Building intelligent electrical fire monitoring method and system

Abstract

The invention discloses a building intelligent electrical fire monitoring method and system, and relates to the field of fire monitoring, and the method comprises the steps: obtaining distribution box parameters of different floors in a target monitoring area in a preset time period, and the distribution box parameters comprise a leakage current value and a temperature value; performing multi-scale decomposition on the parameters of the distribution box to obtain a plurality of intrinsic mode functions and residual trends; performing feature extraction on each intrinsic mode function and the residual trend to obtain an electrical fire feature vector; constructing a fire risk assessment model according to the electrical fire feature vector; and predicting a fire risk level according to the fire risk assessment model. Aiming at neglecting a nonlinear relation between leakage current and temperature of a distribution box in the prior art, the building electrical fire monitoring precision is improved based on multivariable time-delay analysis, ensemble empirical mode decomposition and the like.

Description

Technical Field

[0001] The present application relates to the field of fire monitoring, and in particular to a method and system for intelligent electrical fire monitoring in a building. Background Art

[0002] Traditional building electrical fire monitoring methods mainly rely on regular manual inspections and simple threshold alarm systems, which are lagging and passive. With the development of the Internet of Things and artificial intelligence technologies, the new generation of electrical fire monitoring systems have begun to use intelligent sensor networks to collect electrical parameters in real time, but there are still technical bottlenecks in data processing and risk assessment, especially the limited ability to extract features and intelligently identify complex electrical signals.

[0003] On the one hand, traditional monitoring systems usually monitor and analyze parameters such as leakage current and temperature independently, ignoring the nonlinear coupling relationship between these parameters. In fact, before electrical equipment fails, there is often a complex time delay and mutually reinforcing relationship between leakage current and temperature anomalies. This nonlinear relationship is a key indicator for early warning of electrical fires. On the other hand, existing monitoring systems generally use traditional signal processing methods such as Fourier transform or wavelet analysis to perform frequency domain analysis on electrical parameters. These methods have obvious limitations when dealing with non-stationary and nonlinear electrical parameters, and it is difficult to accurately extract fault features. In particular, although the traditional empirical mode decomposition (EMD) technology can be used for non-stationary signal analysis, it is prone to modal aliasing, which causes electrical fire features to be incorrectly dispersed into different modal functions, reducing the accuracy of fault identification.

[0004] For example, the relevant patent CN118781734A discloses a method for intelligent electrical fire monitoring in a building, including: obtaining the leakage current value and temperature value of the distribution box at multiple predetermined time points on different floors in the target monitoring area within a predetermined time period; arranging the leakage current value and temperature value of the distribution box at multiple predetermined time points on different floors according to the time dimension and the sample dimension to obtain the electrical fire monitoring parameter time series input tensor; feature encoding the electrical fire monitoring parameter time series input tensor to obtain the electrical fire monitoring parameter dynamic change time series feature diagram; feature emphasizing the electrical fire monitoring parameter dynamic change time series feature diagram to obtain the electrical fire monitoring emphasized feature diagram; based on the electrical fire monitoring emphasized feature diagram, obtaining the alarm result. However, this scheme arranges the leakage current value and temperature value according to the time dimension and the sample dimension, ignoring the nonlinear time series correlation between the two, so the monitoring accuracy of this scheme needs to be further improved. Summary of the invention

[0005] In view of the fact that the prior art ignores the nonlinear relationship between the leakage current and temperature of the distribution box, the present application provides a building intelligent electrical fire monitoring method and system, which improves the accuracy of building electrical fire monitoring by multivariate time-lag analysis and ensemble empirical mode decomposition.

[0006] One aspect of the present application provides a method for intelligent electrical fire monitoring in a building, comprising: S1, obtaining distribution box parameters of different floors in a target monitoring area within a preset time period, wherein the distribution box parameters include leakage current values ​​and temperature values; S2, performing multi-scale decomposition on the distribution box parameters to obtain multiple intrinsic mode functions and residual trends; S3, performing feature extraction on each intrinsic mode function and residual trend to obtain an electrical fire feature vector; S4, constructing a fire risk assessment model based on the electrical fire feature vector; S5, predicting the fire risk level based on the fire risk assessment model.

[0007] Furthermore, the intrinsic mode function represents the different frequency oscillation components in the distribution box parameters and the corresponding time-varying characteristics; the residual trend represents the long-term variation baseline and overall evolution trend of the distribution box parameters.

[0008] Furthermore, S2 performs multi-scale decomposition on the distribution box parameters to obtain multiple intrinsic mode functions and residual trends, including: S21, constructing a multivariate time series matrix of leakage current values ​​and temperature values, and calculating the time-lag mutual correlation coefficients of the leakage current values ​​and temperature values; S22, determining the coupling time window between the leakage current value and the temperature value according to the peak value of the time-lag mutual correlation coefficient; combining the leakage current value and the temperature value into a two-dimensional vector signal according to the coupling time window; S23, performing ensemble empirical mode decomposition, generating multiple noise auxiliary signals by adding white noise of different amplitudes to the two-dimensional vector signal multiple times, and each noise auxiliary signal serves as a ensemble member; by adding white noise of different amplitudes and constructing ensemble members, the mode aliasing problem in traditional empirical mode decomposition is effectively solved, and the stability of signal decomposition is improved, so that the extracted intrinsic mode function can more accurately reflect the different frequency oscillation components of the distribution box parameters, and the detection sensitivity of high-frequency weak leakage anomalies is improved.

