, improve the precision and economy of line loss management, and improve the operation safety and intelligent level of the power distribution system.

[0007] To achieve the above-mentioned purpose, the application adopts the following technical solutions: According to the first aspect of the application, a low-voltage load fault identification method is provided, comprising the following steps: Obtaining multi-dimensional data of a low-voltage load circuit, the multi-dimensional data including three-phase current data, three-phase voltage data and temperature data; Based on the multi-dimensional data, real-time characteristic quantities representing the load state are calculated, the characteristic quantities representing the load state including fundamental wave characteristic quantities, waveform characteristic quantities, frequency spectrum characteristic quantities and thermal characteristic quantities; Based on the historical characteristic quantities representing the load state of the low-voltage load circuit within a specified time length, an adaptive threshold is generated through load operation baseline self-learning; Comparing the real-time characteristic quantities representing the load state with the corresponding adaptive threshold, and performing multi-level logical judgment of time sequence correlation analysis based on the comparison result to determine the fault priority; Based on the fault priority, fault warning and fault protection are performed.

[0008] By using the above technical solution, the key characteristic quantities of the low-voltage load circuit are extracted, and comprehensive decision is made in combination with the adaptive threshold and multi-level logical judgment of time sequence correlation analysis. Through lightweight multi-dimensional feature fusion and hierarchical decision mechanism, early and accurate identification and hierarchical protection of multiple fault types are realized under the premise of ensuring low computational complexity, which is particularly suitable for resource-limited embedded micro control unit (MCU).

[0009] By using multi-feature quantity fusion analysis and judgment mechanism based on time sequence causal logic, overload, open phase, unbalance and various early potential faults can be effectively distinguished and located. The adaptive threshold mechanism can significantly reduce the false alarm and missed alarm rate under complex working conditions.

[0010] According to an embodiment of the application, the fundamental wave characteristic quantities include three-phase current effective value, three-phase voltage effective value and three-phase unbalance degree; The waveform characteristic quantities include current wave crest factor; The frequency spectrum characteristic quantities include total harmonic distortion rate and amplitude proportion of specific harmonic; The thermal characteristic quantities include current temperature difference and temperature rise rate.

[0011] Among them, the fundamental characteristic quantity can obtain the effective value of three-phase current, the effective value of three-phase voltage through the conventional calculation method (for example, the root mean square calculation method), and calculate the three-phase imbalance degree from the effective value of the current or voltage.

[0012] The current crest factor is the ratio of the instantaneous peak value of the waveform to its root mean square.

[0013] The spectral characteristic quantity is obtained by performing Fast Fourier Transform (FFT) on the three-phase current data.

[0014] The current temperature difference is the difference between the current temperature of the low-voltage load circuit and the ambient temperature. The temperature rise rate is the magnitude of the temperature rise per unit time of the low-voltage load circuit. In calculating the thermal characteristic quantity, the step of performing first-order low-pass filtering on the calculation results of the current temperature difference and the temperature rise rate to smooth the data fluctuations is also included.

[0015] According to an embodiment of the present application, based on the historical characteristic quantity representing the load state of the low-voltage load circuit within a specified time length, the adaptive threshold value step generated by the load operation baseline self-learning includes: Based on the historical characteristic quantity representing the load state of the low-voltage load circuit within a specified time length, an initial statistical baseline of each characteristic quantity is established, and the initial statistical baseline includes the mean and the standard deviation.

[0016] In subsequent operation, the initial statistical baseline is periodically updated using a sliding window and an exponential smoothing method, and the adaptive threshold value is obtained based on the updated baseline, and the adaptive threshold value includes a pre-warning threshold value and an alarm threshold value.

[0017] According to an embodiment of the present application, the step of periodically updating the initial statistical baseline using a sliding window and an exponential smoothing method, and obtaining an adaptive threshold value based on the updated baseline includes: Periodically set the parameters of the sliding window, and the parameters of the sliding window include the size of the sliding window and the update period; In each cycle, set a ring buffer, and the size of the ring buffer is the same as the size of the sliding window; continuously store the values of each characteristic quantity in the ring buffer, and update the data in the ring buffer in order from old to new; Update the mean and standard deviation of each characteristic quantity using the exponential smoothing method; Recalculate the pre-warning threshold value and the alarm threshold value based on the updated mean and standard deviation of each characteristic quantity.

[0018] By running the adaptive threshold value of the baseline, different load characteristics and long-term running drifts can be adapted, improper threshold value settings can be reduced to cause misoperation, and long-term reliability and applicability can be improved.

