System and method for monitoring operation state of voiceprint acquisition equipment
Through the trend analysis and risk coefficient model of voiceprint collection equipment, the data fluctuations and delay determination problems in wind unit fault monitoring are solved, the fault positioning and risk assessment are achieved, the maintenance sequence is optimized, and the operation of wind unit is improved Efficiency and fault handling efficiency.
Patent Information
- Application Number
- CN202411190426.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-28
- Publication Date
- 2024-11-12
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing wind turbine status monitoring system has data fluctuations and delay determination problems in fault monitoring, making it difficult to accurately determine the fault area and cause of the fault, and cannot conduct predictive analysis to evaluate the fault risk level.
The operating status monitoring system of voiceprint collection equipment is adopted. By obtaining the electrical signal set of wind turbines, the trend analysis is performed, the trend chart is compared with the preset sudden failure trend chart, the judgment label is generated and risk prediction is carried out, and the failure risk is evaluated using the risk coefficient model. level.
It realizes accurate positioning and analysis of the cause of failures in wind turbines, evaluates the degree of impact of failures on unit operation, and optimizes the maintenance sequence, improving the efficiency and accuracy of fault handling.
Smart Images

Figure CN118934497A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of wind turbine monitoring, and more specifically, to a system and method for monitoring the operating status of a voiceprint collection device. Background Art
[0002] A wind turbine is a device used to convert wind energy into electrical energy. A wind turbine captures wind energy through a wind rotor, thereby driving the main shaft to rotate, and then driving the generator to generate electricity. It is widely used in various wind power plants. With the popularization of new energy sources, the installation of wind turbines has continued to expand, becoming one of the transformation directions in the global energy transformation, which has led to an increase in the number of wind turbines installed in a region, and thus an increase in the difficulty of fault monitoring of wind turbines. If the fault of a wind turbine during operation cannot be discovered, the operating efficiency of the wind turbine will decrease, and the wind power conversion efficiency will decrease, thereby affecting the stability of energy supply. In addition, if the fault of a wind turbine is not identified and handled in a timely manner, it will also cause sudden failure of the wind turbine, which will lead to extended downtime and repair time of the wind turbine in the future.
[0003] At present, the existing wind turbine status monitoring system mainly uses various sensor acquisition devices for status monitoring, such as voiceprint acquisition devices, temperature acquisition devices, and vibration acquisition devices, but the data directly obtained by the sensor is easily affected by the environment and fluctuates, and it is easy to obtain erroneous data; for example, the Chinese patent with the authorization publication number CN105179179B discloses a full-state monitoring method and system for a wind turbine generator set, and the Chinese patent with the authorization publication number CN108021026B discloses a fault warning and control parameter online optimization method for a wind turbine generator set. Although the above methods and prior arts can realize the status monitoring of wind turbine sets, research and practical application of the above methods and prior arts have found that the above methods and prior arts have at least the following defects:
[0004] (1) When a sudden fault occurs in a wind turbine, due to the delay in the manifestation of the fault, that is, when the fault occurs, the wind turbine does not show obvious characteristics, and it is difficult to timely determine whether the wind turbine has a fault by using sensor collection equipment to analyze the electrical signal, temperature or vibration of the wind turbine. When the wind turbine has fault manifestations, the fault area has continued to operate for a period of time, thereby aggravating the impact of the wind turbine fault. Furthermore, when there is no fault manifestation, it is difficult to determine the fault area and fault cause;
[0005] (2) In the process of status monitoring of the wind turbine, it is impossible to perform predictive analysis based on the monitoring data of the wind turbine, so it is impossible to obtain the risk level of the wind turbine continuing to operate after a sudden failure occurs, and further it is impossible to determine the extent of the impact of the sudden failure on the wind turbine. Summary of the invention
[0006] The purpose of the present invention is to disclose and provide a system and method for monitoring the operating status of a voiceprint collection device to address the deficiencies in the background technology.
[0007] A voiceprint collection equipment operation status monitoring system, comprising:
[0008] The first analysis module is used to obtain the electrical signal set of each monitoring area on the wind turbine within a given time, and perform trend analysis on the electrical signal set to obtain an electrical signal trend graph;
[0009] Trend chart comparison module: used to compare the electrical signal trend chart of each monitoring area with the preset sudden fault electrical signal trend chart for similarity, so as to obtain the monitoring area with sudden faults and the corresponding sudden fault causes;
[0010] The second analysis module is used to obtain the number of monitoring areas where sudden faults exist, determine the number of areas and a predetermined area number threshold, and generate a determination label for the wind turbine, wherein the determination label includes "perform risk prediction" or "do not perform risk prediction";
[0011] Coefficient prediction module: used to collect data from the wind turbine that generates the "risk prediction" judgment result, obtain the temperature data of the monitoring area of the wind turbine, and perform image stitching on the electrical signal trend graphs of each monitoring area to generate comprehensive image data, and send the generated temperature data and comprehensive image data to the preset risk coefficient prediction model to obtain the risk coefficient of the wind turbine;
[0012] Risk determination module: used to compare the risk factor with the predetermined risk factor threshold to obtain the risk level of the wind turbine after continuing to operate under sudden failure
[0013] Furthermore, the monitoring area includes a bearing area, a gearbox area, a generator area, and a wind rotor area.
[0014] Further, the performing trend analysis on the electrical signal set includes:
[0015] Collecting an electrical signal set of each monitoring area in real time according to a time series, wherein the electrical signal set includes a plurality of voltage values / current values;
[0016] Divide the electrical signal set into time periods according to the order of acquisition time to obtain electrical signal subsets of at least three time periods, and mark the three time periods as a, b, and c respectively;
[0017] The medians Med1, Med2, and Med3 of the electrical signal subsets in time periods a, b, and c are obtained respectively, and the maximum trend value and the minimum trend value are calculated based on the medians Med1, Med2, and Med3, so as to obtain the maximum trend value Amax and the minimum trend value Amin in time period a, the maximum trend value Bmax and the minimum trend value Bmin in time period b, and the maximum trend value Cmax and the minimum trend value Cmin in time period c respectively. According to the maximum trend value and the minimum trend value in time periods a, b, and c, the electrical signal trend graphs corresponding to time periods a, b, and c are obtained.
