Multi-channel signal processing layout optimization method and device suitable for substation reactor room

Abstract

The application discloses a multi-channel signal processing layout optimization method and device suitable for a transformer substation reactor room, relates to the technical field of power system monitoring, and significantly enhances the signal-to-noise ratio and phase consistency of a synthesized signal by constructing similar channel pairs and performing phase calibration summation, effectively suppresses the phase cancellation effect caused by random noise and phase mismatch, and realizes closed-loop adaptive optimization of the sensor layout spacing and similarity threshold based on sequential splicing and linear fitting of the phase average values in the units after clustering, thereby avoiding the problems of redundant sampling or spatial undersampling caused by traditional experience layout.

Description

Technical Field

[0001] This invention relates to the field of power system monitoring technology, and in particular to a method and apparatus for optimizing the deployment of multi-channel signal processing in substation reactor rooms. Background Technology

[0002] The deployment parameters of sensor arrays directly affect the correlation and consistency of multi-channel signals. Existing deployment strategies mostly rely on fixed spacing or empirical values, lacking adaptive optimization methods based on measured signal characteristics, resulting in sensor redundancy or insufficient signal correlation.

[0003] For example, Chinese patent CN121163592A discloses a reactor vibration and partial discharge monitoring system. This system uses a three-dimensional sensor unit arranged axially along the annular oil channel between the reactor core and windings. Temperature compensation is used to correct gas concentration readings and ultrasonic wave propagation speed, thereby achieving dissolved gas analysis and early warning, and three-dimensional time-difference location of partial discharge points. This technology focuses on online monitoring of internal insulation degradation and partial discharge in the reactor. The sensor placement is pre-fixed by the internal structure, and the spacing between adjacent sensors is determined by mechanical installation, without any adaptive optimization of the sensor array's spatial distribution. Furthermore, the different sensor types in this scheme have clearly defined functions; multi-channel signals are only used for independent compensation calculations or time-difference location, lacking correlation clustering analysis based on phase difference and amplitude ratio between channels, and unable to dynamically adjust placement parameters based on measured signals to improve signal consistency.

[0004] Chinese patent CN114580242A discloses a noise source analysis method for oil-immersed reactors. Based on finite element simulation, it calculates the vibration response of the reactor tank under magnetostriction and electromagnetic force excitation through an electromagnetic-structure-sound field coupling model. It then uses the Helmholtz integral equation or long-wave / short-wave approximation to correlate the normal velocity of the vibrating surface with the radiated sound pressure distribution, ultimately obtaining a numerical solution for the external sound field of the tank. This technique is a pure simulation analysis method; the placement of its sensors (or excitation points) is entirely determined by mesh generation and preset boundary conditions, without addressing the optimization of the actual physical sensor array placement. Furthermore, the simulation assumes fixed positions for both force source point excitation and line source acoustic excitation, failing to consider the feedback adjustment of placement positions based on the phase characteristics of the measured signals. While this method can simulate noise radiation under different boundary conditions, it cannot solve the problem of phase mismatch or amplitude inconsistency in multi-channel signals caused by improper sensor spacing in actual deployment. Summary of the Invention

[0005] The purpose of this section is to outline some aspects of the embodiments of the present invention and to briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section, as well as in the abstract and title of the present application, to avoid obscuring the purpose of this section, the abstract and title of the invention. Such simplifications or omissions shall not be used to limit the scope of the present invention.

[0006] In view of the aforementioned existing problems, the present invention is proposed.

[0007] Therefore, the technical problem solved by the present invention is that in the existing reactor room monitoring technology, the placement parameters of the sensor array are usually fixed in advance based on the reactor geometry or engineering experience, and there is a lack of an adaptive optimization mechanism based on the internal characteristics of measured multi-channel signals.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a multi-channel signal processing deployment optimization method applicable to the reactor room of a substation, characterized by comprising: deploying multiple sensors to collect signals from each channel; using one channel as a reference channel; calculating the phase difference of each of the remaining channels relative to the reference channel; and recording the relative phase angle of each channel; constructing a set of similar channel pairs based on the comparison results of the phase differences between channels with a first preset threshold and the amplitude ratio between channels; dividing the channels into several differential matrix units based on the set of similar channel pairs; and summing the real-time signals within each differential matrix unit after phase calibration. The process involves generating unit composite signals; reading the composite signals of each differential matrix unit from memory in ascending order of the average relative phase angle of all channels within each differential matrix unit, concatenating them into a continuous data stream, and extracting the voltage amplitude of each differential matrix unit; performing linear fitting between the unit number and the voltage amplitude in the data stream to obtain the fitting slope; adjusting the first preset threshold based on the deviation of the fitting slope from the preset standard slope range, and adjusting the spacing between adjacent monitoring points based on the direction of the deviation of the fitting slope; using the updated spacing between adjacent monitoring points and the updated first preset threshold for the next deployment, until the termination condition is met.

[0009] As a preferred embodiment of the present invention, the method of deploying multiple sensors to collect signals from each channel includes: deploying multiple sensors at equal intervals according to a preset initial adjacent monitoring point spacing in the reactor room, synchronously collecting signals from each channel at a preset sampling rate, and performing filtering and noise reduction processing on the signals from each channel.

[0010] As a preferred embodiment of the present invention, the construction of the similar channel pair set includes: taking one of the channels as a reference channel, calculating the phase difference of each of the remaining channels relative to the reference channel and recording the relative phase angle of each channel; calculating the absolute value of the phase difference and the amplitude ratio between any two channels, wherein the amplitude ratio is the ratio of the effective values ​​of the two channel signals; and recording all channel pairs whose absolute value of the phase difference is less than a first preset threshold and whose amplitude ratio is within a preset amplitude range as the similar channel pair set.

[0011] As a preferred embodiment of the present invention, the step of dividing the channel into several difference matrix units based on the set of similar channel pairs includes: dividing the channel into several difference matrix units using a connected component extraction algorithm, wherein there is a path consisting of similar channel pairs between any two channels in each difference matrix unit, and there are no similar channel pairs between different difference matrix units; when there are no similar channel pairs, each channel is treated as an independent unit.

[0012] In a preferred embodiment of the present invention, the generation of the synthesized signal by the unit includes: for each differential matrix unit, first performing analog-to-digital conversion on the real-time voltage signals of each channel to obtain discrete digital signals; determining the channel with the smallest relative phase angle within the unit based on the recorded relative phase angle of each channel, and using the channel as the reference channel within the unit; for each other channel within the unit except the reference channel, performing a Hilbert transform on the discrete digital signals in the digital domain to construct an analytic signal, multiplying the analytic signal by a complex rotation factor and taking the real part to achieve phase rotation; summing the discrete digital signals of the reference channel within the unit with all the rotated digital signals of the other channels with equal weights to generate the unit synthesized signal; when there is only one channel within the unit, directly using the discrete digital signal of that channel as the synthesized signal.

[0013] In a preferred embodiment of the present invention, the phase angle of the complex rotation factor is equal to the negative of the difference between the relative phase angle of the reference channel and the relative phase angle of the channel, such that the phase of the channel signal after rotation is consistent with the phase of the reference channel.

[0014] As a preferred embodiment of the present invention, the fitting slope includes: constructing a Cartesian coordinate system with the sequential numbering of the difference matrix units corresponding to each synthesized signal in the data stream as the abscissa and the voltage amplitude of each synthesized signal as the ordinate; when there is only one difference matrix unit, if the number of iterations has reached the maximum number of iterations, the current configuration is output and a warning is issued; otherwise, the first preset threshold is increased to a preset multiple of the original value, and the process directly jumps to the next step of deploying multiple sensors, skipping the subsequent linear fitting and parameter adjustment steps of this round; when there are at least two difference matrix units, linear fitting is performed on all data points and the fitting slope is calculated.