[0019] According to an embodiment of the present application, the fault priority comprises a first priority, a second priority and a third priority; The first priority is that the comparison result of the real-time characteristic quantity representing the load state and the corresponding adaptive threshold value meets the immediate action condition; the immediate action condition comprises the open-phase condition and / or the ultra-high temperature condition; The second priority is that the comparison result of the real-time characteristic quantity representing the load state and the corresponding adaptive threshold value meets the delayed action condition, and the delayed action condition comprises the overload condition and / or the serious imbalance condition; The third priority is that the comparison result of the real-time characteristic quantity representing the load state and the corresponding adaptive threshold value meets the early warning condition, and the early warning condition comprises the harmonic abnormal condition, the crest factor abnormal condition and / or the temperature rise rate abnormal condition.

[0020] According to an embodiment of the present application, the multi-level logic judgment based on the time sequence correlation analysis of the comparison result is used to determine the fault priority step, comprising: The fault priority is estimated based on the comparison result; If the estimated fault priority is the second priority or the third priority, the observation window is started to monitor the time sequence change of the associated characteristic quantity; the observation window can be configured for a period of time; Based on the time sequence change of the associated characteristic quantity, whether it is a typical feature event chain is identified by using a time sequence correlation analyzer according to a preset rule base; the preset rule base comprises an associated event constituting a typical feature event chain with the comparison result, and a time sequence change, a fault type and a confidence degree of the associated characteristic quantity corresponding to the associated event; The fault priority is determined according to the judgment result of the time sequence correlation analyzer.

[0021] Through the verification of the feature event chain, the accuracy of fault identification is further ensured, and the false alarm and omission rates under complex working conditions are further reduced.

[0022] According to the second aspect of the present application, a low-voltage load fault identification device is provided, comprising: A data acquisition module is configured to acquire multi-dimensional data of a low-voltage load circuit, and the multi-dimensional data comprises three-phase current data, three-phase voltage data and temperature data; A characteristic quantity calculation module is configured to calculate real-time characteristic quantities representing the load state based on the multi-dimensional data, and the characteristic quantities representing the load state comprise fundamental wave characteristic quantities, waveform characteristic quantities, frequency spectrum characteristic quantities and thermal characteristic quantities; The threshold determination module is configured to generate an adaptive threshold value through self-learning of a load operation baseline based on historical characteristic quantities of the load state of the low-voltage load circuit in a specified time length; The priority confirmation module is configured to compare the real-time characteristic quantities of the load state with the corresponding adaptive threshold value, perform multi-level logical judgment of time sequence correlation analysis based on a comparison result, and determine a fault priority; The fault early warning module is configured to perform fault early warning and fault protection based on the fault priority.

[0023] According to a third aspect of the present application, a computer device is provided, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the low-voltage load fault identification method of any of the above-mentioned embodiments when executing the computer program.

[0024] According to a fourth aspect of the present application, a computer readable storage medium is provided, which stores a computer program, and the computer program implements the low-voltage load fault identification method of any of the above-mentioned embodiments when executed by a processor.

[0025] Compared with the prior art, the present application has at least the following beneficial effects: 1. The present application adopts multi-feature quantity fusion analysis and multi-level judgment based on time sequence causal logic, which can effectively distinguish and locate overload, open-phase, unbalance, and various early potential faults. Through adaptive threshold value based on operation baseline, the system can adapt to different load characteristics and drift caused by long-term operation, reduce false actions caused by improper threshold setting, and improve the long-term reliability and applicability of the system.

[0026] 2. The present application introduces feature quantities sensitive to early faults (such as harmonics and temperature rise rate) and combines time sequence correlation analysis, which can capture the associated feature event chain in the early stage of faults, realize the leap from "passive protection" to "active early warning" and "precise diagnosis", and not only can early warning, but also can indicate possible fault types (such as bearing wear and insulation deterioration). The adaptive threshold value mechanism and the feature event chain verification double guarantee significantly reduce the false alarm and missed alarm rate under complex working conditions.

[0027] 3. The present application is based on classical signal processing and rule judgment, and is deeply embedded and optimized (such as fixed-point FFT and simplified sampling points), without complex machine learning models, small calculation amount, and very suitable for real-time operation in resource-limited MCUs, breaking the barrier between high-precision algorithms and low-cost hardware.

[0028] The low-voltage load multi-dimensional obstacle early identification and protection device, the electronic device, and the computer readable storage medium provided by the present application also solve the problems proposed in the background part. BRIEF DESCRIPTION OF DRAWINGS

[0029] The drawings constituting a part of this application provide further understanding of the application, the illustrative embodiments of the application and their descriptions serve to explain the application, and do not constitute improper limitations on the application. In the drawings: Figure 1 A flow chart of a low-voltage load fault identification method of an embodiment of the application; Figure 2 A structural schematic diagram of a low-voltage load fault identification device of an embodiment of the application; Figure 3 A structural schematic diagram of an electronic device of an embodiment of the application.

