A distribution box and a distribution box power monitoring system thereof

Abstract

The application relates to the technical field of power monitoring, in particular to a distribution box and a distribution box power monitoring system thereof. First, various data are acquired through a data acquisition module; further, a feature extraction module is used to identify the center time of a current load event and extract a voltage response waveform; according to the differences between the voltage response waveforms, a voltage response difference matrix is constructed and voltage response form dispersion is acquired; further, the cumulative vibration energy between the center times is acquired; according to the correlation between the cumulative vibration energy and the corresponding values in the voltage response difference matrix, a vibration-response correlation coefficient is acquired, and a vibration-response feature pair set is constructed; further, according to the distribution difference of the vibration-response feature pair set of adjacent periods, an electromechanical characteristic distribution offset is acquired; finally, a health state early warning and a sudden event early warning are carried out through a state early warning module, early fault modes are accurately identified, and clear fault mechanism diagnosis information is provided for maintenance personnel.

Description

Technical Field

[0001] This invention relates to the field of power monitoring technology, specifically to a distribution box and its power monitoring system. Background Technology

[0002] Mobile heavy-duty equipment, such as mining trucks and port gantry cranes, operates their power distribution systems in environments characterized by intense mechanical vibration and fluctuating load currents. Inside the distribution boxes of these devices, the bolted connections (cable joints) between the power cables and output terminal blocks experience fretting wear on the contact surfaces under continuous mechanical vibration; simultaneously, the Joule heating effect generated by high current flow accelerates oxidation of the contact surfaces.

[0003] In the early stages of degradation, the signal characteristics are extremely weak. Drastic fluctuations in load current create a strong noise background, completely drowning out the voltage drop and temperature rise signals generated by minute increases in contact resistance, making accurate early warning of early degradation difficult. Existing monitoring methods using infrared thermal imaging can only detect anomalies in the later stages of degradation, when contact resistance increases significantly and a noticeable temperature rise occurs. These methods lack early warning capabilities and cannot predict remaining service life. Offline contact resistance measurement methods require shutting down and powering off the equipment, making continuous monitoring impossible. Summary of the Invention

[0004] To address the technical problem of accurately predicting early degradation of cable joints in distribution boxes for mobile heavy-duty equipment, the present invention aims to provide a distribution box and its power monitoring system. The specific technical solution adopted is as follows:

[0005] A distribution box and a power monitoring system for the distribution box, the system comprising:

[0006] The data acquisition module is used to acquire current sequence, voltage sequence and normal acceleration sequence at preset periods;

[0007] The feature extraction module is used to identify the center moment of the current load event and extract the voltage response waveform within the current cycle based on the fluctuation characteristics of the current sequence; construct a voltage response difference matrix and obtain the voltage response morphology dispersion based on the differences between the voltage response waveforms; obtain the cumulative vibration energy based on the data distribution within the normal acceleration sequence between the center moments; obtain the vibration-response correlation coefficient based on the correlation between the cumulative vibration energy of the binary pairs of all the center moments and the corresponding values ​​in the voltage response difference matrix, and construct a set of vibration-response feature pairs; and obtain the electromechanical characteristic distribution offset based on the distribution differences of the vibration-response feature pairs set in adjacent cycles.

[0008] The status early warning module is used to provide health status early warning based on the distribution of the voltage response morphology dispersion and the vibration-response correlation coefficient in the current period; and to determine whether to provide emergency event early warning based on the distribution of the electromechanical characteristic distribution offset.

[0009] Furthermore, the method for issuing health status early warnings includes:

[0010] For the current cycle, when the voltage response pattern dispersion is less than a preset degradation threshold and the vibration-response correlation coefficient is less than a preset mechanical sensitivity threshold, the cable joint is determined to be in a healthy state.

[0011] When the voltage response pattern dispersion is greater than or equal to a preset degradation threshold, and the vibration-response correlation coefficient is less than a preset mechanical sensitivity threshold, the cable joint is determined to be in an electrical degradation state, and a first-level warning is issued.

[0012] When the voltage response pattern dispersion is less than a preset deterioration threshold and the vibration-response correlation coefficient is greater than or equal to a preset mechanical sensitivity threshold, the cable joint is determined to be mechanically loose, and a secondary warning is issued.

[0013] When the voltage response morphology dispersion is greater than or equal to a preset degradation threshold, and the vibration-response correlation coefficient is greater than or equal to a preset mechanical sensitivity threshold, the cable joint is determined to be in a composite failure state, and a level three warning is issued.

[0014] Furthermore, after determining whether to issue an emergency warning, the process also includes predicting the remaining service life of the cable joint based on the trend of the voltage response pattern dispersion over time.

