A mine power monitoring system and method

Abstract

The present invention relates to a mining power monitoring system and method, the method comprising: obtaining multi-dimensional power data in a mining power system, identifying abnormal power data using an RRCF algorithm, and calculating the local fluctuation degree of the current abnormal power data; using the difference between the local fluctuation degrees of the multi-dimensional abnormal power data as the correlation coefficient of each dimension, and using a negative exponential function to attenuate the variance of the correlation coefficient to obtain the abnormal correlation degree; correcting the original abnormal score of the abnormal power data according to the abnormal correlation degree to obtain a significant abnormal score, and identifying abnormalities in the abnormal power data to determine whether the power data of the mining power system is a real abnormality. The present invention can more accurately identify and distinguish real abnormalities from false abnormalities caused by electromagnetic interference by using the RRCF algorithm in combination with the calculation of the local fluctuation degree and the abnormal correlation degree, thereby improving the accuracy of detection.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and more specifically, to a mining power monitoring system and method. Background Art

[0002] The mine power system includes transformers, generators, motors, switchgear and other components. There are many devices, complex models and different operating modes, and real-time monitoring of the operating status of these devices is crucial to ensure system safety. Aging, improper operation or sudden failure of equipment may cause the power system to shut down or cause accidents, which in turn poses risks to mine production safety. By monitoring mine power equipment, it is possible to promptly detect hidden faults of electrical equipment (such as overload, current, voltage abnormalities, temperature rise, etc.) and issue an alarm to avoid large-scale power outages or equipment damage.

[0003] The existing Chinese patent application document with publication number CN118898363A discloses a mining power intelligent operation monitoring system and method. The method includes the following steps: obtaining a spatial distribution planning map of mining power equipment and performing spatial simulation analysis and regional sensor deployment key point determination to obtain the mining power equipment sensor monitoring deployment key points in each equipment distribution area; by deploying power monitoring sensors to perform real-time power parameter acquisition and regional power interactive change curve analysis, obtain the power operation work intensity-temperature impact interactive fluctuation change sub-curves in each area; based on the power operation work intensity-temperature impact interactive fluctuation change sub-curves in each area, perform power operation state abnormal monitoring and positioning and intelligent response control, generate mining power equipment operation abnormal intelligent response control strategy, so as to perform corresponding mining power equipment operation adjustment control work.

[0004] The application document can realize the intelligent monitoring process of the operation of mining power equipment. At present, electromagnetic interference is a common and serious influencing factor in the mine environment, especially when measuring and monitoring in the mine. Electromagnetic interference may come from electrical equipment, transmission lines, mine cars and other mechanical equipment in the mine, which can have a significant impact on the measuring instrument, and may cause the data collected by the sensor to be offset or interfere with the signal, resulting in distortion of some original data, thereby affecting the subsequent calculation of abnormal scores. Summary of the invention

[0005] In order to solve the problem that the data collected by the sensors of mining power equipment may be offset or interfered with by signals, resulting in distortion of part of the original data, thereby affecting the accuracy of subsequent detection, the present invention provides solutions in the following aspects.

[0006] In a first aspect, a mining power monitoring method comprises: obtaining multi-dimensional power data in a mining power system, wherein the power data comprises: current, voltage and power; identifying anomalies in the power data using an RRCF algorithm to obtain abnormal power data, and calculating the local fluctuation degree of the abnormal power data according to the change difference of the current abnormal power data; using the difference between the local fluctuation degrees of the multi-dimensional abnormal power data as the correlation coefficient of each dimension, exponentially decaying the variance of the correlation coefficient of each dimension using a negative exponential function to obtain the abnormal correlation degree of each dimension in the power data; correcting the original abnormal score of the abnormal power data according to the abnormal correlation degree to obtain a significant abnormal score, identifying anomalies in the abnormal power data, and determining whether the power data of the mining power system is a real abnormality; wherein the significant abnormal score Satisfies the following relationship: , where Indicates The original anomaly score of abnormal power data, Indicates The local fluctuation degree of abnormal power data, Indicates The abnormal correlation degree of each dimension in the abnormal power data, Represented by natural numbers An exponential function with base .