[0030] Reference signs: electronic device 100; memory 101; processor 102; computer program 103; communication bus 104. DETAILED DESCRIPTION

[0031] The application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments. It should be noted that the embodiments in the application and the features in the embodiments can be combined with each other without conflict.

[0032] The following detailed description is exemplary description, which aims to provide further detailed description of the application. Unless otherwise specified, all technical terms used in the application have the same meaning as that generally understood by the general technical personnel in the field to which the application belongs. The terms used in the application are only for the purpose of describing the specific embodiments, and are not intended to limit the exemplary embodiments according to the application.

[0033] Embodiment 1 A low-voltage load fault identification method for embedded terminals, as shown in Figure 1 , includes the following steps: Obtaining multi-dimensional data of a low-voltage load circuit, the multi-dimensional data including three-phase current data, three-phase voltage data and temperature data; Calculating real-time characteristic quantities representing load states based on the multi-dimensional data, the characteristic quantities representing load states including fundamental wave characteristic quantities, waveform characteristic quantities, frequency spectrum characteristic quantities and thermal characteristic quantities; Generating adaptive thresholds through load operation baseline self-learning based on historical characteristic quantities representing load states of the low-voltage load circuit within a specified time length; Comparing the real-time characteristic quantities representing load states with the corresponding adaptive thresholds, and performing multi-level logical judgment of time sequence correlation analysis based on the comparison results to determine fault priority; Performing fault warning and fault protection based on the fault priority.

[0034] Adopting the technical scheme, key characteristic quantities of a low-voltage load loop are extracted, and a multi-level logic judgment of adaptive threshold and time sequence correlation analysis is combined for comprehensive decision-making, and a multi-feature quantity fusion analysis and a judgment mechanism based on time sequence causality logic can effectively distinguish and locate overload, open-phase, unbalance and various early potential faults. The adaptive threshold mechanism can significantly reduce the false alarm and missed alarm rates under complex working conditions.

[0035] Through lightweight multi-dimensional feature fusion and hierarchical decision-making mechanism, early and accurate identification and hierarchical protection of various fault types are realized under the premise of low computational complexity, which is particularly suitable for resource-limited embedded MCUs. The resource occupation of the embedded MCU is extremely low, and the real-time performance is good. The above method is based on classical signal processing and rule judgment, and is deeply embedded and optimized. Without complex machine learning models, the calculation amount is small, and the real-time operation can be realized in the MCU with limited resources, breaking the barrier between high-precision algorithms and low-cost hardware.

[0036] Specifically, the steps of the low-voltage load fault identification method are as follows: S1. Obtain multi-dimensional data of the low-voltage load loop.

[0037] First, system initialization is performed.

[0038] The process of system initialization includes: configuring the clock, GPIO, ADC, timer, communication interface (such as UART, RS485) of the MCU; initializing variables, flags and buffer areas; reading preset threshold parameters (such as rated current I_rated, overload multiple, harmonic threshold, etc.) from non-volatile memory (such as EEPROM).

[0039] After completing the system initialization, data acquisition and preprocessing are performed, and the steps are as follows: (1) Make the timer interrupt, and the timing period is T sample (for example, corresponding to a sampling rate of 10 kHz, the period is 100 μs).

[0040] (2) Obtain multi-dimensional data of the low-voltage load loop through the ADC, and the multi-dimensional data includes three-phase current data, three-phase voltage data and temperature data. The three-phase current is represented by Ia, Ib and Ic, the single-phase voltage is represented by Ua, Ub and Uc, and the environmental temperature is represented by T. Specifically, at each timer interrupt, the ADC is started to sample the Ia, Ib, Ic, Ua, Ub, Uc and T channels of the low-voltage load loop in turn.

[0041] (3) Store the sampling values in a circular buffer area with a size of N (for example, N=200, corresponding to a period of 50 Hz signal).

[0042] (4) Software digital filtering (e.g. using a first-order low-pass filter) is performed on the sampled values to reduce noise interference.

[0043] S2. Calculate real-time characteristic quantities representing the load state based on multi-dimensional data.

[0044] The characteristic quantities representing the load state include fundamental characteristic quantities, waveform characteristic quantities, frequency spectrum characteristic quantities, and thermal characteristic quantities In the main loop, it is constantly checked whether the ring buffer is full for one cycle. Once it is full for one cycle, the following characteristic quantities are calculated in turn: Fundamental characteristic quantities, including three-phase current effective value, three-phase voltage effective value, and three-phase unbalance degree. Traverse the N points in the buffer, calculate the root mean square (RMS) value of the three-phase current, and then calculate the unbalance degree.