[0015] When the total number of current load events in a period is greater than or equal to the minimum current load event threshold, it is marked as an effective period; within the current historical sliding window, when the cable joint status of the effective period is electrically deteriorated for a number of consecutive preset duration periods, and the voltage response pattern dispersion shows a monotonically increasing trend, a linear model of the voltage response pattern dispersion over time is fitted to the historical sliding window.

[0016] The remaining time when the voltage response morphology dispersion reaches a preset critical failure threshold is obtained based on the linear model, and this time is taken as the remaining service life.

[0017] Furthermore, the method for issuing early warnings for emergencies includes:

[0018] When the offset of the electromechanical characteristic distribution is greater than or equal to the preset state transition threshold, an emergency warning is issued.

[0019] Furthermore, the method for obtaining the electromechanical characteristic distribution offset includes:

[0020] The bulldozer distance algorithm is used to calculate the bulldozer distance between the set of vibration-response characteristic pairs in the current period and the adjacent previous period, which is used as the electromechanical characteristic distribution offset.

[0021] Furthermore, the method for obtaining the vibration-response correlation coefficient includes:

[0022] Based on the Pearson correlation coefficient analysis, the correlation between the cumulative vibration energy of the binary tuples at all the central moments and the corresponding values ​​in the voltage response difference matrix is ​​obtained to obtain the vibration-response correlation coefficient.

[0023] Furthermore, the method for obtaining the accumulated vibrational energy includes:

[0024] The squared values ​​of the data in the normal acceleration sequence between the center times of any two current load events are integrated, and the integrated value is used as the cumulative vibration energy between the corresponding two current load events.

[0025] Furthermore, the method for obtaining the voltage response difference matrix includes:

[0026] The DTW distance values ​​between different voltage response waveforms within the current period are used to construct a voltage response difference matrix.

[0027] Furthermore, the method for obtaining the voltage response morphology dispersion includes:

[0028] The voltage response morphological dispersion is obtained based on the overall characteristics of all element values ​​within the voltage response difference matrix.

[0029] Furthermore, the method for obtaining the center time includes:

[0030] The current change rate sequence of the current sequence of the current period is obtained and the maximum value of the absolute value of the current change rate is extracted as the event candidate point. When the absolute value of the current change rate of the event candidate point is greater than the preset current change threshold, it is determined to be a current load event, and the time corresponding to the event candidate point is marked as the center time.

[0031] The present invention has the following beneficial effects:

[0032] First, various data are acquired through the data acquisition module to provide a foundation for subsequent data analysis. Then, the feature extraction module identifies the center moment of current load events, providing independent observation samples and extracting voltage response waveforms as carriers of joint health information, providing data sample support for subsequent statistical analysis. Next, based on the differences between voltage response waveforms, a voltage response difference matrix is ​​constructed to comprehensively demonstrate the inconsistencies among all responses within a cycle, and the voltage response morphology dispersion is obtained to characterize the instability of the joint's electrical response, providing a basis for accurate condition warnings. Furthermore, the cumulative vibration energy is acquired to characterize the total mechanical impact intensity between two current load events. Based on the correlation between the cumulative vibration energy of the binary pairs at all center moments and the corresponding values ​​in the voltage response difference matrix, the vibration-response correlation coefficient is obtained to characterize the cable joint's sensitivity to mechanical vibration, and a set of vibration-response feature pairs is constructed to comprehensively demonstrate the electromechanical coupling relationship. Finally, based on the distribution differences of the vibration-response feature pairs set between adjacent cycles, the electromechanical characteristic distribution offset is obtained to capture and quantify sudden changes in degradation modes, providing more evidence for accurate warnings. Finally, the condition warning module provides health status warnings and emergency event warnings. This invention transforms the difficult-to-measure contact resistance increment into observable dynamic response behavior characteristics by synchronously acquiring electromechanical signals. It effectively solves the problem of difficulty in online monitoring of early deterioration of cable joints under strong noise, and realizes early warning of deterioration state, fault mechanism differentiation and sudden fault capture. Attached Figure Description

[0033] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 A system block diagram of a distribution box and its power monitoring system provided in one embodiment of the present invention;

[0035] Figure 2 This is a flowchart for predicting remaining useful life, provided as an embodiment of the present invention. Detailed Implementation

[0036] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a distribution box and its power monitoring system according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0037] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0038] The following description, in conjunction with the accompanying drawings, details a specific scheme for a distribution box and its power monitoring system provided by the present invention.

[0039] Please see Figure 1 The diagram illustrates a system block diagram of a distribution box and its power monitoring system according to an embodiment of the present invention. The system includes: a data acquisition module 101, a feature extraction module 102, and a status early warning module 103.

[0040] The data acquisition module 101 is used to acquire current sequence, voltage sequence and normal acceleration sequence at a preset period.