[0007] The effect is as follows: by acquiring multi-dimensional power data such as current, voltage and power, the state of the mine power system can be fully monitored. The RRCF algorithm is used to identify abnormal power data, and combined with the calculation of the local fluctuation degree, the abnormal situation in the power system can be detected more accurately, and a high accuracy can be maintained even in the presence of electromagnetic interference; by calculating the correlation coefficient of each dimension and using a negative exponential function to exponentially decay the abnormal correlation degree, the original abnormal score is effectively corrected to obtain a significant abnormal score, which helps to reduce false alarms caused by interference and improves the ability to identify real abnormal situations. By comprehensively considering the local fluctuation degree and abnormal correlation degree of multi-dimensional data, the robustness of the power monitoring system is enhanced.

[0008] Preferably, the identifying abnormality of the power data includes:

[0009] The power data within a preset time before the current power data collection moment is used as a data set, and the data set is divided into subsets;

[0010] Take any subset as the target subset, and use the median in the target subset as the split point. Nodes in the target subset that are smaller than the median are placed on the left side of the binary tree. Otherwise, they are placed on the right side of the binary tree. Iterate the above steps until all nodes in the subset are placed and the random binary tree is constructed.

[0011] Insert the current power data into each random binary tree, calculate the complexity change of each random binary tree, and use the mean of the complexity change as the original anomaly score of the current power data, where the complexity change is the difference between the mean of the depths of all nodes in the binary tree before and after the node is inserted;

[0012] In response to the original abnormality score of the current power data being greater than the abnormality threshold, the current power data is marked as abnormal data, otherwise it is normal data.

[0013] The effect is as follows: by dividing the power data within a preset time into multiple subsets and constructing a random binary tree, a large amount of data can be efficiently processed and analyzed. By using the median as the split point, outliers in the data set can be adaptively processed. By calculating the complexity change of each random binary tree after inserting the current power data and taking the mean as the original anomaly score, the original anomaly score can be compared with the preset anomaly threshold to dynamically mark abnormal data, which helps to adapt to different working environments and conditions and improve the flexibility and accuracy of anomaly detection.

[0014] Preferably, the calculating of the local fluctuation degree of abnormal power data includes:

[0015] Taking any abnormal power data as the marked abnormal data, calculating the ratio of the difference between the marked abnormal data and the power data at the previous moment and the marked abnormal data, and obtaining the degree of change of the marked abnormal data;

[0016] The absolute difference between the marked abnormal data and the average value of the data set within a preset time before the marked abnormal data is used as the relative change degree of the marked abnormal data;

[0017] The square root of the sum of the square of the degree of change and the square of the relative degree of change is taken as the local fluctuation degree of the marked abnormal data.

[0018] The effect is that by calculating the difference ratio between the marked abnormal data and the data at the previous moment, it is helpful to accurately capture the instantaneous changes of the marked abnormal data, and by calculating the absolute difference between the marked abnormal data and the average value of the data set within the previous preset time, it is possible to evaluate the significance of the marked abnormal data relative to the normal operating state.

[0019] Preferably, the calculating of the local fluctuation degree of abnormal power data further includes:

[0020] Taking any abnormal power data as marked abnormal data, calculating the average value of the marked abnormal data within a preset time and the local standard deviation within the same preset time;

[0021] The square root of the difference between the marked abnormal data and the mean value and the square of the local standard deviation is taken as the local fluctuation degree of the marked abnormal data.

[0022] The effect is that by combining the mean and standard deviation, normal fluctuations and abnormal fluctuations can be better distinguished, which is conducive to identifying false anomalies that may be caused by random fluctuations or noise, thus having greater robustness.

[0023] Preferably, the correlation coefficient includes:

[0024] The difference between the local fluctuation degree of the current dimension and the local fluctuation degree of the voltage dimension and the power dimension is calculated respectively, and the reciprocal of the square root of the sum of the squares of the difference is taken as the correlation coefficient of the current dimension;

[0025] The difference between the local fluctuation degree of the voltage dimension and the local fluctuation degree of the current dimension and the power dimension is calculated respectively, and the reciprocal of the square root of the sum of the squares of the difference is taken as the correlation coefficient of the voltage dimension;

[0026] The differences between the local fluctuation degree of the power dimension and the local fluctuation degrees of the current dimension and the voltage dimension are calculated respectively, and the reciprocal of the square root of the sum of the squares of the differences is taken as the correlation coefficient of the power dimension.