[0015] In a preferred embodiment of the present invention, when there are at least two difference matrix units, adjusting the spacing between adjacent monitoring points according to the deviation direction of the fitted slope includes: comparing the fitted slope with a preset standard slope range; when the fitted slope is within the preset standard slope range, terminating the iteration and outputting the current placement parameters as the optimization result; when the fitted slope is higher than the upper limit of the preset standard slope range, multiplying the currently used first preset threshold by a preset scaling factor to obtain a new first preset threshold, and reducing the spacing between adjacent monitoring points by a preset fixed step size but not lower than a preset minimum spacing; when the fitted slope is lower than the lower limit of the preset standard slope range, increasing the first preset threshold by the original value divided by a preset scaling factor, and increasing the spacing between adjacent monitoring points by a preset fixed step size but not exceeding a preset maximum spacing; after each adjustment, limiting the first preset threshold to between a preset minimum threshold and a preset maximum threshold.

[0016] In a preferred embodiment of the present invention, the updated spacing between adjacent monitoring points is used for the next step of deploying multiple sensors, the updated first preset threshold is used for the next step of constructing a set of similar channel pairs, and a new round of iteration begins; the loop terminates when the fitted slope falls within the preset standard slope range or the number of iterations reaches the preset maximum number of iterations.

[0017] On the other hand, the present invention also provides the following technical solution: a multi-channel signal processing deployment optimization device suitable for substation reactor rooms, comprising: a similarity pair construction module, which deploys multiple sensors to collect signals from each channel, uses one channel as a reference channel, calculates the phase difference of each of the remaining channels relative to the reference channel, and records the relative phase angle of each channel; constructs a set of similar channel pairs based on the comparison result of the phase difference between channels with a first preset threshold and the amplitude ratio between channels; a unit synthesis module, which divides the channels into several differential matrix units based on the set of similar channel pairs, performs phase calibration on the real-time signals in each differential matrix unit, and sums them to generate a unit synthesized signal; and a collection module. The amplitude module reads the synthesized signals of each differential matrix unit from memory in ascending order of the average relative phase angle of all channels within each differential matrix unit, splices them into a continuous data stream, and extracts the voltage amplitude of each differential matrix unit. The slope fitting module performs linear fitting between the unit number and the voltage amplitude in the data stream to obtain the fitting slope. The parameter adjustment module adjusts the first preset threshold according to the deviation of the fitting slope from the preset standard slope range, and adjusts the spacing between adjacent monitoring points according to the direction of the deviation of the fitting slope. The iteration termination module uses the updated spacing between adjacent monitoring points and the updated first preset threshold for the next deployment until the termination condition is met.

[0018] The beneficial effects of this invention are as follows: By constructing similar channel pairs and performing phase calibration and summation, this invention significantly enhances the signal-to-noise ratio and phase consistency of the synthesized signal, effectively suppressing the destructive effect caused by random noise and phase mismatch. Furthermore, based on the sequential splicing and linear fitting of the average phase values ​​within the clustered cells, it achieves closed-loop adaptive optimization of sensor placement spacing and similarity threshold, avoiding redundant sampling or spatial undersampling problems caused by traditional empirical placement. This invention can dynamically adjust placement parameters according to the measured signal characteristics, enabling the spacing between adjacent sensors and phase criteria to quickly converge to the optimal configuration, thereby significantly improving the spatial correlation and amplitude stability of multi-channel signals. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments 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. Wherein: Figure 1 This is a flowchart of the multi-channel signal processing and deployment optimization method applicable to the reactor room of a substation, as described in this invention. Figure 2 This is a schematic diagram of the sensor deployment environment and system module connection in an embodiment of the present invention. Detailed Implementation

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0021] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort should fall within the scope of protection of this invention.

[0022] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0023] According to an embodiment of the present invention, in combination Figures 1-2 As shown, a multi-channel signal processing deployment optimization method suitable for substation reactor rooms includes: S1: Deploy multiple sensors to collect signals from each channel. Use one channel as a reference channel, calculate the phase difference of each of the remaining channels relative to the reference channel, and record the relative phase angle of each channel. Based on the comparison results of the phase difference between channels with the first preset threshold and the amplitude ratio between channels, construct a set of similar channel pairs.

[0024] This embodiment aims to optimize the placement and spacing of sensors in reactor room monitoring to improve the consistency of multi-channel signals.

[0025] S1.1: When implementing this embodiment, the first step is to deploy signal acquisition front-ends in the reactor room of the substation. Specifically, based on the geometric dimensions and electromagnetic field distribution characteristics of the reactor room, an initial spacing between adjacent monitoring points is pre-set. This initial spacing is typically one-tenth to one-fifth of the reactor height; for example, for a reactor with a height of 2 meters, the initial spacing can be set to 0.3 meters. Multiple sensors are then deployed at equal intervals along a straight path along the reactor's axial or circumferential direction, using this initial spacing. As a preferred on-site deployment scenario, the actual structure of a 110kV substation reactor room can be referenced: the reactor room has louvers, with dimensions of 2350mm in height and 550mm in width, and noise mainly leaks outward through the windows. Therefore, the sensor deployment path is preferentially arranged horizontally along the window edges, ensuring that the signal acquisition direction of each channel faces the main noise leakage path. The sensors are used to sense the signals generated by the reactor and convert them into voltage signals for output. Figure 2 This is a schematic diagram of the reactor room and sensor placement environment in an embodiment of the present invention. The reactor room is a closed space with a window (louver) on one of the outer walls. The reactors are arranged inside the room, and the sensors are arranged at equal intervals along the edge of the window.

[0026] Due to strong electromagnetic interference within the reactor room, the sensors often contain high-frequency noise and power frequency harmonic interference. Therefore, the initial signal of each channel is filtered and denoised immediately after acquisition. This embodiment uses a bandpass digital filter, with the passband range set to 0.8 to 1.2 times the reactor's rated frequency. For example, for a 50Hz system, the passband is 40Hz to 60Hz, to retain effective signal components and suppress out-of-band noise. The filtered signal is temporarily stored in the buffer unit of the central acquisition unit for subsequent analysis.

[0027] To ensure that the relative phase relationship between channels remains unchanged before and after filtering, this embodiment uses a linear phase FIR filter, such as an equal-ripple filter based on a Hamming window design, to ensure that the signals in each channel undergo the same filtering process and have consistent group delays, thus keeping the phase difference unaffected. The phase angle calculation is performed after filtering.

[0028] S1.2: After obtaining the filtered initial signals of each channel, it is necessary to quantify and evaluate the phase and amplitude relationships between the channels. This embodiment uses the following sub-steps: (1) Randomly select one channel from all deployed channels as the reference channel. For ease of explanation, this embodiment defaults to setting the first deployed channel, i.e., channel 1, as the reference channel. For each of the remaining channels, the phase difference between the channel signal and the reference channel signal is calculated using the zero-crossing detection method or the phase extraction method based on the fast Fourier transform. Specifically, the waveforms of two channels within the same time window are captured, the time delay difference is determined by the peak position of the cross-correlation function, and then converted into a phase difference; or the phase angles of the fundamental components of the two channels are calculated separately, and the difference is taken. The calculated phase difference value is normalized to be between -180° and 180°. The phase angle of the reference channel itself is recorded as 0°, then the relative phase angle of each channel i is... Defined as: =0, =Calculated phase difference values ​​(for i≥2). This yields a list of relative phase angles for all channels.

[0029] To avoid systematic deviations in the relative phase angle caused by an anomaly in a single reference channel, the reference channel can be selected as the channel corresponding to the median of the phase of all channel signals, or it can be determined by averaging the results of multiple random selections of different reference channels.

[0030] (2) For any two channels j and k (j≠k), first calculate the absolute value of their phase difference. For cases exceeding 180°, a supplementary angle adjustment is performed, i.e., 360° is subtracted from the difference to ensure that the absolute value of the phase difference does not exceed 180°; then the amplitude ratio is calculated. The amplitude ratio, in this context, is the physical meaning of the ratio of the effective values ​​of the two channel signals. Specifically, for the filtered voltage signal of each channel, the root mean square value within a complete power frequency cycle (e.g., 20ms) is first calculated. Let the effective value of channel j be... The effective value of channel k is Then define the amplitude ratio To ensure symmetry, the actual judgment should take... and The larger one is used as the comparison benchmark, which is equivalent to the requirement. It lies between the reciprocal of the preset amplitude range and that range.