[0041] Considering the differences in mobile heavy-duty equipment models and their specific tasks, using fixed threshold parameters would limit the applicability of the solution. In one embodiment of the present invention, upon initial operation or receipt of a reset command, a self-tuning phase for operating parameters is first initiated to automatically learn and determine the key thresholds required for subsequent monitoring, thereby achieving adaptation to the specific operating characteristics of the equipment.

[0042] As an example, the self-tuning phase of the operating parameters includes: setting the self-tuning time (e.g., 300 seconds), and the system continuously and synchronously acquiring the normal acceleration sequence installed on the base of the output terminal block of the distribution box. and the instantaneous current sequence flowing through the terminal. .

[0043] Real-time system calculation The first derivative is used to obtain the rate of change of current, and the local maxima of the absolute values ​​of all current rates of change during this period are recorded; simultaneously, the absolute values ​​of all acceleration sequences are recorded. The local maxima.

[0044] After the self-tuning mode ends (after the self-tuning duration is reached), the system performs statistical analysis on the two sets of local maxima recorded. The absolute value of the current change rate at the 95th percentile of the absolute value of the local maxima of all current change rates is used as the current change threshold. The absolute value of the acceleration at the 99th percentile of the local maxima of the absolute values ​​of all acceleration sequences is used as the threshold for mechanical impact events. These two thresholds are stored in non-volatile memory for use in subsequent long-term monitoring steps.

[0045] It should be noted that the data sampling frequency is 5kHz to 20kHz. In one embodiment of the present invention, it is set to 10kHz. The method for determining the percentile and the method for obtaining the first derivative are well known technologies. In other embodiments of the present invention, the implementer can adjust the selection of the percentile and the data acquisition frequency, which will not be described in detail here.

[0046] In one embodiment of the present invention, the length of a period is set to 1 second, and the period index is k. The continuously and synchronously collected data is transformed into discrete current, voltage, and normal acceleration sequences with unique indices, providing a foundation for subsequent data analysis.

[0047] It should be noted that, in one embodiment of the present invention, the data analysis frequency is set to once every 1 to 10 cycles, preferably once after each cycle; in other embodiments of the present invention, the implementer can adjust the cycle length and the data analysis frequency according to actual needs; the self-tuning stage of the operating parameters is performed when the cable joint is confirmed to be in a healthy state, for example, during the new installation and commissioning stage of the equipment or after the shutdown maintenance is completed and the joint is confirmed to be tight, so as to ensure that the various thresholds learned can distinguish the normal operating state of the equipment.

[0048] The feature extraction module 102 is used to identify the center moment of the current load event and extract the voltage response waveform within the current cycle based on the fluctuation characteristics of the current sequence; construct a voltage response difference matrix and obtain the voltage response morphology dispersion based on the differences between the voltage response waveforms; obtain the cumulative vibration energy based on the data distribution within the normal acceleration sequence between center moments; obtain the vibration-response correlation coefficient based on the correlation between the cumulative vibration energy of the binary pairs of all center moments and the corresponding values ​​in the voltage response difference matrix, and construct a set of vibration-response feature pairs; and obtain the electromechanical characteristic distribution offset based on the distribution differences of the vibration-response feature pairs set in adjacent cycles.

[0049] Considering that under continuous vibration and Joule heating, the microstructure of the internal conductive contact surface of a cable joint becomes uneven and unstable due to oxidation and wear. This causes the voltage response waveform to change from highly uniform in a healthy state to random and discrete in a deteriorated state when a large current pulse with similar morphology flows through it. Therefore, by quantitatively assessing the morphological differences between multiple voltage response waveforms over a short period, the microscopic instability of the joint contact state can be directly reflected, thereby quantifying the degree of deterioration.

[0050] Considering that the deterioration of cable joints is reflected in their dynamic response, the load changes are analyzed based on the fluctuation characteristics of the current sequence within the current cycle. The center moment of the current load event is identified, and the continuous current fluctuations are discretized into independent events, providing independent observation samples. The voltage response waveform is extracted as a carrier of joint health information, providing data sample support for subsequent statistical analysis.

[0051] The analysis process for each cycle after the self-tuning stage of the operating parameters is the same. Here, we will only use the latest current cycle in the time domain as an example for description, and will not repeat the explanation.

[0052] In a preferred embodiment of the present invention, considering that a current load event corresponds to a significant change in current, the first derivative of the current sequence of the current period is calculated to obtain the current change rate sequence of the current sequence of the current period, and the maximum value of the absolute value of the current change rate is extracted as the event candidate point to identify the data point with the most drastic local change in current, and the fluctuation characteristics of the current sequence are expressed by means of the change rate.

