A method and system for multi-source monitoring of the operating state of a box-type substation

Abstract

The application relates to the technical field of electric variable state monitoring, and discloses a box-type substation operation state multi-source monitoring method and system, wherein the method comprises the following steps: acquiring a neutral point current signal, a vibration acceleration signal and a three-phase load current signal of a box-type substation, and outputting multi-source filtering data based on synchronous acquisition and frequency division filtering; constructing a current magnetic bias characteristic sequence and a vibration magnetic bias characteristic sequence; obtaining load decoupling discrimination data through load regression decoupling; and performing multi-scale transient synchronous analysis and layered operation state discrimination according to the load decoupling discrimination data and outputting an operation state alarm result. Since the current, vibration and load information are cooperated, the reliability of distinguishing the direct current magnetic bias, load harmonic interference and mechanical looseness of the box-type substation is improved, and the alarm result simultaneously reflects the influence of the current magnetic bias characteristic, the vibration magnetic bias characteristic and the load change on the monitoring result.

Description

Technical Field

[0001] This invention relates to the field of electrical variable status monitoring technology, and in particular to a multi-source monitoring method and system for the operating status of a prefabricated substation. Background Technology

[0002] Currently, in the single-polarity return line operation mode of DC transmission, the DC current returns through the ground and enters the neutral-point grounded transformer in the prefabricated substation, causing half-cycle saturation of the transformer core. This results in excitation current distortion into a spiked wave rich in even harmonics, while the magnetostriction of the core intensifies, leading to harmonic vibrations and abnormal noise at a fundamental frequency of 100Hz. Existing technologies typically deploy current harmonic monitoring and vibration monitoring independently: the current side indirectly reflects the degree of magnetization by analyzing the total harmonic distortion rate or even harmonic content of the neutral point current; the vibration side determines the mechanical condition of the core or windings by detecting the amplitude of the 100Hz component of the vibration signal from the transformer tank wall. Each of these two methods forms a single physical quantity early warning path.

[0003] For example, in the actual operation of prefabricated substations, when power electronic devices, electric arc furnaces, or capacitor switching are connected to the load side, high levels of even-order harmonics such as the second and fourth orders are generated in the power grid. Relying solely on current harmonic monitoring cannot distinguish whether the harmonics originate from DC bias magnetization or load harmonic interference, easily leading to false alarms. Similarly, loose transformer base bolts, the start / stop of cooling fans, or load current fluctuations can cause an abnormal increase in the 100Hz vibration component. Vibration monitoring cannot identify whether the surge in vibration is fundamentally due to DC bias magnetostriction or mechanical loosening, leading to misjudgments. Both single-measurement methods lack directional verification of the physical root cause of DC bias magnetization, resulting in inaccurate fault location and blind power outage maintenance.

[0004] Therefore, there is an urgent need for a multi-source monitoring method that can accurately identify DC bias events and reliably distinguish them from load harmonic interference or mechanical loosening faults through joint analysis of the characteristics of even harmonic groups of excitation current and the time-frequency characteristics of body vibration under complex operating conditions such as load fluctuation, background harmonic interference of power grid and mechanical loosening of the substation, so as to improve the accuracy and robustness of the operation status diagnosis of prefabricated substations. Summary of the Invention

[0005] To address the aforementioned technical shortcomings, the purpose of this invention is to propose a multi-source monitoring method for the operating status of prefabricated substations. This method aims to solve the problem that existing technologies, which rely solely on current harmonic monitoring or vibration monitoring, are easily affected by other operating conditions, making it impossible to confirm DC bias as the root cause. This can lead to misjudgments as load harmonics or mechanical loosening. In particular, under conditions where the neutral-point grounded substation is subjected to DC bias and both excitation current harmonics and body vibration are abnormal, the method cannot automatically distinguish between DC bias events.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The present invention provides a multi-source monitoring method for the operating status of a prefabricated substation.

[0007] The multi-source monitoring method for the operating status of a prefabricated substation includes: Step S10: Acquire the neutral point current signal, vibration acceleration signal and three-phase load current signal of the box-type substation, and perform multi-source signal preprocessing task based on the common sampling clock using synchronous acquisition and frequency division filtering mechanism, and output multi-source filtered data; Step S20: Based on the multi-source filtered data, the half-wave asymmetry extraction and adaptive window harmonic analysis mechanism are used to perform the current bias magnetization feature construction task and output the current bias magnetization feature sequence. Step S30: Based on the current bias magnetization feature sequence and multi-source filtered data, the vibration bias magnetization feature construction task is performed using synchronous compressed wavelet transform and frequency doubling amplitude ratio extraction mechanism, and the vibration bias magnetization feature sequence is output. Step S40: Based on the current bias magnetic characteristic sequence, vibration bias magnetic characteristic sequence and multi-source filtered data, the load regression decoupling construction mechanism is used to perform the load influence elimination and discrimination parameter construction tasks, and output the load decoupling discrimination data; Step S50: Perform multi-scale transient synchronization analysis and hierarchical operation status discrimination based on the load decoupling discrimination data, and output the operation status alarm results; Preferably, step S10, which involves acquiring the neutral point current signal, vibration acceleration signal, and three-phase load current signal of the prefabricated substation, performing multi-source signal preprocessing based on a synchronous acquisition and frequency division filtering mechanism using a common sampling clock, and outputting multi-source filtered data, specifically includes: Step S101: Use a common sampling clock to synchronously sample the neutral point current signal, vibration acceleration signal and three-phase load current signal, and configure timestamps for the corresponding sampled values ​​according to each sampling time to obtain a multi-channel synchronous sampling sequence; Step S102: Perform anti-aliasing low-pass filtering on the neutral point current sampling sequence and vibration acceleration sampling sequence in the multi-channel synchronous sampling sequence, respectively. Perform band-pass filtering on the neutral point current sampling sequence after anti-aliasing low-pass filtering to retain the power frequency component and low-order harmonic components. Perform band-pass filtering on the vibration acceleration sampling sequence after anti-aliasing low-pass filtering to retain the power frequency harmonic vibration component. Calculate the effective value sequence of the three-phase load current based on the three-phase load current sampling sequence. Step S103: Time-align the filtered neutral point current sampling sequence, the filtered vibration acceleration sampling sequence, and the three-phase load current RMS value sequence according to the timestamp to obtain multi-source filtered data. The multi-source filtered data includes filtered neutral point current data, filtered vibration acceleration data, and time-aligned three-phase load current RMS value sequence.