[0031] Before performing the similar channel pair judgment, first set the initial values ​​of the system parameters, and the first preset threshold. The initial angle is 15°, and the preset amplitude range is [0.8, 1.2]; the initial value of the distance between adjacent monitoring points is... The reactor height is set to one-tenth to one-fifth of the reactor height, and in this embodiment, it is 0.3 meters.

[0032] (3) For each pair of channels (j,k), simultaneously test the absolute value of the phase difference. ≤ Amplitude ratio If the channel pair is ∈[0.8, 1.2], and both conditions are met, the channel pair is considered a similar channel pair and is recorded in the similar channel pair set. This set is stored in the form of unordered pairs, for example, {(1,3), (2,5), ...}.

[0033] Once the set of similar channel pairs is constructed, if the set is empty (i.e., no pair of channels satisfies the similarity condition), then each channel will become an independent unit in subsequent clustering steps. The handling of this marginal case has been clearly explained in the abstract and will not be repeated here.

[0034] Before performing the above calculations, this embodiment preprocesses the initial signals of all channels by amplitude normalization to eliminate systematic errors caused by differences in the sensitivity of different sensors. Specifically, when installing the sensor in each channel, its factory calibration coefficient is recorded, and the acquired voltage value is divided by this coefficient during the signal processing stage. The normalized signal is then used for phase difference and amplitude ratio calculation, which can improve the accuracy of similarity judgment. Regarding the selection of the sampling window length, this embodiment recommends at least three complete power frequency cycles to ensure the stability of the root mean square value calculation. The central acquisition unit is equipped with a clock synchronization module to ensure that the sampling times of all channels are strictly aligned, with a time base error of less than 1 microsecond, thereby avoiding false phase differences introduced by sampling asynchrony.

[0035] It should be noted that, unlike conventional multi-channel signal processing methods, such as directly averaging or principal component analysis of all channels, this embodiment constructs similar channel pairs using a dual criterion of phase difference threshold and amplitude ratio range, achieving spatially meaningful grouping of correlations. This grouping does not depend on the specific geometry of the reactor chamber or the load size, but is entirely driven by the intrinsic characteristics of the measured signal, exhibiting strong adaptability and robustness.

[0036] S2: Based on the set of similar channel pairs, the channel is divided into several difference matrix units. The real-time signals in each difference matrix unit are phase-calibrated and then summed to generate the unit synthesized signal.

[0037] S2.1: Treat all the channels corresponding to the deployed sensors as a set of vertices V in graph theory, with each channel corresponding to one vertex. Treat the set of similar channel pairs obtained in step S1 as a set of edges E of the graph, where each edge connects two channels that are determined to be similar. This constructs an undirected graph G(V, E). Due to the symmetry of similarity relationships, this graph is undirected.

[0038] Furthermore, a connected component extraction algorithm is used to partition graph G. Specifically, starting from any unvisited vertex, a breadth-first search or depth-first search is performed to traverse all vertices reachable via similar edges; these vertices constitute a connected component. This process is repeated until all vertices have been visited, and each connected component is a difference matrix unit.

[0039] It should be noted that the properties of connected components guarantee that for any two channels within the same unit, there exists a path consisting of similar channel pairs, meaning that the two are directly or indirectly connected through a series of similar relationships, indicating that the signals of these channels have good continuity and consistency in phase and amplitude; and for any two channels between different units, there are no similar channel pairs, meaning that at least one of the phase difference or amplitude ratio between the two does not meet the preset condition, indicating that their signal characteristics are significantly different.

[0040] If the set of similar channel pairs is empty, meaning there are no two channels that satisfy both the phase difference and amplitude ratio conditions, then graph G contains no edges. In this case, each vertex constitutes an independent connected component, and correspondingly, each channel is treated as a separate difference matrix unit.

[0041] S2.2: Since the subsequent phase calibration operation is performed in the digital domain, the analog voltage signal of each channel first needs to be converted into a discrete digital signal. The multi-channel synchronous analog-to-digital converter integrated in the central acquisition unit samples and quantizes the real-time voltage signal of each channel at the sampling rate set in step S1.1 (e.g., 6.4kHz). The quantization accuracy is not less than 12 bits to ensure that the signal amplitude resolution meets the accuracy requirements of phase calibration. The converted digital signal is stored in the buffer of the central acquisition unit in the form of a binary data stream, with each channel corresponding to a set of time series data.

[0042] For a difference matrix cell currently being processed, first obtain the relative phase angles of all channels within that cell that have been calculated and recorded in step S1.2. .

[0043] From all channels within the unit, select the channel with the smallest relative phase angle as the internal reference channel for that unit. It should be noted that the channel with the smallest phase angle is chosen because, in a real reactor room, phase lead usually corresponds to a spatial proximity to the sound source. Using this channel as the reference allows other channels to achieve phase alignment through phase rotation, without requiring additional processing of the reference channel itself. If multiple channels within the unit have the same smallest relative phase angle, any one of them can be randomly selected as the reference, or the first one can be chosen according to the channel number order.

[0044] Furthermore, for each channel within the unit other than the reference channel, a complex twitch factor needs to be calculated. The phase angle of this twitch factor is equal to the negative of the difference between the relative phase angle of the reference channel and the relative phase angle of this channel. Specifically, let the relative phase angle of the reference channel be denoted as... The relative phase angle of the channel to be rotated is Then the phase shift corresponding to the rotation factor is:

[0045] This embodiment employs an analytic signal method based on digital Hilbert transform to achieve phase rotation. Specifically, a digital Hilbert transform is performed on the discrete digital signal x[n] of each channel to obtain orthogonal components. Thus constructing an analytic signal Based on the relative phase angle difference between the reference channel and the channel to be calibrated. Calculate the complex rotation factor ;Analyze the signal Multiply by the rotation factor to obtain the rotated analytic signal. ;Pick The real part is used as the phase-calibrated real signal. This operation precisely achieves phase shifting at any angle in the digital domain without altering the signal's amplitude-frequency characteristics.

[0046] Furthermore, after completing the phase rotation of all non-reference channels, these rotated real signals are... The original digital signal from the reference channel is arithmetically summed with the original signal from the reference channel. Each channel is weighted equally during the summation. This summation operation can be performed either as an accumulation instruction in a digital signal processor or as a parallel adder in a field-programmable gate array (FPGA).

[0047] The numerical sequence obtained after summation is the unit synthesized signal of the difference matrix unit. The synthesized signal retains the main frequency components of the channel signals in the unit. At the same time, since the phase has been aligned with the phase of the reference channel, the signals of each channel are superimposed in phase during summation, and the effective signal amplitude is enhanced. Meanwhile, random noise partially cancels each other out due to the different phases, thereby improving the signal-to-noise ratio.

[0048] In this embodiment, the digital Hilbert transform is implemented using an FIR filter based on a window function design. The filter order is 63, and the passband frequency range is 45Hz~55Hz (corresponding to a power frequency of 50Hz). Hamming window weighting is used. The filtered orthogonal components are aligned with the original signal after delay to construct an analytic signal. To reduce boundary effects, the data segment acquired by each channel is extended by 100 sampling points before and after the transformation, and the effective portion is then truncated.

[0049] If a difference matrix unit contains only one channel, there is no object to perform phase calibration and summation with other channels. In this case, the discrete digital signal after analog-to-digital conversion of that channel is directly output as the unit synthesized signal without any rotation or summation operations.

[0050] Finally, the central data acquisition unit traverses all the difference matrix units obtained in step S2.1, and generates its own unit composite signal for each unit in sequence according to the above process. Each composite signal is marked with the number of its unit and temporarily stored in memory.