[0053] Not all current fluctuations correspond to load events of analytical value. Therefore, when the absolute value of the current change rate at a candidate event point is greater than the preset current change threshold, it is determined to be a current load event, and the time corresponding to the candidate event point is marked as the center time, effectively distinguishing significant load events of analytical value.

[0054] Using each center time as the center, a waveform slice is extracted from the voltage sequence. For example, 10 milliseconds are extracted on both sides of the center time (immediately adjacent). The extracted slices, together with the voltage waveform corresponding to the voltage data at the center time, are used as the voltage response waveform.

[0055] In calculating the first derivative, the central difference method (a well-known technique) is used to obtain the rate of change of current at each time point within the current cycle;

[0056] To eliminate spurious events caused by high-frequency noise, the system merges multiple adjacent event candidate points with time intervals less than a preset de-jitter duration (e.g., 5 milliseconds) into a single event. Specifically, the merged event candidate points are connected, and 10 milliseconds are truncated to the left (nearest) of the center moment with the shortest timing and 10 milliseconds to the right (nearest) of the center moment with the longest timing. The voltage waveform corresponding to the voltage data between the truncated slices and all connected event candidate points is taken as the voltage response waveform. The moment corresponding to the point with the largest absolute value of current change rate is taken as the center moment of the current load event. When there are multiple points with the largest absolute value of current change rate, the point with the shortest timing is taken. The timing increases to the right (to the right is the positive direction of the time domain) and decreases to the left.

[0057] In other embodiments of the present invention, the implementer may adjust the length of the preset de-jitter duration and the length of the voltage response waveform. The duration of the voltage response waveform should be able to completely cover the voltage transient response process caused by a typical current step. The specific value can be set by analyzing the waveform data collected during the self-tuning stage, observing and determining the typical duration of the transient process (for example, using the half-width of the voltage peak at the center moment as the response duration, taking the 95th percentile). This will not be elaborated further.

[0058] In one embodiment of the present invention, considering that when there are too few current load events in a cycle, it indicates that there are insufficient effective load events under the current cycle's operating conditions, and a reliable stability assessment cannot be performed, it is necessary to determine whether the current load events are effective after identifying them.

[0059] As an example, setting a minimum current load event threshold , If the total number of current load events in the current cycle is less than If the current load event in the current cycle is greater than or equal to a certain value, it is considered an invalid cycle, and subsequent analysis processes will not be executed. Furthermore, any feature values ​​acquired within the current cycle will be marked as invalid. If so, continue the analysis and mark it as a valid period.

[0060] Considering that the differences between voltage response waveforms reflect the impact of cable joint degradation on voltage response, a voltage response difference matrix is ​​constructed to comprehensively show the inconsistency among all responses within the period, obtain the voltage response morphology dispersion, characterize the instability of the joint's electrical response, and provide a basis for accurate condition warning in the future.

[0061] Preferably, in one embodiment of the present invention, considering that the variable frequency drive system of mobile heavy-duty equipment may cause slight frequency drift or phase jitter in load events, the Dynamic Time Warping (DTW) algorithm can effectively measure the morphological similarity of waveforms by nonlinearly distorting the time axis to find the optimal matching path, and is not sensitive to slight distortions on the time axis.

[0062] Therefore, we pair different voltage response waveforms within the current period, resulting in waveform pairs. ,in k is the periodic index number. This represents the number of current load events within the k-th cycle, and also the number of voltage response waveforms. This represents the voltage response waveform of the m-th period within the k-th period. This represents the voltage response waveform within the k-th cycle;

[0063] Obtain the DTW distance values ​​between different voltage response waveforms within the current period, and construct a voltage response difference matrix with m as the horizontal axis and n as the vertical axis.

[0064] The larger the DTW distance value, the greater the difference between the two corresponding voltage response waveforms; since the DTW distance is symmetrical, the DTW distance between itself is 0, so only the DTW distance between itself and itself is calculated. The DTW distance value of the waveform pair under the given conditions, corresponding to the upper triangular part, can quantify the differences between all different voltage response waveforms. The diagonal position and the symmetrical triangular part (lower triangular part) in the matrix can be ignored.

[0065] Considering that the overall characteristics of all element values ​​in the voltage response difference matrix reflect the inconsistency among all responses, the voltage response morphology dispersion is obtained based on the overall characteristics of all element values ​​in the voltage response difference matrix.

[0066] As an example, the average value is used to represent the overall characteristics of the data, and the average value of all elements in the voltage response difference matrix is ​​used as the voltage response morphological dispersion.

[0067] In other embodiments of the present invention, the implementer may also fill the entire voltage response difference matrix with symmetry (symmetry about the diagonals with equal horizontal and vertical coordinates), but when calculating the voltage response morphological dispersion and subsequent analysis, only the upper or lower triangular part is taken.