[0008] Preferably, step S20, which involves constructing current bias characteristics based on multi-source filtered data using a half-wave asymmetry extraction and adaptive window harmonic analysis mechanism, and outputting a current bias characteristic sequence, specifically includes: Step S201: Perform zero-crossing detection on the filtered neutral point current data, and divide the filtered neutral point current data into positive half-cycle current sequences and negative half-cycle current sequences according to adjacent zero-crossing points, and calculate the first zero-crossing point of each sequence. The average absolute values ​​of the positive half-cycle current sequence and the average absolute values ​​of the negative half-cycle current sequence within each analysis window; when the sum of the average absolute values ​​of the positive half-cycle current sequence and the average absolute values ​​of the negative half-cycle current sequence is not zero, the neutral point current half-wave asymmetry is calculated according to the following formula:

[0009] In the formula, For analysis window number, For the first The neutral point current half-wave asymmetry corresponding to each analysis window For the first The average absolute value of the positive half-cycle current sequence within each analysis window. For the first The average absolute value of the negative half-cycle current sequence within each analysis window; Step S202: Detect the fundamental frequency of the filtered neutral point current data within each analysis window, and calculate the fundamental frequency and the rated power frequency. The frequency offset between the two values ​​is used to select a spectrum analysis window function based on the comparison between the absolute value of the frequency offset and a preset frequency offset threshold. The selected spectrum analysis window function is then used to perform spectrum analysis on the filtered neutral point current data to obtain the second harmonic content. Fourth harmonic content and sixth harmonic content The harmonic content rate is the ratio of the amplitude of the corresponding harmonic to the amplitude of the fundamental frequency. Step S203: Calculate the following formula: Even-order harmonic group weighted features corresponding to each analysis window:

[0010] In the formula, For the first The weighted features of even-order harmonic groups corresponding to each analysis window are 0.65, 0.25, and 0.10, which are the preset weights corresponding to the second harmonic content, fourth harmonic content, and sixth harmonic content, respectively. The weighted features of even-order harmonic groups and the half-wave asymmetry of neutral point current corresponding to each analysis window are combined to form a current bias feature vector. The current bias feature vectors are arranged in the time order of the analysis windows to obtain a current bias feature sequence, which includes the weighted feature sequence of even-order harmonic groups and the half-wave asymmetry sequence of neutral point current.

[0011] Preferably, in step S202, the preset frequency offset threshold is 0.1Hz; when the absolute value of the frequency offset is not greater than the preset frequency offset threshold, the Blackman-Harris window function is selected; when the absolute value of the frequency offset is greater than the preset frequency offset threshold, the Kaiser window function is selected, and the shape parameter of the Kaiser window function is set to 3.5.

[0012] Preferably, step S30, which involves constructing vibration bias magnetization features based on the current bias magnetization feature sequence and multi-source filtered data using a synchronous compressed wavelet transform and octet amplitude ratio extraction mechanism, and outputting the vibration bias magnetization feature sequence, specifically includes: Step S301: Based on the analysis window corresponding to each current bias feature vector in the current bias feature sequence, extract the vibration analysis data segment corresponding to the time position from the filtered vibration acceleration data, and use Morlet continuous wavelet to perform synchronous compression wavelet transform on each vibration analysis data segment. Obtain the synchronous compression coefficient through instantaneous frequency estimation and frequency ridge redistribution. The center frequency parameter of Morlet continuous wavelet is 1, and the bandwidth parameter is 1.5. Step S302: Extract the instantaneous amplitude values ​​at the frequency ridges corresponding to the second, fourth, and sixth times power frequency from the synchronous compression coefficients, and smooth each instantaneous amplitude value using a sliding root mean square with a window length of 100ms to obtain the vibration amplitude values ​​of the second, fourth, and sixth times power frequency. Step S303: When the amplitude of the second-harmonic vibration is not zero, the ratio of the fourth-harmonic vibration amplitude to the second-harmonic vibration amplitude is determined as the first harmonic amplitude ratio, and the ratio of the sixth-harmonic vibration amplitude to the second-harmonic vibration amplitude is determined as the second harmonic amplitude ratio; the second-harmonic vibration amplitude, the first harmonic amplitude ratio, and the second harmonic amplitude ratio are combined to form a vibration bias feature vector, and the vibration bias feature vector is arranged according to the time sequence of each vibration analysis data segment to obtain a vibration bias feature sequence, which includes the second-harmonic vibration amplitude sequence, the first harmonic amplitude ratio sequence, and the second harmonic amplitude ratio sequence.

[0013] Preferably, step S40, which involves using a load regression decoupling construction mechanism to perform load influence elimination and discrimination parameter construction tasks based on the current bias magnetic characteristic sequence, vibration bias magnetic characteristic sequence, and multi-source filtered data, and outputting load decoupling discrimination data, specifically includes: Step S401: Construct a rolling load window with the current analysis window as the end, calculate the average value of the three-phase load current effective value within the rolling load window, and obtain the load current value; construct a quadratic polynomial load regression model with the load current value as the independent variable and the even-order harmonic group weighted characteristics as the dependent variable, and construct a cubic polynomial load regression model with the load current value as the independent variable and the twice-power-frequency vibration amplitude as the dependent variable; update the regression coefficients of the quadratic polynomial load regression model and the cubic polynomial load regression model using the recursive least squares method, and obtain the even-order harmonic group weighted characteristics load benchmark estimate and the twice-power-frequency vibration amplitude load benchmark estimate. Step S402: Determine the difference between the weighted characteristic of the even harmonic group and the load benchmark estimate of the weighted characteristic of the even harmonic group as the net current increment, and determine the difference between the twice power frequency vibration amplitude and the twice power frequency vibration amplitude load benchmark estimate as the net vibration increment. Arrange the net current increment and the net vibration increment according to the time order of each analysis window to obtain the net current increment sequence and the net vibration increment sequence. Step S403: Within the sliding baseline window, determine the local medians and absolute deviations of the medians relative to the corresponding local medians for the net current increment sequence, the neutral point current half-wave asymmetry sequence, and the net vibration increment sequence, respectively; multiply each absolute deviation of the median by 1.4826 to obtain the robust noise standard deviations of the net current increment, the neutral point current half-wave asymmetry, and the vibration increment; determine the sum of each local median and 3.5 times the corresponding robust noise standard deviation as the dynamic thresholds for the net current increment, the neutral point current half-wave asymmetry, and the vibration increment, respectively; combine the net current increment sequence, the neutral point current half-wave asymmetry sequence, the vibration increment sequence, the first harmonic amplitude ratio sequence, the second harmonic amplitude ratio sequence, and the corresponding dynamic thresholds as load decoupling discrimination data.

[0014] Preferably, step S50, which involves performing multi-scale transient synchronization analysis and hierarchical operation status discrimination based on load decoupling discrimination data, and outputting operation status alarm results, specifically includes: Step S501: Extract the net current increment sequence and the net vibration increment sequence from the load decoupling discrimination data. Perform continuous wavelet transform on the net current increment sequence and the net vibration increment sequence respectively. Calculate the wavelet cross spectrum and phase consistency within the modulation frequency range of 0.1Hz to 2Hz, determine the time delay corresponding to the peak value of the wavelet cross spectrum, and calculate the multi-scale transient synchronization index according to the following formula:

[0015] In the formula, For the first The multi-scale transient synchronization index corresponds to each analysis window. For the first The average phase coherence coefficient within the modulation frequency range of 0.1Hz to 2Hz for each analysis window. For the first The time delay corresponding to the peak value of the wavelet cross spectrum within each analysis window. The preset time delay attenuation coefficient, Values , It is a natural exponential function; Step S502: Extract the dynamic threshold of net current increment, the dynamic threshold of neutral point current half-wave asymmetry, and the dynamic threshold of net vibration increment from the load decoupling discrimination data. Compare the net current increment with the dynamic threshold of net current increment, the neutral point current half-wave asymmetry with the dynamic threshold of neutral point current half-wave asymmetry, and the net vibration increment with the dynamic threshold of net vibration increment. When the net current increment is greater than the dynamic threshold of net current increment, the neutral point current half-wave asymmetry is greater than the dynamic threshold of neutral point current half-wave asymmetry, and the net vibration increment is greater than the dynamic threshold of net vibration increment, the first layer of bias magnetization candidate conditions is determined to be met. Step S503: Set the preset first harmonic amplitude ratio range to 0.15 to 0.35, the preset second harmonic amplitude ratio range to 0.05 to 0.20, the preset synchronization index threshold to 0.75, and the preset duration threshold to 0.5s; when the first harmonic amplitude ratio is within the preset first harmonic amplitude ratio range and the second harmonic amplitude ratio is within the preset second harmonic amplitude ratio range, the second-level harmonic ratio condition is determined to be met; when the multi-scale transient synchronization index is greater than the preset synchronization index threshold, the third-level synchronization condition is determined to be met; when the first-level bias candidate condition and the second-level... When the frequency multiplication ratio condition and the third-level synchronization condition are both met simultaneously and the duration of their continuous occurrence exceeds the preset duration threshold, a DC bias alarm is output. When the net current increment is greater than the dynamic threshold for net current increment, the neutral point current half-wave asymmetry is greater than the dynamic threshold for neutral point current half-wave asymmetry, and the net vibration increment is not greater than the dynamic threshold for net vibration increment, a load harmonic interference alarm is output. When the net vibration increment is greater than the dynamic threshold for net vibration increment, the net current increment is not greater than the dynamic threshold for net current increment, and the neutral point current half-wave asymmetry is not greater than the dynamic threshold for neutral point current half-wave asymmetry, a mechanical looseness alarm is output.

[0016] The present invention also provides a multi-source monitoring system for the operating status of a prefabricated substation, the system comprising: The multi-source signal preprocessing module is used to acquire the neutral point current signal, vibration acceleration signal and three-phase load current signal of the box-type substation. Based on the common sampling clock, it adopts a synchronous acquisition and frequency division filtering mechanism to perform multi-source signal preprocessing tasks and output multi-source filtered data. The current bias feature construction module is used to perform the current bias feature construction task based on multi-source filtered data using half-wave asymmetry extraction and adaptive window harmonic analysis mechanism, and outputs the current bias feature sequence. The vibration bias magnetization feature construction module is used to perform the vibration bias magnetization feature construction task based on the current bias magnetization feature sequence and multi-source filtered data, using synchronous compressed wavelet transform and frequency doubling amplitude ratio extraction mechanism, and outputs the vibration bias magnetization feature sequence. The load decoupling discrimination data construction module is used to perform load influence elimination and discrimination parameter construction tasks based on current bias magnetic characteristic sequence, vibration bias magnetic characteristic sequence and multi-source filtered data using load regression decoupling construction mechanism, and output load decoupling discrimination data. The operation status discrimination module is used to perform multi-scale transient synchronization analysis and hierarchical operation status discrimination based on load decoupling discrimination data, and output operation status alarm results.

[0017] The present invention also provides a multi-source monitoring device for the operating status of a prefabricated substation. The multi-source monitoring device for the operating status of the prefabricated substation includes: a memory, a processor, and a multi-source monitoring program for the operating status of the prefabricated substation stored in the memory and executable on the processor. When the multi-source monitoring program for the operating status of the prefabricated substation is executed by the processor, it implements the above-mentioned method.

[0018] The present invention also provides a computer program product, which includes a multi-source monitoring program for the operating status of a prefabricated substation. When the multi-source monitoring program for the operating status of the prefabricated substation is executed by a processor, it implements the above-described method.

[0019] The beneficial effects of this invention are as follows: This invention synchronously samples the neutral point current signal, vibration acceleration signal, and three-phase load current signal using a common sampling clock, and performs frequency division filtering on the current side and vibration side respectively, outputting multi-source filtered data so that the subsequent current characteristics, vibration characteristics, and load reference are all on the same time axis.

[0020] By extracting half-wave asymmetry and using an adaptive window harmonic analysis mechanism, the difference between the positive and negative half-cycles of the neutral point current and the variation of even-order harmonic groups are jointly constructed into a current bias magnetic characteristic sequence, so that the current-side bias magnetic characterization no longer depends on a single harmonic amplitude.

[0021] By using synchronous compressed wavelet transform and frequency octave amplitude ratio extraction mechanism, the second, fourth and sixth power frequency related features are extracted from the vibration data segment corresponding to the current analysis window, and the vibration bias magnetic feature sequence is output, so that the magnetostrictive vibration side evidence can correspond to the current side bias magnetic features.

[0022] By constructing a load regression decoupling mechanism, current and vibration load benchmarks are established using the effective value of three-phase load current. The output includes load decoupling discrimination data containing net current increment, net vibration increment, octave amplitude ratio, and dynamic threshold, reducing the impact of normal load changes on anomaly discrimination.

[0023] By combining and judging the net current increment, net vibration increment, half-wave asymmetry, frequency doubling amplitude ratio and synchronization index through multi-scale transient synchronization analysis and hierarchical operation status discrimination, the alarm results, load harmonic interference alarm results or mechanical loosening alarm results are output, making the alarm type more consistent with the mechanism of multi-source signals. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating the first embodiment of a multi-source monitoring method for the operating status of a prefabricated substation according to the present invention. Detailed Implementation

[0025] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0026] Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] Example 1: As Figure 1 The diagram shown is a flowchart of the first embodiment of a multi-source monitoring method for the operating status of a prefabricated substation according to the present invention.

[0028] In the first embodiment, the multi-source monitoring method for the operating status of a prefabricated substation includes: Step S10: Acquire the neutral point current signal, vibration acceleration signal and three-phase load current signal of the box-type substation, and perform multi-source signal preprocessing task based on the common sampling clock using synchronous acquisition and frequency division filtering mechanism, and output multi-source filtered data; The neutral point current signal of a prefabricated substation refers to the current change signal collected by the neutral point grounding loop current sensor, used to reflect the transformer excitation state and current waveform distortion caused by DC bias. The vibration acceleration signal refers to the mechanical vibration signal collected by the acceleration sensor arranged on the transformer body or enclosure structure of the prefabricated substation, used to reflect the vibration response caused by magnetostriction, structural loosening, or other mechanical disturbances. The three-phase load current signal refers to the three-phase operating current obtained by the low-voltage side or feeder side current acquisition device, used to characterize the current load level. A common sampling clock is used to provide a unified sampling cycle for the above multi-source signals, avoiding the deviation in the sequence of events caused by different acquisition devices timing themselves.

