Intelligent monitoring device for carbon brush

Abstract

The present application relates to the technical field of state early warning, in particular to a carbon brush intelligent monitoring device, which comprises a time sequence synchronous acquisition module, a contact performance discrimination module, an asynchronous feature recognition module, a feature event alignment module and a running risk early warning module; based on the time sequence pulse sent by the main control board, the response time of the carbon brush current and temperature sensor is judged.The present application, when collecting multiple parameters of the carbon brush, actively constructs the synchronous matching feature of the current and temperature, integrates the contact performance fluctuation factor under the periodic disturbance, dynamically outputs the synchronous grouping and misplacement proportion of the wear vibration event through the synchronous judgment of the wear mutation and vibration event, establishes a structure abnormal risk grading model, forms a multi-level, time-sequenced and correlated carbon brush health evaluation system, supports the trend monitoring, early warning and risk grading of key hidden dangers, enhances the intelligent diagnosis and data traceability ability under complex working conditions, and improves the fault identification and safety control level in the running process of the carbon brush.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of state early warning, and particularly relates to a carbon brush intelligent monitoring device. BACKGROUND

[0002] The field of state early warning involves real-time monitoring of the state of key components during the operation of equipment or systems and timely issuing of abnormal warnings, including collecting equipment operation data, identifying potential faults of key components and conveying early warning information to operation and maintenance personnel, and covering systematic technical content such as equipment safety monitoring, state identification and automatic early warning. Among them, the traditional carbon brush intelligent monitoring device refers to a monitoring and diagnosis system for the running state of carbon brushes in generators or motors in the context of thermal power plants and hydropower plants. It aims at the fault phenomena such as wear, fracture and poor contact of carbon brushes in long-term operation, and analyzes and judges whether the carbon brushes have abnormal states in combination with real-time data.

[0003] The existing technology relies on static collection of single signals and single parameter alarm mode, and there is often a deviation in the time sequence of collection and response. Independent parameters are difficult to reveal periodic changes and collaborative trends between parameters, and multi-dimensional characteristics such as sudden wear of carbon brushes, synchronous vibration and contact performance are difficult to comprehensively reflect, resulting in that the monitoring results cannot truly reflect the dynamic running state, the risk early warning of equipment under high load fluctuation or abnormal working conditions is easy to be missed, periodic hidden dangers and structural abnormalities are difficult to be managed in stages, and operation and maintenance work faces the problems of data isolation, state distortion and judgment lag. SUMMARY

[0004] The purpose of the present application is to solve the shortcomings in the prior art and to provide a carbon brush intelligent monitoring device.

[0005] In order to achieve the above purpose, the technical scheme adopted by the present application is as follows: a carbon brush intelligent monitoring device, the device comprises:

[0006] The time sequence synchronous acquisition module judges the response time of the carbon brush current and temperature sensor based on the time sequence pulse issued by the main control board, arranges the current and temperature sequence under the same pulse, compares the time tags of the collected data, calculates the synchronization characteristics, and obtains the current temperature rise matching characteristics;

[0007] The contact performance discrimination module compares the change of the contact resistance of the carbon brush in each period based on the current temperature rise matching characteristics, analyzes the resistance fluctuation curve before and after the disturbance, calculates the change amplitude and rate, judges the corresponding relationship with the stability reference, determines the difference between the periods, and obtains the contact performance fluctuation factor;

[0008] The asynchronous feature recognition module screens the carbon brush wear data of the corresponding period based on the contact performance fluctuation factor, judges the frequency and distribution of wear mutation phenomena, analyzes the abnormal event time of the vibration signal, compares the wear mutation and vibration abnormal time sequence in the period, identifies the synchronization, and obtains the wear vibration synchronization event.

[0009] The feature event alignment module analyzes the time distribution of wear and vibration mutation events based on the wear vibration synchronization event, identifies the time difference in the same period, screens the event combination exceeding the interval, judges the proportion, induces the abnormal misplacement phenomenon, and obtains the mutation misplacement proportion index.

[0010] The current temperature rise matching feature includes current and temperature correlation parameters, synchronization statistical indicators, and time sequence consistency factors, the contact performance fluctuation factor includes contact state fluctuation characteristics, period response dispersion, and stability evaluation results, the wear vibration synchronization event includes wear event identification, vibration event identification, and synchronization correlation grouping, and the mutation misplacement proportion index includes period misplacement event proportion, time sequence offset interval, and distribution proportion statistics.

[0011] The timing synchronization acquisition module includes:

[0012] The response judgment submodule judges the synchronization of the starting acquisition actions of the carbon brush current sensor and the temperature sensor based on the timing pulse issued by the main control board, compares the starting sequence of the sensors under the same signal trigger, determines whether there is a starting delay in each channel, and generates a response delay amplitude index;

[0013] The synchronization characteristic calculation submodule compares the time marks of the carbon brush current and temperature sensor acquisition data according to the response delay amplitude index, calculates the data acquisition time interval difference between channels in multiple periods, induces the time difference change trend, and obtains the time synchronization stability level;

[0014] The current temperature rise pairing submodule screens the data with consistent time sequence according to the time synchronization stability level, compares the distribution state of the carbon brush current and temperature data, analyzes the correlation between the current data and the temperature data, establishes a pairing relationship, and obtains the current temperature rise matching feature.

[0015] The contact performance discrimination module includes:

[0016] The response difference extraction submodule compares the change process of the carbon brush contact resistance before and after the disturbance period based on the current temperature rise matching feature, integrates the data sequence at the difference time, optimizes the data consistency of each monitoring point in the same period, judges the trend of the resistance change track, screens the typical interval, and obtains the disturbance response difference sequence.

[0017] The resistance fluctuation calculation sub-module compares the change range of the contact resistance in the key period within the disturbance period based on the difference sequence of the disturbance response, analyzes the fluctuation of the resistance curve under the action of the disturbance, calculates the rise and fall amplitude of the difference interval resistance curve, judges the rate correlation between the continuous fluctuation data, and obtains resistance fluctuation characteristic data;

[0018] The key characteristic discrimination sub-module judges whether the fluctuation characteristic exceeds the operation stability benchmark range according to the resistance fluctuation characteristic data, obtains a resistance offset strength index, analyzes the relationship with the benchmark interval, identifies the continuous deviation phenomenon, and obtains a contact performance fluctuation factor.

[0019] The asynchronous feature recognition module comprises:

[0020] The wear mutation extraction sub-module analyzes the carbon brush wear data in the correlation period based on the contact performance fluctuation factor, judges the rate change characteristic of the thickness change in the continuous period, compares the data sequence of adjacent time periods, calculates the fluctuation of the wear change rate, and obtains a wear mutation frequency sequence.

