Unit evaporative cooling system working condition monitoring method based on MFCC and grey relational degree

Abstract

The invention relates to a unit evaporative cooling system working condition monitoring method based on MFCC and grey relational degree. The method comprises the steps of collecting voiceprint signals of positions such as a condenser and a liquid collecting ring pipe of an evaporative cooling system; the data and temperature, liquid level, flow and pressure data of the evaporative cooling system are used as input data of the unit evaporative cooling system state evaluation model; constructing a state evaluation model based on the grey correlation degree, taking standard sequence data of the unit evaporative cooling system under each working condition state as a reference sequence, taking characteristic data of an evaluation object changing along with time as a comparison sequence for comparison, and calculating the correlation degree of the comparison sequence and the reference sequence; and obtaining the grey relational degree of the evaluation object and each working condition state, and taking the working condition state corresponding to the maximum grey relational degree as the evaluation result of the working condition of the current group of evaporative cooling system. Real-time monitoring of the abnormal working condition state of the unit evaporative cooling system is achieved, and the accuracy of abnormal monitoring of the unit evaporative cooling system is improved.

Description

Technical Field

[0001] The invention belongs to the field of fault monitoring of hydropower units, and in particular relates to a method for monitoring the operating condition of an evaporative cooling system of a unit based on MFCC and grey correlation. Background Art

[0002] The reliable and stable operation of the hydro-turbine generator set is closely related to the healthy state of the evaporative cooling system. Common faults of the evaporative cooling system include low cooling water flow, condenser leakage, condenser cooling water interruption, evaporative cooling medium leakage, etc. If the stator temperature of the generator rises abnormally, it will cause the unit to trip. If people inhale excessive evaporative cooling medium hydrofluorocarbon Freon refrigerant, it will cause dizziness, convulsions, suffocation, irregular heartbeat, loss of consciousness and other symptoms. Therefore, it is of great significance to ensure the healthy operation of the evaporative cooling system and to promptly detect and eliminate its abnormal operating conditions.

[0003] However, the existing fault monitoring methods for the evaporative cooling system of the unit mainly rely on the monitoring of parameters such as cooling water flow and temperature, and fail to fully cover all factors that may affect the performance of the evaporative cooling system. This may lead to some potential faults not being discovered in time. At the same time, the existing fault monitoring equipment and technology may have certain measurement errors, resulting in deviations between the monitored data and the actual situation. This deviation may affect the accuracy of fault judgment. In addition, the existing fault monitoring methods cannot achieve real-time and continuous monitoring, resulting in the inability to detect and take measures in time at the early stage of the fault. Summary of the invention

[0004] The purpose of the present invention is to provide a method for monitoring the working condition of an evaporative cooling system of a unit based on Mel Frequency Cepstral Coefficients (MFCC) and grey correlation, to collect the soundprint signals of the condenser, liquid collecting ring pipe and other positions of the evaporative cooling system; to use the soundprint signals together with the temperature, liquid level, flow rate and pressure data of the evaporative cooling system as the input data of the state evaluation model of the unit evaporative cooling system, so that the monitoring data of the unit evaporative cooling system is more comprehensive, so as to improve the accuracy of abnormal monitoring of the unit evaporative cooling system; and to construct a state evaluation model based on grey correlation, to use the standard sequence data under each working condition of the unit evaporative cooling system as the reference sequence, to use the characteristic data of the evaluation object changing with time as the comparison sequence for comparison, to calculate the correlation between the comparison sequence and the reference sequence, to obtain the grey correlation between the evaluation object and each working condition, and to use the working condition corresponding to the maximum grey correlation as the evaluation result of the working condition of the current unit evaporative cooling system, so as to improve the continuity and accuracy of abnormal monitoring of the unit evaporative cooling system.