[0029] The synchronous acquisition and frequency division filtering mechanism first obtains a multi-channel synchronous sampling sequence according to a common sampling clock and configures the sampling time for each sample value. Then, anti-aliasing low-pass filtering is applied to the neutral point current sampling sequence and the vibration acceleration sampling sequence respectively; specifically, the neutral point current data undergoes band-pass filtering to retain the power frequency component and low-order harmonic components, while the vibration acceleration data undergoes band-pass filtering to retain the vibration components related to the power frequency harmonics. The three-phase load current sampling sequence is used to calculate the effective value sequence of the three-phase load current. Finally, the system times-aligns the filtered neutral point current sampling sequence, the filtered vibration acceleration sampling sequence, and the effective value sequence of the three-phase load current according to the timestamps to obtain multi-source filtered data.

[0030] In scenarios influenced by external DC bias, a current waveform with asymmetrical positive and negative half-cycles will appear on the neutral grounding wire, and the transformer body may simultaneously experience enhanced vibrations at twice the power frequency and its harmonics. The system collects neutral point current, vibration acceleration, and three-phase load current using the same sampling clock, and separately retains the power frequency and low-order harmonics in the current, and the power frequency harmonics response in the vibration. After time alignment, the filtered neutral point current data, filtered vibration acceleration data, and the effective value sequence of the three-phase load current within a certain sampling window are simultaneously entered into subsequent steps, avoiding misattributing load changes or vibration shocks from different time periods to the same bias event.

[0031] Specifically, in terms of on-site data flow, prefabricated substations often experience simultaneous interference sources such as feeder load switching, enclosure fan operation, cabinet door opening and closing, and external construction vibrations. If only the neutral point current and vibration acceleration are saved without aligning the effective values ​​of the three-phase load currents, it becomes impossible to determine whether the enhanced current harmonics originate from bias magnetization or from an increase in load level. If only a uniform low-pass filter is applied, the low-order harmonics of interest on the current side and the harmonic responses of interest on the vibration side will be processed under the same frequency band logic, resulting in weakened effective information or the retention of irrelevant disturbances. Therefore, this step treats synchronous acquisition, frequency division filtering, and timestamp alignment as a single preprocessing chain, ensuring that each analysis window has three types of inputs: current waveform, vibration response, and load background, providing a traceable data foundation for subsequent feature construction.

[0032] Step S20: Based on the multi-source filtered data, the half-wave asymmetry extraction and adaptive window harmonic analysis mechanism are used to perform the current bias magnetization feature construction task and output the current bias magnetization feature sequence. The current bias characteristic sequence refers to the set of current-side bias characteristics arranged in chronological order within the analysis window. It includes the even-order harmonic group weighted characteristic sequence and the neutral point current half-wave asymmetry sequence. Half-wave asymmetry extraction first involves zero-crossing detection of the filtered neutral point current data, and then dividing the current into positive and negative half-cycle sequences based on adjacent zero-crossing points. Within each analysis window, the system calculates the average absolute values ​​of the positive and negative half-cycle current sequences, compares the differences between them, and normalizes the results using the average values ​​of both half-cycles to obtain the neutral point current half-wave asymmetry. This index characterizes the degree to which DC bias causes imbalance between the positive and negative half-cycles of the excitation current.

[0033] The adaptive window harmonic analysis mechanism detects the fundamental frequency within each analysis window and determines its offset relative to the rated power frequency. When the absolute value of the frequency offset is no greater than 0.1 Hz, the system selects the Blackman-Harris window function; when the absolute value of the frequency offset is greater than 0.1 Hz, the system selects the Kaiser window function with a shape parameter of 3.5. Subsequently, the system performs spectral analysis on the filtered neutral point current data to obtain the content rates of the second, fourth, and sixth harmonics relative to the fundamental frequency, and combines them according to weights to form a weighted feature of the even-order harmonic group. The weighted feature of the even-order harmonic group corresponding to each analysis window and the neutral point current half-wave asymmetry are combined to form a current bias feature vector, which is then arranged in chronological order to form a current bias feature sequence.

[0034] This step ensures that the current-side bias characteristics no longer rely solely on a single harmonic amplitude. DC bias typically introduces both excitation current half-wave asymmetry and even-order harmonic group enhancement, while nonlinear load input can also lead to local harmonic increases. Half-wave asymmetry provides a basis for waveform directional shift, and even-order harmonic group weighting provides a basis for frequency domain distortion; the combination of these two factors makes them more suitable as the current-side input for load decoupling in step S40 and hierarchical discrimination in step S50. The adaptive window selection can also reduce the impact of spectral leakage on harmonic content when there are slight fluctuations in the grid frequency, avoiding abnormal jitter in the current bias characteristic sequence due to power frequency offset.

[0035] Within a certain analysis window, if the average absolute value of the neutral point current during the positive half-cycle is significantly greater than that during the negative half-cycle, the system will exhibit a higher half-wave asymmetry. If the second, fourth, and sixth harmonic content also increases synchronously within the same window, the weighted characteristic of the even-order harmonic group will increase accordingly. If the fundamental frequency of this window is close to the rated power frequency, the system employs a Blackman-Harris window to reduce sidelobe interference; if the frequency shift exceeds 0.1 Hz, a Kaiser window is used to accommodate harmonic analysis under the shift. Ultimately, the half-wave asymmetry and even-order harmonic group weighted characteristic corresponding to this window are written into the current bias characteristic sequence, providing a basis for subsequent judgment on whether current anomalies are synchronized with vibration anomalies.

[0036] This current-side processing can also handle two types of easily confused operating segments. The first type occurs when the load contains rectifiers, frequency converters, or other nonlinear loads, in which case a certain even harmonic of the neutral point current may rise briefly, but the difference in the average absolute value of the positive and negative half-cycles does not necessarily continue to increase. The second type occurs when DC bias gradually forms, in which the waveform half-cycle asymmetry and even harmonic groups usually exhibit continuous changes within adjacent windows. The system writes the half-cycle differences obtained from zero-crossing division and the even harmonic groups obtained from harmonic analysis into the same feature vector, which enables step S40 to retain the directional information of current distortion during load regression, and also enables step S50 to avoid issuing an alarm based on only a single harmonic peak value during hierarchical discrimination.

[0037] Step S30: Based on the current bias magnetization feature sequence and multi-source filtered data, the vibration bias magnetization feature construction task is performed using synchronous compressed wavelet transform and frequency doubling amplitude ratio extraction mechanism, and the vibration bias magnetization feature sequence is output. The vibration bias magnetization feature sequence refers to the set of vibration-side bias magnetization features corresponding to the current bias magnetization feature sequence within the analysis window. This includes the second harmonic vibration amplitude sequence, the first harmonic amplitude ratio sequence, and the second harmonic amplitude ratio sequence. The synchronous compressed wavelet transform and harmonic amplitude ratio extraction mechanism first extracts vibration analysis data segments corresponding to the time positions from the filtered vibration acceleration data based on the analysis window corresponding to each current bias magnetization feature vector in the current bias magnetization feature sequence. In this way, vibration features and current features are constructed within the same time window, avoiding cross-window mismatches.