[0051] It should be noted that in practical engineering implementation, to improve calculation accuracy, the digital signals of each channel can be de-DC biased before phase rotation by subtracting the average value of the signal over one period to eliminate the effect of zero drift. Furthermore, for the calculation accuracy of the rotation factor, this embodiment uses fixed-point arithmetic, quantizing the phase angle to the 0.1-degree level, and representing the corresponding sine and cosine values ​​using 16-bit integers. This ensures sufficient phase resolution while also considering the computational resource limitations of embedded devices. When the number of channels within a unit is large, such as more than eight, direct summation may result in an excessively large amplitude of the synthesized signal, causing subsequent data overflow. Therefore, this embodiment can scale the synthesized signal after summation by dividing it by the number of channels within the unit, keeping the amplitude of the synthesized signal at the level of a single channel signal. This scaling operation does not affect the subsequent root mean square value calculation and linear fitting results because the signals of all units use the same scaling strategy.

[0052] It should be noted that this invention employs a digital domain complex rotation calibration method based on the channel with the smallest relative phase angle within the unit, achieving precise phase alignment of signals in each channel. Compared to the analog domain variable phase shifter scheme, digital phase rotation has the advantages of high precision, no temperature drift, and reprogrammability. Furthermore, compared to simple direct summation, phase calibration avoids signal cancellation caused by phase inconsistency, ensuring the signal-to-noise ratio gain of the synthesized signal.

[0053] S3: Read the synthesized signals of each differential matrix unit from memory in ascending order of the average relative phase angle of all channels within each differential matrix unit, splice them into a continuous data stream, and extract the voltage amplitude of each differential matrix unit.

[0054] S3.1: All signal lines from the original sensor channels are connected to the corresponding input ports of the central data acquisition unit. After the central data acquisition unit completes the analog-to-digital conversion, it performs differential matrix unit division, phase calibration, and synthesis in the digital domain. The synthesized unit signals are directly stored in memory without involving additional physical transmission.

[0055] To suppress interference, the signal line can be a twisted-pair shielded cable, with the shielding layer grounded at one end on the central acquisition unit side.

[0056] S3.2: Since the synthesized signals of all differential matrix units have already been generated and stored in the digital domain within the central acquisition unit, there is no need for re-acquisition. The central acquisition unit directly reads the time-domain waveform data of the synthesized signals of each unit from memory in ascending order of the average relative phase angle of the original channels within each unit. The data length of each unit is... The corresponding number of sampling points. If the average relative phase angle of multiple difference matrix units is equal, they are sorted from largest to smallest according to the number of original channels within the unit; if the number of channels is also the same, they are sorted from smallest to largest according to the smallest channel number within the unit.

[0057] in, Take an integer multiple of the reactor's power frequency period (e.g., 50Hz corresponds to 0.02 seconds), such as 10 periods, which is 0.2 seconds, to ensure the stability of the effective value calculation.

[0058] These data segments are sequentially concatenated into a continuous data stream, with unit number identifiers and separators inserted into the stream. Then, the root mean square value of each data segment is calculated as the voltage amplitude of the synthesized signal for that unit.

[0059] The average value is the average phase of the signals from each channel within the unit, representing the overall phase delay of the spatial region where the unit is located relative to the reference channel. Assuming that the signal phase monotonically changes with spatial position along the sensor deployment path, if the fitted slope is negative after sorting by the average phase value, it indicates that the sorting direction is opposite to the actual spatial propagation direction. Reversing the sorting order will make the fitted slope positive. This invention automatically corrects the sorting direction by the sign of the fitted slope, without requiring prior knowledge of the absolute spatial order.

[0060] To accurately distinguish the data segments of each unit, during the splicing process, the central data acquisition unit inserts a unit number identifier before each unit's data segment and a separator mark between adjacent unit data segments. The unit number identifier uses a preset fixed-length byte sequence (such as a 16-bit binary number), sequentially assigned starting from 1; the separator mark uses special codewords that will not appear in normal waveform data (such as a sequence of all 1s). The spliced ​​continuous data stream is stored in the central data acquisition unit's buffer.

[0061] It should be noted that the embodiments of the present invention read and splice the data in the order of the average relative phase angle, so that the spatial-temporal mapping relationship of the data stream is clear. In the subsequent linear fitting, the unit sequence number is used as the horizontal axis, which implies the phase change law along the reactor space. The fitting slope can directly reflect the trend of voltage amplitude changing with phase.

[0062] S3.3: After acquiring the continuous data stream, the central data acquisition unit segments the data into independent waveform data segments corresponding to each unit based on the unit number identifier and separator mark. For the data segment of the k-th unit (including...) (Number of sampling points), the effective voltage value is calculated using the root mean square formula: The product of the sampling rate

[0063] in, The voltage value at the nth sampling point of the synthesized signal of the kth unit. for The product of the sampling rate and the sampling rate. This calculation is performed by the digital signal processor built into the central acquisition unit. To avoid accumulated errors, double-precision floating-point accumulation of intermediate results is used, and the root mean square value is finally obtained by taking the square root.

[0064] Furthermore, the calculated The voltage amplitude is used as the synthesized signal of the differential matrix unit. Here, the voltage amplitude is the energy equivalent DC level of the synthesized signal within one complete cycle, and its magnitude is closely related to the effective value of the original signals of each channel within the unit and the phase consistency between channels. After phase calibration, the signals of each channel within the unit are superimposed in phase, and the amplitude of the synthesized signal is approximately equal to the sum of the amplitudes of the original signals of each channel, significantly improving the signal-to-noise ratio.

[0065] In actual data processing, sensor malfunctions or interference may cause abnormal synthesized signals in a certain unit. This embodiment performs a validity check before calculating the root mean square value: calculate the crest factor (peak value / root mean square value). If it exceeds the normal range, such as 1.414 for a normal sine wave and a set range of 1.2~1.8, the data of that unit is determined to be invalid, the data segment is discarded and replaced by interpolation with an adjacent valid unit, or an alarm signal is issued.

[0066] Since the voltage amplitude output by the sensor is proportional to the sound pressure amplitude, this voltage amplitude can be directly used to characterize the noise signal intensity in the spatial region where the unit is located. Ultimately, the voltage amplitudes of all units... The values ​​are stored as an amplitude array in a sequence that is consistent with the reading order, i.e., in the order of the unit numbers.

[0067] S4: Perform a linear fit between the cell number and the voltage amplitude in the data stream to obtain the fitting slope.

[0068] S4.1: The central data acquisition unit parses the cell number and corresponding voltage amplitude of each differential matrix unit from the continuous data stream generated in step S3. The cell number is assigned sequentially according to the reading order determined in step S3.2, i.e., in ascending order of the average relative phase angle of the original channels within each cell, with values ​​ranging from 1, 2, ..., M, where M is the total number of differential matrix units. The voltage amplitude is the root mean square value of the synthesized signal of each unit calculated in step S3.3, denoted as . (k=1,2,...,M).

[0069] Ideally, a two-dimensional Cartesian coordinate system is constructed, with the element number k as the abscissa and the corresponding voltage amplitude as the ordinate. The vertical axis is used as the y-axis. Since the cell numbers are a discrete integer sequence, the points on the horizontal axis are evenly spaced, with a spacing of 1. (Vertical axis) These are continuous positive real numbers, and their magnitude reflects the overall intensity of the noise signal within the spatial region where the corresponding difference matrix unit is located.

[0070] In actual operation, there may be a situation where the total number of difference matrix units M=1, that is, only one unit is obtained after partitioning in step S2.1. When M=1, there is only one data point in the coordinate system, making linear fitting impossible. To address this edge case, this embodiment stipulates that when M=1, if the number of iterations has reached the maximum number of iterations, the current configuration is output and a warning is issued; otherwise, the first preset threshold is increased to a preset multiple of the original value, and the process returns to step S1. In this case, the number of iterations does not increase, that is, the process after threshold adjustment is re-executed under the same number of iterations. The iteration counter increments by 1 only after completing a complete round of parameter adjustment, that is, after executing the parameter modification in S5.2 or S5.3 and jumping.