[0027] The effect is that by calculating the difference between the local fluctuations of different dimensions, the correlation between the dimensions can be quantified. The correlation coefficient helps to improve the accuracy of anomaly detection. If the local fluctuations of all dimensions are similar, the correlation coefficient is high, which may indicate that the anomaly in the power system is real and affects multiple dimensions. On the contrary, if the local fluctuations of a certain dimension are significantly different from those of other dimensions, the correlation coefficient will be low, which may indicate that the anomaly in this dimension is false and may be caused by measurement errors or external interference. By considering the correlations between different dimensions, this method enhances the robustness of the power monitoring system. The system can better adapt to different working environments and maintain stable operation even in the face of complex electromagnetic interference.

[0028] Preferably, the correlation coefficient further includes:

[0029] Taking any abnormal power data as the marked abnormal data, respectively calculating the sum of the covariances between the current dimension and the voltage dimension, and the current dimension and the power dimension of the marked abnormal data within a preset time, respectively calculating the variance of the current dimension, the variance of the voltage dimension, and the variance of the power dimension, respectively calculating the sum of the square root of the product of the variance of the current dimension and the variance of the voltage dimension and the square root of the product of the variance of the current dimension and the variance of the power dimension, and taking the ratio of the sum of the covariances to the sum of the square roots as the correlation coefficient of the current dimension of the marked abnormal data within the preset time;

[0030] The sum of the covariances between the voltage dimension and the current dimension, and between the voltage dimension and the power dimension of the marked abnormal data within the preset time is calculated respectively, and the sum of the square root of the product of the variance of the voltage dimension and the variance of the current dimension and the square root of the product of the variance of the voltage dimension and the variance of the power dimension are calculated respectively, and the ratio of the sum of the covariances to the sum of the square roots is used as the correlation coefficient of the voltage dimension of the marked abnormal data within the preset time;

[0031] The sum of the covariances between the power dimension and the current dimension, and the sum of the covariances between the power dimension and the voltage dimension of the marked abnormal data within the preset time are calculated respectively, and the sum of the square root of the product of the variance of the power dimension and the variance of the current dimension and the square root of the product of the variance of the power dimension and the variance of the voltage dimension are calculated respectively, and the ratio of the sum of the covariances to the sum of the square roots is used as the correlation coefficient of the power dimension of the marked abnormal data within the preset time.

[0032] Preferably, the determining whether the power data of the mining power system is truly abnormal includes:

[0033] In response to the significant anomaly score and the original anomaly score being greater than the anomaly threshold, the mining power data is true anomaly data, otherwise the anomaly label is changed to a normal label;

[0034] In response to the mine power data being real abnormal data, the system triggers an early warning mechanism and sends an alarm to relevant staff.

[0035] In a second aspect, a mine power monitoring system includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned mine power monitoring method is implemented.

[0036] The present invention has the following effects:

[0037] 1. The present invention acquires multi-dimensional power data in the mining power system and uses the RRCF algorithm in combination with the calculation of the local fluctuation degree and the abnormal correlation degree, so as to more accurately identify and distinguish between real anomalies and false anomalies caused by electromagnetic interference, and reduce the influence of the original data distortion caused by sensor data offset or interference signal on the accuracy of anomaly detection.

[0038] 2. The present invention calculates the significant anomaly score and corrects the original anomaly score of the abnormal power data in combination with the local fluctuation degree and the abnormal correlation degree, which helps to remove the error caused by electromagnetic interference, ensures that the detected abnormal data is more accurate, avoids misjudgment of normal operation, and thus enhances the system's ability to recognize real abnormal patterns. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0040] Figure 1 It is a method flow chart of a mining power monitoring method according to an embodiment of the present invention.

[0041] Figure 2 It is a structural block diagram of a mining power monitoring system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0043] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0044] Specific implementation scenario: In a mine power monitoring system and method, the RRCF algorithm is applied to identify abnormal power data in the mine power system. When abnormal power data is identified, an early warning device is triggered to monitor the mine power.