[0068] Since mechanical vibration is another core stress that leads to joint deterioration, and mechanical vibration is directly reflected in the normal acceleration data, the cumulative vibration energy is obtained by analyzing the data distribution in the normal acceleration sequence between the center moments. This characterizes the total mechanical impact intensity between the center moments of the two current load events, providing a basis for subsequent analysis of the correlation between vibration and voltage response.

[0069] Considering that the decrease in bolt tightening force or fretting wear on the contact surface is a key mechanism in the joint deterioration process, this leads to an increased sensitivity of the joint to external mechanical shocks. Therefore, based on the correlation between the cumulative vibration energy of the binary tuples at all center moments and the corresponding values ​​in the voltage response difference matrix, the vibration-response correlation coefficient is obtained to characterize the sensitivity of the cable joint to mechanical vibration, and a set of vibration-response characteristic pairs is constructed to comprehensively demonstrate the electromechanical coupling relationship.

[0070] Preferably, in one embodiment of the present invention, considering that the overall intensity of the acceleration signal reflects the total intensity of the mechanical impact, the squared values ​​of the data in the normal acceleration sequence between the center moments of any two current load events are integrated, and the integrated value is used as the cumulative vibration energy between the corresponding two current load events.

[0071] By iterating through all the central time pairs, the cumulative vibrational energy between all two current load events is obtained.

[0072] Considering a healthy, tight joint, its electrical response characteristics should not be affected by normal vibration; however, for a deteriorated, loose joint, a single significant vibration may cause instantaneous reconstruction of its internal conductive spots, thereby changing its subsequent electrical response characteristics. The Pearson correlation coefficient can measure the degree of linear correlation between two variables. Therefore, based on the Pearson correlation coefficient, we analyze the correlation between the cumulative vibration energy of the binary pair at all center moments and the corresponding values ​​in the voltage response difference matrix to obtain the vibration-response correlation coefficient.

[0073] Specifically, since the cumulative vibration energy and the corresponding value (DTW distance value) in the voltage response difference matrix corresponding to the binary pair at the center time are determined, and the Pearson correlation coefficient calculation only depends on the overall distribution of the data values ​​and does not depend on the order of the data points in the sequence, the binary pairs at the center time are arranged in any order. Based on the order of the binary pairs at the center time, a sequence of cumulative vibration energy is constructed as the vibration sequence, and a sequence of DTW distance values ​​is constructed as the response sequence. The absolute value of the Pearson correlation coefficient between the vibration sequence and the response sequence is used as the vibration-response correlation coefficient of the current period.

[0074] The binary pairs corresponding to the center time are sorted according to the order of the voltage response waveform pairs analyzed when constructing the voltage response difference matrix. Regardless of whether the correlation is positive or negative, as long as its absolute value is high, it indicates that there is a strong systematic dependence between vibration and electrical response, which is the sign of mechanical loosening fault. Therefore, the absolute value of Pearson correlation coefficient is taken.

[0075] Furthermore, the cumulative vibration energy and the corresponding values ​​(DTW distance values) in the voltage response difference matrix corresponding to the binary pair at the center time are paired as vibration-response feature pairs, forming a set of vibration-response feature pairs.

[0076] It should be noted that the DTW algorithm and Pearson correlation coefficient are both well-known technologies and will not be elaborated further.

[0077] Considering that joint degradation is not always gradual, but may be abrupt or phased due to a strong impact or internal damage, such abrupt changes will be reflected in the alteration of its overall "electromechanical coupling behavior mode". Therefore, by obtaining the distribution offset of electromechanical characteristics based on the distribution differences of vibration-response characteristics of adjacent cycles, it is possible to capture and quantify the sudden changes in degradation mode, providing more basis for accurate early warning in the future.

[0078] Preferably, in one embodiment of the present invention, considering that the Earth Mover's Distance (EMD) algorithm can comprehensively and robustly quantify the overall morphological differences between two multidimensional empirical distributions, the vibration-response feature pairs of the current period and the adjacent previous (effective) period are regarded as empirical two-dimensional probability distributions describing electromechanical coupling characteristics. The bulldozer distance between the vibration-response feature pairs of the current period and the adjacent previous (effective) period is calculated using the bulldozer distance algorithm as the electromechanical characteristic distribution offset.

[0079] If the current period is the first valid period after the system starts up (including restarts after a reset command), there are no comparable historical states. The electromechanical characteristic distribution offset is set to 0, and the vibration-response characteristics of the current period are overwritten and updated. This provides a historical benchmark for the calculation and analysis of the next effective cycle.