[0005] To achieve the above object, the technical solution of the present invention is: The method for monitoring the operating condition of the unit evaporative cooling system based on MFCC and grey correlation degree includes the following steps: Step 1: Collect the soundprint signals of the condenser, liquid collecting ring pipe, and gas collecting ring pipe when the evaporative cooling system is in working state; and obtain the temperature, liquid level, flow rate, and pressure data of the evaporative cooling system; Step 2: Using the Mel frequency cepstral coefficient method, extracting a feature vector from the voiceprint signal; Step 2.1: sound preprocessing, pre-emphasis, framing and windowing of the voiceprint signal; Step 2.2: Fast Fourier transform, using the Fourier transform method to deconstruct the preprocessed time domain signal into a frequency domain signal; Step 2.3: Mel filtering, filtering the obtained spectrum through a Mel filter, and dividing the spectrum into multiple Mel filter groups; Step 2.4: Take the logarithm and discrete Fourier transform, calculate the logarithmic energy of the output of the filter bank, and then perform discrete cosine transform to obtain the Mel-frequency cepstrum coefficients; Step 3: Construct a state assessment model based on grey correlation; the state assessment model determines the sequence data under each working condition as the reference sequence, takes the characteristic data of the assessment object changing over time as the comparison sequence for comparison, calculates the correlation between the comparison sequence and the reference sequence, and obtains the grey correlation between the assessment object and each working condition; Step 4: After preprocessing the temperature, liquid level, flow rate and pressure data of the evaporative cooling system obtained in step 1 and the characteristic vector of the voiceprint signal obtained in step 2, a comparison vector, i.e., a comparison sequence, is obtained for evaluation and comparison, and the comparison vector is input into the state evaluation model to calculate the grey correlation coefficient between the comparison sequence and the reference sequence of different working conditions; Step 5: According to the grey correlation coefficient obtained in step 4, the operating state corresponding to the maximum grey correlation coefficient is selected as the operating condition evaluation result of the unit evaporative cooling system, and the fault level is determined according to the current operating state obtained by the evaluation.

[0006] Preferably, in step 1, the temperature, liquid level, flow and pressure data of the evaporative cooling system include the unit stator bar temperature, condenser water inlet temperature, condenser water outlet temperature, evaporative cooling medium level, condenser cooling water flow and evaporative cooling system pressure.

[0007] Preferably, in step 2.1, a Hamming window is used to window each frame of the signal after framing, and the calculation formula of the Hamming window is:

[0008] In the formula represents the window function, N represents the number of frames, is the coefficient and n is the index of the sample point.

[0009] Furthermore, in step 2.2, the formula of the Fourier transform method is:

[0010] In the formula Represents the input sound signal, represents the number of Fourier transform points, k is the frequency index, Represents a frequency domain signal.

[0011] Preferably, in step 2.3, the filter calculation formula is:

[0012] Where k is the frequency index; m is the serial number of the Mel filter; f is the actual frequency; Represents the filter function.

[0013] Preferably, in step 2.4, the formula for converting frequency to Mel scale is:

[0014] In the formula, represents the frequency after conversion to Mel scale, Indicates the actual frequency.

[0015] Preferably, it is characterized in that, in step 2.4, the calculation formula of the Mel frequency cepstral coefficient is:

[0016] In the formula represents the Mel-frequency cepstral coefficient; represents the order of the Mel frequency cepstrum coefficient; s(m) represents the mth sample value of the speech signal after pre-emphasis, framing and windowing preprocessing; M represents the number of points under the logarithmic scale, is the order number of the Mel-frequency cepstrum coefficients.

[0017] Preferably, in step 3, the calculation formula for the degree of correlation between the comparison sequence and the reference sequence is:

[0018] In the formula For comparison vector For the reference vector The degree of association, where the resolution coefficient .

[0019] Furthermore, in step 3, the reference sequence, i.e., the reference vector, is:

[0020] In the formula Represents the condenser soundprint signal, Represents the soundprint signal of the liquid collecting ring pipe, Represents the soundprint signal of the gas collecting ring pipe. represents the stator bar temperature, represents the condenser inlet water temperature, Represents the condenser outlet water temperature, Represents the evaporative cooling medium level, represents the condenser cooling water flow, Represents the evaporative cooling system pressure; The comparison sequence is , t represents the time, are the comparison vectors of the evaluation object at different times; Calculate separately and degree of association, and and The average value of the correlation degree is taken as the grey correlation degree.