[0071] It should be noted that, based on the spatial correlation of the signal field inside the reactor, when there is only one differential matrix unit, it indicates that the current phase difference threshold is too strict, causing all channels to be judged as dissimilar. To avoid slow convergence due to excessively small threshold adjustment during iteration, and to prevent the threshold from exceeding the optimal value due to excessive adjustment, experimental statistics show that multiplying the threshold by 1.5 can bring the threshold to a reasonable range within 3 to 5 iterations. This scaling factor can also be adjusted between 1.2 and 2.0 according to the actual scenario; this embodiment uses 1.5 as an example for explanation.

[0072] S4.2: When the total number of difference matrix units M≥2, the central data collector collects M data points (k, Linear fitting is performed. In this embodiment, the least squares method is used as the fitting algorithm. This algorithm has the advantages of simple calculation, unbiased estimation, and optimal error for normal distribution. It is suitable for the actual situation where the voltage amplitude measurement value contains random noise in this method.

[0073] Specifically, the horizontal axis is = i (i=1,2,...,M), with the ordinate as The slope of the fitted line The calculation formula is:

[0074]

[0075]

[0076] in, The arithmetic mean of the x-axis. The arithmetic mean of the voltage amplitude. Let be the effective voltage value of the synthesized signal of the i-th unit. To efficiently implement this calculation in an embedded system and avoid multiple calls to floating-point division, an incremental recursive formula or a lookup table method can be used. As a preferred implementation, the central data acquisition unit first calculates the average value of the horizontal axis. Then, iterate through all i, accumulating the numerator and denominator terms. The denominator term depends only on M, and can be pre-calculated and stored in a constant table. After accumulation, perform a division operation to obtain the slope. .when When positive, it indicates that the voltage amplitude increases with the increase of the cell number (corresponding to an increase in phase lag); when... When it is negative, it indicates that the voltage amplitude is decreasing; when When the slope is close to 0, it indicates that the voltage amplitude remains essentially constant. The absolute value of the slope reflects the degree of non-uniformity in the signal field distribution within the reactor chamber. The larger the absolute value, the more significant the energy difference between different spatial regions, which may suggest that the current deployment spacing or similarity criterion settings are unreasonable and require optimization and adjustment.

[0077] Preferably, to improve the reliability of the fitting results, this embodiment calculates the slope... At the same time, the coefficient of determination is also calculated. The coefficient of determination is used to evaluate the goodness of fit of a linear model. The formula for calculating the coefficient of determination is:

[0078] in, The intercept of the fitted line can be obtained from... get. The value range of is [0,1]. The closer the value is to 1, the more closely the data points are distributed around the fitted line, and the stronger the linear relationship; if... If the value is below a preset threshold (e.g., 0.7), it indicates that there may not be a significant linear relationship between the voltage amplitude and the unit number. In this case, the central data acquisition unit can issue a warning message, prompting the user to check the signal quality or increase the sampling time. It should be noted that... It is used only to assist in diagnosis and does not participate in subsequent optimization decisions.

[0079] The calculated fitted slope Stored in a designated register of the central data collector, marked as This slope value will serve as the core criterion for subsequent threshold adjustment and deployment spacing optimization. Simultaneously, the central data acquisition unit can send the fitting results to the host computer monitoring system via a communication interface (e.g., RS-485, Ethernet, or wireless module) for real-time display and recording.

[0080] Optionally, in actual reactor operation, the voltage amplitude measurement accuracy of different differential matrix units may vary due to factors such as sensor aging and local temperature differences. To further improve the accuracy of the fitting, this embodiment optionally assigns a weighting coefficient to each data point using a weighted least squares method. For example, the reciprocal of the number of channels in the unit can be used, or the signal-to-noise ratio (SNR) estimate of the synthesized signal in the unit can be taken. In this embodiment of the invention, the SNR estimation is performed using the following method: for the synthesized signal of each unit, the total energy is calculated, and the ratio of the residual energy between the original signal and the synthesized signal of each channel in the unit is calculated; or the ratio of the energy of the autocorrelation function of the synthesized signal at non-zero delay to the energy at zero delay is used. Specifically, for the i-th unit, the SNR is:

[0081] in, For the fundamental component extracted by the bandpass filter, the weights are... Pick The normalized value, or the reciprocal of the number of channels in the cell, can be used as a simplification method.

[0082] Furthermore, the formula for calculating the slope of the weighted fit is:

[0083]

[0084] Weighted processing can reduce the adverse effects of low-quality data points on the fitting results. Before fitting, this embodiment can also perform a gross error detection. Using the Laida criterion (3σ criterion), the mean and standard deviation of all voltage amplitudes are calculated. If the 3σ criterion is met, the point is determined to be an outlier and removed from the fitting dataset. After removal, if the number of remaining data points is still ≥2, the remaining points are used for fitting; if the number of remaining points is less than 2, the process switches to single-unit edge case handling and terminates the optimization process. This operation can effectively eliminate outliers caused by transient electromagnetic pulses or hardware failures, improving the robustness of the fitting slope.

[0085] As can be seen, this invention compresses discrete, multi-unit voltage amplitude information into a fitting slope, which greatly reduces the computational complexity of subsequent optimization decisions. Compared with directly analyzing all amplitude data, using the slope as a criterion allows threshold adjustment and spacing optimization to be completed with only a numerical comparison.

[0086] S5: Adjust the first preset threshold according to the deviation between the fitted slope and the preset standard slope range, and adjust the spacing between adjacent monitoring points according to the direction of the deviation of the fitted slope.

[0087] Before performing this step, based on the design parameters of the reactor chamber, such as the reactor's rated power, dimensions, and allowable energy distribution non-uniformity, a standard slope range is pre-defined, denoted as . ,in and All are positive numbers, and < When the fitting slope is within this interval, it indicates that the rate of change of the voltage amplitude of each differential matrix unit with the spatial phase order under the current deployment scheme is within an acceptable engineering range, the signal field distribution is relatively uniform, and no further optimization is needed; when the fitting slope is higher than this interval, it indicates that the voltage amplitude of each differential matrix unit with the spatial phase order is within an acceptable engineering range, the signal field distribution is relatively uniform, and no further optimization is needed. When the voltage amplitude increases too rapidly with increasing cell number, it indicates that the energy in the phase lag region is abnormally high, possibly due to insufficient spatial sampling caused by excessively large sensor spacing or an overly broad similarity criterion; when the fitting slope is lower than This indicates that the voltage amplitude decreases too rapidly as the unit number increases, meaning the energy in the phase-leading region is too high and the energy in the lagging region is too low. This may be due to signal redundancy caused by excessively small sensor spacing or overly strict similarity criteria. Specific values ​​can be obtained through offline simulation or on-site calibration tests. As a preferred embodiment of this practice, for a typical power frequency reactor room, a setting can be made... = 0.02, = 0.08, the unit is volts per unit number. This value is only for illustrative purposes and can be set according to actual needs. The embodiments of this invention only provide typical values.

[0088] S5.1: The central data acquisition unit will use the fitted slope obtained in step S4. Compared with the preset standard slope range Compare the results and execute one of the following branches: First branch: ∈ At this point, it is determined that the current deployment parameters already meet the optimization objective, and no further adjustments are needed. The central data acquisition unit terminates the iterative loop and outputs the currently used adjacent monitoring point spacing, the first preset threshold, and other relevant parameters as the final optimization result. The output format can be a human-machine interface display, data file storage, or uploading to the monitoring system via a communication interface; simultaneously, the central data acquisition unit sends an optimization completion prompt signal to the operator; Second branch: > At this point, if the fitting slope is higher than the upper limit of the standard range, it indicates that the voltage amplitude increases too rapidly with the unit number. It is necessary to simultaneously tighten the similarity criterion to make subsequent clustering more refined, and reduce the sensor placement spacing to increase spatial sampling density; Third branch: < If the fitting slope is found to be below the lower limit of the standard range, it indicates that the voltage amplitude decays too quickly with the unit number. It is necessary to relax the similarity criterion to allow more channels to be assigned to the same unit and increase the sensor placement spacing to expand the spatial sampling range.