[0009] S24, for each set member, screen the corresponding local extreme points, and construct upper and lower envelopes based on the local extreme points; calculate the mean of the upper and lower envelopes of each set member, subtract the corresponding mean from the set member to obtain the candidate component; S25, calculate the orthogonality index between the candidate component and the corresponding two-dimensional vector signal; obtain the candidate component whose orthogonality index is lower than the preset orthogonality index threshold as the first intrinsic mode function; S26, subtract the extracted intrinsic mode function from the two-dimensional vector signal obtained in S22, repeat S23 to S25 for the remaining signal, and extract the next intrinsic mode function in turn; S27, repeat S26 until the remaining signal is a monotonic function, and take the monotonic function as the residual trend.

[0010] Further, S21, constructs a multivariate time series matrix of leakage current values ​​and temperature values, and calculates the time-lag mutual correlation coefficient of the leakage current values ​​and the temperature values, including: obtaining the leakage current data sequence and the temperature data sequence of different floors within a preset time period; aligning the leakage current data sequence and the temperature data sequence according to the time label to construct a two-dimensional time series matrix; setting a maximum time lag range according to the leakage current acquisition frequency of the distribution box, and the maximum time lag range reflects the delay relationship between the temperature anomaly and the leakage current in the distribution box; within the set maximum time lag range, sequentially moving the temperature data sequence with a preset step size to form data pairs under different time lag conditions; calculating the mutual information coefficient between the data pairs under different time lag conditions as the time-lag mutual correlation coefficient of the leakage current value and the temperature value.

[0011] Further, S22, determining the coupling time window between the leakage current value and the temperature value according to the peak value of the time-delay mutual correlation coefficient; combining the leakage current value and the temperature value into a two-dimensional vector signal according to the coupling time window, including: constructing a mutual correlation coefficient curve according to the time-delay mutual correlation coefficient; extracting the time lag value corresponding to the peak value of the mutual correlation coefficient curve ; With time lag value As the center, obtain the continuous time lag interval whose time lag correlation coefficient is greater than the preset threshold value as the coupling time window between the leakage current value and the temperature value; according to the coupling time window, the leakage current data sequence is offset relative to the temperature data sequence time steps, and obtain the time-aligned leakage current data sequence and temperature data sequence; the time-aligned leakage current data sequence is used as the first dimension component, and the temperature data sequence is used as the second dimension component to form a two-dimensional vector signal. By calculating the time-delay mutual correlation coefficient between the leakage current and temperature values, the nonlinear coupling relationship and delay characteristics between the two are accurately identified, which is more in line with the temperature-leakage physical change law when the electrical component fails than the traditional method.

[0012] Further, S3, feature extraction is performed on each intrinsic mode function and residual trend to obtain an electrical fire feature vector, including: S31, Hilbert-Huang transform is performed on each intrinsic mode function to obtain frequency domain features of each intrinsic mode function; S32, segmented fitting is performed on the residual trend to obtain trend features; S33, frequency domain features and trend features are combined to obtain an electrical fire feature vector.

[0013] Further, S31, Hilbert-Huang transform is performed on each intrinsic mode function to obtain the frequency domain characteristics of each intrinsic mode function, including: using the mirror symmetric extension method to extend the two ends of each intrinsic mode function to N times the length of the original signal; performing Hilbert transform on each intrinsic mode function after extension to calculate the corresponding analytical signal; according to the analytical signal, extracting the instantaneous amplitude of each intrinsic mode function and instantaneous phase; calculate the derivative of the instantaneous phase as the instantaneous frequency of each eigenmode function, delete the instantaneous frequency corresponding to the extended part, and only retain the instantaneous frequency within the original signal interval ; According to the retained instantaneous frequency and the corresponding instantaneous amplitude , construct the Hilbert energy spectrum of each eigenmode function ; According to the Hilbert energy spectrum , calculate the frequency domain characteristics of each eigenmode function.

[0014] Furthermore, the value range of N is 1.2 to 1.6.

[0015] Furthermore, S4 constructs a fire risk assessment model based on the electrical fire feature vector, including: collecting a sample set of electrical fire feature vectors corresponding to electrical fire levels in different buildings; using the electrical fire feature vectors in the sample set as input and the actual fire level as output, to construct a fire risk assessment model based on random forests.

[0016] The present application also provides a building intelligent electrical fire monitoring system, including: a data acquisition module, which collects distribution box parameters on different floors in a target monitoring area within a preset time period, the distribution box parameters including leakage current value and temperature value; a signal decomposition module, which performs multi-scale decomposition on the distribution box parameters to obtain multiple intrinsic mode functions and residual trends; a feature extraction module, which performs feature extraction on each intrinsic mode function and residual trend to obtain an electrical fire feature vector; and a risk prediction module, which predicts the fire risk level of the electrical fire feature vector according to a fire risk assessment model.

[0017] Compared with the prior art, the advantages of this application are: Traditional empirical mode decomposition (EMD) is prone to modal aliasing, which causes electrical fire features to be incorrectly dispersed into different IMFs, reducing the accuracy of fault identification. In addition, leakage current and temperature anomalies often have temporal correlation, but the prior art often decomposes the two independently, ignoring the mutual correlation. This application calculates the time-delay mutual correlation coefficient between the leakage current and temperature values, identifies the coupling time window between the two, and combines the two into a two-dimensional vector signal for joint analysis; it uses ensemble empirical mode decomposition technology to construct ensemble members by adding multiple groups of white noises of different amplitudes to the signal, effectively overcoming the modal aliasing problem; combined with the Hilbert-Huang transform and segmented fitting technology, it realizes the comprehensive extraction of the multi-scale characteristics of the leakage current signal, greatly improving the accuracy and foresight of electrical fire risk assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The present application will be further described in the form of exemplary embodiments, which will be described in detail by the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same number represents the same structure, wherein: Figure 1 is an exemplary flow chart of a building intelligent electrical fire monitoring method according to some embodiments of the present application; Figure 2 is an exemplary flow chart of a method for extracting a first intrinsic mode function according to some embodiments of the present application; Figure 3 is an exemplary flow chart of constructing an electrical fire feature vector according to some embodiments of the present application. DETAILED DESCRIPTION