[0021] Compared with the prior art, the beneficial effects of the present invention include: 1) The present invention collects the soundprint signals of the condenser, liquid collecting ring pipe and gas collecting ring pipe under the working state of the unit's evaporative cooling system, and uses them together with the unit's stator bar temperature, condenser water inlet temperature, condenser water outlet temperature, evaporative cooling medium liquid level, condenser cooling water flow and evaporative cooling system pressure as input data of the unit's evaporative cooling system status evaluation model, comprehensively covering all factors that may affect the performance of the evaporative cooling system, performing data-level fusion processing and analysis on the collected multi-feature data, realizing real-time monitoring of abnormal working conditions of the unit's evaporative cooling system, and greatly improving the accuracy of abnormal monitoring of the unit's evaporative cooling system.

[0022] 2) The present invention uses the Mel-frequency cepstral coefficient method to extract feature vectors from the soundprint signal of the unit's evaporative cooling system, normalizes them with the temperature, liquid level, flow rate and pressure data of the evaporative cooling system respectively, and fuses them after unifying the dimensions, thereby realizing the fusion of multi-feature data of the evaporative cooling system, which is beneficial to improving the comprehensiveness and accuracy of abnormal monitoring.

[0023] 3) The present invention constructs a state assessment model based on grey correlation, takes the standard sequence data under each operating state of the unit's evaporative cooling system as a reference sequence, takes the characteristic data of the assessment object that changes with time as a comparison sequence for comparison, calculates the correlation between the comparison sequence and the reference sequence, obtains the grey correlation between the assessment object and each operating state, takes the operating state corresponding to the maximum grey correlation as the assessment result of the current unit's evaporative cooling system operating state, and realizes real-time and continuous monitoring of the unit's evaporative cooling system operating state, which not only improves the continuity of the unit's evaporative cooling system abnormality monitoring, but also improves the accuracy of the operating state monitoring, can identify the abnormality of the unit's evaporative cooling system as early as possible in the early stage of the fault, and is convenient for timely taking measures to avoid unplanned shutdown of the hydropower unit.

[0024] 4) The present invention utilizes a state assessment model based on grey correlation to monitor the operating state of the evaporative cooling system of the unit in real time and continuously, and can monitor and discover abnormal development trends of the evaporative cooling system in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0026] Figure 1 It is a structural schematic diagram of the evaporative cooling system of the unit according to an embodiment of the present invention.

[0027] Figure 2 Schematic diagram of a monitoring data acquisition system according to an embodiment of the present invention.

[0028] Figure 3 Schematic diagram of a method for monitoring the operating condition of an evaporative cooling system of a unit according to an embodiment of the present invention.

[0029] Figure 4 It is a frame waveform diagram of the voiceprint signal according to an embodiment of the present invention.

[0030] Figure 5 1 is a diagram of the Hamming window and windowed waveform of a voiceprint signal according to an embodiment of the present invention.

[0031] Figure 6 FIG. 4 is a spectrum diagram of a voiceprint signal according to an embodiment of the present invention.

[0032] Figure 7 4 is a logarithmic energy waveform diagram of a voiceprint signal according to an embodiment of the present invention.

[0033] Figure 8 MFCC waveform diagram of the voiceprint signal according to an embodiment of the present invention.

[0034] Explanation of the reference numerals: condenser 1, upper gas collecting ring pipe 2, lower liquid collecting ring pipe 3, stator wire rod 4, condenser pickup 5, gas collecting pickup 6, liquid collecting pickup 7, stator wire rod temperature sensor 8, water inlet temperature sensor 9, water outlet temperature sensor 10, equalizing tube 11, upper guide tube 12, upper insulating drainage tube 13, lower insulating drainage tube 14. DETAILED DESCRIPTION