[0089] S5.2: For the second branch, reduce the first preset threshold to its original value. Multiply by a preset scaling factor ,in For positive numbers less than 1, that is: For the third branch, the first preset threshold is increased to its original value divided by the same preset scaling factor β (equivalent to multiplying by 1 / β), that is: .in, The typical value range is 0.5 to 0.9. The specific value is set by the user during system initialization based on the size of the reactor chamber and the electromagnetic environment, or a default value of 0.8 can be used. A larger value results in a smoother threshold adjustment; a smaller value leads to faster convergence but may cause oscillations. A smaller value... Values ​​(such as 0.5) can achieve fast convergence, but may cause oscillations; larger values... Values ​​(such as 0.95) converge smoothly but slowly. The specific values ​​can be set during system initialization based on the actual size of the reactor room and the environment; this embodiment does not impose any restrictions.

[0090] It should be noted that the preset multiple when M=1 is only used to solve the startup problem where there are no units that can be clustered. Once normal iteration begins (M≥2), all subsequent iterations will use the multiple. The adjustment, where M=1, is only used as the initial value for the next iteration and does not participate in the subsequent iterations. The cumulative effect of adjustments. It is also stipulated that two types of adjustments will not occur simultaneously in the same iteration round.

[0091] Furthermore, to prevent the first preset threshold from diverging during the iteration process and losing its physical meaning—for example, being too small to identify any similar channel pairs, or too large to classify all channel pairs as similar—this embodiment adjusts the threshold. Apply boundary constraints: Preset a minimum threshold, for example, 1°. When the calculated... When <1°, forced removal = 1°. This lower bound ensures that even under the most stringent conditions, channel pairs with a phase difference of less than 1° can still be identified as similar, preventing algorithm degradation due to empty sets; a maximum threshold is preset, for example, 45°. When the calculated... When the angle is >45°, forced removal = 45°. This upper limit prevents physically dissimilar channels (such as inverted signals) from being incorrectly grouped into the same cell due to excessively large thresholds. The specific values ​​of the minimum and maximum thresholds should be preset before the method is executed based on the reactor's operating frequency (e.g., phase resolution corresponding to 50Hz) and engineering experience, and stored in the non-volatile memory of the central acquisition unit.

[0092] It can be seen that by linking the adjustment of the first preset threshold with the direction and degree of deviation of the fitting slope, adaptive closed-loop control of the similarity criterion is achieved. When the fitting slope is too high, decreasing the threshold makes the phase consistency requirement more stringent, and the phases between channels within the clustered cells are closer, thereby suppressing spurious amplitude changes caused by excessively large cell spans. When the fitting slope is too low, increasing the threshold allows channels with slightly larger phase differences to be merged, avoiding statistical fluctuations caused by excessively small cells. This bidirectional adjustment capability enables the method to automatically adapt to the energy distribution characteristics of different reactor chambers, significantly improving the robustness of deployment optimization.

[0093] S5.3: For the second branch, output the first deployment optimization instruction, which instructs to reduce the spacing between adjacent monitoring points by a preset fixed step size. Let the current spacing be D, then: ,in, This represents the current distance between adjacent monitoring points. This represents the adjusted spacing. The purpose of reducing the spacing is to increase the density of spatial sampling points, i.e., to place more sensors within the same length, thereby improving the ability to capture spatial changes in the signal field and suppressing the fitting slope.

[0094] For the third branch, a second deployment optimization instruction is output, which instructs that the spacing between adjacent monitoring points be increased by a preset fixed step size. ,Right now: The purpose of increasing the spacing is to reduce redundant sampling and avoid the inability to reflect the true spatial change trend due to the high similarity of signals in adjacent channels caused by overly dense sensors.

[0095] in, Typical values ​​can be set based on the overall dimensions of the reactor chamber and the number of sensors. For example, if the reactor height is 2 meters and the initial spacing is 0.3 meters, then the following values ​​can be set: = 0.05 meters (5 centimeters). It should not be too small, otherwise convergence will be slow; it should not be too large, otherwise it may skip the optimal solution.

[0096] The preset minimum and maximum spacing can be pre-set according to the size of the reactor chamber and the physical dimensions of the sensor. For example, for a reactor chamber with a height of 2 meters, the minimum spacing can be set to 0.1 meters and the maximum spacing to 0.5 meters. When the adjusted spacing is lower than the minimum spacing, the minimum spacing will be forced to be used; when it is higher than the maximum spacing, the maximum spacing will be forced to be used. The specific values ​​can be set according to actual needs; only an example is given here.

[0097] By adjusting the sensor spacing in the opposite direction based on the deviation of the fitted slope, a negative feedback control loop is constructed: a high slope leads to a smaller spacing, resulting in denser spatial sampling and a lower fitted slope in the next iteration; a low slope leads to a larger spacing, resulting in sparser spatial sampling and a higher fitted slope in the next iteration. This closed-loop mechanism automatically drives the deployment scheme to approach the optimal spacing without requiring repeated manual trial and error.

[0098] S5.4: After completing the adjustment of the first preset threshold and the spacing between adjacent monitoring points, the central data acquisition unit will update the... and The updated parameters are stored in a local register, overwriting the old values ​​used in the previous iteration. The updated parameters will be used for the configuration in the next execution of step S1.

[0099] After the parameter update is completed, the central acquisition unit performs an unconditional jump, returns to the starting step of this method, and begins a new round of iteration. In the new round of iteration, the central acquisition unit rereads the signals of each channel using the updated deployment spacing, re-judges similar channel pairs using the updated first preset threshold, and executes steps S2, S3, and S4 in sequence to recalculate the fitting slope and proceed to step S5 for comparison.

[0100] The iterative process will continue until any of the following termination conditions are met: In the first branch of step S5.1, the fitting slope falls within the preset standard slope range, at which point the method has obtained a deployment configuration that meets the engineering requirements. As a typical result, in the actual optimization of the reactor room of a 110kV substation, when the distance between adjacent monitoring points converges to 0.2m and the first preset threshold converges to 8°, the fitting slope is 0.045 V / number, which is within the preset standard slope range [0.02, 0.08]. At this time, the sensor array covers the entire effective area of ​​the louvers, and the voltage amplitude of each differential matrix unit changes uniformly with the spatial phase.

[0101] The preset maximum number of iterations is typically 15, but not less than 3. When the maximum number of iterations is reached, the central data collector forcibly terminates the iteration and outputs the parameter whose most recent fitting slope is closest to the standard range as the optimization result. At the same time, it issues an alarm message indicating that the maximum number of iterations has been reached and the optimization has not fully converged.

[0102] S6: Use the updated spacing between adjacent monitoring points and the updated first preset threshold for the next deployment until the termination condition is met.

[0103] S6.1: Updated spacing between adjacent monitoring points This needs to be converted into actual sensor position adjustment commands. Based on the interface type between the central data acquisition unit and the sensor deployment mechanism, this embodiment provides the following two implementation methods: In the automated deployment mode, the central data acquisition unit sends position commands to the electric slide or robotic arm via a digital control interface, such as a stepper motor driver interface or an industrial fieldbus. These commands include the target spacing value. And the coordinates of the reference starting point. The actuator drives each sensor to move along a preset guide rail according to instructions, ensuring that the physical distance between adjacent sensors is precisely equal to... After the movement is completed, the actuator sends a confirmation signal to the central data collector. To ensure adjustment accuracy, this embodiment uses closed-loop control, that is, the actual position of the displacement sensor is fed back in real time, compared with the command value, and the error is corrected until the deviation is less than the preset tolerance (e.g., ±1 mm).