[0019] The method and system provided in the embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0020] like Figure 1 As shown, the distribution box parameters of different floors in the target monitoring area within the preset time period are obtained, and the distribution box parameters include leakage current value and temperature value; the distribution box parameters are multi-scale decomposed to obtain multiple intrinsic mode functions and residual trends; the features of each intrinsic mode function and residual trend are extracted to obtain the electrical fire feature vector; according to the electrical fire feature vector, a fire risk assessment model is constructed; and the fire risk level is predicted according to the fire risk assessment model.

[0021] S1, firstly, the intelligent sensor network is deployed in the distribution boxes on each floor of the target building to realize the automatic collection of distribution box parameters. The specific implementation method is as follows: a leakage current transformer and a temperature sensor are deployed in the distribution box on each floor. The leakage current transformer adopts a magnetic core ring design with a sensitivity of not less than 0.5mA; the temperature sensor adopts a thermocouple array, and multiple collection points are set at key positions inside the distribution box to ensure the comprehensiveness of temperature field distribution monitoring. According to the development characteristics of electrical fires, the leakage current sampling frequency is set to 10Hz-100Hz, which can capture transient leakage changes; the temperature data sampling frequency is set to 0.1Hz-1Hz to meet the needs of heat change monitoring. The collected raw data is standardized, including removing power supply interference, electromagnetic radiation interference and sensor noise, and completing missing values ​​to ensure data quality. The collected data is stored in the format of "floor number-distribution box number-parameter type-timestamp-parameter value" to form a standardized time series data set for subsequent time series analysis.

[0022] like Figure 2 As shown in S2, the distribution box parameters are decomposed at multiple scales to obtain multiple intrinsic mode functions and residual trends; the intrinsic mode functions represent the different frequency oscillation components in the distribution box parameters and the corresponding time-varying characteristics; the residual trend represents the long-term change baseline and overall evolution trend of the distribution box parameters.

[0023] S21, construct a multivariate time series matrix and calculate the time-lag correlation coefficient, extract the leakage current and temperature data series within a specific time period (usually 1-7 days) from the database, and keep the sampling time consistency of the two types of data. For inconsistent sampling frequencies, the two types of data are aligned in the time dimension through interpolation. The aligned leakage current series is recorded as , the temperature series is recorded as , construct a two-dimensional matrix ,in, represents a two-dimensional time series matrix; t represents a time point. According to the characteristics of electrical faults, the maximum time delay range is set to 24 hours. This range covers the time delay from the occurrence of leakage to the obvious temperature change of most electrical faults. Within the set time delay range, the temperature series is sequentially Moving backward, we get , τ represents the time lag value. For each τ, calculate and The mutual information between: ,in, is the joint probability density, and is the marginal probability density, calculated by the kernel density estimation method.

[0024] S22, determine the coupling time window between the leakage current value and the temperature value according to the peak value of the time lag mutual correlation coefficient; combine the leakage current value and the temperature value into a two-dimensional vector signal according to the coupling time window. Specifically, the different time lag values ​​τ (usually in the range of 0-24 hours) calculated in S21 and their corresponding mutual information coefficients Organized into a corresponding relationship table. With the time lag value τ as the horizontal axis, the mutual information coefficient The cross-correlation coefficient curve is drawn with y as the ordinate. To ensure the smoothness of the curve, the cubic spline interpolation method is used to continuously interpolate between discrete sampling points to form a smooth cross-correlation coefficient curve. To eliminate possible high-frequency noise interference, the Savitzky-Golay filter is applied to the original cross-correlation coefficient curve, with the window width set to 7-11 sampling points and the polynomial order set to 3, so as to retain the main features of the curve while reducing the influence of noise.

[0025] A multi-scale peak detection algorithm is used to identify local maxima at different time scales. First, the main peak area is located at a coarse scale, and then the peak position is precisely located at a fine scale. The detected peak point is verified by calculating the second-order derivative of the curve to ensure that the second-order derivative at the peak point is negative and the absolute value is greater than the preset threshold (such as 0.05). From the verified peak points, the point with the largest mutual information coefficient is selected, and its corresponding time lag value is For multiple peaks, if the difference between the highest peak and the second highest peak is less than 20%, the time lag values ​​of the two peaks are weighted averaged (the weight is proportional to the peak value) as the value. Usually in the 6-12 hour range.

[0026] The preset threshold is set to 70% of the peak mutual information coefficient, that is, The threshold can be adjusted according to the characteristics of the specific building electrical system. It can be set to 60% for high sensitivity scenarios and 80% for high specificity scenarios. At the beginning, search in the left and right directions respectively to find the time lag point where the mutual information coefficient is lower than the preset threshold for the first time, which are recorded as and . Interval This is the coupling time window of leakage current and temperature. If the identified window is too wide (>12 hours), an adaptive threshold method is used to increase the threshold to 75%-85% and recalculate the window boundary to ensure that the captured coupling relationship has sufficient specificity.