[0038] For each vibration analysis data segment, the system uses Morlet continuous wavelet for synchronous compression wavelet transform and obtains synchronous compression coefficients through instantaneous frequency estimation and frequency ridge redistribution. The center frequency parameter of the Morlet continuous wavelet is 1, and the bandwidth parameter is 1.5. The system extracts the instantaneous amplitudes at the frequency ridges corresponding to the second, fourth, and sixth harmonic frequencies from the synchronous compression coefficients, and smooths each instantaneous amplitude using a moving root mean square with a window length of 100ms to obtain the vibration amplitudes at the second, fourth, and sixth harmonic frequencies. When the second harmonic vibration amplitude is not zero, the system determines the amplitude ratio of the fourth harmonic to the second harmonic, and the amplitude ratio of the sixth harmonic to the second harmonic, and combines the second harmonic vibration amplitude and the amplitude ratios of the two harmonics into a vibration bias feature vector, forming a vibration bias feature sequence in chronological order.

[0039] This step extracts the bias-magnetic-related mechanical response from the background of ordinary vibration. Traditional vibration monitoring, if only observing broadband vibration amplitude, can lead to increased amplitude due to mechanical loosening, fan startup, enclosure impact, or external construction vibration; if only observing the second-harmonic vibration, it is difficult to distinguish the normal magnetostriction enhancement caused by load changes. This step simultaneously extracts the second-harmonic vibration amplitude and the amplitude ratios of the fourth and sixth harmonics relative to the second harmonic, preserving both vibration intensity and harmonic structure morphology. This vibration bias-magnetic characteristic sequence is analyzed synchronously with the net current increment sequence in step S50, which can distinguish the electromagnetic and mechanical coupling response truly caused by bias-magnetism from simple mechanical loosening.

[0040] In operating scenarios with continuously increasing DC bias, the second-harmonic ridge of the filtered vibration acceleration data will strengthen with changes in excitation state, and the fourth-harmonic and sixth-harmonic ridges may also show corresponding increases. The system extracts vibration data within the same time period as the current analysis window, extracts each harmonic ridge after synchronous compressed wavelet transform, and suppresses transient noise using a 100ms sliding root mean square. If the amplitude ratio of the fourth harmonic to the second harmonic falls near a predefined range, and the amplitude ratio of the sixth harmonic to the second harmonic also falls near the corresponding range, then the vibration bias characteristics of this window are more consistent with biased magnetostrictive response than random mechanical shock.

[0041] Compared to conventional spectrum analysis, synchronous compressed wavelet transform is suitable for handling non-stationary segments of vibration in prefabricated substations. For example, when the load changes rapidly, the vibration harmonic components may shift or intensify within several short windows; directly averaging the spectrum of the entire vibration signal would flatten out this variation. When the cabinet is subjected to a short-term impact, the broadband energy may increase but will not stably conform to the 2nd, 4th, and 6th power frequency ridges. The system extracts the harmonic ridges from vibration data segments aligned with the current analysis window, and then uses the harmonic amplitude ratio to describe the harmonic structure. This ensures that the output vibration bias characteristics can reflect both the periodic response caused by magnetostriction and maintain the ability to distinguish between random impacts and broadband mechanical noise.

[0042] Step S40: Based on the current bias magnetic characteristic sequence, vibration bias magnetic characteristic sequence and multi-source filtered data, the load regression decoupling construction mechanism is used to perform the load influence elimination and discrimination parameter construction tasks, and output the load decoupling discrimination data; Load decoupling discrimination data refers to the discrimination input set formed after removing the load influence from the current bias magnetic characteristic sequence, vibration bias magnetic characteristic sequence, and multi-source filtered data. This includes the net current increment sequence, neutral point current half-wave asymmetry sequence, vibration net increment sequence, first harmonic amplitude ratio sequence, second harmonic amplitude ratio sequence, and corresponding dynamic thresholds. The load regression decoupling construction mechanism constructs a rolling load window with the current analysis window as the end point, calculates the average of the effective values ​​of the three-phase load current within this window, and obtains the load current value.

[0043] The system constructs a quadratic polynomial load regression model with load current as the independent variable and even-order harmonic group weighted characteristics as the dependent variable, and a cubic polynomial load regression model with load current as the independent variable and twice the power frequency vibration amplitude as the dependent variable. Recursive least squares is used to update the coefficients of the two regression models, obtaining the load baseline estimates for even-order harmonic group weighted characteristics and twice the power frequency vibration amplitude. Subsequently, the system defines the portion of the even-order harmonic group weighted characteristics exceeding its load baseline estimate as the net current increment, and the portion of the twice the power frequency vibration amplitude exceeding its load baseline estimate as the net vibration increment, arranged in chronological order according to the analysis window. The system also determines the local median and median absolute deviation of the net current increment sequence, the neutral point current half-wave asymmetry sequence, and the net vibration increment sequence within the sliding baseline window, and constructs corresponding dynamic thresholds using robust noise standard deviation.

[0044] The purpose of this step is to separate the normal current harmonic rise and vibration enhancement caused by load changes from the anomaly discrimination parameters. When the load of a prefabricated substation increases, even without DC bias, the weighted characteristics of even-order harmonic groups and the amplitude of double-frequency vibration may rise with changes in operating current; if the original characteristics are used directly for limit judgment, load peaks will cause false alarms. By forming a baseline estimate through a load regression model, the system only focuses on the net increment exceeding the interpretable range of the load. The dynamic threshold is constructed using the local median and the absolute deviation of the median, which is insensitive to short-term spikes and anomalies, allowing the threshold to be adjusted according to the background noise level.

[0045] During peak evening load periods, the effective value of the three-phase load current continuously increases, and both the weighted characteristic of even-order harmonic groups and the amplitude of double-frequency vibration show a certain increase. If the regression model determines that this increase is consistent with the change in load current, the net current increment and net vibration increment will not significantly exceed the dynamic threshold, and the bias candidate condition will not be triggered. If the neutral point current half-wave asymmetry suddenly increases under the same load level, and the weighted characteristic of even-order harmonic groups and the amplitude of double-frequency vibration also exceed their respective load references, the system will obtain the net current increment and net vibration increment, and write them together with the dynamic threshold into the load decoupling discrimination data, providing input for step S50 to determine the anomaly type.

[0046] The key to load regression decoupling lies in separating the "part that changes reasonably with the load" from the "part that remains abnormally prominent under similar loads." A quadratic polynomial is used to characterize the bending trend of the weighted characteristics of even-order harmonic groups as the load changes, while a cubic polynomial is used to characterize the nonlinear changes that the amplitude of the second power frequency vibration may exhibit in different load ranges. Recursive least squares updates allow the model to slowly adapt to the long-term operating background of the prefabricated substation, rather than using a static baseline from the initial commissioning phase. When local medians and median absolute deviations are used to construct dynamic thresholds, the spikes in a single window do not directly push up the threshold, nor do they completely swallow up the real anomalies in consecutive windows, thus ensuring that the discrimination data received in step S50 simultaneously includes net increments, frequency doubling structures, and threshold boundaries.

[0047] Step S50: Perform multi-scale transient synchronization analysis and hierarchical operation status discrimination based on the load decoupling discrimination data, and output the operation status alarm results; The operational status alarm results refer to the hierarchical status conclusions output based on the load decoupling discrimination data, including DC bias alarms, load harmonic interference alarms, and mechanical loosening alarms. Multi-scale transient synchronization analysis first extracts the net current increment sequence and the net vibration increment sequence from the load decoupling discrimination data, performs continuous wavelet transforms on them respectively, and calculates the wavelet cross spectrum and phase consistency within the modulation frequency range of 0.1Hz to 2Hz. The system determines the multi-scale transient synchronization index based on the time delay corresponding to the peak value of the wavelet cross spectrum and the average phase consistency coefficient; this index is higher when the net current increment and net vibration increment are synchronized and the time delay is small, and lower when they are asynchronous or the delay is large.