[0045] Reference Figure 1 A mining power monitoring method includes steps S1 to S4, which are as follows:

[0046] S1: Acquire multi-dimensional power data in the mine power system, where the power data includes: current, voltage and power.

[0047] It should be noted that sensors are deployed in the power equipment in the mine to obtain power data of the mine power equipment, which includes but is not limited to: current, voltage, and power data, and the same collection frequency (such as 1Hz) is set, and the current, voltage, and power data of the mine power equipment at the same moment are recorded as the power data at that moment.

[0048] S2: Use the RRCF algorithm to identify abnormalities in power data, obtain abnormal power data, and calculate the local fluctuation degree of the abnormal power data based on the change difference of the current abnormal power data.

[0049] In this embodiment, the RRCF algorithm is a machine learning algorithm for anomaly detection, and its basic principle is: by constructing multiple random binary trees to evaluate the degree of abnormality of data points, when identifying abnormal data, it is judged based on the degree of change in tree complexity caused by inserting nodes into each random binary tree. The average of the degree of change in tree complexity caused by inserting nodes into each random binary tree is the original anomaly score of the node. When the original anomaly score exceeds the set threshold, the node is marked as an anomaly point; the RRCF algorithm is a well-known technology in the art and will not be described in detail.

[0050] Identify abnormalities in power data, including:

[0051] The power data within a preset time before the current power data collection moment is used as a data set, and the data set is divided into subsets;

[0052] It should be noted that, in this embodiment, the preset time is 1 minute, which can be adjusted according to the specific implementation situation. The power data within 1 minute before the current power data collection moment is used as the data set, and the data set is divided into subsets of length 15, and a random binary tree is constructed for each subset.

[0053] Take any subset as the target subset, and use the median in the target subset as the split point. Nodes in the target subset that are smaller than the median are placed on the left side of the binary tree. Otherwise, they are placed on the right side of the binary tree. Iterate the above steps until all nodes in the subset are placed and the random binary tree is constructed.

[0054] Insert the current power data into each random binary tree, calculate the complexity change of each random binary tree, and use the mean of the complexity change as the original anomaly score of the current power data, where the complexity change is the difference between the mean of the depths of all nodes in the binary tree before and after the node is inserted;

[0055] It should be noted that the calculation of node depth is a well-known technology to those skilled in the art.

[0056] In response to the original abnormality score of the current power data being greater than the abnormality threshold, the current power data is marked as abnormal data, otherwise it is normal data.

[0057] In this embodiment, the abnormal threshold is 0.62, and the implementer can adjust it according to the specific situation.

[0058] That is to say, since the RRCF algorithm identifies abnormal data based on the degree of complexity change when nodes are inserted into different binary trees, this identification method does not take into account some characteristics of the data itself. Therefore, in the process of correcting the original abnormal score of abnormal power data, it can be analyzed based on its local fluctuation degree and the degree of correlation between internal dimensions.

[0059] The reason for the analysis is that local volatility refers to the degree of change in abnormal power data in time or space, which can be manifested as a sudden jump in data points, a sharp fluctuation in data sequences, etc. The greater the local volatility, the more significant the data has changed at a certain moment or in a certain area, which may be caused by equipment failure or interference from the external environment.

[0060] Therefore, local volatility can be used as an important indicator to evaluate the degree of abnormality and to correct the original abnormality score. The higher the local volatility of the abnormal power data, the more credible the abnormal power data is, and the smaller the degree of correction of the original abnormality score is. At the same time, the abnormal power data here is multi-dimensional data, including three dimensions: current, voltage, and power. These dimensions within the power data usually have a certain correlation. If the data of a certain dimension of the abnormal power data is abnormal, and the data of other related dimensions do not change accordingly, this may indicate that the abnormality is forged. Then, the calculation of the local volatility includes the following steps:

[0061] Taking any abnormal power data as the marked abnormal data, calculating the ratio of the difference between the marked abnormal data and the power data at the previous moment and the marked abnormal data, and obtaining the degree of change of the marked abnormal data;

[0062] It should be noted that the reason why the power data at the previous moment is selected to calculate the degree of change of the marked abnormal data is that: the power data is time series data and has continuity. Selecting the data at the previous moment can reflect the direct predecessor state of the marked abnormal data in time, thereby capturing the immediate changes before the abnormality occurs. By comparing the marked abnormal data point with its previous moment data, the change trend of the power parameters before the abnormality occurs can be analyzed. This comparison helps to identify the suddenness of the abnormality, that is, whether the abnormality is caused by a sharp change.