[0035] In the embodiment, three soundprint collectors are arranged on the evaporative cooling system of the hydropower unit, wherein the condenser pickup 5 is installed on the condenser 1, the gas collection pickup 6 is installed on the upper gas collection ring pipe 2, and the liquid collection pickup 7 is arranged on the lower liquid collection ring pipe 3; the stator wire rod temperature sensor 8 is arranged on the stator wire rod 4, which is used to collect the stator wire rod temperature signal data; the inlet water temperature sensor 9 and the outlet water temperature sensor 10 are respectively arranged at the water inlet and the water outlet of the condenser 1, which are used to collect the condenser water inlet and outlet temperature signal data; a liquid level sensor is installed on the evaporative cooling medium device, which is used to collect the evaporative cooling medium liquid level signal data; a flow sensor is installed in the condenser 1, which is used to collect the condenser cooling water flow signal data; a pressure sensor is arranged in the evaporative cooling system pipeline, which is used to collect the evaporative cooling system pressure signal data, such as Figure 1 and Figure 2 shown.

[0036] In the embodiment, the voiceprint sensor adopts ZKHY-IVOD-013 industrial diagnostic pickup, with a collection frequency of 20HZ-20KHZ; the temperature sensor adopts PT100 thermistor sensor, with a measurement range of -200℃~850℃; the liquid level sensor adopts HH9C-JZB-S float-type water level sensor; the flow sensor adopts HWLDE-200 electromagnetic flow sensor; the pressure sensor adopts PX305-100V pressure sensor. The voiceprint collection edge gateway terminal integrated machine ZKHY-IVH-002, the collection card has 8 audio inputs, and the input voltage is DC 9~36V. The voiceprint server adopts the cloud engine server ZKHY-VCES-2000.

[0037] like Figure 3 As shown, the method for monitoring the operating condition of the unit evaporative cooling system based on MFCC and grey correlation degree includes the following steps: Step 1: Collect the real-time soundprint signals of the condenser, liquid collecting ring pipe, and gas collecting ring pipe in the working state of the evaporative cooling system; and obtain the real-time temperature, liquid level, flow rate, and pressure data of the evaporative cooling system. The temperature, liquid level, flow rate, and pressure data of the evaporative cooling system include the unit stator bar temperature, condenser water inlet temperature, condenser water outlet temperature, evaporative cooling medium level, condenser cooling water flow rate, and evaporative cooling system pressure.

[0038] Step 2: Use the Mel-frequency cepstral coefficient method to extract feature vectors from the voiceprint signal.

[0039] The specific process of extracting feature vectors from voiceprint signals in the embodiment is as follows: S1: Voice signal reading: Use the voiceprint sensor to collect the voiceprint signal of the evaporative cooling system when it is running at a sampling rate of 16kHz and a sampling accuracy of 24bit, and use the audio processing library to read the input 50ms voice signal.

[0040] S2: Process the sound samples; Preprocessing: Perform pre-emphasis on the audio signal, and set the pre-emphasis coefficient to 0.97. Then, divide the audio signal into frames, each frame length is 25ms, and the frame shift is 10ms. Finally, perform windowing on each frame, and select the Hamming window as the window function.

[0041] In the embodiment, a frame signal output by a voiceprint sensor is as follows: Figure 4 As shown, the waveform of the frame signal after Hamming window processing is as follows Figure 5 shown.

[0042] S3: Fast Fourier Transform: Perform fast Fourier transform on each frame of speech signal to obtain its spectrum representation. The number of fast Fourier transform points is set to 512. Figure 5 The spectrum of the windowed signal is shown as Figure 6 shown.

[0043] S4: Mel filtering: Design a Mel filter bank with 24 filters and a filter frequency range covering the entire spectrum range of the speech signal. Filter the spectrum obtained by fast Fourier transform through the Mel filter bank to obtain a Mel spectrum.

[0044] S5: Take logarithm: perform logarithmic transformation on the Mel spectrum to obtain the logarithmic Mel spectrum. The logarithmic transformation uses a natural logarithmic function. Figure 5 The logarithmic energy waveform of the windowed signal after Mel filtering is shown in Figure 7 shown.