[0027] According to the coupling time window , determine the effective data analysis time range, and intercept the data series of the two parameters within this time range. Offset Right time unit, that is This operation aligns the leakage current signal with the temperature signal in time, and directly reflects the physical cause-effect relationship in the data sequence. The offset operation may cause invalid data to appear at the sequence boundary. The "mirror extension" method is used to deal with the boundary problem, that is, adding data points symmetrical to the boundary points at both ends of the sequence to maintain the continuity and integrity of the sequence.

[0028] Leakage current sequence after alignment and temperature series Normalization is performed to unify the scales of the two. Normalization uses the Z-score standardization method: , where X represents the original data point; Z represents the normalized data point; μ is the sequence mean, and σ is the sequence standard deviation. The normalized leakage current sequence is used as the first dimension component and the temperature sequence is used as the second dimension component to construct a two-dimensional vector signal ,in, represents the normalized leakage current series; Represents the normalized temperature sequence; by calculating the autocorrelation function of each component of the two-dimensional vector, it is ensured that the constructed vector signal maintains the main time characteristics of the original data. At the same time, the alignment effect is verified by calculating the cross-correlation function between the two components. This application takes into account the time delay characteristics between the abnormal leakage current and the temperature rise during the development of electrical component failures, which conforms to the physical laws of electrical fire development. Compared with the traditional method of analyzing each parameter separately, it can better reflect the real fault evolution process.

[0029] S23, performs ensemble empirical mode decomposition, and uses a high-quality pseudo-random number generator to generate white noise to ensure that the noise sequence has a uniform spectral distribution. The noise amplitude is set using an adaptive method and dynamically adjusted based on signal strength: the white noise amplitude is set to α×σ, where σ is the standard deviation of the original two-dimensional vector signal and α is the proportional coefficient. For high signal-to-noise ratio scenarios, the α value range is 0.1 to 0.2; for low signal-to-noise ratio scenarios, the α value range is 0.2 to 0.3. In order to obtain more comprehensive noise coverage, a linear increasing sequence of α values ​​is designed, increasing from 0.1 to 0.3, with a total of 10 to 20 levels.

[0030] For each noise amplitude level, 5 to 10 independent white noise sequences are generated, forming a total of 50 to 200 noise auxiliary set members. ,in, Represents the processed leakage current series; white noise is added to the two components respectively: ;in, represents the kth noise auxiliary set member; and is the noise sequence of the kth set member, and the two are independent but have the same amplitude. A parallel computing architecture is used to process multiple set members to improve the decomposition efficiency. According to the number of server cores, the set members are grouped and each group is assigned to a computing core for simultaneous processing.

[0031] This application introduces disturbances in different frequency ranges of the signal by adding white noise of different amplitudes, which causes components with close frequencies to be separated into different IMFs. After the white noise is introduced, the close frequency components that may have been mistakenly merged are locally modulated by the noise during the decomposition process, which enhances their differences in the distribution of extreme points. By adding random white noise multiple times and averaging, the randomness of the noise is offset in the final averaging process, and the different frequency components separated by the noise can be retained in the corresponding IMF.

[0032] S24, using a multi-scale local extremum detection algorithm, for each set member Processing: First, detect local maxima and minima on the original sampling scale, that is, satisfy The point is the maximum point, satisfying The point is the minimum point. Anti-interference processing is used: if the number of sampling points between two adjacent maximum (or minimum) points is less than 3, the amplitudes of the two maximum (or minimum) values ​​are compared, and the point with the larger (or smaller) amplitude is retained. For high sampling rate data (>50Hz), sliding window pre-smoothing is used, and the window width is an odd number, usually 5 to 11 sampling points, to reduce excessive extreme points caused by noise.

[0033] The upper and lower envelopes are constructed using cubic spline interpolation to ensure smooth transition of the envelope in the entire time domain: Upper envelope Constructed through all local maximum points. Lower envelope Constructed through all local minimum points. The boundary processing adopts the mirror extension method: 3-5 mirror extreme points are added at both ends of the signal to ensure the smoothness and accuracy of the envelope in the edge area. The spline interpolation adopts the "not-a-knot condition" to ensure the smooth transition of the envelope between extreme points and avoid excessive oscillation.

[0034] Compute the point-wise means of the upper and lower envelopes: ,in, Represents the mean of the upper and lower envelopes of the kth set member; subtract the mean from the currently processed signal to obtain the candidate component: ; The upper limit of the number of iterations is usually set to 10 to prevent excessive iterations. If the IMF condition is still not met after 10 iterations, the current candidate component is forced to be accepted. Verify the candidate component Whether the following IMF conditions are met: Condition 1: The absolute value of the difference between the total number of extreme points and the total number of zero-crossing points does not exceed 1. Condition 2: Within the entire data range, the relative deviation SD of the envelope mean from the zero value is lower than the threshold: If the above conditions are not met, It is considered as a new input signal and the screening process is repeated.

[0035] S25, the orthogonality index (OI) is used to measure the degree of orthogonality between the candidate components and the original signal: ;in is the candidate component obtained in the current iteration, is the kth member of the set. The orthogonality index range is [0, 1], and the smaller the value, the better the orthogonality. The orthogonality index threshold is dynamically adjusted based on the signal complexity, and the complexity is calculated by the spectral entropy of the signal: for low-complexity signals (spectral entropy <0.6), the threshold is set to 0.05. For medium-complexity signals (spectral entropy 0.6-0.8), the threshold is set to 0.075. For high-complexity signals (spectral entropy>0.8), the threshold is set to 0.1. Through experimental verification, the leakage current-temperature coupling signal of the distribution box is usually a medium-complexity signal.