[0080] It should be noted that, in one embodiment of the present invention, the maximum mean difference (MMD) algorithm can be used to calculate the offset of the electrical characteristic distribution. The method of measuring the difference between two (set) distributions, which is the same as that of the bulldozer distance algorithm, is a well-known technical means and will not be described in detail here.

[0081] The status warning module 103 is used to provide health status warnings based on the distribution of voltage response morphology dispersion and vibration-response correlation coefficient in the current cycle; and to determine whether to provide emergency warnings based on the distribution of electromechanical characteristic distribution offset.

[0082] Considering that the current cycle voltage response morphology dispersion reflects the microscopic instability and severity of degradation of the joint's conductivity state, and that the vibration-response correlation coefficient reflects whether the instability is dominated by mechanical vibration, combining the distributions of both can achieve accurate identification of early fault modes, thereby providing accurate health status warnings and clear fault mechanism diagnostic information for maintenance personnel. Considering that the distribution of electromechanical characteristic offset reflects the possibility of sudden changes in degradation modes, it is also necessary to determine whether to issue a warning for sudden events. This can capture the step-like degradation of joint performance caused by strong external impacts or internal damage, which cannot be predicted by the progressive model, providing the system with a rapid response capability to sudden faults and a complete early warning system.

[0083] Preferably, in one embodiment of the present invention, considering that the greater the dispersion of the voltage response pattern, the more it conforms to the random dispersion under the deterioration state, indicating that the cable joint is more likely to experience electrical deterioration, and when the vibration-response correlation coefficient is greater, the cable joint is more sensitive to mechanical vibration, indicating that the cable joint is more likely to become loose, therefore, by combining the distribution of the two, a two-dimensional diagnostic space of "deterioration severity - mechanical sensitivity" can be constructed to clearly distinguish the states of "health", "pure electrical deterioration", "mechanical loosening warning" and "composite failure";

[0084] Based on this, for the current cycle, when the voltage response pattern dispersion is less than a preset degradation threshold and the vibration-response correlation coefficient is less than a preset mechanical sensitivity threshold, the cable joint is considered healthy. This condition indicates that the joint has stable electrical contact and is mechanically secure.

[0085] When the voltage response morphology dispersion is greater than or equal to a preset degradation threshold, and the vibration-response correlation coefficient is less than a preset mechanical sensitivity threshold, the cable joint is determined to be in an electrical degradation state, and a Level 1 warning is issued. This state indicates that the joint may be undergoing uniform oxidation or wear, but the mechanical structure is intact.

[0086] When the voltage response morphology dispersion is less than a preset degradation threshold, and the vibration-response correlation coefficient is greater than or equal to a preset mechanical sensitivity threshold, the cable joint is determined to be mechanically loose, and a level-two warning is issued. This state is an early warning, indicating that the joint has become mechanically loose and sensitive to vibration, but has not yet developed into significant electrical instability.

[0087] When the voltage response morphology dispersion is greater than or equal to a preset degradation threshold, and the vibration-response correlation coefficient is greater than or equal to a preset mechanical sensitivity threshold, the cable joint is determined to be in a combined failure state, triggering a level three warning. This state indicates that the joint simultaneously suffers from severe electrical instability and mechanical loosening, resulting in a high risk of failure.

[0088] Furthermore, while performing two-dimensional state diagnosis, the electromechanical characteristic distribution offset is independently judged. When the electromechanical characteristic distribution offset is greater than or equal to the preset state transition threshold, it indicates that the joint performance has deteriorated in a step manner, and an emergency warning is issued, triggering a second-level warning. The joint state changes before and after the state transition are recorded to provide clues for fault root cause analysis.

[0089] It should be noted that the preset degradation level threshold, preset mechanical sensitivity threshold, and preset state transition threshold are obtained during the self-tuning phase of the operating parameters. As an example, the system statistically analyzes the voltage response morphology dispersion, vibration-response correlation coefficient, and electromechanical characteristic distribution offset recorded during the self-tuning phase of the operating parameters. The voltage response morphology dispersion at the 95th percentile of all voltage response morphology dispersions is used as the preset degradation level threshold; the vibration-response correlation coefficient at the 99th percentile of all vibration-response correlation coefficients is used as the preset mechanical sensitivity threshold; and the electromechanical characteristic distribution offset at the 99th percentile of all electromechanical characteristic distribution offsets is used as the preset state transition threshold.

[0090] It should be noted that no warning is issued when the status is healthy. The specific forms of the first, second and third level warnings can be flashing status lights of different colors or different symbols, which can be set by the implementer. In other embodiments of the present invention, the implementer can adjust the selection of percentiles, which will not be elaborated further.