[0063] The absolute difference between the marked abnormal data and the average value of the data set within a preset time before the marked abnormal data is used as the relative change degree of the marked abnormal data;

[0064] The square root of the sum of the square of the degree of change and the square of the relative degree of change is taken as the local fluctuation degree of the marked abnormal data.

[0065] Specifically, the local fluctuation degree satisfies the following relationship:

[0066] ;

[0067] In the formula, Indicates The local fluctuation degree of abnormal power data, Indicates The size of abnormal power data, Indicates The size of the power data of the previous moment adjacent to the abnormal power data, It indicates the mean value of the power data size in the data set corresponding to the power data within the preset time before the current abnormal power data collection moment.

[0068] That is to say, Indicates The greater the value, the greater the change of the abnormal power data compared with the previous power data. The greater the degree of change of abnormal power data.

[0069] It represents the mean difference between the size of the previous abnormal power data and the size of the power data in the corresponding data set within 1 minute before the current abnormal power data collection moment. The greater the mean difference, the higher the relative volatility of the abnormal power data. The data set corresponding to the power data within the preset time before the current abnormal power data collection moment is used here because the current abnormal power data is identified based on the random binary tree composed of the data in the data set.

[0070] Specifically, the calculation method of the size of the abnormal power data satisfies the following relationship:

[0071] ;

[0072] In the formula, Indicates The size of abnormal power data, Indicates The current, voltage and power of the abnormal power data are Represents the inverse tangent function.

[0073] It should be noted that the purpose of using the inverse tangent function is to map the input value to The range of the data can be reduced, thereby reducing the impact of different dimensions (current, voltage, power) on the comprehensive measurement, making the data of different dimensions more fair when compared.

[0074] In addition, the characteristic value of the data set corresponding to the power data within 1 minute before the abnormal power data collection time can also be replaced by the variance of the power data size in the data set. Specifically, the local fluctuation degree satisfies the following relationship:

[0075] ;

[0076] In the formula, Indicates The local fluctuation degree of abnormal power data, Indicates The size of abnormal power data, Indicates The size of the power data of the previous moment adjacent to the abnormal power data, represents the variance function.

[0077] In addition, another embodiment further includes:

[0078] Taking any abnormal power data as marked abnormal data, calculating the average value of the marked abnormal data within a preset time and the local standard deviation within the same preset time;

[0079] The square root of the difference between the marked abnormal data and the mean value and the square of the local standard deviation is taken as the local fluctuation degree of the marked abnormal data.

[0080] Specifically, the local fluctuation degree satisfies the following relationship:

[0081] ;

[0082] In the formula, Indicates The local fluctuation degree of abnormal power data, Indicates The size of abnormal power data, Indicates The average value of abnormal power data within a preset time period, Indicates The local standard deviation of abnormal power data within a preset time.

[0083] That is, the average value within the preset time represents the average operating status of the power system over a period of time, which helps to identify and filter out short-term random fluctuations. The local standard deviation represents the volatility of the power system around the preset time. A high standard deviation means that the system has experienced large fluctuations within the preset time, which may be caused by load changes, equipment failures or other external factors.

[0084] S3: The difference between the local fluctuation degrees of multi-dimensional abnormal power data is used as the correlation coefficient of each dimension, and the variance of the correlation coefficient of each dimension is exponentially decayed using a negative exponential function to obtain the abnormal correlation degree of each dimension in the power data.

[0085] It should be noted that the variance of the correlation coefficient of each dimension represents the The variance of the correlation coefficients of the three dimensions in the abnormal power data. The larger the variance, the greater the difference in the correlation coefficients between the three dimensions. The lower the abnormal correlation degree of each dimension in the abnormal power data.