[0021] The vibration anomaly recognition sub-module analyzes the vibration signal amplitude data in the corresponding period according to the wear mutation frequency sequence, optimizes the analysis mode of the signal fluctuation, judges the abnormal fluctuation of the vibration amplitude in the difference period, compares the vibration amplitude change in the continuous interval, and obtains a vibration amplitude anomaly sequence.

[0022] The event synchronization recognition sub-module compares the time distribution of events according to the vibration amplitude anomaly sequence, analyzes the time interval between event combinations, calculates a time synchronization index, and obtains a wear vibration synchronous event.

[0023] The feature event alignment module comprises:

[0024] The time difference analysis sub-module analyzes the occurrence time of the wear mutation and the vibration mutation event in the same monitoring period based on the wear vibration synchronous event, compares the time labels of the corresponding events, calculates the time interval between each group of events, judges the time distribution characteristics of the events, and obtains an event time offset sequence.

[0025] The misalignment event screening sub-module screens the event combination with a time interval greater than the synchronization judgment interval based on the event time offset sequence, analyzes the synchronization of the difference events, and summarizes the events that are not synchronized, and obtains a misalignment event combination set.

[0026] The proportion index calculation sub-module compares the maximum time interval and the average time interval in the misalignment events based on the misalignment event combination set, judges the proportion of the misalignment events in the period and the offset fluctuation characteristics, and obtains a mutation misalignment proportion index.

[0027] The device further comprises:

[0028] The running risk early warning module judges whether the misplacement ratio and the contact performance fluctuation factor exceed the abnormality discrimination standard at the same time based on the mutation misplacement proportion index, analyzes the period section where the abnormality occurs and the risk change trend, and determines that the carbon brush currently has an abnormal running hidden danger if the conditions are met, to obtain a structure abnormality risk classification.

[0029] The structure abnormality risk classification comprises a risk classification mark, a risk category identifier, and a running abnormality state indication.

[0030] The running risk early warning module comprises:

[0031] The misplacement discrimination sub-module analyzes the event distribution reflected in the period based on the mutation misplacement proportion index, compares the state change amplitude of the period corresponding to the contact performance fluctuation factor, judges whether consistent trends are shown in the same period, and obtains a misplacement and fluctuation overlapping segment.

[0032] The risk trend calculation sub-module calculates the continuation and change in a plurality of continuous periods according to the misplacement and fluctuation overlapping segment, screens the synchronous change section between the periods, analyzes the trend increasing or decreasing situation, and obtains a periodic risk trajectory.

[0033] The structure risk classification sub-module judges the classification boundary features in the structure abnormality risk standard according to the periodic risk trajectory, compares the distribution range of the current period under each grade division, analyzes the risk category identifier and the occurrence frequency of the running abnormality state, and obtains the structure abnormality risk classification.

[0034] Compared with the prior art, the application has the advantages and positive effects that:

[0035] In the application, when collecting multiple parameters of the carbon brush, synchronous matching features of the current and the temperature are actively constructed, the contact performance fluctuation factor under period disturbance is integrated, synchronous grouping of the wear and vibration events and the misplacement proportion are dynamically output through synchronous judgment of the wear mutation and the vibration events, a structure abnormality risk classification model is established, a multi-level, time-sequenced and correlated carbon brush health evaluation system is formed, trend monitoring, early warning and risk classification of key hidden dangers are supported, intelligent diagnosis and data traceability ability under complex working conditions are enhanced, and the fault recognition and safety control level in the carbon brush running process are improved. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 The device flowchart of the application;

[0037] Figure 2 The flowchart of the time-sequenced synchronous acquisition module in the application;

[0038] Figure 3 Flow chart of the contact performance discrimination module in the application;

[0039] Figure 4 Flow chart of the asynchronous feature recognition module in the application;

[0040] Figure 5 Flow chart of the feature event alignment module in the application;

[0041] Figure 6 Flow chart of the operation risk early warning module in the application. DETAILED DESCRIPTION

[0042] In order to make the objects, technical solutions and advantages of the application clearer, the application will be further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the application and do not limit the application.

[0043] In the description of the application, it should be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only used to facilitate the description of the application and simplify the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the application. In addition, in the description of the application, the meaning of "a plurality of" is two or more, unless otherwise explicitly and specifically limited. Examples:

[0044] Please refer to Figure 1 The application provides a technical solution: a carbon brush intelligent monitoring device comprising:

[0045] The time sequence synchronous acquisition module judges the acquisition response time of the carbon brush current sensor and the temperature sensor based on the time sequence pulse issued by the main control board, arranges the acquisition sequence of the carbon brush current and the temperature under the same pulse signal, compares the time tags of the acquisition data at each time, calculates the synchronization characteristics of the two groups of data on the time axis, and arranges the corresponding relationship between the carbon brush running current and the temperature, to obtain the current temperature rise matching feature;

[0046] The contact performance discrimination module compares the change process of the contact resistance of the carbon brush in the difference disturbance period based on the current temperature rise matching feature, analyzes the fluctuation curve of the contact resistance of the carbon brush before and after the disturbance, calculates the change amplitude and change rate of the resistance curve, judges the corresponding degree of the change amplitude and the running stability reference, and determines whether there is a key difference in the resistance characteristics between periods, to obtain the contact performance fluctuation factor;

[0047] The asynchronous feature recognition module screens the carbon brush wear data in the corresponding period based on the contact performance fluctuation factor, judges the frequency and time distribution of the mutation phenomenon in the wear information, analyzes the occurrence time of the amplitude abnormal event of the vibration signal, identifies the synchronization by comparing the time sequence of the wear mutation and the vibration abnormality in the period, and obtains the wear-vibration synchronous event;

[0048] The feature event alignment module analyzes the time distribution of the wear mutation event and the vibration mutation event based on the wear-vibration synchronous event, identifies the time difference of the wear mutation event and the vibration mutation event in the same period, screens the event combination whose time difference exceeds the limited interval, judges the proportion of such combination in all events, summarizes the abnormal misplacement phenomenon in the period, and obtains the mutation misplacement proportion index.

[0049] The operation risk early warning module judges whether the misplacement ratio and the contact performance fluctuation factor exceed the abnormality discrimination standard at the same time based on the mutation misplacement proportion index, analyzes the period segment where the abnormality occurs and the risk change trend, determines that the carbon brush currently has an abnormal operation hidden danger if the conditions are met, and obtains the structure abnormality risk grading.

[0050] The current temperature matching feature includes the current-temperature correlation parameter, the synchronization statistical index, and the time sequence consistency factor. The contact performance fluctuation factor includes the contact state fluctuation feature, the period response dispersion, and the stability evaluation result. The wear-vibration synchronous event includes the wear event identifier, the vibration event identifier, and the synchronous correlation grouping. The mutation misplacement proportion index includes the period misplacement event proportion, the time sequence offset interval, and the distribution proportion statistics. The structure abnormality risk grading includes the risk grading mark, the risk category identifier, and the operation abnormality state indication.