[0036] For each set member, obtain the candidate component with an orthogonality index lower than the preset threshold as the first IMF of the member: If the candidate component of a set member still does not meet the orthogonality requirement after MAX_ITER iterations (usually set to 10), the current candidate component is forced to be adopted and the reliability of the IMF is marked as low in the result.

[0037] Take the ensemble average of the first IMF obtained by all ensemble members: , where N represents the total number of set members; represents the first intrinsic mode function extracted from the kth set member; in order to reduce the influence of outliers, a modified weighted average method can be used: calculate each The IMF assigns a lower weight (0.3) to those with a deviation of more than 3 standard deviations, and a standard weight (1.0) to the rest. The weighted average is recalculated based on the weights.

[0038] S26, subtract the extracted i-th IMF from the original two-dimensional vector signal to obtain the residual signal: ;in, represents the i-th intrinsic mode function, which represents the final IMF result obtained after ensemble averaging; is the original two-dimensional vector signal .Will As a new input signal; Represents the remaining signal after the i-1th decomposition; repeat steps S23 to S25 to extract the next IMF. To improve the computational efficiency, the remaining signal can be downsampled: as the IMF number increases, the signal frequency decreases, and the number of sampling points can be gradually reduced to reduce the computational complexity. Downsampling uses anti-aliasing filtering to ensure that no valid information is lost.

[0039] The recursive process is terminated when any of the following conditions are met: Condition 1: Remaining signal It becomes a monotonic function, that is, the total number of extreme points is ≤ 2. Condition 2: The energy ratio of the remaining signal is lower than the threshold: ,in Represents the original two-dimensional vector signal, which consists of the normalized leakage current series and temperature series. Condition 3: The number of extracted IMFs reaches the preset upper limit MAX_IMF (usually 8 to 10). Condition 4: The spectral correlation coefficient of two consecutive IMFs is higher than 0.85, indicating that the decomposition has become redundant.

[0040] S27, repeat S26 until the remaining signal is a monotonic function, and take the monotonic function as the residual trend. The final remaining signal As the residual trend, it represents the long-term change baseline of the distribution box parameters. In order to enhance the smoothness and physical meaning of the trend, a low-pass filter is applied to the residual trend, and the cut-off frequency is set to 0.01Hz to filter out the high-frequency components that may be mixed in. For the two-dimensional vector signal, the residual trends of the two components are extracted respectively to obtain the long-term evolution trend of the leakage current and temperature parameters of the distribution box.

[0041] For distribution box monitoring applications, different IMFs have clear physical meanings: to Usually corresponds to high-frequency noise and transient leakage characteristics (frequency range: 1 to 100 Hz), reflecting transient states such as equipment switching and load fluctuations. to It usually corresponds to medium-frequency oscillation components (frequency range: 0.1-1Hz), reflecting gradual faults such as deterioration of electrical components and poor contact. The low-frequency components (frequency < 0.1 Hz) after the aging of the insulation and the humidity of the environment reflect the long-term influencing factors. The residual trend reflects the equipment load baseline and seasonal change trend.

[0042] like Figure 3 As shown, S3, feature extraction is performed on each intrinsic mode function and residual trend to obtain the electrical fire feature vector. S31, before performing the Hilbert transform, it is necessary to solve the distortion problem of the traditional Hilbert transform at the signal boundary. The specific implementation method is as follows: for each intrinsic mode function , the mirror symmetric extension method is used at both ends of the time domain to extend. At the left end of the signal, the mirror reflection about the point t=0 is used to form ; At the right end of the signal, a mirror reflection about point t=T (T is the length of the original signal) is used to form .

[0043] The extension coefficient N is usually selected between 1.2 and 1.6, and the selection of N value follows the following rules: for high-frequency IMF (the first 1-3 IMFs), select a smaller N value (1.2-1.3) to reduce the computational burden; for medium-frequency IMF (the 4th-6th IMF), select a medium N value (1.3-1.5); for low-frequency IMF (the 7th and later IMFs), select a larger N value (1.5-1.6) to better handle low-frequency boundary effects. To avoid discontinuity at the extension seam, the Hanning window function is applied for smooth transition at the connection between the original signal and the extension part. The window function length is 5% of the original signal length to ensure the smooth continuity of the entire extended signal.

[0044] Analytical signal calculation: After extension Perform Hilbert transform to obtain the analytical signal: ,in, yes The Hilbert transform of is calculated as: ; Wherein, j represents the imaginary unit; represents the integration variable.

[0045] Instantaneous amplitude calculation: Instantaneous amplitude To resolve the modulus of the signal: A moving median filter (with a window length of 5-7 sampling points) was applied to the calculated results to eliminate possible numerical fluctuations.

[0046] Instantaneous phase calculation: Instantaneous phase To resolve the signal's argument: To avoid phase jumps, a phase unwrapping algorithm is used to ensure continuous phase changes.

[0047] Instantaneous frequency calculation: Instantaneous frequency is the time derivative of the phase: , the derivative calculation uses the five-point central difference formula to improve the accuracy of numerical calculation: ,in, is the sampling interval.

[0048] Boundary data processing: Delete the instantaneous characteristic data corresponding to the extended part, and only retain the instantaneous amplitude in the original signal interval [0, T] and instantaneous frequency To avoid the inaccuracy of 2 to 3 edge points caused by differential calculation, linear extrapolation is used to reconstruct the values ​​of these points.