[0086] The correlation coefficient includes the steps:

[0087] The difference between the local fluctuation degree of the current dimension and the local fluctuation degree of the voltage dimension and the power dimension is calculated respectively, and the reciprocal of the square root of the sum of the squares of the difference is taken as the correlation coefficient of the current dimension;

[0088] Specifically, the correlation coefficient of the current dimension satisfies the following relationship:

[0089] ;

[0090] In the formula, Indicates The correlation coefficient of the current dimension in the abnormal power data, Indicates The local fluctuation degree of the current dimension in the abnormal power data, Indicates The local fluctuation degree of voltage dimension in abnormal power data, Indicates The local fluctuation degree of the power dimension in the abnormal power data.

[0091] in, Indicates The greater the degree of local fluctuation of the current dimension in the abnormal power data is compared with the degree of local fluctuation of the other two dimensions of voltage and power, the lower the correlation coefficient of the current dimension is.

[0092] The difference between the local fluctuation degree of the voltage dimension and the local fluctuation degree of the current dimension and the power dimension is calculated respectively, and the reciprocal of the square root of the sum of the squares of the difference is taken as the correlation coefficient of the voltage dimension;

[0093] Specifically, the correlation coefficient of the voltage dimension satisfies the following relationship:

[0094] ;

[0095] In the formula, Indicates The correlation coefficient of the voltage dimension in the abnormal power data, Indicates The local fluctuation degree of the current dimension in the abnormal power data, Indicates The local fluctuation degree of voltage dimension in abnormal power data, Indicates The local fluctuation degree of the power dimension in the abnormal power data.

[0096] The differences between the local fluctuation degree of the power dimension and the local fluctuation degrees of the current dimension and the voltage dimension are calculated respectively, and the reciprocal of the square root of the sum of the squares of the differences is taken as the correlation coefficient of the power dimension.

[0097] Specifically, the correlation coefficient of the power dimension satisfies the following relationship:

[0098] ;

[0099] In the formula, Indicates The correlation coefficient of the power dimension in the abnormal power data, Indicates The local fluctuation degree of the current dimension in the abnormal power data, Indicates The local fluctuation degree of voltage dimension in abnormal power data, Indicates The local fluctuation degree of the power dimension in the abnormal power data.

[0100] In addition, another embodiment further includes:

[0101] Taking any abnormal power data as the marked abnormal data, respectively calculating the sum of the covariances between the current dimension and the voltage dimension, and the current dimension and the power dimension of the marked abnormal data within a preset time, respectively calculating the variance of the current dimension, the variance of the voltage dimension, and the variance of the power dimension, respectively calculating the sum of the square root of the product of the variance of the current dimension and the variance of the voltage dimension and the square root of the product of the variance of the current dimension and the variance of the power dimension, and taking the ratio of the sum of the covariances to the sum of the square roots as the correlation coefficient of the current dimension of the marked abnormal data within the preset time;

[0102] Specifically, the correlation coefficient of the current dimension satisfies the following relationship:

[0103] ;

[0104] In the formula, Indicates The correlation coefficient of the current dimension in the abnormal power data, Indicates The local fluctuation degree of the current dimension in the abnormal power data, Indicates The local fluctuation degree of voltage dimension in abnormal power data, Indicates The local fluctuation degree of the power dimension in the abnormal power data, represents the covariance function, represents the variance function.

[0105] The sum of the covariances between the voltage dimension and the current dimension, and between the voltage dimension and the power dimension of the marked abnormal data within the preset time is calculated respectively, and the sum of the square root of the product of the variance of the voltage dimension and the variance of the current dimension and the square root of the product of the variance of the voltage dimension and the variance of the power dimension are calculated respectively, and the ratio of the sum of the covariances to the sum of the square roots is used as the correlation coefficient of the voltage dimension of the marked abnormal data within the preset time;

[0106] Specifically, the correlation coefficient of the voltage dimension satisfies the following relationship:

[0107] ;

[0108] In the formula, Indicates The correlation coefficient of the voltage dimension in the abnormal power data, Indicates The local fluctuation degree of the current dimension in the abnormal power data, Indicates The local fluctuation degree of voltage dimension in abnormal power data, Indicates The local fluctuation degree of the power dimension in the abnormal power data, represents the covariance function, represents the variance function.