[0051] In module 1, the time sequence pulse refers to the periodic electrical signal sent by the master control board, which is used to trigger the downstream sensor to start collecting monitoring parameters at the same time reference, ensuring the time consistency of various data. The collection response time refers to the specific time when the carbon brush current sensor and the temperature sensor actually start collecting data after receiving the time sequence pulse signal. This time is used to judge whether the sensor response is synchronized. The collection sequence refers to a set of continuous data collected by the sensor in sequence according to the time sequence pulse. Each collection corresponds to a pulse trigger, forming an ordered data set. The time label refers to the time identifier attached to each set of collected data, which is used to accurately mark the specific time when the data is generated and is the basis for subsequent multi-source data alignment and comparison. The synchronization characteristic refers to the alignment degree of the collected data of multiple sensors at the same time sequence pulse in time, which is used to evaluate whether the signals of each channel are strictly synchronized and provides support for data fusion. The corresponding relationship refers to pairing the current and temperature collection data one by one according to time to form a complete parameter pair of the current and temperature rise of the carbon brush at a certain time, reflecting the working state of the carbon brush.

[0052] In module 2, the difference disturbance period refers to the period of time when the system artificially applies different pulse load disturbances, facilitating the comparison of the change law of the carbon brush contact resistance under different working conditions before and after the disturbance; the contact resistance change process refers to the entire dynamic process of the contact resistance from the initial state to the change under the disturbance and then to the recovery, reflecting the contact quality; the fluctuation curve refers to the curve of the contact resistance value changing with time, which intuitively shows the fluctuation of the contact resistance with the disturbance; the change amplitude refers to the difference between the maximum and minimum values of the contact resistance in the disturbance period; the change rate refers to the speed of resistance change, which can be measured by the resistance change amount in the time interval; the running stability benchmark refers to the change range or state of the contact resistance during normal operation of the equipment or system set by industry experience, experimental results, etc., which is used to identify abnormalities; the key difference refers to the state of the contact resistance that is obviously different from the benchmark or the previous period found through period comparison, which usually indicates abnormal contact performance of the carbon brush.

[0053] In module 3, the wear data refers to the record of the carbon brush consumption changing with time measured by displacement, thickness, and other wear sensors, which is the basis for determining the service life and failure risk of the carbon brush; the sudden change phenomenon refers to the case where the wear data suddenly rises or drops sharply in a short time, which often means abnormal physical wear of the carbon brush or external impact; the vibration signal amplitude abnormal event refers to the event where the amplitude of the collected vibration signal suddenly increases or decreases, and such fluctuations reflect problems such as mechanical structure loosening and abnormal vibration of the carbon brush; the time sequence refers to arranging the specific time of the wear sudden change event and the vibration abnormal event to form a data time axis sequence; the synchronicity situation refers to the proximity, overlap, and sequence of the wear sudden change event and the vibration abnormal event in the same period, which is used to reveal the potential causal relationship.

[0054] In module 4, the time distribution situation refers to the arrangement of the time points of all wear sudden change and vibration sudden change events in a monitoring period, which can be used to analyze the concentration or dispersion trend of the events; the time difference refers to the interval length on the time axis between the wear sudden change and the vibration sudden change event in the same period, which is a key parameter for measuring synchronization or misalignment; the limited interval refers to the time interval range (not a specific value) for determining whether the events are "synchronous" or "misaligned", i.e., greater than this range is considered misaligned, and less than this range is considered synchronous; the abnormal misalignment phenomenon refers to the case where the time difference between the wear and vibration sudden change events exceeds the limited interval, which usually reflects abnormal operating conditions such as structure loosening or system disturbance.

[0055] In the module 5, the abnormality discrimination criterion refers to a judgment criterion for identifying the abnormality of the carbon brush state, such as the dislocation proportion of wear and vibration mutation, the boundary line of contact performance fluctuation, which are system preset standards; the abnormality cycle section refers to the cycle interval in which the index is abnormal in the continuous monitoring cycle, and the cycle corresponds to the stage in which the equipment working state is unstable and there is a fault hidden danger; the risk change trend refers to the change direction and rate of the risk level through the analysis of the continuous cycle monitoring data, which can help to judge the fault development trend.

[0056] Please refer to Figure 2 , the time sequence synchronous acquisition module comprises:

[0057] The response judgment sub-module judges the synchronization of the starting acquisition actions of the carbon brush current sensor and the temperature sensor based on the time sequence pulse issued by the main control board, compares the starting sequence of the sensors under the same signal trigger, determines whether there is a starting delay in each channel, and generates a response delay amplitude index;

[0058] The pulse is set as a periodic electrical signal, for example, the period is set to 100 milliseconds, and the voltage amplitude is fixed at 5 volts. The acquisition actions of the carbon brush current sensor and the temperature sensor are triggered uniformly by the pulse. After each sensor receives the pulse, it starts data acquisition through an internal trigger mechanism, and then records the actual starting time point of each sensor in the form of a time stamp, which is generated by the built-in clock of the main control board. The starting time points of all acquisitions are compared with the time when the pulse signal occurs, and the acquisition starting delay values of each channel are obtained in sequence. In actual operation, if the current sensor records the acquisition starting time as 21.5 milliseconds after the pulse trigger, and the temperature sensor records the acquisition starting time as 20.1 milliseconds, then the current sensor has a delay of 1.4 milliseconds relative to the temperature sensor. The delay value is stored in the response delay data set, and a preset time synchronization judgment threshold is called to judge whether the delay value is less than the set threshold of 2 milliseconds. If the delay value is less than the set threshold of 2 milliseconds, it is marked as "synchronous start"; if the delay value exceeds 2 milliseconds, it is marked as "non-synchronous start". The judgment result will be continuously counted based on the starting delay record after each sensor acquisition, and a sequence of response delay amplitude indexes in multiple acquisition cycles will be generated as the basis for time synchronization evaluation.

[0059] The synchronous characteristic calculation sub-module compares the time marks of the carbon brush current and temperature sensor acquisition data according to the response delay amplitude index, calculates the data acquisition time interval difference between the channels in multiple cycles, summarizes the time difference change trend, and obtains the time synchronization stability level.

[0060] The current sensor and the temperature sensor are extracted cycle by cycle, and the collection time tags after each time sequence pulse trigger are compared. In actual operation, 5 continuous monitoring cycles can be set, and the recorded time difference values are 1.2 milliseconds, 1.6 milliseconds, 1.4 milliseconds, 2.3 milliseconds and 3.1 milliseconds in turn. All time difference values are uniformly distributed in the preset delay interval, for example, 0 to 2 milliseconds is a synchronous interval, and 2 to 5 milliseconds is a non-synchronous interval. The number of cycles of collected data in each interval is counted to calculate the proportion of different delay levels, so as to obtain the distribution trend of the collection time difference between channels in continuous cycles. Then, the sliding calculation method is used to evaluate the interval fluctuation range of the time difference value set, and the fluctuation degree and change trend of the set are extracted. In an actual example, if the average collection delay difference in the first 50 cycles is 1.92 milliseconds and the fluctuation amplitude is 0.78 milliseconds, according to the set stability judgment standard, the average value and the fluctuation amplitude are compared with the preset stability threshold value. If the average difference is below 2 milliseconds and the fluctuation value is less than 1 millisecond, it is evaluated as "medium stability", otherwise it is judged as "low stability" or "high stability". Finally, the time synchronization stability level label of the channel combination is output, and the synchronization performance of each cycle is identified for grading, which is used as the basis for screening paired data.