[0018] Furthermore, the acquisition logic of the electrical signal trend graph corresponding to the time periods a, b, and c is as follows:
[0019] The electrical signal subsets in time periods a, b, and c are divided into high-order data sets and low-order data sets according to the medians Med1, Med2, and Med3, respectively;
[0020] The variance of the high-order data set and the low-order data set is calculated to obtain the maximum trend value and the minimum trend value of the electrical signal subset in time periods a, b, and c. The calculation formula for the maximum trend value is: γ is a, b or c; the calculation formula for the minimum trend value is γ is a, b or c;
[0021] The maximum trend value and the minimum trend value in the time periods a, b, and c are sorted out, and a two-dimensional plane diagram is constructed according to the time-trend value to obtain the electrical signal trend diagram of the time periods a, b, and c.
[0022] Furthermore, the logic for comparing the similarity between the electric signal trend graph and the preset sudden fault electric signal trend graph is:
[0023] Compare the similarity between the electric signal trend graph and the preset sudden fault electric signal trend graph according to the similarity algorithm. If the similarity is greater than the preset first similarity threshold, it is determined that there is a sudden fault in the monitoring area corresponding to the electric signal trend graph; extract the monitoring area with the sudden fault and the corresponding sudden fault cause;
[0024] If the similarity is less than the preset first similarity threshold, no processing is performed.
[0025] Further, the number of monitoring areas where sudden faults exist is obtained, and the number of areas is compared with a predetermined area number threshold, including:
[0026] Obtain the number of monitoring areas Xf where sudden failures occur;
[0027] The number of monitoring areas with sudden faults Xf is compared with a predetermined area number threshold Xt, where Xt is equal to 1;
[0028] If the number of regions Xf is greater than or equal to the predetermined region number threshold Xt, a judgment label of "perform risk prediction" is generated; if the number of regions Xf is less than the predetermined region number threshold Xt, a judgment label of "do not perform risk prediction" is generated.
[0029] Furthermore, the temperature data is obtained by averaging the temperature values collected in each monitoring area; the generation logic of the risk coefficient prediction model is:
[0030] Acquire historical wind turbine prediction data, and divide the historical wind turbine risk prediction data into a test set and a training set, wherein the historical wind turbine prediction data includes comprehensive image data, temperature data, and corresponding risk coefficients;
[0031] Construct a regression network, use the comprehensive image data and temperature data in the training set as the input of the regression network, use the corresponding risk coefficient as the output of the regression network, train the regression network, and obtain an initial risk coefficient prediction network;
[0032] The test set is used to verify the model of the initial risk coefficient prediction network, and the initial risk coefficient prediction network with a value less than or equal to the preset test error threshold is output as the risk coefficient prediction model.
[0033] Furthermore, the logic for generating the risk coefficient is:
[0034] Acquire the voiceprint data of each monitoring area within a specified time span through a voiceprint acquisition device, wherein the voiceprint data includes a voiceprint frequency fluctuation value or a voiceprint amplitude fluctuation value, and the voiceprint data includes bearing voiceprint data, gearbox voiceprint data, generator voiceprint data, and wind wheel voiceprint data;
[0035] Obtain the fluctuation pre-calibration value of each monitoring area, and compare the fluctuation pre-calibration value with the voiceprint data to calculate the fluctuation coefficient of each monitoring area;
[0036] Integrate the fluctuation coefficients of each monitoring area to obtain the risk coefficient;
[0037] The fluctuation coefficients of each monitoring area are integrated, including:
[0038] The risk coefficient is calculated based on the generated bearing soundprint data, gearbox soundprint data, engine soundprint data and wind wheel soundprint data, and the calculation formula is: Among them, Rkc is the risk factor of the wind turbine.
[0039] Furthermore, the risk level includes a first-level risk level, a second-level risk level and a third-level risk level, and the risk coefficient threshold includes a first risk threshold and a second risk threshold, wherein the first risk threshold is less than the second risk threshold, and the first risk threshold and the second risk threshold are both integers greater than 0;
[0040] The logic for generating the risk level is:
[0041] The risk factor is compared with a predetermined risk factor threshold, and if the risk factor is less than or equal to a first risk threshold, the wind turbine is marked as a first-level risk level;
[0042] If the risk factor is greater than the first risk threshold, and the risk factor is less than or equal to the second risk threshold, the wind turbine is marked as a second-level risk level;
[0043] If the risk factor is greater than or equal to the second risk threshold, the wind turbine is marked as a third-level risk level.
[0044] A method for monitoring the operating status of a voiceprint collection device, comprising:
[0045] Obtaining an electrical signal set of each monitoring area on the wind turbine within a given time, and performing trend analysis on the electrical signal set to obtain an electrical signal trend graph;
[0046] Compare the electrical signal trend graph of each monitoring area with the preset sudden fault electrical signal trend graph for similarity, so as to obtain the monitoring area with sudden fault and the corresponding sudden fault cause;
[0047] Obtaining the number of monitoring areas where sudden faults exist, comparing the number of areas with a predetermined threshold of the number of areas, and generating a determination label for the wind turbine, wherein the determination label includes "performing risk prediction" or "not performing risk prediction";
[0048] Data collection is performed on the wind turbines that generate the "risk prediction" judgment result to obtain the temperature data of the monitoring area of the wind turbines, and the electrical signal trend graphs of each monitoring area are image stitched to generate comprehensive image data, and the generated temperature data and comprehensive image data are sent to a preset risk coefficient prediction model to obtain the risk coefficient of the wind turbines;
[0049] The risk factor is compared with a predetermined risk factor threshold to obtain the risk level of the wind turbine after continuing to operate under a sudden failure.