[0049] Construct the Hilbert energy spectrum based on the retained instantaneous frequencies and the instantaneous amplitude , construct the Hilbert energy spectrum: , where δ is the Dirac function, indicating that the energy at time t is only concentrated at the frequency Department; Represents the angular frequency variable; in practical applications, the time-frequency plane is discretized into a grid with a grid resolution of: Time resolution: original sampling interval , frequency resolution: (maximum frequency - minimum frequency) / 100, usually 0.1 to 1 Hz. Falling into the frequency range When Distribute it to adjacent frequency points in proportion to achieve smooth energy distribution.

[0050] Calculate the frequency domain characteristic parameters and marginal spectrum: Integrate the Hilbert energy spectrum in the time dimension to obtain the marginal spectrum: When implemented discretely, the trapezoidal integration method is used: .

[0051] Energy concentration calculation: Energy concentration EC is used to quantify the degree of aggregation of frequency distribution: The larger the EC value, the more concentrated the energy is at a specific frequency. The EC of a normal distribution box is usually 0.1 to 0.3, and can reach 0.5 to 0.8 in abnormal situations. represents the marginal spectrum; Represents discrete frequency points.

[0052] Main frequency and bandwidth calculation: Main frequency point , is the frequency corresponding to the maximum value of the marginal spectrum, and the frequency bandwidth BW is defined as the minimum frequency interval width containing 80% of the energy. The calculation steps are: Normalized to Make ,right Sort from largest to smallest , find the smallest k such that , calculate the maximum and minimum differences of these k frequency points as the bandwidth BW.

[0053] Extract electrical fault characteristic parameters and divide the frequency range into three frequency bands: low frequency (0 to 10 Hz), medium frequency (10 to 100 Hz), and high frequency (>100 Hz); calculate the proportion of marginal spectrum energy in each frequency band to the total energy: . In a normal distribution box, low-frequency energy usually dominates ( ), the proportion of medium and high frequency energy increases significantly during electrical faults, Indicates the energy proportion of the low frequency band (0 to 10Hz); Indicates the energy proportion of the mid-frequency band (10 to 100 Hz); Indicates the energy ratio of high and low frequency bands (>100Hz); Calculate the frequency change rate index, the time derivative of the instantaneous frequency: , calculated using the central difference method; calculate The mean and variance ; Frequency change rate index: , characterizes the severity of frequency fluctuation. Calculate the frequency fluctuation index and calculate the instantaneous frequency The mean and standard deviation ; Frequency Fluctuation Index: ; Usually indicates the presence of abnormal fluctuations. The above features are combined into a frequency domain feature vector: , this eigenvector is calculated for each IMF, and together they form a full-frequency domain feature matrix for subsequent fault type identification.

[0054] Specifically, the marginal spectrum :The marginal spectrum reflects the energy distribution of the leakage current signal on each frequency component. In the building electrical system, the leakage current during normal operation usually has a relatively stable spectrum characteristic, which is mainly concentrated on the grid base frequency (50Hz or 60Hz) and its harmonics. When there are early faults such as insulation aging and poor contact, abnormal energy increases will occur in specific frequency bands. For example, partial discharge before insulation breakdown will generate high-frequency noise; loose connections will generate medium-frequency fluctuations.

[0055] Energy concentration EC: Energy concentration quantifies the degree of frequency distribution. For electrical equipment that is operating normally, the leakage current spectrum energy is usually concentrated on certain specific frequencies, and the EC value is high. When electrical equipment begins to fail, the spectrum becomes dispersed, which is manifested as a decrease in EC value. For example, insulation aging can cause the spectrum energy to spread to a wider frequency band, generating more random noise.

[0056] Main frequency point And frequency bandwidth BW: The main frequency point is the frequency with the most concentrated energy, and the bandwidth reflects the frequency concentration of the signal. In the building electrical system, the main frequency under normal operation is usually fixed and related to the operating characteristics of the equipment. When a fault precursor appears, the main frequency may shift or the bandwidth may increase significantly. For example, motor bearing wear will cause the main frequency to shift; cable insulation aging will cause the frequency band to become wider.

[0057] Energy proportion of each frequency band :Divide the spectrum into three intervals: low frequency (0 to 10Hz), medium frequency (10 to 100Hz) and high frequency (>100Hz), and calculate the energy proportion of each frequency band. In the building electrical system, different types of faults will show energy anomalies in different frequency bands: low frequency (0 to 10Hz): corresponds to slow changes in system load, and abnormal values ​​may mean continuous leakage of large current; medium frequency (10 to 100Hz): contains the grid fundamental frequency and low-order harmonics, and abnormalities are usually related to conductor connection problems or line faults; high frequency (>100Hz): contains high-order harmonics and noise, and abnormal values ​​often reflect discharge or arc precursors before insulation breakdown.

[0058] Frequency change rate (FVR): measures how fast and how violently the instantaneous frequency changes over time. When the electrical system is operating normally, the instantaneous frequency usually changes slowly, corresponding to the load change. When the system has intermittent faults or load mutations, the frequency change rate increases significantly. For example, poor conductor contact can cause frequency mutations; the start and stop of equipment can cause rapid frequency changes.

[0059] Frequency Fluctuation Index FI: measures the relative fluctuation of frequency and reflects frequency stability. A normally operating electrical system has stable frequency characteristics and a low FI value. Early fault conditions such as poor contact and insulation aging can lead to frequency instability, which is manifested as an increase in FI value. For example, loose contact points can cause impedance fluctuations, which in turn cause frequency instability.