[0061] The current temperature rise pairing sub-module selects data with consistent time sequence according to the time synchronization stability level, compares the distribution state of the carbon brush current and temperature data, analyzes the correlation between the current data and the temperature data, establishes a pairing relationship, and obtains the current temperature rise matching characteristics.

[0062] The data screening only retains the acquisition cycle data with synchronization stability of "high stability" or "medium stability" level, and all acquisition data in the cycle determined as "low stability" will be excluded. Then the retained current and temperature acquisition data are reordered according to the acquisition time sequence, and two one-to-one corresponding parameter sequences are constructed. Each current value is paired with the temperature value collected at the same time point to form a parameter pair. After completing the parameter pairing, the difference of each group is extracted and recorded to form a difference sequence representing the correspondence between current and temperature changes. In practical application, the average value and the maximum value in the difference sequence are calculated as the consistency reference basis of the paired data. If the difference value is small, it indicates that the paired relationship is stable and consistent. If the difference value is large, it indicates that there is an inconsistent response between temperature and current changes in the pairing. For example, the average difference value calculated in a group of data is 0.8, and the maximum difference value is 2.3. According to the set matching classification standard, it is set that strong pairing needs to satisfy the average difference value less than 0.5 and the maximum difference value less than 1, otherwise it is marked as weak pairing or abnormal pairing. Finally, the proportion information of the pairing strength level in multiple acquisition cycles is output, for example, the strong pairing cycle accounts for 60%, the weak pairing cycle accounts for 30%, and the abnormal pairing accounts for 10%. The pairing level sequence is output as the current temperature rise matching feature.

[0063] Please refer to Figure 3 The contact performance discrimination module includes:

[0064] The response difference extraction submodule compares the change process of the carbon brush contact resistance before and after the disturbance cycle based on the current temperature rise matching feature, integrates the data sequence at the difference time, optimizes the data consistency of each monitoring point in the same cycle, judges the trend of the resistance change trajectory, selects the typical interval, and obtains the disturbance response difference sequence.

[0065] The monitoring data before and after the disturbance period is called, the continuous collection data of the carbon brush contact resistance 10 seconds before and after the disturbance period is extracted, the data is divided into 200 groups of contact resistance values at a frequency of 10 samples per second, the disturbance starting point is set as t=0, the resistance sequence from t=-10 seconds to t=0 seconds and the resistance sequence from t=0 seconds to t=10 seconds are respectively named as the pre-disturbance sequence and the post-disturbance sequence, in the comparison process, the average resistance of each second in the two periods is subjected to difference operation and the difference sequence is recorded, in the example, if the resistance of the pre-disturbance sequence is 0.032 ohms at t=-3 seconds and the resistance of the post-disturbance sequence is 0.041 ohms at t=3 seconds, the difference is 0.009 ohms, in this way, the difference time point list is constructed, the points with significant difference are extracted, the criterion is that when the difference is greater than 0.005 ohms, it is marked as a "significant difference point", the threshold is derived from the upper limit of the resistance standard fluctuation range obtained by statistical analysis during stable operation of the equipment, the current, temperature and vibration data corresponding to all significant difference points are integrated to form the data sequence of the difference time, then according to the monitoring point collection time in the same period in the data sequence, the time alignment operation is performed, even if there is a small difference in the sampling interval of different channel data, the linear interpolation filling or direct rejection of the misaligned point is performed by comparing the collection time label, in this way, the time consistency between the data in the same period is optimized, then the trend of the resistance change trajectory is judged, the peak and valley positions are extracted on each resistance time sequence, the increase / decrease amplitude and duration between the peak and valley are calculated, whether the trend is continuous rise, continuous fall or fluctuation is judged, in the example, if the pre-disturbance is a gentle curve and the post-disturbance is a rapid rising trend, it is determined that the change trajectory of this period is "rising type", then the interval with a change duration of more than 3 seconds is selected as a typical interval to obtain the disturbance response difference sequence.

[0066] The resistance fluctuation calculation sub-module compares the change range of the contact resistance in the key period in the disturbance period based on the disturbance response difference sequence, analyzes the fluctuation of the resistance curve under the action of the disturbance, calculates the rise / fall amplitude of the resistance curve in the difference interval, judges the rate correlation between the continuous fluctuation data, and obtains the resistance fluctuation characteristic data;

[0067] The resistance time period marked as a typical interval in each perturbation cycle is selected, the resistance values at the starting point and the ending point of the interval are taken respectively, a difference calculation is performed to obtain a change range, if the starting resistance is 0.035 ohms and the ending resistance is 0.046 ohms, then the change range is 0.011 ohms, then the resistance value data of each second in the entire typical interval is recorded as a fluctuation curve sequence, in the example, if the length of the typical interval is 5 seconds, the resistance value sequence in 5 seconds is extracted, then the resistance difference between any two adjacent seconds in the sequence is calculated as the resistance change rate per unit time, and the maximum value, the minimum value and the average value of the rate set are extracted, the fluctuation amplitude and the fluctuation trend are counted, and the judgment standard is that if the maximum change rate is greater than 0.004 ohms / second and the minimum rate is less than 0.001 ohms / second, then it is marked as a rate dispersion type fluctuation, if all the change rates are between 0.002 and 0.003 ohms / second, then it is marked as a rate balanced type fluctuation, then the change rates of any two consecutive data are compared by difference, if the difference of the change rates of three consecutive rates is less than 0.001 ohms / second, then it is considered to have a rate correlation, otherwise it is a non-associated rate, the above operation is performed on each perturbation response difference sequence resistance curve, and finally the resistance fluctuation characteristic data including the resistance change range, the fluctuation type and the rate correlation label are obtained.

[0068] The key characteristic discrimination sub-module judges whether the fluctuation characteristics exceed the running stability reference range according to the resistance fluctuation characteristic data, and adopts the formula:

[0069] ;

[0070] The resistance offset strength index is obtained , the relationship with the reference interval is analyzed, the continuous deviation phenomenon is identified, and the contact performance fluctuation factor is obtained, wherein, represents the maximum sampling number of the contact resistance in the perturbation cycle, that is, the maximum data point of the contact resistance monitored in the cycle, represents the minimum sampling number of the contact resistance in the perturbation cycle, that is, the minimum data point of the contact resistance monitored in the cycle, represents the average sampling number of the contact resistance in the perturbation cycle, that is, the average of the contact resistance data of all monitoring points, represents the time interval standard deviation between the collection time points in the perturbation cycle, which is used to measure the dispersion degree of the time interval of each sampling point, and reflects the uniformity of the time sequence distribution in the collection process, represents the time corresponding to the th collection time point in the perturbation cycle, represents the total number of collection time points in the perturbation cycle.