[0104] Manual deployment method: The central data collector will update the spacing. The information is displayed on a human-machine interface, such as an LCD screen or touchscreen, and accompanied by audio and visual prompts, guiding the operator to readjust the sensor position according to the new spacing. After the operator completes the adjustment, they can confirm the adjustment by pressing a confirmation button to notify the central data acquisition unit to continue with the subsequent steps. This method is suitable for situations without automatic actuators or during laboratory commissioning.

[0105] The first preset threshold is a software parameter used for similar channel pair judgment in step S1.2, and its update does not involve hardware changes. The central acquisition unit will use the threshold calculated in step S5. The specified variable address is written to the internal random access memory, overwriting the old threshold. In subsequent iterations of step S1.2, the similarity channel directly reads the current value of this variable as a comparison benchmark for the judgment logic. To ensure that parameter updates are not lost after a power outage, the central data collector simultaneously... Backup to non-volatile storage (such as EEPROM or flash memory) so that the device can directly use the best threshold obtained from the last optimization after restarting, without having to iterate again.

[0106] To prevent abnormal interruptions during parameter updates, such as power outages or hardware failures that could cause old and new parameters to become mixed, this embodiment employs a transaction mechanism: ... and The storage operation is executed as an atomic transaction. The update flag is only set to valid after both parameters have been successfully written to the target storage medium; if either parameter fails to write, the original parameter remains unchanged, an error log is logged, and the system waits for a reset or manual intervention. This mechanism guarantees the atomicity of paired parameter updates, preventing abnormal behavior in subsequent iterations due to parameter inconsistencies.

[0107] S6.2: After completing parameter passing and configuration, the central data acquisition unit performs an unconditional jump, returning to the starting step of this method. Specifically, the central data acquisition unit resets the program counter to the entry address of step S1, or re-executes the main control loop function via a function call. Before the jump, the central data acquisition unit needs to perform the following cleanup and increment operations, including clearing the intermediate data cache generated in steps S2, S3, and S4 to avoid mixing old and new data; incrementing the iteration counter by 1; and reinitializing the analog-to-digital converter, timer, and communication interface to ensure the timing accuracy of the new round of reading.

[0108] After the jump, the central data collector uses the updated sensor spacing. Re-execute the sensor deployment. If the automatic deployment has already completed the physical movement in the previous step, there is no need to repeat the movement, but the signal needs to be read again. Then, the steps of signal reading, similarity judgment, clustering, synthesis, transmission, amplitude extraction, and linear fitting are executed in sequence. Then, step S5 is entered again to compare the slope and adjust the parameters.

[0109] S6.3: The loop process is not infinite. After each parameter adjustment in step S5, before jumping to the next iteration, and when a single cell edge is detected in step S4, the central data collector needs to determine whether the termination condition is met. This embodiment defines the following two independent termination conditions: (1) When the first branch of step S5.1 is executed, the central data acquisition unit immediately determines that the optimization goal has been achieved. At this time, no parameter adjustment is performed, nor are the parameter updates and jumps in steps S6.1 and S6.2 executed. Instead, it directly enters the output stage. The satisfaction of this condition indicates that the current deployment spacing and similarity threshold have made the rate of change of voltage amplitude with unit number within the acceptable range for engineering, and further iteration is unnecessary.

[0110] (2) In order to prevent the method from failing to converge and falling into an infinite loop due to improper parameter settings, abnormal electromagnetic environment or sensor failure, this embodiment presets a maximum allowed number of iterations. If the iteration count is greater than the maximum allowed number of iterations, the loop is forcibly terminated. If M=1 twice in a row and the first preset threshold has reached the upper limit, the loop is forcibly terminated, and the parameter that most recently satisfies the fitting slope closest to the standard range is output as the suboptimal solution, and a warning message is issued at the same time.

[0111] In real-world engineering environments, the mechanical movement of a deployment mechanism can take anywhere from several seconds to tens of seconds. The central data acquisition unit sets a timeout timer for each physical operation; for example, the maximum waiting time for confirmation of arrival after a movement command is 30 seconds. If no confirmation signal is received within the timeout period, the actuator is considered faulty, the loop is terminated, and an error is reported. Similarly, corresponding software timeouts are set for signal reading and data processing steps to prevent indefinite waiting due to task deadlock. Figure 2 As shown, in this embodiment, multiple sensors are placed at equal intervals along the edge of the reactor's outdoor window. The sensor signals are processed by the central acquisition unit and sequentially go through similar channel pair construction, differential matrix unit division, phase calibration and synthesis, amplitude extraction and fitting. Finally, the parameter optimization module adjusts the first preset threshold and the distance between adjacent monitoring points, and realizes iterative looping through feedback arrows until the optimization termination condition is met.

[0112] It should be noted that this invention realizes the closed-loop transmission of optimization results to the actuator. Whether it is automatic or manual deployment, the updated spacing can physically change the sensor position, while the updated threshold can software-change the subsequent judgment logic. The two work together to drive the evolution of the optimization process.

[0113] The present invention also includes a multi-channel signal processing and deployment optimization device suitable for substation reactor rooms, comprising: The similarity pair construction module deploys multiple sensors to read signals from each channel. Taking one channel as a reference channel, it calculates the phase difference of each of the remaining channels relative to the reference channel and records the relative phase angle of each channel. Based on the comparison results of the phase difference between channels with a first preset threshold and the amplitude ratio between channels, a set of similar channel pairs is constructed. The unit synthesis module divides the channel into several difference matrix units based on the set of similar channel pairs. After performing phase calibration on the real-time signals in each difference matrix unit, the signals are summed to generate the unit synthesized signal. The bus amplitude module reads the synthesized signals of each differential matrix unit from memory in ascending order of the average relative phase angle of all channels within each differential matrix unit, splices them into a continuous data stream, and extracts the voltage amplitude of each differential matrix unit. The slope fitting module performs a linear fit between the cell number and the voltage amplitude in the data stream to obtain the fitting slope. The parameter adjustment module adjusts the first preset threshold according to the deviation between the fitted slope and the preset standard slope range, and adjusts the spacing between adjacent monitoring points according to the direction of the deviation of the fitted slope. The iteration termination module uses the updated spacing between adjacent monitoring points and the updated first preset threshold for the next deployment until the termination condition is met.

[0114] The device also includes one or more processors and memory.

[0115] The memory is used to store operable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations, including the flow of the multi-channel signal processing deployment optimization method for substation reactor rooms as described in the foregoing embodiments, particularly... Figure 1 The flowchart of the method is shown.

[0116] Other aspects disclosed in the embodiments of the present invention also propose a computer-readable medium for storing software including instructions executable by one or more computers, which, upon execution, cause the one or more computers to perform operations including the flow of the multi-channel signal processing deployment optimization method applicable to substation reactor rooms as described in the foregoing embodiments, particularly... Figure 1 The flowchart of the method is shown.

[0117] It should be recognized that embodiments of the present invention may be implemented or carried out by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer-readable storage medium.

[0118] The method can be implemented using standard programming techniques, including a non-transitory computer-readable storage medium configured with a computer program in the computer program, wherein the storage medium is configured such that the computer operates in a specific and predefined manner.

[0119] Each program can be implemented in a high-level procedural or object-oriented programming language to communicate with the computer system; however, if required, the program can be implemented in assembly or machine language.

[0120] In any case, the language can be either compiled or interpreted.

[0121] Furthermore, for this purpose, the program can run on programmed application-specific integrated circuits.

[0122] The processes described herein (or variations and / or combinations thereof) can be executed under the control of one or more computer systems configured with executable instructions, and can be implemented by hardware or a combination thereof as code (e.g., executable instructions, one or more computer programs, or one or more applications) that commonly executes on one or more processors. The computer program includes a plurality of instructions executable by one or more processors.

[0123] Furthermore, the method can be implemented in any suitable computing platform, including but not limited to personal computers, minicomputers, mainframes, workstations, networked or distributed computing environments, standalone or integrated computer platforms, or in communication with charged particle tools or other imaging devices.