[0045] Waveform characteristic quantities, including current crest factor. Find the maximum value of the absolute value of the current instantaneous value in the cycle, and divide it by the corresponding RMS value.

[0046] Frequency spectrum characteristic quantities, including total harmonic distortion rate and amplitude ratio of specific harmonics, obtained through FFT operation. Perform 64-point or 128-point FFT operation on the N points in the one-phase current buffer. Use CMSIS-DSP library or similar optimized library to perform 64-point or 128-point fixed-point FFT operation to obtain harmonic data with extremely low resource overhead. Calculate the amplitude of the fundamental wave (50Hz) and specific harmonics, and calculate their percentage relative to the fundamental wave.

[0047] Specifically, when using 64-point FFT operation on one-phase current signal, sampling is performed at a fixed sampling rate of 3.2kHz, and when using 128-point FFT operation, sampling is performed at a fixed sampling rate of 6.4kHz, and the 64 or 128 sample data obtained is stored in the buffer. Use the CMSIS-DSP library to call its optimized Q-format fixed-point FFT function (such as arm_cfft_q15) to transform the buffer data. The CMSIS-DSP library uses a rotation factor table for calculation, which greatly reduces the operation overhead. Perform spherical operation on the attached results of the FFT output to obtain the amplitude of each frequency component. By finding the corresponding Pu County, the amplitude of the specific harmonic is accurately calculated. In this embodiment, the specific harmonics are the 2nd, 3rd, 5th, and 7th harmonics. Finally, the percentage of each harmonic relative to the fundamental wave is calculated. The entire algorithm process of sampling, FFT operation, and amplitude calculation can be easily completed within a 10ms time slice of an STM32L4 series MCU after optimization.

[0048] Thermal features, including current temperature difference and temperature rise rate. Read the ADC value of the temperature sensor and convert it to Celsius. Compare with the temperature value of the last cycle, calculate the temperature rise rate dT / dt, and first-order low-pass filter it to smooth the data fluctuations.

[0049] S3. Based on the historical features of the low-voltage load circuit representing the load state within a specified time period, generate adaptive thresholds through load operation baseline self-learning.

[0050] Based on the historical features of the low-voltage load circuit representing the load state within a specified time period, establish the initial statistical baseline of each feature, including the mean and standard deviation.

[0051] Further, the specified time period is at least 24 hours. The collected historical features of the low-voltage load circuit representing the load state within the specified time period are data under normal state operation, including fundamental feature, waveform feature, frequency spectrum feature, and thermal feature.

[0052] During the data collection process within the specified time period, the values of each feature are recorded at a fixed sampling interval. Then, the mean μ and standard deviation σ of each feature are calculated separately to form the initial baseline. For example, for the current effective value, the μ and σ of phase A, phase B, and phase C are calculated respectively; for harmonic features, the μ and σ of total harmonic distortion (THD), 2nd harmonic ratio, etc. are calculated, and these baseline values are used for adaptive threshold generation.

[0053] In subsequent operation, the initial statistical baseline is periodically updated using a sliding window and exponential smoothing method, and the adaptive threshold is obtained based on the updated baseline, including the pre-warning threshold and the alarm threshold.

[0054] The specific steps are as follows: Step 1: Set the sliding window parameters.

[0055] S(1-1). Define the size of the sliding window, such as the data of the last 24 hours (or a fixed number of data points, such as 8640 points, assuming sampling every 10ms). The window size should cover a complete load operation cycle (such as daily cycle).

[0056] S(1-2). Set the update period, for example, update the baseline every 1 hour (or every 360 data points).

[0057] Step 2: Data collection and buffering.

[0058] During the running, the feature value is continuously stored in a circular buffer, and the size of the buffer is equal to the size of the sliding window. When new data arrives, the oldest data is overwritten. That is, the value of each feature is continuously stored in the circular buffer during each cycle, and the size of the circular buffer is the same as the size of the sliding window; and the data in the circular buffer is updated in order from old to new.

[0059] Step 3: Update the mean μ and the standard deviation σ by the exponential smoothing method.

[0060] S(3-1). For each feature, the exponential smoothing method is used to update μ and σ instead of recalculating the statistics of the entire window to reduce the computational complexity. The exponential smoothing method is a recursive method suitable for real-time updating of embedded systems.

[0061] S(3-2). Update the mean μ: μ1=α×P + (1 -α)×μ0; Where μ1 is the updated mean, μ0 is the mean before updating; P is the sample mean of the current window; α is the smoothing factor (0<α<1), usually a small value (such as 0.05-0.1) is taken to ensure that the baseline slowly adapts to long-term changes. The sample mean of the current window can be the arithmetic mean of the last N data in the sliding window.