[0071] The resistance deviation intensity index refers to a comprehensive statistical data for reflecting the overall fluctuation amplitude and rate change degree of the carbon brush contact resistance in the time sequence change process in the disturbance period in the process of intelligent monitoring of the carbon brush, which can directly reflect whether the fluctuation range of the contact resistance of the carbon brush is abnormally expanded and whether the change trend becomes unstable under the abnormal disturbance period, and then assist in judging whether the contact performance of the carbon brush in the period is obviously abnormal.

[0072] Whether the fluctuation characteristics exceed the running stability reference range needs to be jointly analyzed based on the resistance fluctuation amplitude, time coverage range and balance of the resistance change process in the disturbance period. In the specific implementation process, the key values extracted from the resistance fluctuation curve are first confirmed. Taking the current collection as an example, the maximum sampling number of the contact resistance in the disturbance period is set to , the minimum sampling number is , each time point in the period is , , , , , the sum is , the resistance sampling number is , and the standard deviation of the five sampling intervals is calculated by the standard formula . In order to ensure the adaptability of different physical quantity units in unified expression, each participating item is converted by dimensionless conversion according to the normalization standard. The normalized parameters are as follows:

[0073] , , , , ;

[0074] The above values are substituted into the formula:

[0075] ;

[0076] The value is the resistance deviation intensity index of the contact resistance in the period. The numerical result can be compared and analyzed with the set stability reference range. For example, if the index value exceeds the upper limit, it is determined that the contact state in the period exists a sustained deviation characteristic. According to the above result, it can be identified that the contact resistance state of the carbon brush in the disturbance period exists abnormal fluctuation behavior, and it needs to enter the subsequent fluctuation behavior classification and structural stability judgment process, so as to obtain the contact performance fluctuation factor. The formula couples the dynamic change range of the contact resistance, the total span of the sampling time and the discrete level of the fluctuation time sequence at the same time, constructs a multi-factor fusion judgment mechanism, and can comprehensively judge the real influence degree of the periodic disturbance on the contact performance.

[0077] Please refer to Figure 4The asynchronous feature recognition module comprises:

[0078] The wear mutation extraction submodule analyzes the carbon brush wear data in the associated period based on the contact performance fluctuation factor, judges the rate change characteristics of the thickness change in the continuous time period, compares the data sequences of adjacent time periods, calculates the fluctuation of the wear change rate, and obtains the wear mutation frequency sequence.

[0079] The corresponding monitoring period number is extracted, the carbon brush wear thickness data sequence in the period is called, continuous thickness data points are obtained at a rate of 1 sample per second, for example, 30 thickness sample values are obtained when the period is 30 seconds, then the data is divided into multiple equal-length time periods according to the collection order, each period contains 5 seconds of data, the total amount of thickness change in each period is calculated and divided by the duration of the period to obtain the wear rate sequence, in the actual example, the thickness decreases from 15.20 mm to 15.17 mm in the first 5-second segment, so the wear rate is 0.003 mm / s, the thickness decreases from 15.17 mm to 15.12 mm in the second 5-second segment, so the rate is 0.01 mm / s, and so on to obtain the wear rates of multiple time periods. Then, the difference between adjacent two rate values is calculated to form a rate change sequence, for example, the rate difference from 0.003 to 0.01 is 0.007 mm / s, the next difference is extracted and a complete change amplitude sequence is constructed, and the reference threshold for judging mutation is set to 0.006 mm / s, which is derived from the maximum rate fluctuation amplitude in the 95% confidence range of the first 500 period data. When the rate change amplitude is greater than the threshold, it is recorded as a mutation point. Each detection of a mutation point increases the count in the wear mutation frequency record table, forming a wear mutation frequency sequence corresponding to each monitoring period, which is used for subsequent alignment and comparison with vibration information.

[0080] The vibration anomaly recognition submodule analyzes the vibration signal amplitude data in the period corresponding to the wear mutation frequency sequence, optimizes the analysis method of signal fluctuation, judges the abnormal fluctuation of vibration amplitude in the difference time period, compares the vibration amplitude change in the continuous interval, and obtains the vibration amplitude anomaly sequence.

[0081] The period with a non-zero mutation frequency is screened as an analysis period, and the vibration signal amplitude data in the period is extracted. The original vibration signal is processed. First, time periods are divided by a fixed sliding window, for example, every 3 seconds is an analysis window, and the difference between the maximum value and the minimum value of the vibration signal in the window is calculated as the amplitude fluctuation value of the current window. In the example, the maximum value of the vibration signal in a 3-second window is 2.5, and the minimum value is 0.8. The fluctuation amplitude is 1.7. If the fluctuation amplitude difference in the continuous 3 analysis windows exceeds 1.2, and the minimum window amplitude is less than 0.8, and the maximum window amplitude is greater than 2.0, then the current period vibration is marked as abnormal fluctuation. The fluctuation amplitudes of all analysis windows in the current period are arranged to form a vibration fluctuation sequence, and the time window with the most significant difference is determined and marked as a difference time period. The vibration amplitude value of each second in the difference time period is extracted and compared with the average value of the adjacent window. If the amplitude value of a time point is greater than 1.0 than the average value of the adjacent window, it is determined to be an abnormal time point. The abnormal time point is recorded in the abnormal sequence. The number of abnormal time points in each period is the vibration abnormality frequency. Further, the points with abnormal amplitude values exceeding 2.0 in each period are arranged to form a final vibration amplitude abnormal sequence, which is used for subsequent synchronization analysis and alignment with the wear mutation event.

[0082] The event synchronization identification submodule compares the time distribution of the events according to the vibration amplitude abnormal sequence, analyzes the time interval between the event combinations, and uses the formula:

[0083]

[0084] Calculate the time synchronization index to obtain the wear vibration synchronization event, wherein represents the total number of synchronization events, represents the time point of the th wear mutation event, represents the time point of the vibration abnormal event corresponding to the th wear event, represents the time point fluctuation compensation factor of the th event pair, represents the time distribution dispersion coefficient of the th event pair.

[0085] The time synchronization index is a parameter for measuring the consistency and synchronization degree of the time distribution of the carbon brush wear mutation event and the vibration abnormal event in the same period, which is used to determine whether the wear and vibration abnormal events are related in time sequence, and is an important basis for subsequent fault judgment, risk identification and structural abnormality analysis. For example, if the wear and vibration abnormalities often appear synchronously, it indicates a special risk of mechanical structure or operating state.