[0045] S6: Discrete cosine transform: Perform discrete cosine transform on the logarithmic Mel spectrum to obtain the MFCC feature vector. Discrete cosine transform can compress the highly correlated features in the Mel spectrum, reduce the feature dimension, and retain important information that affects sound recognition.

[0046] Figure 5 The Mel frequency cepstrum coefficients of the windowed signal shown in FIG. 1 are obtained after Mel filtering, logarithmic transformation and discrete cosine transformation. Figure 8 shown.

[0047] Step 3: Construct a state assessment model based on grey correlation. The state assessment model determines the sequence data under each working condition as the reference sequence, takes the characteristic data of the assessment object that changes over time as the comparison sequence for comparison, calculates the correlation between the comparison sequence and the reference sequence, and obtains the grey correlation between the assessment object and each working condition.

[0048] The calculation formula for the degree of correlation between the comparison sequence and the reference sequence is:

[0049] In the formula For comparison vector For the reference vector The degree of association, where the resolution coefficient , generally the value is 0.5.

[0050] Step 4: After preprocessing the temperature, liquid level, flow rate and pressure data of the evaporative cooling system obtained in step 1 and the characteristic vector of the voiceprint signal obtained in step 2, a comparison vector, i.e., a comparison sequence, is obtained for evaluation and comparison, and the comparison vector is input into the state evaluation model to calculate the grey correlation coefficient between the comparison sequence and the reference sequence of different working conditions.

[0051] The reference sequence, i.e. the reference vector, is:

[0052] In the formula Represents the condenser soundprint signal, Represents the soundprint signal of the liquid collecting ring pipe, Represents the soundprint signal of the gas collecting ring pipe. represents the stator bar temperature, represents the condenser inlet water temperature, Represents the condenser outlet water temperature, Represents the evaporative cooling medium level, represents the condenser cooling water flow, Represents the evaporative cooling system pressure; The comparison sequence is , t represents the time, are the comparison vectors of the evaluation object at different times; Calculate separately and degree of association, and and The average value of the correlation degree is taken as the grey correlation degree.

[0053] Step 5: According to the grey correlation coefficient obtained in step 4, the operating state corresponding to the maximum grey correlation coefficient is selected as the evaluation result of the real-time operating state of the unit evaporative cooling system.

[0054] In the embodiment, after obtaining the real-time operating state of the unit evaporative cooling system, if the real-time operating state of the unit evaporative cooling system is abnormal, the comparison sequence of the current unit evaporative cooling system is used as the feature sequence to be diagnosed, and the gray correlation between the feature sequence to be diagnosed and the typical fault feature sequence of each evaporative cooling system fault type is calculated, and the gray correlation is sorted, and the fault type with the largest gray correlation is the fault type of the feature sequence to be diagnosed. For the obtained fault type, an alarm is triggered.

Claims

1. The method for monitoring the working condition of the unit evaporative cooling system based on MFCC and grey correlation is characterized in that: The following steps are involved: Step 1: Collect the soundprint signals of the condenser, liquid collecting ring pipe, and gas collecting ring pipe when the evaporative cooling system is in working state; and obtain the temperature, liquid level, flow rate, and pressure data of the evaporative cooling system; Step 2: Using the Mel frequency cepstral coefficient method, extracting a feature vector from the voiceprint signal; Step 2.1: Pre-process the voiceprint signal by pre-emphasis, framing and windowing; Step 2.2: Deconstruct the preprocessed time domain signal into frequency domain signal using Fourier transform method; Step 2.3: Filter the obtained spectrum through a Mel filter to divide the spectrum into multiple Mel filter groups; Step 2.4: Calculate the logarithmic energy of the output of the filter bank, and then perform discrete cosine transform to obtain the Mel frequency cepstrum coefficients; Step 3: Construct a state assessment model based on grey relational degree; The state assessment model determines the sequence data under each working state as a reference sequence, takes the characteristic data of the assessment object changing over time as a comparison sequence for comparison, calculates the correlation between the comparison sequence and the reference sequence, and obtains the grey correlation between the assessment object and each working state; Step 4: After preprocessing the temperature, liquid level, flow rate and pressure data of the evaporative cooling system obtained in step 1 and the characteristic vector of the voiceprint signal obtained in step 2, a comparison vector, i.e., a comparison sequence, is obtained for evaluation and comparison, and the comparison vector is input into the state evaluation model to calculate the grey correlation coefficient between the comparison sequence and the reference sequence of different working conditions; Step 5: According to the grey correlation coefficient obtained in step 4, the operating state corresponding to the maximum grey correlation coefficient is selected as the operating condition evaluation result of the unit evaporative cooling system.