[0050] Compared with the prior art, the present invention has the following beneficial effects:
[0051] The present invention discloses a system and method for monitoring the operation status of a voiceprint collection device, comprising: obtaining an electrical signal set of each monitoring area on a wind turbine within a predetermined time, performing trend analysis on the electrical signal set to obtain an electrical signal trend graph; performing similarity comparison between the electrical signal trend graph of each monitoring area and a preset sudden fault electrical signal trend graph to obtain monitoring areas with sudden faults and corresponding sudden fault causes; obtaining the number of monitoring areas with sudden faults, determining the number of areas with a predetermined area number threshold, and generating a determination label for the wind turbine; performing data collection on a wind turbine that generates a "risk prediction" determination result to obtain temperature data of the monitoring area of the wind turbine, and performing image stitching on the electrical signal trend graphs of each monitoring area to generate a comprehensive graph The image data is generated, and the generated temperature data and comprehensive image data are sent to a preset risk factor prediction model to obtain the risk factor of the wind turbine; the risk factor is compared with a preset risk factor threshold to obtain the risk level of the wind turbine after continuing to operate under a sudden fault; the present invention realizes the accurate positioning of the sudden fault monitoring area and the cause of the fault during the fault inspection and repair process by comparing the electrical signal trend graph with the sudden fault electrical signal trend graph; the present invention also generates a risk factor for predicting that the wind turbine continues to operate after a sudden fault occurs through a risk factor prediction model, and generates a risk level for judging the degree of influence of the sudden fault on the wind turbine according to the risk factor and the preset risk factor threshold, and optimizes the maintenance sequence of the wind turbine according to the risk level. BRIEF DESCRIPTION OF THE DRAWINGS
[0052] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.
[0053] Figure 1 A flow chart of a method for monitoring the operating status of a voiceprint collection device provided in Example 1 of the present invention;
[0054] Figure 2 A flow chart of a method for monitoring the operating status of a voiceprint collection device provided in Embodiment 2 of the present invention;
[0055] Figure 3 This is a module diagram of the voiceprint collection device operation status monitoring system provided in Embodiment 3 and Embodiment 4 of the present invention. DETAILED DESCRIPTION
[0056] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0057] Example 1
[0058] See also Figure 1 As shown, this embodiment discloses a method for monitoring the operating status of a voiceprint collection device, the method comprising:
[0059] S110: Obtaining an electrical signal set of each monitoring area on the wind turbine within a predetermined time, and performing trend analysis on the electrical signal set to obtain an electrical signal trend graph;
[0060] Specifically, the monitoring area includes a bearing area, a gearbox area, a generator area, and a wind rotor area;
[0061] It should be noted that the monitoring area is an important functional execution area in the process of wind power generation, and faces a variety of unfavorable conditions during operation, which makes these areas prone to failure; among them, in the bearing area, the bearing bears the load of the wind wheel, gearbox and generator, and the operating stress is large. Wind turbines are usually in harsh environments, such as high humidity, low temperature or dust, which can easily lead to poor lubrication and increased wear in the bearing area. In addition, the bearing works at a high speed and has high friction, which can easily lead to overheating and wear. The gearbox area is composed of multiple gears, bearings and other mechanical components, and has a complex structure. The failure of a single component will affect the operation of the entire gearbox. The gearbox also needs to convert the low-speed and high-torque of the wind wheel into the high-speed and low-torque required by the generator. The transmission load is large, and vibration and impact are easily generated during the meshing process of the gears in the gearbox area, increasing the risk of wear. The generator area is easily affected by current shock, short circuit or insulation aging, and the rotating parts in the generator area work under high load and high speed, which is prone to mechanical stress fatigue. In the wind wheel area, the wind wheel is exposed to the harsh natural environment and is easily affected by wind, rain, ice and snow, which leads to wear of the wind wheel during long-term use. At the same time, the wind wheel is subjected to huge dynamic loads and centrifugal forces during rotation, which is prone to operational failures.
[0062] The predetermined time is the time range for collecting the electrical signal set after artificial experimental analysis;
[0063] Specifically, the trend analysis of the electrical signal set includes:
[0064] S111: collecting an electrical signal set of each monitoring area in real time according to a time series, wherein the electrical signal set includes a plurality of voltage values / current values;
[0065] S112: Divide the electrical signal set into time periods according to the order of acquisition time to obtain electrical signal subsets of at least three time periods, and mark the three time periods as a, b, and c respectively;
[0066] It should be noted that: the acquisition time of the electric signal subset in the time period a takes precedence over the acquisition time of the electric signal subset in the time period b, and the acquisition time of the electric signal sub-data in the time period b takes precedence over the acquisition time of the electric signal subset in the time period c;
[0067] S113: respectively obtaining the medians Med1, Med2, and Med3 of the electrical signal subsets in time periods a, b, and c, and calculating the maximum trend value and the minimum trend value according to the medians Med1, Med2, and Med3, respectively obtaining the maximum trend value Amax and the minimum trend value Amin in time period a, the maximum trend value Bmax and the minimum trend value Bmin in time period b, and the maximum trend value Cmax and the minimum trend value Cmin in time period c, and obtaining the electrical signal trend graphs corresponding to time periods a, b, and c according to the maximum trend value and the minimum trend value in time periods a, b, and c;
[0068] Specifically, the acquisition logic of the electrical signal trend graph corresponding to the time periods a, b, and c is:
[0069] The electrical signal subsets in time periods a, b, and c are divided into high-order data sets and low-order data sets according to the medians Med1, Med2, and Med3, respectively;
[0070] It should be understood that: the high-order data set is a subset of electrical signals greater than the medians Med1, Med2, and Med3 in time periods a, b, and c; the low-order data set is a subset of electrical signals less than the medians Med1, Med2, and Med3 in time periods a, b, and c;
[0071] The variance of the high-order data set and the low-order data set is calculated to obtain the maximum trend value and the minimum trend value of the electrical signal subset in time periods a, b, and c. The calculation formula for the maximum trend value is: γ is a, b or c; the calculation formula for the minimum trend value is γ is a, b or c;
[0072] The maximum trend value and the minimum trend value in the time periods a, b, and c are sorted out, and a two-dimensional plane diagram is constructed according to the time-trend value to obtain the electrical signal trend diagram of the time periods a, b, and c;
[0073] It should be understood that: the trend value includes a maximum trend value and a minimum trend value.