[0124] Various aspects of the present invention can be implemented in machine-readable code stored on a non-transitory storage medium or device, whether portable or integrated into a computing platform, such as a hard disk, optical read and / or write storage medium, RAM, ROM, etc., such that it can be read by a programmable computer, and when the storage medium or device is read by the computer, it can be used to configure and operate the computer to perform the processes described herein.

[0125] Furthermore, machine-readable code, or parts thereof, can be transmitted via wired or wireless networks.

[0126] When such media includes instructions or programs that combine with a microprocessor or other data processor to implement the steps described above, the invention described herein includes these and other different types of non-transitory computer-readable storage media.

[0127] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for optimizing the deployment of multi-channel signal processing in substation reactor rooms, characterized by: include: Multiple sensors are deployed to collect signals from each channel. One channel is used as a reference channel. The phase difference of each of the remaining channels relative to the reference channel is calculated, and the relative phase angle of each channel is recorded. Based on the comparison results between the phase difference between channels and the first preset threshold, as well as the amplitude ratio between channels, a set of similar channel pairs is constructed. Based on the set of similar channel pairs, the channel is divided into several difference matrix units. The real-time signals in each difference matrix unit are phase-calibrated and then summed to generate the unit synthesized signal. The synthesized signals of each unit are read from memory in ascending order of the average relative phase angle of all channels in each differential matrix unit, spliced ​​into a continuous data stream, and the voltage amplitude of each differential matrix unit is extracted. A linear fit is performed between the cell number and the voltage amplitude in the data stream to obtain the fitting slope; The first preset threshold is adjusted according to the deviation between the fitted slope and the preset standard slope range, and the spacing between adjacent monitoring points is adjusted according to the direction of the deviation of the fitted slope. The updated spacing between adjacent monitoring points and the updated first preset threshold are used for the next deployment until the termination condition is met.

2. The multi-channel signal processing and deployment optimization method for substation reactor rooms as described in claim 1, characterized in that: The deployment of multiple sensors to collect signals from each channel includes: Multiple sensors are placed at equal intervals in the reactor room according to the preset initial adjacent monitoring point spacing. The signals of each channel are collected synchronously at a preset sampling rate, and the signals of each channel are filtered and denoised.

3. The multi-channel signal processing and deployment optimization method for substation reactor rooms as described in claim 2, characterized in that: The construction of the similar channel pair set includes: Using one of the channels as a reference channel, calculate the phase difference of each of the remaining channels relative to the reference channel and record the relative phase angle of each channel; Calculate the absolute value of the phase difference and the amplitude ratio between any two channels. The amplitude ratio is the ratio of the effective values ​​of the two channel signals. All channel pairs whose absolute phase difference is less than a first preset threshold and whose amplitude ratio is within a preset amplitude range are recorded as a set of similar channel pairs.

4. The multi-channel signal processing and deployment optimization method for substation reactor rooms as described in claim 3, characterized in that: The method of dividing channels into several difference matrix units based on a set of similar channel pairs includes: The connected component extraction algorithm is used to divide the channel into several difference matrix units. There is a path consisting of similar channel pairs between any two channels in each difference matrix unit, and there are no similar channel pairs between different difference matrix units. When no similar channel pairs exist, each channel is treated as an independent unit.

5. The multi-channel signal processing and deployment optimization method for substation reactor rooms as described in claim 4, characterized in that: The signal synthesized by the generation unit includes: For each differential matrix unit, the real-time voltage signal of each channel is first converted from analog to digital to obtain a discrete digital signal; Based on the recorded relative phase angle of each channel, determine the channel with the smallest relative phase angle within the unit, and use the channel as the reference channel within the unit; For each channel in the unit other than the reference channel, the discrete digital signal is transformed by Hilbert in the digital domain to construct an analytic signal. The analytic signal is then multiplied by a complex rotation factor and the real part is taken to achieve phase rotation. The discrete digital signal of the reference channel within the unit is summed with all the rotated digital signals of other channels using equal weights to generate the unit composite signal. When there is only one channel in the unit, the discrete digital signal of the channel is directly used as the synthesized signal.

6. The multi-channel signal processing and deployment optimization method for substation reactor rooms as described in claim 5, characterized in that: The phase angle of the complex rotation factor is equal to the negative of the difference between the relative phase angle of the reference channel and the relative phase angle of the channel, so that the phase of the channel signal after rotation is consistent with the phase of the reference channel.

7. The multi-channel signal processing and deployment optimization method for substation reactor rooms as described in claim 1, characterized in that: The fitting slope includes: A Cartesian coordinate system is constructed with the differential matrix units corresponding to each synthesized signal in the data stream numbered sequentially as the abscissa and the voltage amplitude of each synthesized signal as the ordinate. When there is only one difference matrix unit, if the number of iterations has reached the maximum number of iterations, the current configuration is output and a warning is issued; otherwise, the first preset threshold is increased to a preset multiple of the original value, and the process jumps directly to the next step of deploying multiple sensors, skipping the subsequent linear fitting and parameter adjustment steps in this round. When there are at least two difference matrix units, perform a linear fit on all data points and calculate the fit slope.

8. The multi-channel signal processing and deployment optimization method for substation reactor rooms as described in claim 1, characterized in that: When there are at least two difference matrix units, adjusting the spacing between adjacent monitoring points according to the deviation direction of the fitting slope includes: Compare the fitted slope with a preset standard slope range; When the fitted slope is within the preset standard slope range, the iteration is terminated and the current placement parameters are output as the optimization result. When the fitting slope is higher than the upper limit of the preset standard slope range, the currently used first preset threshold is multiplied by the preset scaling factor to become the new first preset threshold, and the distance between adjacent monitoring points is reduced by a preset fixed step size but not lower than the preset minimum distance. When the fitting slope is lower than the lower limit of the preset standard slope range, the first preset threshold is increased to the original value divided by the preset scaling factor, and the distance between adjacent monitoring points is increased by a preset fixed step size but not exceeding the preset maximum distance. After each adjustment, the first preset threshold is limited to between the preset minimum threshold and the preset maximum threshold.

9. The multi-channel signal processing and deployment optimization method for substation reactor rooms as described in claim 8, characterized in that: The updated spacing between adjacent monitoring points will be used in the next step of deploying multiple sensors, and the updated first preset threshold will be used in the next step of building a set of similar channel pairs, and a new round of iteration will begin. The loop terminates when the fitted slope falls within the preset standard slope range or the number of iterations reaches the preset maximum number of iterations.

10. A multi-channel signal processing deployment optimization device for substation reactor rooms, based on the multi-channel signal processing deployment optimization method for substation reactor rooms according to any one of claims 1 to 9, characterized in that: Also includes: The similar pair construction module deploys multiple sensors to collect signals from each channel. Using one channel as a reference channel, it calculates the phase difference of each of the remaining channels relative to the reference channel and records the relative phase angle of each channel. Based on the comparison results between the phase difference between channels and the first preset threshold, as well as the amplitude ratio between channels, a set of similar channel pairs is constructed. The unit synthesis module divides the channel into several difference matrix units based on the set of similar channel pairs. After performing phase calibration on the real-time signals in each difference matrix unit, the signals are summed to generate the unit synthesized signal. The bus amplitude module reads the synthesized signals of each differential matrix unit from memory in ascending order of the average relative phase angle of all channels within each differential matrix unit, splices them into a continuous data stream, and extracts the voltage amplitude of each differential matrix unit. The slope fitting module performs a linear fit between the cell number and the voltage amplitude in the data stream to obtain the fitting slope. The parameter adjustment module adjusts the first preset threshold according to the deviation between the fitted slope and the preset standard slope range, and adjusts the spacing between adjacent monitoring points according to the direction of the deviation of the fitted slope. The iteration termination module uses the updated spacing between adjacent monitoring points and the updated first preset threshold for the next deployment until the termination condition is met.

Images