A concentrator intelligent fusion terminal with equipment operation status monitoring function

Abstract

This invention relates to the field of data processing technology, and in particular to a concentrator intelligent fusion terminal with equipment operation status monitoring function. The terminal includes a processor and a memory. The processor executes a computer program in the memory to perform the following steps: acquiring the equipment non-mechanical state coefficient and equipment mechanical state coefficient at each moment within a reference operating period of the concentrator; dividing the reference operating period into at least two time periods; clustering the equipment mechanical state coefficient and equipment non-mechanical state coefficient at each moment within each time period into two-dimensional feature values ​​to obtain a multi-dimensional clustering result index for each time period; based on the differences in the multi-dimensional clustering result indexes for each time period, acquiring a clustering reference time period for each moment within a preset future operating cycle; acquiring the two-dimensional feature values ​​of any moment and each moment within its clustering reference time period and clustering them to obtain the operation status monitoring result, thereby increasing the accuracy of equipment operation status monitoring.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a concentrator intelligent fusion terminal with equipment operation status monitoring function. Background Technology

[0002] Against the backdrop of the accelerated construction of smart grids, distribution substations, as the "last mile" of the power system, face core challenges in their intelligent transformation. The intelligent fusion terminal of the concentrator, as the data nerve center of the distribution substation, mainly uses the threshold comparison method for equipment status monitoring (such as triggering simple alarms by the tilt angle of drop-out fuses or the temperature of transformer shells). However, this threshold comparison method has a lag in fault early warning in the field of equipment status monitoring. It is difficult to capture the gradual process of equipment deterioration based on a single parameter threshold, resulting in fault warnings often reaching a critical stage. Moreover, this technical architecture is difficult to meet the requirements of new power systems for full perception of equipment status, proactive fault assessment, and dynamic optimization of power supply quality.

[0003] Existing technologies typically use unsupervised learning algorithms to address the problems of threshold comparison methods. Unsupervised learning algorithms (such as DBSCAN, GMM, and Self-Organizing Maps SOM) can automatically determine the cluster structure based on data density or probability distribution, adapting to the time-varying characteristics of equipment operating status. However, when using such clustering algorithms, multidimensional data is often used for clustering. Although concentrator equipment status data is diverse, multidimensional data is difficult to use directly, and as equipment ages, the distribution of some data may change. This means that the presence of old data may affect the classification results of new data, leading to inaccurate monitoring results of equipment operating status.

[0004] Therefore, how to reduce the impact of old data in the concentrator and increase the accuracy of equipment operation status monitoring has become an urgent problem to be solved. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a concentrator intelligent fusion terminal with equipment operation status monitoring function to solve the problem of how to reduce the impact of old data in the concentrator and increase the accuracy of equipment operation status monitoring.

[0006] This invention provides a concentrator intelligent fusion terminal with device operation status monitoring function, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it performs the following steps:

[0007] Acquire multidimensional mechanical data and multidimensional non-mechanical data at every moment during the operation of the concentrator;

[0008] Obtain the reference runtime of the concentrator, and based on the multidimensional non-mechanical data at each moment within the reference runtime, obtain the equipment non-mechanical state coefficient at each moment. Based on the multidimensional mechanical data at each moment within the reference runtime, obtain the equipment mechanical state coefficient at each moment.

[0009] The reference runtime is divided into at least two time periods. For any given time period, the mechanical state coefficient and non-mechanical state coefficient of the equipment at each moment within that time period are combined to form a two-dimensional feature value for that moment. The two-dimensional feature values ​​at each moment are clustered to obtain a multi-dimensional clustering result index for that given time period. The multi-dimensional clustering result index for each time period is obtained. Based on the differences in the multi-dimensional clustering result index for each time period, the attenuation coefficient and attenuation window value are obtained.

[0010] Based on the attenuation coefficient and the attenuation window value, a clustering reference time period is obtained for each moment in a preset future operating cycle after the reference operating period. For any moment in the preset future operating cycle, the two-dimensional feature values ​​of the any moment and each moment in the clustering reference time period are obtained and clustered to obtain the monitoring result of the concentrator's operating status at any moment.

[0011] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows:

[0012] The process involves acquiring multidimensional mechanical and non-mechanical data at each moment during the concentrator's operation; obtaining a reference operating period for the concentrator; acquiring the equipment non-mechanical state coefficient at each moment based on the multidimensional non-mechanical data within the reference operating period; acquiring the equipment mechanical state coefficient at each moment based on the multidimensional mechanical data within the reference operating period; dividing the reference operating period into at least two time periods; for any given time period, constructing a two-dimensional feature value for each moment by combining the equipment mechanical state coefficient and the equipment non-mechanical state coefficient; clustering the two-dimensional feature values ​​at each moment to obtain a multidimensional clustering result index for that time period; acquiring the multidimensional clustering result index for each time period; and acquiring an attenuation coefficient and an attenuation window value based on the differences in the multidimensional clustering result indexes for each time period; and based on the attenuation coefficient and the attenuation window value, acquiring a clustering reference time period for each moment within a preset future operating cycle after the reference operating period; acquiring the two-dimensional feature values ​​of that moment and each moment within the preset future operating cycle, and clustering them to obtain the concentrator's operating status monitoring result at that moment. Specifically, based on the multidimensional non-mechanical data at each moment within the reference operating period, the non-mechanical state coefficient of the equipment at each moment is obtained, and based on the multidimensional mechanical data at each moment within the reference operating period, the mechanical state coefficient of the equipment at each moment is obtained. This transforms the multimodal data into a unified measurement standard, preparing for cluster analysis. Based on the differences in the multidimensional clustering results indicators for each time period, the attenuation coefficient and attenuation window value are obtained. This allows for the acquisition of a clustering reference time period for each moment within a preset future operating cycle after the reference operating period. When monitoring each moment within the preset future operating cycle, the influence of old data in the concentrator is reduced, increasing the accuracy of equipment operating status monitoring. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a flowchart of a concentrator intelligent fusion method with equipment operation status monitoring function provided in Embodiment 1 of the present invention. Detailed Implementation

[0015] Embodiments of this disclosure are described in detail below, with examples of these embodiments illustrated in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this disclosure, and should not be construed as limiting it.