[0060] S32, performing segmented fitting on the residual trend to obtain trend characteristics, including: performing sliding window analysis on the residual trend, determining the appropriate window length according to the changing characteristics of the leakage current signal; applying segmented linear fitting technology to model the residual trend, and determining the optimal segmentation point position and the slope of each segment by the least squares method; extracting key trend characteristic parameters, including: the slope and duration of each linear segment, reflecting the rate of change of the leakage current; the slope mutation between adjacent linear segments, characterizing the degree of change of the leakage characteristics; the time position and amplitude of the key inflection point, indicating the significant change of the leakage mode; combining the extracted parameters to form a trend characteristic vector , where k represents the slope, The slope change, t represents the inflection point time, and v represents the inflection point amplitude, which reflects the long-term evolution characteristics of the distribution system. Through the sliding window analysis and piecewise linear fitting of the residual trend, the long-term evolution law of the leakage current is accurately captured, especially the extracted slope mutation and key inflection point characteristics, which can reflect the slow accumulation process of leakage current under insulation aging and humid environment.

[0061] S33, merging the frequency domain feature vector of each IMF calculated in S31 with the residual trend feature vector calculated in S32, and forming an original feature space of (8×n+4n-4) dimensions for n IMFs.

[0062] S4, collect data records of buildings where electrical fires have occurred, extract the complete sequence of distribution box parameters 72 hours before the fire; obtain electrical fire accident investigation reports and parameter records from the fire safety department; establish a standardized electrical fire case library, including typical fault data of different types of buildings (commercial, residential, industrial). Simulate typical electrical fault scenarios in a laboratory environment, including insulation aging, line overload, poor contact, etc.; set different environmental conditions (temperature 20 to 40°C, relative humidity 40 to 90%), and record the entire process of fault development; collect at least 50 sets of complete parameter evolution data for each fault type.

[0063] Apply S1 to S3 processes to the original data to extract standardized electrical fire feature vectors; each sample contains 8×n-dimensional IMF frequency domain features and 4n-4-dimensional residual trend features, where n is the number of IMFs (usually 6 to 10); perform normalization on all feature vectors so that the eigenvalue range of each dimension is in the interval [0, 1].

[0064] Establish a 5-level risk assessment system: Level 0 (safe): all parameters are within the normal range, no abnormalities; Level 1 (low risk): slight abnormalities occur, but are within the acceptable range; Level 2 (medium risk): obvious abnormalities but have not yet reached a dangerous level; Level 3 (high risk): serious deviations from normal values, requiring emergency intervention; Level 4 (extremely high risk): the system is in a critical state and a fire may be about to occur.

[0065] Historical fire case annotation: the level is determined according to the time from the fire occurrence (0 to 6 hours is level 4, 6 to 24 hours is level 3, etc.); experimental sample annotation: the risk level is determined according to the leakage current value, fault severity and expert evaluation.

[0066] A hierarchical random forest architecture is adopted, including a base layer and an integration layer. The base layer designs three dedicated random forest sub-models: High-frequency feature forest: focuses on the frequency domain features of IMF1 to IMF3; the frequency domain features of IMF1 to IMF3 are sensitive to sudden faults. Medium-frequency feature forest: focuses on the frequency domain features of IMF4 to IMF6; the frequency domain features of IMF4 to IMF6 are sensitive to gradual faults. Trend feature forest: focuses on residual trend features; the integration layer designs a meta-random forest to integrate the output probabilities of the three sub-models.

[0067] Number of trees: 500 trees in the base layer forest and 300 trees in the integrated layer forest; Maximum tree depth: 25-30 layers, tuned by the validation set; Minimum number of samples for nodes: 5 samples for leaf nodes and 12 samples for internal nodes; Feature selection criteria: Gini impurity index; Tree diversity control: Each tree uses the square root of the total number of features as the number of candidate features for splitting. Based on cost-sensitive analysis, the optimal decision threshold is determined for each risk level; a lower threshold (0.6 to 0.7) is used for high risk levels (3 to 4) to improve recall; a higher threshold (0.8 to 0.9) is used for low risk levels (0-1) to improve precision.

[0068] S5, predicting the fire risk level according to the fire risk assessment model. Obtain the real-time leakage current value and temperature value of the distribution box in the target monitoring area, and process according to the process of S2 and S3 to obtain the real-time electrical fire feature vector; input the real-time electrical fire feature vector into the random forest model constructed by S4 to calculate the fire risk probability value; map the fire risk probability value to a preset risk level interval, and use the mapping result as the fire risk level, and the risk level interval includes four levels: low risk, medium risk, high risk and extremely high risk; when the fire risk level is high risk or extremely high risk, send fire risk warning information to the building management system, and the warning information includes risk level, fault type determination result and distribution box location information; store the fire risk level assessment results based on time series, and form an electrical fire risk evolution trend curve for long-term risk analysis and system optimization.

Claims

1. A building intelligent electrical fire monitoring method, characterized in that: include: S1, obtaining the distribution box parameters of different floors in the target monitoring area within a preset time period, wherein the distribution box parameters include leakage current value and temperature value; S2, multi-scale decomposition of distribution box parameters is performed to obtain multiple intrinsic mode functions and residual trends; S3, extracting features from each intrinsic mode function and residual trend to obtain electrical fire feature vectors; S4, constructing a fire risk assessment model based on the electrical fire feature vector; S5, predicting the fire risk level according to the fire risk assessment model.

2. The intelligent building electrical fire monitoring method according to claim 1 is characterized in that: The intrinsic mode function represents the different frequency oscillation components in the distribution box parameters and the corresponding time-varying characteristics; The residual trend represents the long-term variation baseline and overall evolution trend of the distribution box parameters.