[0091] It should be noted that, in one embodiment of the present invention, in order to provide quantitative decision-making basis for predictive maintenance, after determining whether to issue an emergency warning, the remaining service life of the cable joint is predicted based on the trend of voltage response morphology dispersion over time. Please refer to [link to relevant documentation]. Figure 2 It illustrates a flowchart of predicting remaining useful life provided by an embodiment of the present invention, specifically including:

[0092] Step S301: Within the current historical sliding time window, when the cable joint status is electrically degraded for a number of consecutive preset duration periods and the voltage response morphology dispersion shows a monotonically increasing trend, fit a linear model of the voltage response morphology dispersion of the historical sliding time window over time.

[0093] In order to accurately predict the remaining service life, it is necessary to analyze the trend of state changes in the historical effective period. When the joint condition is continuously stable in "electrical degradation" and the voltage response morphology dispersion shows a monotonically increasing trend, it indicates that the joint degradation process is slow, uniform and continuously deteriorating, which meets the basic premise of linear extrapolation prediction. At this time, the system starts baseline prediction and uses linear regression to fit the linear model of the voltage response morphology dispersion over time in the historical sliding window to obtain the average degradation rate of the joint.

[0094] As an example, the historical sliding window is used to store the most recent W historical valid cycles, W=1000, and the preset number of duration cycles is set to 100. Among the valid cycles in which the joint condition is electrically deteriorated, valid cycles with more than 100 duration cycles are selected. When the voltage response morphology dispersion shows a monotonically increasing trend through the Mann-Kendall test, the linear regression method is used to fit the voltage response morphology dispersion of the historical sliding window and the corresponding timestamp.

[0095] The resulting linear model is: In the formula, the slope 'a' represents the average deterioration rate of the joint. For time, The intercept is... This is the predicted value of the voltage response morphology dispersion at time t;

[0096] The voltage response morphology dispersion corresponds to the timestamp of the first sampling point of its respective period. A valid period is defined as the total number of current load events within a period that is greater than or equal to the minimum current load event threshold; this determination is already performed in feature extraction module 102 and will not be repeated here. When multiple valid periods meet the fitting requirements, the latest one is selected for fitting. The Mann-Kendall test for data monotonicity is a well-known technique and will not be elaborated further.

[0097] Step S302: Based on the linear model, obtain the remaining time when the voltage response morphology dispersion reaches the preset critical failure threshold, and use this as the remaining service life.

[0098] As an example, the preset critical failure threshold is twice the preset degradation threshold;

[0099] Using the preset critical failure threshold as Substituting into the linear model, time t is the point in time at which the voltage response morphology dispersion reaches the preset critical failure threshold. ; to the time point of prediction The time interval (remaining time) is taken as the remaining useful life.

[0100] It should be noted that the linear model is only applicable to describing slow, uniform degradation processes. When the system detects a change in health status to "mechanical loosening" or "combined failure," or detects a triggering of a sudden event warning, it indicates that the degradation mechanism of the joint has fundamentally changed, and the original degradation rate is no longer applicable.

[0101] At this point, the system immediately performs a prediction reset operation, discarding the current linear model, clearing the data in the historical sliding window, and starting to accumulate data again from the next effective period to learn new, potentially steeper, degradation rates.

[0102] In other embodiments of the present invention, the implementer may adjust the length of the historical sliding window, the number of preset duration periods, and the preset critical failure threshold.

[0103] In summary, to address the existing technical problem of accurately predicting the early degradation of cable joints in distribution boxes of mobile heavy-duty equipment, this invention proposes a distribution box and its power monitoring system. First, various data are acquired through a data acquisition module. Then, a feature extraction module identifies the center moment of current load events and extracts the voltage response waveform. Based on the differences between voltage response waveforms, a voltage response difference matrix is ​​constructed, and the voltage response morphology dispersion is obtained. Next, the cumulative vibration energy between center moments is acquired. Based on the correlation between the cumulative vibration energy of the binary pairs at all center moments and the corresponding values ​​in the voltage response difference matrix, a vibration-response correlation coefficient is obtained, and a set of vibration-response feature pairs is constructed. Further, based on the distribution differences of the vibration-response feature pairs set in adjacent periods, the electromechanical characteristic distribution offset is obtained. Finally, a status warning module provides health status warnings and emergency event warnings. This invention, by synchronously acquiring electromechanical signals, transforms the difficult-to-measure contact resistance increment into observable dynamic response behavior characteristics, effectively solving the problem of difficult online monitoring of early degradation of cable joints under strong noise, and realizing early warning of degradation status, fault mechanism differentiation, and sudden fault capture.