[0086] ​Based on the vibration amplitude anomaly sequence, the wear mutation time point of each event and its corresponding vibration anomaly time point are extracted and sequentially paired into event time pairs. The time difference under the same event number is compared, and the time interval between all event time pairs is calculated. The corresponding fluctuation compensation factor is then substituted into each pair. and distribution dispersion coefficient To analyze the time synchronization between two types of events, the time points of events 1 to 4 are as follows: wear event time points are 1023ms, 1874ms, 2560ms, and 3120ms, and corresponding vibration event time points are 1050ms, 1902ms, 2610ms, and 3185ms. The calculated time differences are 27ms, 28ms, 50ms, and 65ms, respectively, with corresponding normalized time differences of 0.27, 0.28, 0.50, and 0.65. Each time difference is adjusted and weighted using a fluctuation compensation factor. The values ​​were set to 0.12, 0.15, 0.08, and 0.10 respectively, and the product terms for each term were calculated as follows:

[0087] Group 1: ;

[0088] Group 2: ;

[0089] Group 3: ;

[0090] Group 4: ;

[0091] Adding the above results together, we get the numerator as follows:

[0092] ;

[0093] The denominator uses the normalized distribution dispersion coefficient. The values ​​are set to 0.05, 0.07, 0.04, and 0.06, with corresponding correction terms of 1.05, 1.07, 1.04, and 1.06 respectively. Summing these values ​​yields:

[0094] ;

[0095] Calculate time synchronization index for:

[0096] ;

[0097] The numerical result shows that the wear and vibration abnormal events have a certain time synchronization in the monitoring period. If the synchronization standard interval [0.40, 0.60] is set, it can be judged that the events belong to the medium synchronization level, have a certain cooperativity basis, and can be used for joint analysis and processing of subsequent risk events. The square difference root method is used to measure the absolute distance of time difference in the formula, the fluctuation compensation factor is used to adjust the response strength of the event, and the distribution dispersion coefficient is used to suppress the statistical deviation of the over-concentrated event pair. The overall operation integrates the difference between events and the periodic stability, and reflects the time sequence consistency change characteristics of the structural abnormal state.

[0098] Please refer to Figure 5 The feature event alignment module includes:

[0099] The time difference analysis submodule analyzes the occurrence time of wear mutation and vibration mutation events in the same monitoring period based on wear vibration synchronization events, compares the time labels of the corresponding events, calculates the time interval between each group of events, judges the time distribution characteristics of the events, and obtains the event time offset sequence.

[0100] The event data identified as wear mutation and vibration mutation in the current monitoring period is called, and the occurrence time labels recorded by each event are extracted. For example, in a certain monitoring period, the wear mutation event occurs at 12.3 seconds, 17.8 seconds, and 25.5 seconds, and the vibration mutation event occurs at 12.1 seconds, 18.5 seconds, and 25.9 seconds. The wear mutation time label and the vibration mutation time label are paired according to the time axis order, and the pairing principle is to find the event pair combination with the shortest time distance. The time interval between the paired events is calculated group by group, and the interval is the time interval between the vibration mutation and the corresponding wear mutation. If the first group interval is -0.2 seconds, the second group is 0.7 seconds, and the third group is 0.4 seconds, the interval values are summarized to form a time offset value list, and the time offset list is identified according to the time sign. Positive value indicates that vibration lags behind wear, and negative value indicates that vibration precedes wear. Then the time offset list is divided into intervals and counted. The synchronization interval is set to ±0.3 seconds, the mild offset interval is ±0.3 to ±0.8 seconds, and the serious offset interval is greater than ±0.8 seconds. The 80% concentrated distribution segment obtained in the previous actual test process is used as the synchronization interval threshold, and each time offset value is distributed into the above three intervals. The event number ratio of each interval is counted, for example, if there are 5 pairs of events, 3 pairs in the synchronization interval, 1 pair in the mild offset interval, and 1 pair in the serious offset interval, the synchronization ratio is 60%. Finally, the event time offset sequence is output by period, which indicates the time interval value, offset direction and classification label of each event pair.

[0101] The misalignment event screening submodule screens event combinations with a time interval greater than a synchronization judgment interval based on an event time offset sequence, analyzes synchronization of difference events, and induces unsynchronized events to obtain a misalignment event combination set;

[0102] First, event groups with a time interval greater than a preset synchronization judgment interval are selected, the preset synchronization judgment interval is set to a time difference less than or equal to 0.3 seconds in absolute value, and event combinations exceeding the interval are regarded as unsynchronized event combinations, for example, if the time difference of a certain event pair is 0.65 seconds, the combination is determined to be a misalignment event, all event pairs exceeding ±0.3 seconds are screened again, marked as difference events, and the time difference value and event identification number are recorded, the frequency of occurrence of misalignment events in the period in which the misalignment events are located is counted, if there are more than three misalignment event combinations in a period, the period is marked as an abnormally high-frequency misalignment period, on this basis, the synchronization state of each misalignment event combination is determined, whether there is a long lag, advance, repeated change, and other asymmetric patterns is recorded by comparing the order of occurrence of vibration and wear events, in the example, if vibration mutation always occurs after wear events for 3 seconds, it is marked as a “long lag type misalignment”, if wear precedes vibration by less than 1 second but frequently alternates, it is marked as an “alternating type misalignment”, then all event pairs that do not meet the synchronization interval judgment standard are induced, the time difference, misalignment type, period number, event order and other information are summarized, the information is combined into a misalignment event combination set for further analysis of the proportion and distribution trend of the combination in the period.

[0103] The proportion index calculation submodule compares the maximum time interval and the average time interval in the misalignment event combination set based on the misalignment event combination set, and uses the formula:

[0104] ;

[0105] The proportion index of the misalignment event in the period is determined , wherein, represents the number of misalignment event combination sets, represents the maximum time interval between the wear mutation event and the vibration mutation event in all misalignment event combinations, represents the average time interval of all event pairs, i.e., the average time interval of all wear mutation and vibration mutation event combinations, represents the number of all event pairs, represents the number of the first misalignment event combination, represents the product of the wear mutation amplitude and the vibration mutation amplitude in the first misalignment event combination, which measures the joint strength of abnormal fluctuations in the combination, represents the average of the product of the wear mutation amplitude and the vibration mutation amplitude in all misalignment event combinations.

[0106] The mutation misplacement proportion index is a comprehensive statistical parameter for measuring the proportion and distribution characteristics of the phenomenon that the carbon brush wear mutation event and the vibration mutation event are out of sync (i.e., "misplacement") in a monitoring period, which can directly reflect the severity of the misplacement of abnormal wear and vibration events of the carbon brush and the collector ring and other components in the actual operation process, help to find the risks of device structure loosening, system abnormal interference and the like in advance, and provide a quantitative basis for intelligent monitoring and risk warning.