[0074] S120: comparing the electrical signal trend graph of each monitoring area with the preset sudden fault electrical signal trend graph for similarity, so as to obtain the monitoring area where the sudden fault exists and the corresponding sudden fault cause;
[0075] Specifically, the logic for comparing the similarity between the electric signal trend graph and the preset sudden fault electric signal trend graph is:
[0076] Compare the similarity between the electric signal trend graph and the preset sudden fault electric signal trend graph according to the similarity algorithm. If the similarity is greater than the preset first similarity threshold, it is determined that there is a sudden fault in the monitoring area corresponding to the electric signal trend graph; extract the monitoring area with the sudden fault and the corresponding sudden fault cause;
[0077] If the similarity is less than the preset first similarity threshold, no processing is performed;
[0078] It should be noted that: the similarity algorithm includes but is not limited to perceptual hashing or Euclidean distance, and before the similarity comparison is performed, it needs to be preprocessed, including but not limited to image enhancement and image denoising;
[0079] The sudden fault electrical signal trend diagram and its corresponding sudden fault cause are determined by artificial experimental analysis.
[0080] The above process realizes accurate positioning of the sudden fault monitoring area and the cause of the fault during the fault inspection and repair process by comparing the electrical signal trend graph with the sudden fault electrical signal trend graph;
[0081] It should be noted that the degree of impact of the fault is determined based on the maintenance time, maintenance cost and maintenance difficulty; the impact of the fault on the continued operation of the wind turbine after a sudden fault occurs is specifically as follows: after a sudden fault occurs in the generator area, the continued operation of the wind turbine will cause electrical failure, overheating or burning, thereby affecting the power generation capacity of the wind turbine; after a sudden fault occurs in the bearing area, the continued operation of the wind turbine will cause increased bearing wear, reduced operating efficiency and increased noise; when a sudden fault occurs in the rotor area, the wind turbine continues to operate, the rotor blades will be damaged or unbalanced, and the wind energy capture efficiency will be reduced; when a sudden fault occurs in the gearbox area, the wind turbine continues to operate, the gears will wear or be damaged, affecting the power transmission efficiency.
[0082] Example 2
[0083] Reference Figure 2As shown, according to the number of monitoring areas where sudden faults exist, the risk level of the wind turbine is predicted. Specifically, this embodiment discloses a method for monitoring the operating status of a voiceprint collection device, and the method further includes:
[0084] S130: obtaining the number of monitoring areas where sudden faults exist, comparing the number of areas with a predetermined threshold of the number of areas, and generating a determination label for the wind turbine, wherein the determination label includes “performing risk prediction” or “not performing risk prediction”;
[0085] Specifically, the number of monitoring areas where sudden faults exist is obtained, and the number of areas is compared with a predetermined area number threshold value, including:
[0086] Z1: Get the number of monitoring areas with sudden failures Xf;
[0087] Z2: Compare the number of monitoring areas with sudden faults Xf with the predetermined area number threshold Xt, where Xt is equal to 1;
[0088] Z3: If the number of regions Xf is greater than or equal to the predetermined number of regions threshold Xt, a judgment label of "perform risk prediction" is generated; if the number of regions Xf is less than the predetermined number of regions threshold Xt, a judgment label of "do not perform risk prediction" is generated;
[0089] Among them, there are sudden faults in the four monitoring areas set on the wind turbine generating the "no risk prediction" judgment label, and the wind turbine generating the "risk prediction" judgment label needs to be analyzed in step S140, while the wind turbine generating the "no risk prediction" judgment label is in a normal operating state, and there is no need to perform state analysis on the wind turbine;
[0090] It should be noted that the predetermined area quantity threshold is set through artificial experimental analysis.