[0104] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0105] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A distribution box and a distribution box power monitoring system thereof, characterized in that, The system comprises: a data acquisition module, configured to acquire current sequences, voltage sequences and normal acceleration sequences at a preset period; a feature extraction module, configured to, in a current period, identify center time instants of current load events and extract voltage response waveforms according to fluctuation features of the current sequences; construct a voltage response difference matrix and acquire a voltage response pattern dispersion degree according to differences between the voltage response waveforms; acquire cumulative vibration energy according to data distribution in the normal acceleration sequences between the center time instants; acquire a vibration-response correlation coefficient according to correlation between the cumulative vibration energy of a binary tuple of all the center time instants and a corresponding value in the voltage response difference matrix, and construct a vibration-response feature pair set; acquire an electromechanical characteristic distribution offset according to distribution difference of the vibration-response feature pair set of adjacent periods; a state early warning module, configured to perform health state early warning according to distribution of the voltage response pattern dispersion degree and the vibration-response correlation coefficient of a current period; and determine whether to perform a sudden event early warning according to distribution of the electromechanical characteristic distribution offset.

2. A distribution box and its distribution box power monitoring system according to claim 1, characterized in that, The method for performing health state early warning comprises: for a current period, when the voltage response pattern dispersion degree is less than a preset degradation degree threshold value and the vibration-response correlation coefficient is less than a preset mechanical sensitivity threshold value, determining that a cable joint state is healthy; when the voltage response pattern dispersion degree is greater than or equal to the preset degradation degree threshold value and the vibration-response correlation coefficient is less than the preset mechanical sensitivity threshold value, determining that the cable joint state is electrically degraded, and performing a first-level early warning; when the voltage response pattern dispersion degree is less than the preset degradation degree threshold value and the vibration-response correlation coefficient is greater than or equal to the preset mechanical sensitivity threshold value, determining that the cable joint state is mechanically loose, and performing a second-level early warning; when the voltage response pattern dispersion degree is greater than or equal to the preset degradation degree threshold value and the vibration-response correlation coefficient is greater than or equal to the preset mechanical sensitivity threshold value, determining that the cable joint state is a composite failure, and performing a third-level early warning.

3. A distribution box and its distribution box power monitoring system according to claim 1, characterized in that, After determining whether to perform a sudden event early warning, the method further comprises predicting a remaining service life of the cable joint according to a change trend of the voltage response pattern dispersion degree over time: when a total number of the current load events in a period is greater than or equal to a minimum current load event threshold value, marking as a valid period; in a current historical sliding time window, when cable joint states of a continuous preset number of duration periods of the valid periods are electrically degraded and the voltage response pattern dispersion degree presents a monotone increasing trend, fitting a linear model of the voltage response pattern dispersion degree over time in the historical sliding time window; acquiring, based on the linear model, a remaining time for the voltage response pattern dispersion degree to reach a preset critical failure threshold value as the remaining service life.

4. The power distribution box and power monitoring system thereof according to claim 1, wherein, The method for performing a sudden event early warning comprises: when the electromechanical characteristic distribution offset is greater than or equal to a preset state jump threshold value, performing a sudden event early warning.

5. The power distribution box and power monitoring system thereof according to claim 1, wherein, The method for acquiring the electromechanical characteristic distribution offset comprises: The bulldozer distance between the current period and the adjacent previous period is calculated by using the bulldozer distance algorithm, as the electromechanical characteristic distribution offset.

6. A distribution box and its distribution box power monitoring system according to claim 1, characterized in that, The method for obtaining the vibration-response correlation coefficient comprises: The vibration-response correlation coefficient is obtained by analyzing the correlation between the cumulative vibration energy of all the binary groups of the center time and the corresponding value in the voltage response difference matrix based on the Pearson correlation coefficient.

7. A distribution box and its distribution box power monitoring system according to claim 1, characterized in that, The method for obtaining the cumulative vibration energy comprises: The square value of the data in the normal acceleration sequence between the center time of any two current load events is integrated, and the integral value is used as the cumulative vibration energy between the corresponding two current load events.

8. The power distribution box and power monitoring system thereof according to claim 1, wherein, The method for obtaining the voltage response difference matrix comprises: The DTW distance values between different voltage response waveforms in the current period are constructed into a voltage response difference matrix.

9. The power distribution box and power monitoring system thereof according to claim 1, wherein, The method for obtaining the voltage response form dispersion degree comprises: The voltage response form dispersion degree is obtained according to the overall characteristics of all element values in the voltage response difference matrix.

10. The power distribution box and power monitoring system thereof according to claim 1, wherein, The method for obtaining the center time comprises: The current change rate sequence of the current sequence is obtained, and the maximum value point of the current change rate absolute value is extracted as an event candidate point; when the current change rate absolute value of the event candidate point is greater than a preset current change threshold, the event candidate point is determined as a current load event, and the time corresponding to the event candidate point is marked as the center time.

Images