[0048] The layered operation status judgment is divided into three main layers. The first layer extracts the dynamic thresholds for net current increment, neutral point current half-wave asymmetry, and vibration increment from the load decoupling judgment data, and compares whether the net current increment, neutral point current half-wave asymmetry, and vibration increment exceed the corresponding thresholds. When all three exceed the limits simultaneously, the first-layer bias candidate condition is met. The second layer checks whether the first harmonic amplitude ratio and the second harmonic amplitude ratio are within the preset range to confirm whether the vibration frequency doubling structure conforms to the biased magnetostrictive response. The third layer checks whether the multi-scale transient synchronization index is greater than the preset synchronization index threshold, and checks whether the duration of continuous fulfillment of the three-layer conditions exceeds the preset duration threshold. When all three conditions are met simultaneously, a DC bias alarm is output; when the current side exceeds the limit but the vibration side does not, a load harmonic interference alarm is output; when the vibration side exceeds the limit but the current side and half-wave asymmetry do not, a mechanical loosening alarm is output.

[0049] This step unifies current anomalies, vibration anomalies, frequency doubling structures, and transient synchronization relationships into a hierarchical discrimination logic, avoiding misjudgments caused by single threshold triggers. Load harmonic interference may cause abnormalities in net current increment and half-wave asymmetry, but it does not necessarily lead to a synchronous net increment of twice the power frequency vibration; mechanical loosening may increase the net vibration increment, but the current-side bias characteristic does not hold; DC bias is more consistent with the pattern of current-side asymmetry, even-order harmonic group enhancement, changes in vibration frequency doubling structure, and transient synchronization of current and vibration. The duration threshold can also filter out instantaneous impacts and sampling spikes, making the output alarm more suitable for operation and maintenance handling.

[0050] During a field monitoring session, if the net current increment exceeds the dynamic threshold for net current increment during peak load periods, and the neutral point current half-wave asymmetry also exceeds the corresponding threshold, but the net vibration increment does not exceed the dynamic threshold for net vibration increment, the system outputs a load harmonic interference alarm, prompting priority to investigate nonlinear loads or harmonic sources. If the net vibration increment exceeds the threshold during another period, while the net current increment and half-wave asymmetry do not exceed the limits, the system outputs a mechanical looseness alarm. Only when the net current increment, half-wave asymmetry, and net vibration increment all exceed the limits, the octave amplitude ratio falls within the corresponding range, and the multi-scale transient synchronization index exceeds 0.75 and remains above 0.5s, will the system output a DC bias alarm.

[0051] This hierarchical structure provides an interpretable troubleshooting path for alarm results. The first layer provides an initial screening conclusion as to whether the current and vibration sides simultaneously exceed limits. The second layer checks whether the vibration frequency doubling structure conforms to a biased magnetostrictive response. The third layer further checks whether the net current increment and the net vibration increment have a transient synchronization relationship within the range of 0.1Hz to 2Hz. If only the first layer current-side condition is met, but the vibration structure and synchronization relationship are not met, maintenance personnel can prioritize checking the load harmonic sources. If the vibration side exceeds the limit, but the current side and half-wave asymmetry are not met, fasteners, fans, and enclosure structures can be checked first. If all three layer conditions are met consecutively, it indicates that the current distortion, vibration frequency doubling, and transient synchronization all point to the same operational event, providing more sufficient basis for outputting a DC biased alarm. Subsequent handling can focus on the neutral point grounding loop and external DC intrusion factors.

[0052] Example 2: Furthermore, the present invention provides a multi-source monitoring system for the operating status of a prefabricated substation, employing a multi-source monitoring method for the operating status of a prefabricated substation as described in the above embodiments, which can solve the technical problem of multi-source monitoring of the operating status of a prefabricated substation. The beneficial effects of the multi-source monitoring system for the operating status of a prefabricated substation provided by the present invention are the same as those of the multi-source monitoring method for the operating status of a prefabricated substation provided in the above embodiments, and other technical features of the multi-source monitoring system for the operating status of a prefabricated substation are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0053] Example 3: This invention provides a multi-source monitoring device for the operating status of a prefabricated substation. The device includes at least one processor and a memory communicatively connected to the processor. The memory stores instructions executable by the processor, which are then executed to enable the processor to perform the multi-source monitoring method for the operating status of a prefabricated substation described in Example 1. The multi-source monitoring device for the operating status of a prefabricated substation in this invention may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and vehicle-mounted terminals (e.g., vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. This multi-source monitoring device for the operating status of a prefabricated substation is merely an example and should not limit the functionality or scope of the invention. The device may also include a processing unit (e.g., a central processing unit, a graphics processing unit), which can perform various appropriate actions and processes based on a program stored in a read-only memory or a program loaded from a storage device into a random access memory. The random access memory also stores various programs and data required for the operation of a multi-source monitoring device for the operating status of a prefabricated substation. The processing unit, read-only memory, and random access memory are interconnected via a bus. The I / O interface is also connected to the bus. Typically, the following systems can be connected to the I / O interface: input devices including touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices including liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices including magnetic tapes, hard disks, etc.; and communication devices. The communication device allows the multi-source monitoring device for the operating status of a prefabricated substation to exchange data wirelessly or via wired communication with other devices. While a multi-source monitoring device for the operating status of a prefabricated substation with various systems has been described, it should be understood that it is not required to implement or possess all of the described systems. More or fewer systems can be implemented alternatively.

[0054] Example 4: This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the multi-source monitoring method for the operating status of a prefabricated substation as described above. The computer program product provided by this invention can solve the technical problem of multi-source monitoring of the operating status of a prefabricated substation. Compared with the prior art, the beneficial effects of the computer program product provided by this invention are the same as those of the multi-source monitoring method for the operating status of a prefabricated substation provided in the above embodiments, and will not be repeated here.

[0055] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device, or installed from a read-only memory. When the computer program is executed by a processing device, it performs the functions defined in the methods of the embodiments disclosed in this invention.

[0056] It should be understood that the various parts disclosed in this invention can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0057] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the present invention and its equivalents, the present invention also intends to include these modifications and variations.

Claims

1. A multi-source monitoring method for the operating status of a prefabricated substation, characterized in that, The method includes: Step S10: Acquire the neutral point current signal, vibration acceleration signal and three-phase load current signal of the box-type substation, and perform multi-source signal preprocessing task based on the common sampling clock using synchronous acquisition and frequency division filtering mechanism, and output multi-source filtered data; Step S20: Based on the multi-source filtered data, the half-wave asymmetry extraction and adaptive window harmonic analysis mechanism are used to perform the current bias magnetization feature construction task and output the current bias magnetization feature sequence. Step S30: Based on the current bias magnetization feature sequence and multi-source filtered data, the vibration bias magnetization feature construction task is performed using synchronous compressed wavelet transform and frequency doubling amplitude ratio extraction mechanism, and the vibration bias magnetization feature sequence is output. Step S40: Based on the current bias magnetic characteristic sequence, vibration bias magnetic characteristic sequence and multi-source filtered data, the load regression decoupling construction mechanism is used to perform the load influence elimination and discrimination parameter construction tasks, and output the load decoupling discrimination data; Step S50: Perform multi-scale transient synchronization analysis and hierarchical operation status discrimination based on the load decoupling discrimination data, and output the operation status alarm results.