[0091] S140: collecting data from the wind turbine for which a “risk prediction” determination result is generated, obtaining temperature data of the monitoring area of the wind turbine, and performing image stitching on the electrical signal trend graphs of each monitoring area to generate comprehensive image data, and sending the generated temperature data and comprehensive image data to a preset risk coefficient prediction model to obtain the risk coefficient of the wind turbine;
[0092] Specifically, the temperature data is obtained by calculating the average value of the temperature values collected in each monitoring area;
[0093] It should be noted that: the steps of obtaining temperature data are illustrated by taking an example, the temperature values of the bearing area, gearbox area, generator area and wind wheel area are collected by temperature sensors set in each monitoring area, and the collected temperature values are marked as T1, T2, T3 and T4 respectively, and the average of T1, T2, T3 and T4 is calculated to obtain the average value Tp;
[0094] It is important to understand that: Numpy is used to stitch the electrical signal trend graphs of each monitoring area to generate a comprehensive image. Before stitching the images, the electrical signal trend graphs of each monitoring area need to be preprocessed;
[0095] Specifically, the generation logic of the risk coefficient prediction model is:
[0096] S141: Acquire historical wind turbine prediction data, and divide the historical wind turbine risk prediction data into a test set and a training set, wherein the historical wind turbine prediction data includes comprehensive image data, temperature data, and corresponding risk coefficients;
[0097] Specifically, the logic for generating the risk coefficient is:
[0098] g1: Acquire the voiceprint data of each monitoring area within a specified time span through a voiceprint collection device, wherein the voiceprint data includes a voiceprint frequency fluctuation value or a voiceprint amplitude fluctuation value, and the voiceprint data includes bearing voiceprint data, gearbox voiceprint data, generator voiceprint data, and wind wheel voiceprint data;
[0099] It should be noted that: the voiceprint data can be a voiceprint frequency fluctuation value or a voiceprint amplitude fluctuation value, wherein the voiceprint frequency and voiceprint amplitude are obtained by a voiceprint collection device;
[0100] The voiceprint frequency fluctuation value is calculated by the difference between the maximum voiceprint frequency and the minimum voiceprint frequency within the specified time span, and the voiceprint vibration fluctuation value is calculated by the difference between the maximum voiceprint amplitude and the minimum voiceprint amplitude within the specified time span;
[0101] The soundprint frequency refers to the vibration frequency of the sound wave signal detected in the monitoring area, which can be understood as the number of vibrations of the sound wave signal per second. The soundprint amplitude refers to the strength / amplitude of the sound wave signal detected in the monitoring area. The larger the amplitude, the stronger the energy of the sound wave signal. The bearing soundprint data reflects the overall health status of the bearing operation. High-amplitude bearing soundprint data indicates that the bearing is worn or damaged. Abnormal soundprint frequencies in the bearing soundprint data correspond to different types of faults, such as bearing inner ring, bearing outer ring and bearing rolling element faults. The high-amplitude gearbox soundprint data reflects gear wear or poor meshing, while abnormal frequency gearbox soundprint data indicates gear failure, such as broken teeth, The soundprint amplitude in the generator soundprint data reflects the operating status of the mechanical or electrical components inside the generator. Abnormal soundprint amplitude indicates bearing failure or rotor imbalance of the generator, while abnormal soundprint frequency in the generator soundprint data indicates mechanical looseness, winding short circuit and other problems in the generator; abnormal soundprint amplitude in the wind wheel soundprint data reflects the balance state and operating status of the wind wheel. Abnormal soundprint amplitude indicates blade damage or loose connection, while abnormal soundprint frequency of the wind wheel soundprint data is used to indicate structural damage such as cracks and delamination of the blades; the above data are all collected by soundprint collection equipment, which includes but is not limited to acoustic sensors, vibration sensors and acoustic emission sensors;
[0102] It is important to understand that: the specified time span is set based on artificial experiments;
[0103] g2: Obtain the fluctuation pre-calibration value of each monitoring area, and compare the fluctuation pre-calibration value with the voiceprint data to calculate the fluctuation coefficient of each monitoring area;
[0104] Specifically, the fluctuation predetermined calibration values include bearing fluctuation calibration values, gearbox fluctuation calibration values, generator fluctuation calibration values and wind rotor fluctuation calibration values, and the fluctuation coefficients of each monitoring area include bearing fluctuation coefficients, gearbox fluctuation coefficients, generator fluctuation coefficients and wind rotor fluctuation coefficients;
[0105] The logic for generating the fluctuation coefficient of each monitoring area is:
[0106] The bearing soundprint data is compared with the bearing fluctuation standard value to calculate the bearing fluctuation coefficient. The calculation formula is: Among them, Bcf is the bearing fluctuation coefficient, Bvi is the bearing fluctuation standard value, and Baf is the bearing soundprint data;
[0107] The gearbox soundprint data is compared with the gearbox fluctuation standard value to calculate the gearbox fluctuation coefficient. The calculation formula is: Among them, Gcf is the gearbox fluctuation coefficient, Gvi is the gearbox fluctuation standard value, and Gaf is the gearbox soundprint data;
[0108] The engine soundprint data is compared with the engine fluctuation standard value to calculate the engine fluctuation coefficient. The calculation formula is: Among them, Ecf is the engine fluctuation coefficient, Evi is the engine fluctuation standard value, and Eaf is the engine soundprint data;
[0109] The wind wheel sound pattern data is compared with the wind wheel fluctuation standard value to calculate the wind wheel fluctuation coefficient. The calculation formula is: Rcf is the wind wheel fluctuation coefficient, Rvi is the wind wheel fluctuation standard value, and Raf is the wind wheel sound pattern data;
[0110] g3: Integrate the fluctuation coefficients of each monitoring area to obtain the risk coefficient;
[0111] Specifically, the fluctuation coefficients of each monitoring area are integrated, including:
[0112] The risk coefficient is calculated based on the generated bearing soundprint data, gearbox soundprint data, engine soundprint data and wind wheel soundprint data, and the calculation formula is: Where Rkc is the risk factor of the wind turbine;
[0113] It should be noted that the smaller the risk coefficient Rkc is, the more likely it is that the wind turbine will continue to operate after a sudden failure occurs, and the greater the impact of future failures of the wind turbine.
[0114] S142: constructing a regression network, taking the comprehensive image data and temperature data in the training set as inputs of the regression network, taking the corresponding risk coefficient as outputs of the regression network, training the regression network, and obtaining an initial risk coefficient prediction network;
[0115] S143: Using the test set to perform model verification on the initial risk coefficient prediction network, and outputting an initial risk coefficient prediction network that is less than or equal to a preset test error threshold as a risk coefficient prediction model;
[0116] It should be noted that the risk coefficient prediction model includes but is not limited to support vector machine, random forest and linear regression.