[0062] S(3-3). Update the standard deviation σ: Since the standard deviation calculation involves squaring, to simplify the calculation, the exponential smoothing method is usually used to update the variance (σ²), and then the square root is taken to get σ.

[0063] σ1²=β×P + (1 -β)×σ1² ; Where σ1² is the updated standard deviation, σ0² is the standard deviation before updating; P is the sample mean of the current window; β is another smoothing factor (may be the same as or different from α). The sample variance of the current window can be calculated by the sum of squares of deviations of the data in the sliding window.

[0064] Then, σ1= (σ1²) 1 / 2 ; Note: In embedded systems, to reduce the computational burden, approximate methods may be used, such as using the mean absolute deviation (MAD) instead of the standard deviation, or using a recursive formula to directly update σ.

[0065] Step 4: Threshold update.

[0066] According to the updated μ and σ, the warning threshold and the alarm threshold are recalculated.

[0067] Step 5: Periodic execution.

[0068] The above updating process is automatically triggered at each updating period (e.g. every hour) to ensure that the baseline always reflects the latest normal state of the load.

[0069] This method ensures that the threshold value can adapt to changes in the load while maintaining computational efficiency, suitable for running on resource-constrained MCUs.

[0070] In this embodiment, the early warning threshold is set to μ + 2σ, and the alarm threshold is set to μ + 3σ.

[0071] By running the adaptive threshold of the baseline, it can adapt to different load characteristics and drift caused by long-term operation, reduce false actions caused by improper threshold setting, and help improve long-term reliability and applicability.

[0072] S4. Compare the real-time feature quantity representing the load state with the corresponding adaptive threshold, and perform multi-level logical judgment based on the time sequence correlation analysis to determine the fault priority.

[0073] Compare the feature quantity calculated in S2 with the adaptive threshold generated in S3, and perform multi-level logical judgment of time sequence correlation analysis to determine the fault priority. The fault priority includes first priority, second priority and third priority.

[0074] The first priority corresponds to high priority faults, which is the comparison result of the real-time feature quantity representing the load state and the corresponding adaptive threshold meeting the immediate action condition; the immediate action condition includes the open-phase condition and / or the ultra-high temperature condition; if it is met, the circuit is immediately tripped. The open-phase condition is that the effective value of any phase current is lower than the minimum threshold value and the other phases are normal; the ultra-high temperature condition is that the absolute value of the temperature exceeds the absolute safety threshold value.

[0075] It should be noted that the minimum threshold value of the current effective value is a fixed, pre-set safety threshold value, which is set based on hardware protection or safety standards, used to detect current disappearance (such as wire breakage, etc.), and is not the adaptive threshold generated in step S3. The absolute safety threshold value of the temperature absolute value is also a fixed, pre-set safety threshold value, which is set based on the thermal limit of the element or safety regulations, used to prevent hardware damage, and is not the adaptive threshold generated in step S3. Based on the need, other feature quantities can also set corresponding minimum threshold values. The first priority is basically based on a fixed threshold value, used for the determination of high priority faults, to ensure fast response.

[0076] The second priority corresponds to a medium priority fault, and a comparison result of a real-time feature quantity representing a load state and a corresponding adaptive threshold value satisfies a delay action condition, the delay action condition including an overload condition and / or a serious imbalance condition; if the comparison result satisfies the delay action condition, a reverse time delay is started, and a trip is performed after the delay ends. The overload condition is that a current effective value exceeds a rated threshold value and lasts for a certain time, and the serious imbalance condition is that an unbalance degree exceeds a corresponding adaptive threshold value.

[0077] It should be noted that the rated threshold value of the current effective value is usually a fixed value based on the rated current of the load or the device specification, rather than the adaptive threshold value generated in step S3. The adaptive threshold value used to determine the serious imbalance condition is the alarm threshold value μ + 3σ.

[0078] The third priority corresponds to a low priority early warning, and a comparison result of a real-time feature quantity representing a load state and a corresponding adaptive threshold value satisfies an early warning condition, the early warning condition including a harmonic abnormal condition, a crest factor abnormal condition, and / or a temperature rise rate abnormal condition. The harmonic abnormal condition is that a specific harmonic proportion exceeds a corresponding adaptive threshold value, the crest factor abnormal condition is that a crest factor continuously exceeds a corresponding adaptive threshold value, and the temperature rise rate abnormal condition is that a temperature rise rate exceeds a corresponding adaptive threshold value.

[0079] It should be noted that the determination of the third priority is based on the adaptive threshold value, and each determination condition uses a corresponding early warning threshold value μ + 2σ for determination.