[0107] First, determine the number of misplacement events , the number of all event combinations is , wherein the 8th group and the 10th group are misplacement event combinations, and the wear mutation times thereof are identified as 830 seconds for the 8th group and 1040 seconds for the 10th group, and the vibration mutation times thereof are 825 seconds for the 8th group and 1035 seconds for the 10th group, and the calculated time intervals are both 5 seconds, the maximum time interval is set to , and the average total time interval is , then the wear mutation amplitude and the vibration mutation amplitude are extracted from the monitoring data, the original data is normalized, the wear mutation amplitude of the 8th group is set to 2.5 mm, and the vibration mutation amplitude is set to 2.6 g, and the normalized product is , the wear mutation amplitude of the 10th group is 2.7 mm, the vibration mutation amplitude is 3.0 g, and the normalized product is , the average product is calculated as , and the above parameters are substituted into the formula:

[0108] ;

[0109] ;

[0110] The results show that in the monitoring period, the proportion of misplacement events reaches about 96.15% after adjustment by the combined fluctuation term, reflecting that in the case of a small number of events, the time interval and the fluctuation intensity are concentrated and significant, the distribution of the deviation has a dominant trend, and the generated parameter is defined as the mutation misplacement proportion index, which can be used to comprehensively evaluate the relative deviation degree of the event combination under the non-synchronous characteristics, and assist in identifying the potential abnormal period of carbon brush wear, the formula combines the time interval and the double-parameter amplitude deviation, and constructs the denominator by means of square root and absolute deviation, so that the index has more sensitive response capability when facing high fluctuation and small sample.

[0111] Please refer to Figure 6 , the operation risk warning module comprises:

[0112] The misalignment discrimination submodule analyzes the event distribution reflected in its period based on the sudden misalignment proportion index, compares the state change amplitude of the contact performance fluctuation factor corresponding period, judges whether it shows consistent trend in the same period, and gets the misalignment and fluctuation coincidence segment;

[0113] The event time difference statistics table in the current monitoring period is extracted, the number of event pairs and the number of event combinations falling into the non-synchronous interval recorded therein are used to calculate the misalignment proportion value of the period, the number of misalignment event pairs in each period is divided by the total number of event pairs to get the sudden misalignment proportion index of the period, for example, in the 35th period, there are 10 groups of wear-vibration event pairs, among which 4 groups of time difference exceed the synchronous judgment threshold of 0.3 seconds, so the misalignment proportion is 0.4, then analyze whether the time position of each misalignment event in the period and the contact resistance fluctuation degree in the corresponding period of the contact performance fluctuation factor data coincide, the resistance fluctuation degree is based on the contact resistance change amplitude, the difference between the maximum and minimum values of the resistance curve in the same period is extracted as the fluctuation amplitude, for example, the fluctuation amplitude is 0.012 ohm, set the resistance fluctuation amplitude higher than 0.01 ohm as significant fluctuation, if the current period meets the conditions of misalignment proportion higher than 0.3 and resistance fluctuation amplitude higher than 0.01 ohm at the same time, then the period is marked as the period of misalignment and fluctuation in the same direction, judge whether the situation that both types of characteristic values exceed the set threshold in the same period appears, if it meets, then the period number is included in the misalignment and fluctuation coincidence segment, finally generate the joint result of period number and its corresponding misalignment proportion value and resistance fluctuation amplitude value.

[0114] The risk trend calculation submodule calculates the continuation and change in multiple consecutive periods according to the misalignment and fluctuation coincidence segment, filters the synchronous change segment between periods, analyzes the trend increasing or decreasing situation, and gets the periodic risk trajectory;

[0115] The period numbers recorded as coincident segments in consecutive multiple periods are extracted, arranged in chronological order, and then the continuity between the numbers is calculated. For example, if the segment period numbers are 35, 36, 37, 40, and 41, two consecutive segments are identified as 35 to 37 and 40 to 41, and then the change of the coincident feature in each consecutive segment is judged. The misregistration proportion and resistance fluctuation amplitude value in each period are extracted, and two sequences are constructed in the order of period increase. The difference between adjacent items is calculated and the difference value is recorded. If all the difference values are positive, it is judged to be an increasing trend. If it is negative, it is judged to be a decreasing trend. If the signs alternate, it is judged to be a fluctuation type. In the example, if the misregistration proportion of period 35 is 0.34, the misregistration proportion of period 36 is 0.41, and the misregistration proportion of period 37 is 0.49, the misregistration feature is increasing. If the resistance fluctuation amplitudes are 0.011, 0.013, and 0.018 ohms, respectively, the fluctuation trend is also increasing. This segment is determined to be a synchronous increasing risk segment, and the segment is marked as a risk increasing segment. The same analysis process is performed on the next segment, and the trend type and start and end period numbers of each segment are recorded. Finally, the interval set with trend changes in all monitoring periods is output, forming a periodic risk trajectory. The trajectory structure is arranged in order and is attached with a trend label and feature parameters.

[0116] The structure risk grading submodule judges the grading boundary features in the structure abnormal risk standard according to the periodic risk trajectory, compares the distribution range of the current period under each level division, analyzes the occurrence frequency of the risk category identification and the running abnormal state, and obtains the structure abnormal risk grading;

[0117] A set of structure abnormal risk level boundary value table is called, which defines three risk level division conditions. The first level risk requires that the misregistration proportion is greater than 0.5 and the resistance fluctuation amplitude is greater than 0.02 ohms. The second level risk requires that the misregistration proportion is greater than 0.35 and less than or equal to 0.5, and the resistance fluctuation is between 0.015 and 0.02 ohms. The third level risk requires that the misregistration proportion is between 0.25 and 0.35, the resistance fluctuation amplitude is greater than 0.01 ohms and less than or equal to 0.015 ohms. The parameters of each period in the periodic risk trajectory are compared with the upper and lower limits of each level, respectively. If the misregistration proportion in a certain period is 0.52 and the resistance fluctuation amplitude is 0.023 ohms, it meets the first level risk condition. If it is 0.38 and 0.016 ohms, it belongs to the second level. Then the frequency of each level risk is counted according to the period. In 50 periods, the first level risk occurs 5 times, the second level risk occurs 12 times, and the third level risk occurs 18 times. The rest is normal or undefined level. The occurrence frequency ratio is calculated. The first level risk is 10%, the second level risk is 24%, and the third level risk is 36%. The risk category identification and period number are stored in the structure abnormal grading record table, and the running abnormal state mark field is added to record whether there is a sudden increase in vibration or a sudden rise in current in the corresponding period. Finally, the structure abnormal risk grading result sequence is output.

[0118] The above merely describes the preferred embodiments of the present application, but is not intended to limit the present application in other forms. Any person skilled in the art can make changes or modifications to the above disclosed technical contents into equivalent embodiments with equivalent changes, and apply to other fields. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present application without departing from the technical solution content of the present application still falls within the protection scope of the present application.