[0117] S150: comparing the risk factor with a predetermined risk factor threshold to obtain a risk level of the wind turbine after continuing to operate under a sudden fault;
[0118] Specifically, the risk level includes a first-level risk level, a second-level risk level, and a third-level risk level, and the risk coefficient threshold includes a first risk threshold and a second risk threshold, wherein the first risk threshold is less than the second risk threshold, and the first risk threshold and the second risk threshold are both integers greater than 0;
[0119] The logic for generating the risk level is:
[0120] The risk factor is compared with a predetermined risk factor threshold, and if the risk factor is less than or equal to a first risk threshold, the wind turbine is marked as a first-level risk level;
[0121] If the risk factor is greater than the first risk threshold, and the risk factor is less than or equal to the second risk threshold, the wind turbine is marked as a second-level risk level;
[0122] If the risk factor is greater than or equal to the second risk threshold, the wind turbine is marked as a third-level risk level;
[0123] It should be noted that the future failure impact of a wind turbine marked as a first-level risk level is greater than the future failure impact of a wind turbine marked as a second-level risk level, and so on. The failure impact is the same as the failure impact described in Example 1; the risk coefficient threshold is determined by artificial experimental analysis;
[0124] The risk coefficient prediction model is used to generate a risk coefficient for predicting whether the wind turbine will continue to operate after a sudden failure occurs. According to the risk coefficient and a predetermined risk coefficient threshold, a risk level is generated to determine the degree of impact of the sudden failure on the wind turbine, and the maintenance sequence of the wind turbine is optimized according to the risk level.
[0125] The impact degree of the fault is determined based on the maintenance time, maintenance cost and maintenance difficulty; the logic of maintenance sequence optimization is: the operator shuts down the wind turbines with sudden faults in the monitoring area, and then sorts them according to the third-level risk level to the first-level risk level to obtain the optimized maintenance sequence of the wind turbines;
[0126] It should be understood that: after the maintenance sequence optimization is completed, accurate fault maintenance is performed on the monitoring area with sudden faults according to the corresponding fault causes in Example 1; it can be understood that wind turbines with smaller fault impacts are repaired first, and then the corresponding monitoring area is repaired according to the cause of the sudden fault.
[0127] Example 3
[0128] Reference Figure 3 As shown, based on the same inventive concept of Example 1, this embodiment discloses a system for monitoring the operating status of a voiceprint collection device, the system comprising:
[0129] The first analysis module S210 is used to obtain an electrical signal set of each monitoring area on the wind turbine within a predetermined time, and perform trend analysis on the electrical signal set to obtain an electrical signal trend graph;
[0130] Specifically, the monitoring area includes a bearing area, a gearbox area, a generator area, and a wind rotor area; and the trend analysis of the electrical signal set includes:
[0131] Collecting an electrical signal set of each monitoring area in real time according to a time series, wherein the electrical signal set includes a plurality of voltage values / current values;
[0132] Divide the electrical signal set into time periods according to the order of acquisition time to obtain electrical signal subsets of at least three time periods, and mark the three time periods as a, b, and c respectively;
[0133] The medians Med1, Med2, and Med3 of the electrical signal subsets in time periods a, b, and c are obtained respectively, and the maximum trend value and the minimum trend value are obtained according to the medians Med1, Med2, and Med3, so as to obtain the maximum trend value Amax and the minimum trend value Amin in time period a, the maximum trend value Bmax and the minimum trend value Bmin in time period b, and the maximum trend value Cmax and the minimum trend value Cmin in time period c, respectively; and the electrical signal trend graphs corresponding to time periods a, b, and c are obtained according to the maximum trend value and the minimum trend value in time periods a, b, and c;
[0134] Specifically, the acquisition logic of the electrical signal trend graph corresponding to the time periods a, b, and c is:
[0135] The electrical signal subsets in time periods a, b, and c are divided into high-order data sets and low-order data sets according to the medians Med1, Med2, and Med3, respectively;
[0136] The variance of the high-order data set and the low-order data set is calculated to obtain the maximum trend value and the minimum trend value of the electrical signal subset in time periods a, b, and c. The calculation formula for the maximum trend value is: γ is a, b or c; the calculation formula for the minimum trend value is γ is a, b or c;
[0137] The maximum trend value and the minimum trend value in the time periods a, b, and c are sorted out, and a two-dimensional plane diagram is constructed according to the time-trend value to obtain the electrical signal trend diagram of the time periods a, b, and c;
[0138] Trend graph comparison module S220: used to compare the electrical signal trend graph of each monitoring area with the preset sudden fault electrical signal trend graph for similarity, so as to obtain the monitoring area where the sudden fault exists and the corresponding sudden fault cause;
[0139] Specifically, the logic for comparing the similarity between the electric signal trend graph and the preset sudden fault electric signal trend graph is:
[0140] Compare the similarity between the electric signal trend graph and the preset sudden fault electric signal trend graph according to the similarity algorithm. If the similarity is greater than the preset first similarity threshold, it is determined that there is a sudden fault in the monitoring area corresponding to the electric signal trend graph; extract the monitoring area with the sudden fault and the corresponding sudden fault cause;
[0141] If the similarity is less than the preset first similarity threshold, no processing is performed.