[0080] The multi-level logic determination based on the time sequence correlation analysis of the comparison result determines the fault priority step, including: Estimating the fault priority based on the comparison result; If the estimated fault priority is the second priority or the third priority, an observation window is started to monitor the time sequence change of the associated feature quantity; the observation window can be configured for a certain time (such as 30-60 seconds); Based on the time sequence change of the associated feature quantity, whether it is a typical feature event chain is identified by using a time sequence correlation analyzer according to a preset rule base; the preset rule base includes associated events constituting a typical feature event chain with the comparison result, time sequence changes of associated feature quantities corresponding to the associated events, fault types, and confidence levels; The fault priority is determined according to the judgment result of the time sequence correlation analyzer.

[0081] Through the verification of the feature event chain, the accuracy of fault identification is further ensured, which helps to further reduce the false positive and false negative rates under complex working conditions.

[0082] The time sequence correlation analyzer has a configurable rule base and an observation window, rather than relying on a few fixed rules. The specific parameters in the rule base, such as time window, correlation feature, and fault type, can be flexibly defined according to the fault mechanism of different loads. That is, when any low-priority fault warning is triggered, an observation window with a configurable duration is started, and the time sequence correlation analyzer is called. Based on the preset and configurable expert rule base, the analyzer monitors the time sequence changes of other correlation feature quantities to identify typical "feature event chains".

[0083] Specifically, the preset rule base is a rule set stored in the non-volatile memory of the embedded MCU, and each rule defines the logic of fault diagnosis. The rule base can be flexibly configured according to the fault mechanism of different load types (such as motors and transformers), including the following parameters: Rule ID: unique identifier; Trigger event: low-priority warning event (such as "5th harmonic exceeds") that starts the observation window; Correlation event: other events (such as "abnormal temperature rise rate") that need to be monitored within the observation window; Time window: time length (such as 30 seconds) from the trigger event, during which the correlation event must occur; Fault type: fault or warning type (such as "bearing wear warning") output when the condition is met; Confidence: rule confidence (such as 85%), used for subsequent decision weighting.

[0084] When a low-priority warning is triggered (estimated fault priority is second or third priority), the system dynamically creates an observation window (timer) and starts correlation event monitoring. The window duration is configurable (such as 30-60 seconds).

[0085] Within the observation window, it is continuously checked whether the correlation events defined in the rule base occur (i.e., whether other feature quantities exceed the threshold or meet certain conditions).

[0086] Specifically, in the trigger phase, when any low-priority warning (such as harmonic anomaly, peak factor anomaly, temperature rise rate anomaly) is triggered, the time sequence correlation analyzer is called. Based on the preset rule base, all rules with the warning as the trigger event are found, and an independent observation window (timer) is started for each rule.

[0087] Then enter the monitoring phase, within the observation window, the system monitors the correlation events defined in the rules. For example, for rule example 1 (bearing wear), when "5th harmonic exceeds" is triggered, it is monitored whether "abnormal temperature rise rate" is triggered within 30 seconds. Event monitoring is based on real-time calculation values of feature quantities compared with adaptive thresholds (as described in S3).

[0088] Then enters the decision stage. If all the associated events are triggered in the regular order (logical conditions in the rules, such as AND, OR) within the observation window, the rule condition is satisfied. The corresponding fault warning (such as “bearing wear warning”) is output with the confidence level. If multiple rules are satisfied simultaneously, the system can choose the output with the highest confidence level or make a fusion decision.

[0089] According to the decision result, a low-priority warning can be upgraded to a more specific fault warning, and the corresponding fault code is output. The warning information can be used for maintenance reminders, without immediate tripping.

[0090] An example of rule library configuration is shown in Table 1.

[0091] Table 1. Example of rule library configuration

[0092] With the preset rule library, the same set of diagnostic system can be applied to different loads, only different rule sets need to be loaded. For example, for motor loads, the rules can focus on bearing and insulation faults; for power supply loads, the rules can focus on harmonics and overheating. By correlating the time series changes of multiple feature quantities, early signs of faults can be captured, improving diagnostic accuracy. By encoding expert knowledge into configurable rules using the preset rule library, lightweight multi-dimensional feature fusion is achieved, thereby achieving early fault recognition with low computational complexity. Rule checking is based on simple logic and timing, without the need for complex models, suitable for embedded MCUs.

[0093] S5. Fault warning and fault protection based on fault priority.

[0094] According to the judgment result of S4 on fault priority, the corresponding fault code or warning information is output, and the corresponding operation is performed: for high-priority faults, a trip signal is immediately output; for medium-priority faults, a inverse time delay is started and a trip signal is output after the delay ends; for low-priority warnings or warnings upgraded via correlation analysis, the corresponding warning information is output and continuous monitoring is performed.