Claims

1. A carbon brush intelligent monitoring device, characterized in that, The device includes: The timing synchronization acquisition module determines the carbon brush current and temperature sensor response time based on the timing pulses emitted by the main control board, organizes the current and temperature sequences under the same pulse, compares the time stamps of the acquired data, calculates the synchronization characteristics, and obtains the current-temperature rise matching characteristics. Based on the current-temperature-rise matching characteristics, the contact performance discrimination module compares the changes in carbon brush contact resistance in each cycle, analyzes the resistance fluctuation curves before and after the disturbance, calculates the change amplitude and rate, determines the correspondence with the stability benchmark, determines the difference between cycles, and obtains the contact performance fluctuation factor. The contact performance discrimination module includes: The response difference extraction submodule, based on the current temperature rise matching characteristics, compares the change process of carbon brush contact resistance before and after the disturbance cycle, integrates the data sequence at the time of difference, optimizes the data consistency of each monitoring point under the same cycle, judges the trend of resistance change trajectory, selects typical intervals, and obtains the disturbance response difference sequence. The resistance fluctuation calculation submodule compares the range of change of contact resistance during key periods within the disturbance period based on the disturbance response difference sequence, analyzes the fluctuation of the resistance curve under the disturbance, calculates the rise and fall of the resistance curve in the difference interval, determines the rate correlation between continuous fluctuation data, and obtains resistance fluctuation characteristic data. The key characteristic discrimination submodule determines whether the fluctuation characteristics exceed the operating stability benchmark range based on the resistance fluctuation characteristic data, obtains the resistance offset intensity index, analyzes its relationship with the benchmark range, identifies continuous deviation phenomena, and obtains the contact performance fluctuation factor. The asynchronous feature recognition module, based on the contact performance fluctuation factor, filters carbon brush wear data for the corresponding period, determines the frequency and distribution of wear abrupt changes, analyzes the time of abnormal vibration events, compares the wear abrupt changes and vibration anomalies within the period, identifies the synchronization situation, and obtains wear vibration synchronization events. The asynchronous feature recognition module includes: The wear mutation extraction submodule analyzes the carbon brush wear data within the associated period based on the contact performance fluctuation factor, determines the rate change characteristics of thickness change in continuous time period, compares the data sequences of adjacent time periods, calculates the fluctuation of wear change rate, and obtains the wear mutation frequency sequence. The vibration anomaly identification submodule analyzes the vibration signal amplitude data within the corresponding period based on the wear mutation frequency sequence, optimizes the signal fluctuation analysis method, judges the abnormal fluctuation of vibration amplitude in different time periods, compares the vibration amplitude changes in continuous intervals, and obtains the vibration amplitude anomaly sequence. The event synchronization identification submodule compares the time distribution of events based on the abnormal vibration amplitude sequence, analyzes the time interval between event combinations, calculates the time synchronization index, and obtains the wear vibration synchronization event. Based on the wear and vibration synchronization events, the feature event alignment module analyzes the time distribution of wear and vibration abrupt events, identifies time differences within the same period, filters event combinations that exceed the interval, determines the proportion, summarizes abnormal misalignment phenomena, and obtains the abrupt misalignment proportion index. The feature event alignment module includes: The time difference analysis submodule analyzes the occurrence time of wear abrupt change and vibration abrupt change events within the same monitoring period based on the wear vibration synchronization event, compares the time tags of the corresponding events, calculates the time interval between each group of events, determines the time distribution characteristics of the events, and obtains the event time offset sequence. The misaligned event filtering submodule filters event combinations with time intervals greater than the synchronization judgment interval based on the event time offset sequence, analyzes the synchronization status of the difference events, summarizes the unsynchronized events, and obtains a set of misaligned event combinations. The proportion index calculation submodule, based on the misalignment event combination set, compares the maximum time interval and the average time interval in the misalignment events, determines the proportion of misalignment events and the offset fluctuation characteristics within the period, and obtains the sudden misalignment proportion index. The current-temperature-rise matching characteristics include current-temperature correlation parameters, synchronization statistics, and timing consistency factors. The contact performance fluctuation factors include contact state fluctuation characteristics, periodic response dispersion, and stability assessment results. The wear-vibration synchronization events include wear event identifiers, vibration event identifiers, and synchronization association groups. The abrupt misalignment proportion index includes the proportion of periodic misalignment events, timing offset intervals, and distribution proportion statistics.

2. The intelligent carbon brush monitoring device according to claim 1, characterized in that, The timing synchronization acquisition module includes: The acquisition response judgment submodule judges the synchronization of the start acquisition action of the carbon brush current sensor and the temperature sensor based on the timing pulses sent by the main control board, compares the start order of the sensors under the same signal trigger, determines whether there is a start delay in each channel, and generates a response delay amplitude index. The synchronization characteristic calculation submodule compares the time stamps of the carbon brush current and temperature sensor data based on the response delay amplitude index, calculates the data acquisition time interval difference between channels under multiple cycles, summarizes the time difference change trend, and obtains the time synchronization stability level. The current-temperature-rise matching submodule filters data with consistent time sequence based on the time synchronization stability level, compares the distribution of carbon brush current and temperature data, analyzes the correlation between current data and temperature data, establishes a matching relationship, and obtains current-temperature-rise matching characteristics.

3. The intelligent carbon brush monitoring device according to claim 1, characterized in that, The device further includes: The risk warning module determines whether the misalignment ratio and the contact performance fluctuation factor both exceed the anomaly judgment criteria based on the mutation misalignment ratio index. It analyzes the periodic segments and risk change trends of the anomalies. If the conditions are met, it determines that the carbon brush currently has a potential for abnormal operation and obtains the structural anomaly risk classification. The structural anomaly risk classification includes risk classification markers, risk category identifiers, and operational anomaly status indicators.

4. The intelligent carbon brush monitoring device according to claim 3, characterized in that, The operational risk early warning module includes: The misalignment discrimination submodule analyzes the event distribution reflected in the period based on the mutation misalignment ratio index, compares the state change amplitude of the contact performance fluctuation factor corresponding to the period, and determines whether it shows a consistent trend in the same period, thus obtaining the misalignment and fluctuation overlapping segment. The risk trend calculation submodule calculates the continuation and change in multiple consecutive cycles based on the overlapping segments of misalignment and fluctuation, filters out synchronous change segments between cycles, analyzes the trend of increasing or decreasing, and obtains the periodic risk trajectory. The structural risk classification submodule determines the classification boundary characteristics in the structural anomaly risk standard based on the periodic risk trajectory, compares the distribution range of the current period under each level classification, analyzes the frequency of occurrence of risk category identifiers and abnormal operating states, and obtains the structural anomaly risk classification.

Images