[0142] Example 4
[0143] Reference Figure 3 As shown, according to the number of areas with sudden fault monitoring areas, the risk level of the wind turbine is predicted. The present invention discloses a system for monitoring the operating status of a voiceprint collection device, and the system also includes:
[0144] The second analysis module S230 is used to obtain the number of monitoring areas where sudden faults exist, compare the number of areas with a predetermined area number threshold, and generate a determination label for the wind turbine, wherein the determination label includes "perform risk prediction" or "do not perform risk prediction";
[0145] Specifically, the number of monitoring areas where sudden faults exist is obtained, and the number of areas is compared with a predetermined area number threshold value, including:
[0146] Obtain the number of monitoring areas Xf where sudden failures occur;
[0147] The number of monitoring areas with sudden faults Xf is compared with a predetermined area number threshold Xt, where Xt is equal to 1;
[0148] If the number of regions Xf is greater than or equal to the predetermined region number threshold Xt, a judgment label of "perform risk prediction" is generated; if the number of regions Xf is less than the predetermined region number threshold Xt, a judgment label of "do not perform risk prediction" is generated;
[0149] The coefficient prediction module S240 is used to collect data from the wind turbine that generates the "risk prediction" judgment result, obtain the temperature data of the monitoring area of the wind turbine, and perform image stitching on the electrical signal trend graphs of each monitoring area to generate comprehensive image data, and send the generated temperature data and comprehensive image data to a preset risk coefficient prediction model to obtain the risk coefficient of the wind turbine;
[0150] Specifically, the temperature data is obtained by averaging the temperature values collected in each monitoring area; the generation logic of the risk coefficient prediction model is:
[0151] Acquire historical wind turbine prediction data, and divide the historical wind turbine risk prediction data into a test set and a training set, wherein the historical wind turbine prediction data includes comprehensive image data, temperature data, and corresponding risk coefficients;
[0152] Construct a regression network, use the comprehensive image data and temperature data in the training set as the input of the regression network, use the corresponding risk coefficient as the output of the regression network, train the regression network, and obtain an initial risk coefficient prediction network;
[0153] The initial risk factor prediction network is validated using the test set, and the initial risk factor prediction network with a value less than or equal to a preset test error threshold is output as the risk factor prediction model;
[0154] Specifically, the logic for generating the risk coefficient is:
[0155] Acquire the voiceprint data of each monitoring area within a specified time span through a voiceprint acquisition device, wherein the voiceprint data includes a voiceprint frequency fluctuation value or a voiceprint amplitude fluctuation value, and the voiceprint data includes bearing voiceprint data, gearbox voiceprint data, generator voiceprint data, and wind wheel voiceprint data;
[0156] Obtain the fluctuation pre-calibration value of each monitoring area, and compare the fluctuation pre-calibration value with the voiceprint data to calculate the fluctuation coefficient of each monitoring area;
[0157] Integrate the fluctuation coefficients of each monitoring area to obtain the risk coefficient;
[0158] The fluctuation coefficients of each monitoring area are integrated, including:
[0159] The risk coefficient is calculated based on the generated bearing soundprint data, gearbox soundprint data, engine soundprint data and wind wheel soundprint data, and the calculation formula is: Where Rkc is the risk factor of the wind turbine;
[0160] Risk determination module S250: used to compare the risk factor with a predetermined risk factor threshold value to obtain the risk level of the wind turbine after continuing to operate under a sudden fault;
[0161] Specifically, the risk level includes a first-level risk level, a second-level risk level, and a third-level risk level, and the risk coefficient threshold includes a first risk threshold and a second risk threshold, wherein the first risk threshold is less than the second risk threshold, and the first risk threshold and the second risk threshold are both integers greater than 0;
[0162] The logic for generating the risk level is:
[0163] The risk factor is compared with a predetermined risk factor threshold, and if the risk factor is less than or equal to a first risk threshold, the wind turbine is marked as a first-level risk level;
[0164] If the risk factor is greater than the first risk threshold, and the risk factor is less than or equal to the second risk threshold, the wind turbine is marked as a second-level risk level;
[0165] If the risk factor is greater than or equal to the second risk threshold, the wind turbine is marked as a third-level risk level.
[0166] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center through a wired network or a wireless network. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD) or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.
[0167] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.
[0168] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.
[0169] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a system and method for monitoring the operating status of the voiceprint collection device. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.
[0170] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0171] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
[0172] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.
[0173] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A voiceprint collection equipment operation status monitoring system, It is characterized in that The system comprises: The first analysis module is used to obtain the electrical signal set of each monitoring area on the wind turbine within a given time, and perform trend analysis on the electrical signal set to obtain an electrical signal trend graph; Trend chart comparison module: used to compare the electrical signal trend chart of each monitoring area with the preset sudden fault electrical signal trend chart for similarity, so as to obtain the monitoring area with sudden faults and the corresponding sudden fault causes; The second analysis module is used to obtain the number of monitoring areas where sudden faults exist, compare the number of areas with a predetermined area number threshold, and generate a determination label for the wind turbine, wherein the determination label includes "perform risk prediction" or "do not perform risk prediction"; Coefficient prediction module: used to collect data from the wind turbines that generate the "risk prediction" judgment result, obtain the temperature data of the monitoring area of the wind turbines, and perform image stitching on the electrical signal trend graphs of each monitoring area to generate comprehensive image data, and send the generated temperature data and comprehensive image data to the preset risk coefficient prediction model to obtain the risk coefficient of the wind turbines; Risk determination module: used to compare the risk factor with a predetermined risk factor threshold to obtain the risk level of the wind turbine after continuing to operate under sudden failure.
2. According to claim 1, a voiceprint collection equipment operation status monitoring system, It is characterized in that The monitoring area includes a bearing area, a gearbox area, a generator area, and a wind wheel area.
3. According to claim 2, a voiceprint collection equipment operation status monitoring system, It is characterized in that The trend analysis of the electrical signal set includes: Collecting an electrical signal set of each monitoring area in real time according to a time series, wherein the electrical signal set includes a plurality of voltage values / current values; Divide the electrical signal set into time periods according to the order of acquisition time to obtain electrical signal subsets of at least three time periods, and mark the three time periods as a, b, and c respectively; The medians Med1, Med2, and Med3 of the electrical signal subsets in time periods a, b, and c are obtained respectively, and the maximum trend value and the minimum trend value are calculated based on the medians Med1, Med2, and Med3, so as to obtain the maximum trend value Amax and the minimum trend value Amin in time period a, the maximum trend value Bmax and the minimum trend value Bmin in time period b, and the maximum trend value Cmax and the minimum trend value Cmin in time period c respectively. According to the maximum trend value and the minimum trend value in time periods a, b, and c, the electrical signal trend graphs corresponding to time periods a, b, and c are obtained.
4. According to claim 3, a voiceprint collection equipment operation status monitoring system, It is characterized in that The acquisition logic of the electrical signal trend graph corresponding to the time periods a, b, and c is: The electrical signal subsets in time periods a, b, and c are divided into high-order data sets and low-order data sets according to the medians Med1, Med2, and Med3, respectively; The variance of the high-order data set and the low-order data set is calculated to obtain the maximum trend value and the minimum trend value of the electrical signal subset in time periods a, b, and c. The calculation formula for the maximum trend value is: γ is a, b or c; the calculation formula for the minimum trend value is γ is a, b or c; The maximum trend value and the minimum trend value in the time periods a, b, and c are sorted out, and a two-dimensional plane diagram is constructed according to the time-trend value to obtain the electrical signal trend diagram of the time periods a, b, and c.