[0095] The low-voltage load fault recognition method of the embodiment can capture the associated feature event chain in the early stage of fault, realize the leap from “passive protection” to “active warning” and “precise diagnosis”, not only can warn, but also can indicate the possible fault type (such as bearing wear, insulation deterioration); and can effectively distinguish and locate overload, open-phase, unbalance and various early potential faults, so that the system can adapt to different load characteristics and the drift caused by long-term operation, reduce the misoperation caused by improper threshold setting, and improve the long-term reliability and applicability of the system.

[0096] Embodiment 2 As Figure 2 shown, based on the same inventive concept as the above embodiment, the present application also provides a low-voltage load fault identification device, comprising: a data acquisition module for acquiring multi-dimensional data of a low-voltage load circuit, the multi-dimensional data including three-phase current data, three-phase voltage data and temperature data; a feature quantity calculation module for calculating real-time feature quantities representing load states based on the multi-dimensional data, the feature quantities representing load states including fundamental wave feature quantities, waveform feature quantities, frequency spectrum feature quantities and thermal feature quantities; a threshold value determination module for generating adaptive threshold values through load operation baseline self-learning based on historical feature quantities representing load states of the low-voltage load circuit within a specified time length; a priority confirmation module for comparing the real-time feature quantities representing load states with corresponding adaptive threshold values, performing multi-level logic judgment of time sequence correlation analysis based on the comparison results, and determining fault priority; a fault warning module for performing fault warning and fault protection based on the fault priority.

[0097] Embodiment 3 As Figure 3 shown, the present application also provides an electronic device 100 for realizing analysis and search of power distribution line loss anomaly; The electronic device 100 comprises a memory 101, at least one processor 102, a computer program 103 stored in the memory 101 and executable on the at least one processor 102, and at least one communication bus 104.

[0098] The memory 101 can be used to store the computer program 103, and the processor 102 can realize the steps of the low-voltage load fault identification method of embodiment 1 by running or executing the computer program stored in the memory 101 and calling the data stored in the memory 101.

[0099] The memory 101 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required by a function (such as a sound playing function, an image playing function, etc.), etc.; and the data storage area can store data (such as audio data) created according to the use of the electronic device 100, etc. In addition, the memory 101 can include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device.

[0100] The at least one processor 102 can be a central processing unit (CPU), and can also be other general-purpose processors, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, etc. The processor 102 can be a microprocessor or the processor 102 can also be any conventional processor, etc. The processor 102 is a control center of the electronic device 100, and is connected to various parts of the electronic device 100 through various interfaces and lines.

[0101] The memory 101 in the electronic device 100 stores a plurality of instructions to implement a low-voltage load fault identification method, and the processor 102 can execute the plurality of instructions to implement: Obtaining multi-dimensional data of a low-voltage load circuit, the multi-dimensional data including three-phase current data, three-phase voltage data and temperature data; Calculating real-time characteristic quantities representing load states based on the multi-dimensional data, the characteristic quantities representing load states including fundamental wave characteristic quantities, waveform characteristic quantities, frequency spectrum characteristic quantities and thermal characteristic quantities; Generating an adaptive threshold through load operation baseline self-learning based on historical characteristic quantities representing load states of the low-voltage load circuit within a specified time length; Comparing the real-time characteristic quantities representing load states with corresponding adaptive thresholds, and performing multi-level logical judgment of time sequence correlation analysis based on a comparison result to determine a fault priority; Performing fault warning and fault protection based on the fault priority.

[0102] Embodiment 4 If the modules / units integrated in the electronic device 100 are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, and read-only memory (ROM).

[0103] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0104] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0105] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0106] These computer program instructions can also be loaded into a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flowcharts and / or blocks Figure 1 one or more flowcharts and / or blocks

[0107] In the description of the present specification, the description of the terms "one embodiment", "an example", "a specific example" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application but not to limit it, and although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that the specific embodiments of the present application can be modified or replaced by the equivalent, without departing from the spirit and scope of the present application, and any modification or equivalent replacement should be covered in the protection scope of the claims of the present application.

Claims

1. A low voltage load fault recognition method, characterized by, The method comprises the following steps: acquiring multi-dimensional data of the low-voltage load circuit, the multi-dimensional data comprising three-phase current data, three-phase voltage data and temperature data; calculating real-time characteristic quantities representing the load state based on the multi-dimensional data, the characteristic quantities representing the load state comprising fundamental wave characteristic quantities, waveform characteristic quantities, frequency spectrum characteristic quantities and thermal characteristic quantities; generating adaptive thresholds through load operation baseline self-learning based on historical characteristic quantities representing the load state of the low-voltage load circuit within a specified time length; comparing the real-time characteristic quantities representing the load state with the corresponding adaptive thresholds, performing multi-level logical judgment of time sequence correlation analysis based on the comparison results, and determining the fault priority; performing fault warning and fault protection based on the fault priority.