2. The multi-source monitoring method for the operating status of a prefabricated substation as described in claim 1, characterized in that, Step S10 involves acquiring the neutral point current signal, vibration acceleration signal, and three-phase load current signal of the prefabricated substation, performing multi-source signal preprocessing based on a synchronous acquisition and frequency division filtering mechanism using a common sampling clock, and outputting multi-source filtered data. Specifically, this includes: Step S101: Use a common sampling clock to synchronously sample the neutral point current signal, vibration acceleration signal and three-phase load current signal, and configure timestamps for the corresponding sampled values ​​according to each sampling time to obtain a multi-channel synchronous sampling sequence; Step S102: Perform anti-aliasing low-pass filtering on the neutral point current sampling sequence and vibration acceleration sampling sequence in the multi-channel synchronous sampling sequence, perform band-pass filtering on the neutral point current sampling sequence after anti-aliasing low-pass filtering to retain the power frequency component and low-order harmonic components, perform band-pass filtering on the vibration acceleration sampling sequence after anti-aliasing low-pass filtering to retain the power frequency harmonic vibration component, and calculate the effective value sequence of the three-phase load current based on the three-phase load current sampling sequence; Step S103: Time-align the filtered neutral point current sampling sequence, the filtered vibration acceleration sampling sequence, and the three-phase load current RMS value sequence according to the timestamp to obtain multi-source filtered data. The multi-source filtered data includes filtered neutral point current data, filtered vibration acceleration data, and time-aligned three-phase load current RMS value sequence.

3. The multi-source monitoring method for the operating status of a prefabricated substation as described in claim 2, characterized in that, Step S20, which involves constructing current bias characteristics based on multi-source filtered data using a half-wave asymmetry extraction and adaptive window harmonic analysis mechanism, and outputting a current bias characteristic sequence, specifically includes: Step S201: Perform zero-crossing detection on the filtered neutral point current data, and divide the filtered neutral point current data into positive half-cycle current sequences and negative half-cycle current sequences according to adjacent zero-crossing points, and calculate the first zero-crossing point of each sequence. The average absolute values ​​of the positive half-cycle current sequence and the average absolute values ​​of the negative half-cycle current sequence within each analysis window; when the sum of the average absolute values ​​of the positive half-cycle current sequence and the average absolute values ​​of the negative half-cycle current sequence is not zero, the neutral point current half-wave asymmetry is calculated according to the following formula: In the formula, For analysis window number, For the first The neutral point current half-wave asymmetry corresponding to each analysis window For the first The average absolute value of the positive half-cycle current sequence within each analysis window. For the first The average absolute value of the negative half-cycle current sequence within each analysis window; Step S202: Detect the fundamental frequency of the filtered neutral point current data within each analysis window, and calculate the fundamental frequency and the rated power frequency. The frequency offset between the two values ​​is used to select a spectrum analysis window function based on the comparison between the absolute value of the frequency offset and a preset frequency offset threshold. The selected spectrum analysis window function is then used to perform spectrum analysis on the filtered neutral point current data to obtain the second harmonic content. Fourth harmonic content and sixth harmonic content The harmonic content rate is the ratio of the amplitude of the corresponding harmonic to the amplitude of the fundamental frequency. Step S203: Calculate the following formula: Even-order harmonic group weighted features corresponding to each analysis window: In the formula, For the first The weighted features of even-order harmonic groups corresponding to each analysis window are 0.65, 0.25, and 0.10, which are the preset weights corresponding to the second harmonic content, fourth harmonic content, and sixth harmonic content, respectively. The weighted features of even-order harmonic groups and the half-wave asymmetry of neutral point current corresponding to each analysis window are combined to form a current bias feature vector. The current bias feature vectors are arranged in the time order of the analysis windows to obtain a current bias feature sequence, which includes the weighted feature sequence of even-order harmonic groups and the half-wave asymmetry sequence of neutral point current.

4. The multi-source monitoring method for the operating status of a prefabricated substation as described in claim 3, characterized in that, In step S202, the preset frequency offset threshold is 0.1Hz; when the absolute value of the frequency offset is not greater than the preset frequency offset threshold, the Blackman-Harris window function is selected; when the absolute value of the frequency offset is greater than the preset frequency offset threshold, the Kaiser window function is selected, and the shape parameter of the Kaiser window function is set to 3.

5.

5. The multi-source monitoring method for the operating status of a prefabricated substation as described in claim 3, characterized in that, Step S30, which involves constructing vibration bias magnetization features based on the current bias magnetization feature sequence and multi-source filtered data using a synchronous compressed wavelet transform and octet amplitude ratio extraction mechanism, and outputting the vibration bias magnetization feature sequence, specifically includes: Step S301: Based on the analysis window corresponding to each current bias feature vector in the current bias feature sequence, extract the vibration analysis data segment corresponding to the time position from the filtered vibration acceleration data, and use Morlet continuous wavelet to perform synchronous compression wavelet transform on each vibration analysis data segment. Obtain the synchronous compression coefficient through instantaneous frequency estimation and frequency ridge redistribution. The center frequency parameter of Morlet continuous wavelet is 1, and the bandwidth parameter is 1.

5. Step S302: Extract the instantaneous amplitude values ​​at the frequency ridges corresponding to the second, fourth, and sixth times power frequency from the synchronous compression coefficients, and smooth each instantaneous amplitude value using a sliding root mean square with a window length of 100ms to obtain the vibration amplitude values ​​of the second, fourth, and sixth times power frequency. Step S303: When the amplitude of the second-harmonic vibration is not zero, the ratio of the fourth-harmonic vibration amplitude to the second-harmonic vibration amplitude is determined as the first harmonic amplitude ratio, and the ratio of the sixth-harmonic vibration amplitude to the second-harmonic vibration amplitude is determined as the second harmonic amplitude ratio; the second-harmonic vibration amplitude, the first harmonic amplitude ratio, and the second harmonic amplitude ratio are combined to form a vibration bias feature vector, and the vibration bias feature vector is arranged according to the time sequence of each vibration analysis data segment to obtain a vibration bias feature sequence, which includes the second-harmonic vibration amplitude sequence, the first harmonic amplitude ratio sequence, and the second harmonic amplitude ratio sequence.