[0109] The sum of the covariances between the power dimension and the current dimension, and the sum of the covariances between the power dimension and the voltage dimension of the marked abnormal data within the preset time are calculated respectively, and the sum of the square root of the product of the variance of the power dimension and the variance of the current dimension and the square root of the product of the variance of the power dimension and the variance of the voltage dimension are calculated respectively, and the ratio of the sum of the covariances to the sum of the square roots is used as the correlation coefficient of the power dimension of the marked abnormal data within the preset time.

[0110] Specifically, the correlation coefficient of the power dimension satisfies the following relationship:

[0111] ;

[0112] In the formula, Indicates The correlation coefficient of the power dimension in the abnormal power data, Indicates The local fluctuation degree of the current dimension in the abnormal power data, Indicates The local fluctuation degree of voltage dimension in abnormal power data, Indicates The local fluctuation degree of the power dimension in the abnormal power data, represents the covariance function, represents the variance function.

[0113] The two embodiments in step S3 have the same effect, and the implementation method can be selected according to the specific situation.

[0114] Specifically, the abnormal correlation degree satisfies the following relationship:

[0115] ;

[0116] In the formula, Indicates The abnormal correlation degree of each dimension in the abnormal power data, Indicates The correlation coefficient of the current dimension in the abnormal power data, Indicates The correlation coefficient of the voltage dimension in the abnormal power data, Indicates The correlation coefficient of the power dimension in the abnormal power data, Represented by natural numbers An exponential function with base .

[0117] S4: The original abnormality score of the abnormal power data is modified according to the abnormal correlation degree to obtain a significant abnormality score, and the abnormal power data is identified as abnormal to determine whether the power data of the mine power system is a real abnormality.

[0118] Specifically, the significant abnormality score Satisfies the following relationship:

[0119] ;

[0120] In the formula, Indicates The significant anomaly scores of abnormal power data, Indicates The original anomaly score of abnormal power data, Indicates The local fluctuation degree of abnormal power data, Indicates The abnormal correlation degree of each dimension in the abnormal power data, Represented by natural numbers An exponential function with base .

[0121] In response to the significant anomaly score and the original anomaly score being greater than the anomaly threshold, the mining power data is true anomaly data, otherwise the anomaly label is changed to a normal label;

[0122] In response to the mine power data being real abnormal data, the system triggers an early warning mechanism and sends an alarm to relevant staff.

[0123] It should be noted that the abnormal threshold is 0.62, and implementers can adjust it according to specific circumstances.

[0124] The present invention also provides a mining power monitoring system. Figure 2 As shown, the system includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a mining power monitoring method according to the first aspect of the present invention is implemented.

[0125] The system also includes other components familiar to those skilled in the art, such as a communication bus and a communication interface. The configuration and functions of these components are known in the art and will not be described in detail here.

[0126] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented by computer-readable / executable instructions stored or otherwise maintained by such a computer-readable medium.

[0127] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0128] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A mining power monitoring method, characterized in that: include: Acquire multi-dimensional power data in a mining power system, wherein the power data includes: current, voltage and power; Use the RRCF algorithm to identify abnormalities in power data to obtain abnormal power data, and based on the change difference of the current abnormal power data, take any abnormal power data as the marked abnormal data, calculate the ratio of the difference between the marked abnormal data and the power data at the previous moment and the marked abnormal data, and obtain the change degree of the marked abnormal data; take the absolute difference between the marked abnormal data and the average value of the data set within a preset time before the marked abnormal data as the relative change degree of the marked abnormal data; take the square root of the sum of the square of the change degree and the square of the relative change degree as the local fluctuation degree of the marked abnormal data; or, Taking any abnormal power data as the marked abnormal data, calculating the average value of the marked abnormal data within a preset time and the local standard deviation within the same preset time; adding the square of the difference between the marked abnormal data and the average value and the square of the local standard deviation to obtain the square root as the local fluctuation degree of the marked abnormal data; The difference between the local fluctuation degrees of multi-dimensional abnormal power data is taken as the correlation coefficient of each dimension, and the variance of the correlation coefficient of each dimension is exponentially decayed using a negative exponential function to obtain the abnormal correlation degree of each dimension in the power data; The original abnormal score of abnormal power data is modified according to the abnormal correlation degree to obtain a significant abnormal score, and the abnormal power data is identified as abnormal to determine whether the power data of the mine power system is truly abnormal; Among them, the significant abnormality score Satisfies the following relationship: , where Indicates The original anomaly score of abnormal power data, Indicates The local fluctuation degree of abnormal power data, Indicates The abnormal correlation degree of each dimension in the abnormal power data, Represented by natural numbers An exponential function with base .