2. The low-voltage load fault identification method according to claim 1, wherein the fundamental wave characteristic quantities comprise three-phase current effective values, three-phase voltage effective values and three-phase unbalance degrees; the waveform characteristic quantities comprise current wave crest factors; the frequency spectrum characteristic quantities comprise total harmonic distortion rates and amplitude proportions of specific harmonics; the thermal characteristic quantities comprise current temperature differences and temperature rise rates.

3. The low-voltage load fault identification method according to claim 2, wherein the frequency spectrum characteristic quantities are obtained through fast Fourier transform.

4. The low-voltage load fault identification method according to claim 1, wherein the step of generating adaptive thresholds through load operation baseline self-learning based on historical characteristic quantities representing the load state of the low-voltage load circuit within a specified time length comprises: establishing initial statistical baselines of the characteristic quantities based on historical characteristic quantities representing the load state of the low-voltage load circuit within a specified time length, the initial statistical baselines comprising mean values and standard deviations; in subsequent operation, periodically updating the initial statistical baselines using a sliding window and an exponential smoothing method, and obtaining adaptive thresholds based on the updated baselines, the adaptive thresholds comprising warning thresholds and alarm thresholds.

5. The low-voltage load fault identification method according to claim 4, wherein the step of periodically updating the initial statistical baselines using a sliding window and an exponential smoothing method, and obtaining adaptive thresholds based on the updated baselines comprises: periodically setting parameters of the sliding window, the parameters of the sliding window comprising the size of the sliding window and the update period; in each period, setting a ring buffer with the same size as the sliding window; continuously storing the values of the characteristic quantities in the ring buffer, and sequentially updating the data in the ring buffer from old to new; updating the mean values and standard deviations of the characteristic quantities using the exponential smoothing method; recomputing the warning thresholds and alarm thresholds based on the updated mean values and standard deviations of the characteristic quantities.

6. The low-voltage load fault identification method according to claim 1, wherein the fault priority comprises a first priority, a second priority and a third priority; the first priority is that the comparison result of the real-time characteristic quantities representing the load state and the corresponding adaptive thresholds meets an immediate action condition; the immediate action condition comprises an open-phase condition and / or an ultra-high temperature condition. The second priority is that a comparison result of a real-time feature quantity representing a load state and a corresponding adaptive threshold value meets a delay action condition, and the delay action condition includes an overload condition and / or a serious imbalance condition. The third priority is that the comparison result of the real-time feature quantity representing the load state and the corresponding adaptive threshold value meets an early warning condition, and the early warning condition includes a harmonic abnormal condition, a crest factor abnormal condition and / or a temperature rise rate abnormal condition.

7. The low-voltage load fault identification method according to claim 1, characterized in that, The multi-level logical judgment based on the comparison result and the time sequence correlation analysis determines the fault priority step, including: estimating the fault priority based on the comparison result; if the estimated fault priority is the second priority or the third priority, starting an observation window to monitor the time sequence change of the associated feature quantity; the observation window can be configured for a time length; based on the time sequence change of the associated feature quantity, using a time sequence correlation analyzer to identify whether it is a typical feature event chain according to a preset rule base; the preset rule base includes an associated event composed of the comparison result, and a time sequence change of an associated feature quantity, a fault type and a confidence degree corresponding to the associated event; determining the fault priority according to the judgment result of the time sequence correlation analyzer.

8. A low voltage load fault recognition device, characterized by including: a data acquisition module configured to acquire multi-dimensional data of a low-voltage load circuit, the multi-dimensional data including three-phase current data, three-phase voltage data and temperature data; a feature quantity calculation module configured to calculate real-time feature quantities representing a load state based on the multi-dimensional data, the feature quantities representing the load state including fundamental wave feature quantities, waveform feature quantities, frequency spectrum feature quantities and thermal feature quantities; a threshold value determination module configured to generate adaptive threshold values through load operation baseline self-learning based on historical feature quantities representing the load state of the low-voltage load circuit within a specified time length; a priority confirmation module configured to compare the real-time feature quantities representing the load state with corresponding adaptive threshold values, and perform multi-level logical judgment based on the comparison result and time sequence correlation analysis to determine the fault priority; a fault warning module configured to perform fault warning and fault protection based on the fault priority.

9. An electronic device, comprising: including a processor and a memory, the processor being configured to execute a computer program stored in the memory to implement the low-voltage load fault identification method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores at least one instruction, and the at least one instruction is executed by the processor to implement the low-voltage load fault identification method according to any one of claims 1-7.

Images