6. The multi-source monitoring method for the operating status of a prefabricated substation as described in claim 5, characterized in that, Step S40, which involves using a load regression decoupling construction mechanism to perform load influence elimination and discrimination parameter construction based on the current bias magnetization characteristic sequence, vibration bias magnetization characteristic sequence, and multi-source filtered data, and outputting load decoupling discrimination data, specifically includes: Step S401: Construct a rolling load window with the current analysis window as the end, calculate the average value of the three-phase load current effective value within the rolling load window, and obtain the load current value; construct a quadratic polynomial load regression model with the load current value as the independent variable and the even-order harmonic group weighted characteristics as the dependent variable, and construct a cubic polynomial load regression model with the load current value as the independent variable and the twice-power-frequency vibration amplitude as the dependent variable; update the regression coefficients of the quadratic polynomial load regression model and the cubic polynomial load regression model using the recursive least squares method, and obtain the even-order harmonic group weighted characteristics load benchmark estimate and the twice-power-frequency vibration amplitude load benchmark estimate. Step S402: Determine the difference between the weighted characteristic of the even harmonic group and the load benchmark estimate of the weighted characteristic of the even harmonic group as the net current increment, and determine the difference between the twice power frequency vibration amplitude and the twice power frequency vibration amplitude load benchmark estimate as the net vibration increment. Arrange the net current increment and the net vibration increment according to the time order of each analysis window to obtain the net current increment sequence and the net vibration increment sequence. Step S403: Within the sliding baseline window, determine the local medians and absolute deviations of the medians relative to the corresponding local medians for the net current increment sequence, the neutral point current half-wave asymmetry sequence, and the net vibration increment sequence, respectively; multiply each absolute deviation of the median by 1.4826 to obtain the robust noise standard deviations of the net current increment, the neutral point current half-wave asymmetry, and the vibration increment; determine the sum of each local median and 3.5 times the corresponding robust noise standard deviation as the dynamic thresholds for the net current increment, the neutral point current half-wave asymmetry, and the vibration increment, respectively; combine the net current increment sequence, the neutral point current half-wave asymmetry sequence, the vibration increment sequence, the first harmonic amplitude ratio sequence, the second harmonic amplitude ratio sequence, and the corresponding dynamic thresholds as load decoupling discrimination data.

7. The multi-source monitoring method for the operating status of a prefabricated substation as described in claim 6, characterized in that, Step S50, which involves performing multi-scale transient synchronization analysis and hierarchical operation status discrimination based on load decoupling discrimination data, and outputting operation status alarm results, specifically includes: Step S501: Extract the net current increment sequence and the net vibration increment sequence from the load decoupling discrimination data. Perform continuous wavelet transform on the net current increment sequence and the net vibration increment sequence respectively. Calculate the wavelet cross spectrum and phase consistency within the modulation frequency range of 0.1Hz to 2Hz, determine the time delay corresponding to the peak value of the wavelet cross spectrum, and calculate the multi-scale transient synchronization index according to the following formula: In the formula, For the first The multi-scale transient synchronization index corresponds to each analysis window. For the first The average phase coherence coefficient within the modulation frequency range of 0.1Hz to 2Hz for each analysis window. For the first The time delay corresponding to the peak value of the wavelet cross spectrum within each analysis window. The preset time delay attenuation coefficient, Values , It is a natural exponential function; Step S502: Extract the dynamic threshold of net current increment, the dynamic threshold of neutral point current half-wave asymmetry, and the dynamic threshold of net vibration increment from the load decoupling discrimination data. Compare the net current increment with the dynamic threshold of net current increment, the neutral point current half-wave asymmetry with the dynamic threshold of neutral point current half-wave asymmetry, and the net vibration increment with the dynamic threshold of net vibration increment. When the net current increment is greater than the dynamic threshold of net current increment, the neutral point current half-wave asymmetry is greater than the dynamic threshold of neutral point current half-wave asymmetry, and the net vibration increment is greater than the dynamic threshold of net vibration increment, the first layer of bias magnetization candidate conditions is determined to be met. Step S503: Set the preset first harmonic amplitude ratio range to 0.15 to 0.35, the preset second harmonic amplitude ratio range to 0.05 to 0.20, the preset synchronization index threshold to 0.75, and the preset duration threshold to 0.5s; when the first harmonic amplitude ratio is within the preset first harmonic amplitude ratio range and the second harmonic amplitude ratio is within the preset second harmonic amplitude ratio range, the second-level harmonic ratio condition is determined to be met; when the multi-scale transient synchronization index is greater than the preset synchronization index threshold, the third-level synchronization condition is determined to be met; when the first-level bias candidate condition and the second-level... When the frequency multiplication ratio condition and the third-level synchronization condition are both met simultaneously and the duration of their continuous occurrence exceeds the preset duration threshold, a DC bias alarm is output. When the net current increment is greater than the dynamic threshold for net current increment, the neutral point current half-wave asymmetry is greater than the dynamic threshold for neutral point current half-wave asymmetry, and the net vibration increment is not greater than the dynamic threshold for net vibration increment, a load harmonic interference alarm is output. When the net vibration increment is greater than the dynamic threshold for net vibration increment, the net current increment is not greater than the dynamic threshold for net current increment, and the neutral point current half-wave asymmetry is not greater than the dynamic threshold for neutral point current half-wave asymmetry, a mechanical looseness alarm is output.

8. A multi-source monitoring system for the operating status of a prefabricated substation, applied to the multi-source monitoring method for the operating status of a prefabricated substation as described in any one of claims 1 to 7, characterized in that, The multi-source monitoring system for the operating status of the prefabricated substation includes: The multi-source signal preprocessing module is used to acquire the neutral point current signal, vibration acceleration signal and three-phase load current signal of the box-type substation. Based on the common sampling clock, it adopts a synchronous acquisition and frequency division filtering mechanism to perform multi-source signal preprocessing tasks and output multi-source filtered data. The current bias feature construction module is used to perform the current bias feature construction task based on multi-source filtered data using half-wave asymmetry extraction and adaptive window harmonic analysis mechanism, and outputs the current bias feature sequence. The vibration bias magnetization feature construction module is used to perform the vibration bias magnetization feature construction task based on the current bias magnetization feature sequence and multi-source filtered data, using synchronous compressed wavelet transform and frequency doubling amplitude ratio extraction mechanism, and outputs the vibration bias magnetization feature sequence. The load decoupling discrimination data construction module is used to perform load influence elimination and discrimination parameter construction tasks based on current bias magnetic characteristic sequence, vibration bias magnetic characteristic sequence and multi-source filtered data using load regression decoupling construction mechanism, and output load decoupling discrimination data. The operation status discrimination module is used to perform multi-scale transient synchronization analysis and hierarchical operation status discrimination based on load decoupling discrimination data, and output operation status alarm results.

9. A multi-source monitoring device for the operating status of a prefabricated substation, characterized in that, The multi-source monitoring device for the operating status of the prefabricated substation includes: a memory, a processor, and a multi-source monitoring program for the operating status of the prefabricated substation stored in the memory and executable on the processor. When the multi-source monitoring program for the operating status of the prefabricated substation is executed by the processor, it implements a multi-source monitoring method for the operating status of a prefabricated substation according to any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a multi-source monitoring program for the operating status of a prefabricated substation. When the multi-source monitoring program for the operating status of a prefabricated substation is executed by a processor, it implements a multi-source monitoring method for the operating status of a prefabricated substation as described in any one of claims 1 to 7.

Images