3. The intelligent building electrical fire monitoring method according to claim 2 is characterized in that: S2, multi-scale decomposition of distribution box parameters is performed to obtain multiple intrinsic mode functions and residual trends, including: S21, constructing a multivariate time series matrix of leakage current values ​​and temperature values, and calculating the time lag mutual correlation coefficient of the leakage current values ​​and the temperature values; S22, determining a coupling time window between the leakage current value and the temperature value according to a peak value of a time-delay mutual correlation coefficient; combining the leakage current value and the temperature value into a two-dimensional vector signal according to the coupling time window; S23, performing ensemble empirical mode decomposition, generating multiple noise auxiliary signals by adding white noise of different amplitudes to the two-dimensional vector signal multiple times, each noise auxiliary signal being a member of a set; S24, for each set member, selecting the corresponding local extreme value point, and constructing the upper and lower envelopes according to the local extreme value points; Calculate the mean of the upper and lower envelopes of each set member, subtract the corresponding mean from the set member to obtain the candidate component; S25, calculating the orthogonality index between the candidate component and the corresponding two-dimensional vector signal; Obtaining a candidate component whose orthogonality index is lower than a preset orthogonality index threshold as the first eigenmode function; S26, subtracting the extracted intrinsic mode function from the two-dimensional vector signal obtained in S22, repeating S23 to S25 for the remaining signal, and extracting the next intrinsic mode function in turn; S27, repeat S26 until the remaining signal is a monotonic function, and take the monotonic function as the residual trend.

4. The intelligent building electrical fire monitoring method according to claim 3 is characterized in that: S21, constructing a multivariate time series matrix of leakage current values ​​and temperature values, and calculating the time lag mutual correlation coefficients of leakage current values ​​and temperature values, including: Obtain leakage current data sequences and temperature data sequences of different floors within a preset time period; Align the leakage current data series and the temperature data series according to the time labels and construct a two-dimensional time series matrix; According to the distribution box leakage current collection frequency, a maximum time lag range is set, and the maximum time lag range reflects the delay relationship between the temperature anomaly and the leakage current in the distribution box; Within the set maximum time lag range, the temperature data sequence is moved in sequence with a preset step length to form data pairs under different time lag conditions; The mutual information coefficients between the data pairs under different time lag conditions were calculated as the time lag mutual correlation coefficients of the leakage current value and the temperature value.

5. The intelligent building electrical fire monitoring method according to claim 4 is characterized in that: S22, determining a coupling time window between the leakage current value and the temperature value according to the peak value of the time-delay mutual correlation coefficient; combining the leakage current value and the temperature value into a two-dimensional vector signal according to the coupling time window, including: According to the time-lag mutual correlation coefficient, a mutual correlation coefficient curve is constructed; Extract the lag value corresponding to the peak of the cross-correlation coefficient curve ; By time lag As the center, obtain the continuous time lag intervals whose time lag mutual correlation coefficient is greater than the preset threshold value as the coupling time window between the leakage current value and the temperature value; Shift the leakage current data sequence relative to the temperature data sequence according to the coupled time window time steps, and obtain the leakage current data series and temperature data series after time alignment; The time-aligned leakage current data sequence is used as the first-dimensional component, and the temperature data sequence is used as the second-dimensional component, and they are combined to form a two-dimensional vector signal.

6. The intelligent building electrical fire monitoring method according to claim 3, characterized in that: S3, extract the features of each intrinsic mode function and residual trend to obtain the electrical fire feature vector, including: S31, performing Hilbert-Huang transform on each intrinsic mode function to obtain frequency domain characteristics of each intrinsic mode function; S32, performing piecewise fitting on the residual trend to obtain trend characteristics; S33, combining the frequency domain features and the trend features to obtain an electrical fire feature vector.

7. The intelligent building electrical fire monitoring method according to claim 6, characterized in that: S31, performing Hilbert-Huang transform on each intrinsic mode function to obtain frequency domain characteristics of each intrinsic mode function, including: The mirror symmetric extension method is used to extend the two ends of each intrinsic mode function to N times the length of the original signal; Perform Hilbert transform on each extended intrinsic mode function and calculate the corresponding analytical signal; According to the analytical signal, the instantaneous amplitude of each intrinsic mode function is extracted and instantaneous phase; Calculate the derivative of the instantaneous phase as the instantaneous frequency of each eigenmode function, delete the instantaneous frequency corresponding to the extended part, and only retain the instantaneous frequency within the original signal interval ; According to the instantaneous frequency retained and the corresponding instantaneous amplitude , construct the Hilbert energy spectrum of each eigenmode function ; According to the Hilbert energy spectrum , calculate the frequency domain characteristics of each eigenmode function.

8. The intelligent building electrical fire monitoring method according to claim 7, characterized in that: The value of N ranges from 1.2 to 1.

6.

9. The intelligent building electrical fire monitoring method according to claim 8, characterized in that: S4, based on the electrical fire feature vector, build a fire risk assessment model, including: Collect electrical fire feature vector sample sets corresponding to electrical fire levels in different buildings; The electrical fire feature vector in the sample set is taken as input and the actual fire level is taken as output to construct a fire risk assessment model based on random forest.

10. A building intelligent electrical fire monitoring system, characterized in that: include: The data acquisition module collects the distribution box parameters of different floors in the target monitoring area within a preset time period. The distribution box parameters include leakage current value and temperature value; The signal decomposition module performs multi-scale decomposition on the distribution box parameters to obtain multiple intrinsic mode functions and residual trends; The feature extraction module extracts the features of each intrinsic mode function and residual trend to obtain the electrical fire feature vector; The risk prediction module predicts the fire risk level of the electrical fire feature vector based on the fire risk assessment model.

Images