2. A mining power monitoring method according to claim 1, characterized in that: The identifying abnormality of the power data includes: The power data within a preset time before the current power data collection moment is used as a data set, and the data set is divided into subsets; Take any subset as the target subset, and use the median in the target subset as the split point. Nodes in the target subset that are smaller than the median are placed on the left side of the binary tree. Otherwise, they are placed on the right side of the binary tree. Iterate the above steps until all nodes in the subset are placed and the random binary tree is constructed. Insert the current power data into each random binary tree, calculate the complexity change of each random binary tree, and use the mean of the complexity change as the original anomaly score of the current power data, where the complexity change is the difference between the mean of the depths of all nodes in the binary tree before and after the node is inserted; In response to the original abnormality score of the current power data being greater than the abnormality threshold, the current power data is marked as abnormal data, otherwise it is normal data.

3. A mining power monitoring method according to claim 1, characterized in that: The correlation coefficient includes: The difference between the local fluctuation degree of the current dimension and the local fluctuation degree of the voltage dimension and the power dimension is calculated respectively, and the reciprocal of the square root of the sum of the squares of the difference is taken as the correlation coefficient of the current dimension; The difference between the local fluctuation degree of the voltage dimension and the local fluctuation degree of the current dimension and the power dimension is calculated respectively, and the reciprocal of the square root of the sum of the squares of the difference is taken as the correlation coefficient of the voltage dimension; The differences between the local fluctuation degree of the power dimension and the local fluctuation degrees of the current dimension and the voltage dimension are calculated respectively, and the reciprocal of the square root of the sum of the squares of the differences is taken as the correlation coefficient of the power dimension.

4. A mining power monitoring method according to claim 1, characterized in that: The correlation coefficient also includes: Taking any abnormal power data as the marked abnormal data, respectively calculating the sum of the covariances between the current dimension and the voltage dimension, and the current dimension and the power dimension of the marked abnormal data within a preset time, respectively calculating the variance of the current dimension, the variance of the voltage dimension, and the variance of the power dimension, respectively calculating the sum of the square root of the product of the variance of the current dimension and the variance of the voltage dimension and the square root of the product of the variance of the current dimension and the variance of the power dimension, and taking the ratio of the sum of the covariances to the sum of the square roots as the correlation coefficient of the current dimension of the marked abnormal data within the preset time; The sum of the covariances between the voltage dimension and the current dimension, and between the voltage dimension and the power dimension of the marked abnormal data within the preset time is calculated respectively, and the sum of the square root of the product of the variance of the voltage dimension and the variance of the current dimension and the square root of the product of the variance of the voltage dimension and the variance of the power dimension are calculated respectively, and the ratio of the sum of the covariances to the sum of the square roots is used as the correlation coefficient of the voltage dimension of the marked abnormal data within the preset time; The sum of the covariances between the power dimension and the current dimension, and the sum of the covariances between the power dimension and the voltage dimension of the marked abnormal data within the preset time are calculated respectively, and the sum of the square root of the product of the variance of the power dimension and the variance of the current dimension and the square root of the product of the variance of the power dimension and the variance of the voltage dimension are calculated respectively, and the ratio of the sum of the covariances to the sum of the square roots is used as the correlation coefficient of the power dimension of the marked abnormal data within the preset time.

5. A mining power monitoring method according to claim 1, characterized in that: The determining whether the power data of the mining power system is truly abnormal includes: In response to the significant anomaly score and the original anomaly score being greater than the anomaly threshold, the mining power data is true anomaly data, otherwise the anomaly label is changed to a normal label; In response to the mine power data being real abnormal data, the system triggers an early warning mechanism and sends an alarm to relevant staff.

6. A mining power monitoring system, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the mining power monitoring method according to any one of claims 1 to 5 is implemented.

Images