2. The method for monitoring the operating condition of the unit evaporative cooling system according to claim 1, characterized in that: In step 1, the temperature, liquid level, flow and pressure data of the evaporative cooling system include the unit stator bar temperature, condenser water inlet temperature, condenser water outlet temperature, evaporative cooling medium level, condenser cooling water flow and evaporative cooling system pressure.

3. The method for monitoring the operating condition of the unit evaporative cooling system according to claim 2, characterized in that: In step 2.1, a Hamming window is used to add a window to each frame signal after framing. The calculation formula of the Hamming window is: ； In the formula represents the window function, N represents the number of frames, is the coefficient and n is the index of the sample point.

4. The method for monitoring the operating condition of the unit evaporative cooling system according to claim 3, characterized in that: In step 2.2, the formula of the Fourier transform method is: ； In the formula Represents the input sound signal, represents the number of Fourier transform points, k is the frequency index, Represents a frequency domain signal.

5. The method for monitoring the operating condition of the unit evaporative cooling system according to claim 4, characterized in that: In step 2.3, the filter calculation formula is: ； Where k is the frequency index; m is the serial number of the Mel filter; f is the actual frequency; Represents the filter function.

6. The method for monitoring the operating condition of the unit evaporative cooling system according to claim 5, characterized in that: In step 2.4, the formula for converting frequency to Mel scale is: ； In the formula, Represents the frequency after conversion to Mel scale, Indicates the actual frequency.

7. The method for monitoring the operating condition of the unit evaporative cooling system according to claim 6, characterized in that: In step 2.4, the calculation formula of the Mel frequency cepstral coefficient is: ； In the formula represents the Mel-frequency cepstral coefficient; Indicates the order of Mel-frequency cepstral coefficients; s(m) represents the mth sample value of the speech signal after pre-emphasis, framing and windowing preprocessing; M represents the number of points in the logarithmic scale, is the order number of the Mel-frequency cepstrum coefficients.

8. The method for monitoring the operating condition of the unit evaporative cooling system according to claim 7, characterized in that: In step 3, the calculation formula for the correlation between the comparison sequence and the reference sequence is: ； In the formula For comparison vector For the reference vector The degree of association, where the resolution coefficient .

9. The method for monitoring the operating condition of the unit evaporative cooling system according to claim 8, characterized in that: In step 3, the reference sequence, i.e., the reference vector, is: ； In the formula Indicates the condenser soundprint signal, Indicates the soundprint signal of the liquid collecting ring pipe. Indicates the soundprint signal of the gas collecting ring pipe. represents the stator bar temperature, Indicates the condenser water inlet temperature. Indicates the condenser outlet water temperature. Indicates the evaporative cooling medium level, Indicates the condenser cooling water flow rate, Indicates the evaporative cooling system pressure; The comparison sequence is , t represents the time, are the comparison vectors of the evaluation object at different times; Calculate separately and degree of association, and and The average value of the correlation degree is taken as the grey correlation degree.

10. The method for monitoring the operating condition of the unit evaporative cooling system according to claim 9, characterized in that: In step 5, the grey correlation coefficients of the comparison sequence obtained in step 4 and the reference sequences of different operating conditions are sorted from large to small, the maximum grey correlation coefficient in the grey correlation coefficient sequence is selected, and the operating condition corresponding to the maximum grey correlation coefficient is used as the evaluation result of the operating condition of the current group of evaporative cooling systems.

Images