5. According to claim 4, a voiceprint collection equipment operation status monitoring system, It is characterized in that The logic for comparing the similarity between the electric signal trend graph and the preset sudden fault electric signal trend graph is: Compare the similarity between the electric signal trend graph and the preset sudden fault electric signal trend graph according to the similarity algorithm. If the similarity is greater than the preset first similarity threshold, it is determined that there is a sudden fault in the monitoring area corresponding to the electric signal trend graph; extract the monitoring area with the sudden fault and the corresponding sudden fault cause; If the similarity is less than the preset first similarity threshold, no processing is performed.
6. According to claim 5, a voiceprint collection equipment operation status monitoring system, It is characterized in that The number of monitoring areas with sudden faults is obtained, and the number of areas is compared with a predetermined area number threshold, including: Obtain the number of monitoring areas Xf where sudden failures occur; The number of monitoring areas with sudden faults Xf is compared with a predetermined area number threshold Xt, where Xt is equal to 1; If the number of regions Xf is greater than or equal to the predetermined region number threshold Xt, a judgment label of "perform risk prediction" is generated; if the number of regions Xf is less than the predetermined region number threshold Xt, a judgment label of "do not perform risk prediction" is generated.
7. According to claim 6, a voiceprint collection equipment operation status monitoring system, It is characterized in that The temperature data is obtained by averaging the temperature values collected in each monitoring area; the generation logic of the risk coefficient prediction model is: Acquire historical wind turbine prediction data, and divide the historical wind turbine risk prediction data into a test set and a training set, wherein the historical wind turbine prediction data includes comprehensive image data, temperature data, and corresponding risk coefficients; Construct a regression network, use the comprehensive image data and temperature data in the training set as the input of the regression network, use the corresponding risk coefficient as the output of the regression network, train the regression network, and obtain an initial risk coefficient prediction network; The test set is used to verify the model of the initial risk coefficient prediction network, and the initial risk coefficient prediction network with a value less than or equal to the preset test error threshold is output as the risk coefficient prediction model.
8. According to claim 7, a voiceprint collection equipment operation status monitoring system, It is characterized in that The logic for generating the risk coefficient is: Acquire the voiceprint data of each monitoring area within a specified time span through a voiceprint acquisition device, wherein the voiceprint data includes a voiceprint frequency fluctuation value or a voiceprint amplitude fluctuation value, and the voiceprint data includes bearing voiceprint data, gearbox voiceprint data, generator voiceprint data, and wind wheel voiceprint data; Obtain the fluctuation pre-calibration value of each monitoring area, and compare the fluctuation pre-calibration value with the voiceprint data to calculate the fluctuation coefficient of each monitoring area; Integrate the fluctuation coefficients of each monitoring area to obtain the risk coefficient; The fluctuation coefficients of each monitoring area are integrated, including: The risk coefficient is calculated based on the generated bearing soundprint data, gearbox soundprint data, engine soundprint data and wind wheel soundprint data, and the calculation formula is: Among them, Rkc is the risk factor of the wind turbine.
9. According to claim 8, a voiceprint collection device operation status monitoring system, It is characterized in that The risk level includes a first-level risk level, a second-level risk level and a third-level risk level, and the risk coefficient threshold includes a first risk threshold and a second risk threshold, wherein the first risk threshold is less than the second risk threshold, and the first risk threshold and the second risk threshold are both integers greater than 0; The logic for generating the risk level is: The risk factor is compared with a predetermined risk factor threshold, and if the risk factor is less than or equal to a first risk threshold, the wind turbine is marked as a first-level risk level; If the risk factor is greater than the first risk threshold, and the risk factor is less than or equal to the second risk threshold, the wind turbine is marked as a second-level risk level; If the risk factor is greater than or equal to the second risk threshold, the wind turbine is marked as a third-level risk level.
10. A method for monitoring the operating status of a voiceprint collection device, It is characterized in that The method comprises: Obtaining an electrical signal set of each monitoring area on the wind turbine within a given time, and performing trend analysis on the electrical signal set to obtain an electrical signal trend graph; Compare the electrical signal trend graph of each monitoring area with the preset sudden fault electrical signal trend graph for similarity, so as to obtain the monitoring area with sudden fault and the corresponding sudden fault cause; Obtaining the number of monitoring areas where sudden faults exist, comparing the number of areas with a predetermined threshold of the number of areas, and generating a determination label for the wind turbine, wherein the determination label includes "perform risk prediction" or "do not perform risk prediction"; Data collection is performed on the wind turbines that generate the "risk prediction" judgment result to obtain the temperature data of the monitoring area of the wind turbines, and the electrical signal trend graphs of each monitoring area are image stitched to generate comprehensive image data, and the generated temperature data and comprehensive image data are sent to the preset risk coefficient prediction model to obtain the risk coefficient of the wind turbines; The risk factor is compared with a predetermined risk factor threshold to obtain the risk level of the wind turbine after continuing to operate under a sudden failure.
Citation Information
Patent Citations
Integrated frequency converter operation state monitoring system and monitoring method
CN116593811A
Fan transmission chain state monitoring method and system based on acoustic emission and vibration
CN116892489A
Wind turbine generator fault diagnosis method, system and device and storage medium
CN116988944A
Operation performance anomaly detection method and system for fan unit
CN118030409A
System for monitoring and diagnosing transformer for wind power generation and method thereof
KR1020140033944A
Cited By
Method and system for predicting running state of wind turbine generator
CN119338077A
Mine mining electromechanical equipment fault intelligent diagnosis system and diagnosis method
CN119849945A
Natural gas tiny leakage MFCC detection system and method based on voiceprint coupling
CN120062565A