[0016] It should be noted that the terms "first," "second," etc., used in this disclosure and the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure.

[0017] To illustrate the technical solution of the present invention, specific embodiments are described below.

[0018] This invention provides a concentrator intelligent fusion terminal with equipment operation status monitoring function, including a processor and a memory. The processor executes a computer program in the memory to implement a concentrator intelligent fusion method with equipment operation status monitoring function, such as... Figure 1 As shown, the method includes the following steps:

[0019] Step S101: Obtain multidimensional mechanical data and multidimensional non-mechanical data at each moment during the operation of the concentrator.

[0020] When monitoring the operating status of concentrators, traditional methods primarily consider power status data, such as input voltage, input current, power module temperature, internal temperature, and ambient humidity, due to the multi-dimensional data involved in their operation. However, they neglect mechanical status data, such as vibration and impact data, mechanical deformation, and displacement data. Problems with these data indicate potential malfunctions in the physical structure and dynamic characteristics of moving parts, such as loosening, wear, or deformation. Vibration, loosening, or corrosion of the concentrator's mechanical structure (such as terminals and antenna connections) can lead to poor contact or signal interference. For example, after a concentrator's 485 terminal became loose, the failure rate of meter data acquisition increased to 30%. In another distribution area, a loose antenna connection caused the signal strength to drop to -110dBm, resulting in a data upload success rate of only 70%.

[0021] Therefore, in this embodiment, the acquisition frequency is once per second, which is not limited here and can be set according to the specific implementation scenario. Multidimensional non-mechanical data at every moment of the concentrator's operation is acquired, including but not limited to current data, voltage data, power supply temperature, ambient humidity, and internal device temperature. Current and voltage data are acquired directly by the concentrator; power supply temperature and ambient humidity are acquired through temperature and humidity sensors; and internal device temperature is acquired through built-in temperature sensors, such as built-in temperature sensors (NTC / PTC thermistors) or digital temperature sensors (e.g., DS18B20, MAX31865). Multidimensional mechanical data at every moment of the concentrator's operation is also acquired, including but not limited to vibration data and multidimensional crimping force data. Vibration data is acquired through vibration sensors; multidimensional crimping force data includes but is not limited to crimping force data of terminals and antenna connector components. Terminal crimping force data is acquired by embedding a miniature crimping force sensor into a terminal crimping tool (e.g., crimping pliers) or directly clamping the sensor on both sides of the terminal. Antenna connector crimping force data is acquired by installing a thin-film pressure sensor under the connector housing or fixing nut to increase the reliability and comprehensiveness of the device operation status monitoring results.

[0022] Existing technologies typically use unsupervised learning algorithms to address the lag in fault warnings inherent in traditional threshold comparison methods. Unsupervised learning algorithms (such as DBSCAN, GMM, and Self-Organizing Maps (SOM)) can automatically determine cluster structures based on data density or probability distribution, adapting to the time-varying characteristics of equipment operating status. However, when using such clustering algorithms, multidimensional data is often employed for clustering. Although concentrator equipment status data is diverse, multidimensional data is difficult to use directly, and as equipment ages, the distribution of some data may change. This means that the presence of old data may affect the classification results of new data, leading to inaccurate monitoring results of equipment operating status.

[0023] Therefore, in this embodiment, a reference operating period of the concentrator is obtained, and the non-mechanical state coefficient and mechanical state coefficient of the equipment at each moment within the reference operating period are obtained. The reference operating period is divided into at least two time periods. The mechanical state coefficient and non-mechanical state coefficient of the equipment at each moment within each time period are used to form two-dimensional feature values ​​for clustering, thereby obtaining a multi-dimensional clustering result index for each time period. Based on the difference in the multi-dimensional clustering result index for each time period, a clustering reference time period for each moment within a preset future operating cycle is obtained. The two-dimensional feature values ​​of any moment and each moment within its clustering reference time period are obtained and clustered to obtain the operating status monitoring result, thereby increasing the accuracy of equipment operating status monitoring.

[0024] Step S102: Obtain the reference operating period of the concentrator; obtain the equipment non-mechanical state coefficient at each moment based on the multidimensional non-mechanical data at each moment within the reference operating period; and obtain the equipment mechanical state coefficient at each moment based on the multidimensional mechanical data at each moment within the reference operating period.

[0025] First, the reference runtime of the concentrator is obtained. Since the concentrator is not stable when it first starts running, this embodiment uses three months after the concentrator starts running for 20 minutes as the reference runtime of the concentrator for cluster analysis. There is no restriction here, and it can be set according to the specific implementation scenario.

[0026] Since the measurement scales for the equipment's non-mechanical state coefficients and mechanical state coefficients differ at each moment, feature extraction analysis is required to transform data with different measurement scales into a unified measurement standard. Because this solution aims to monitor equipment operating status, this embodiment obtains the equipment's non-mechanical state coefficients at each moment based on multidimensional non-mechanical data within the reference operating period, and obtains the equipment's mechanical state coefficients at each moment based on multidimensional mechanical data within the reference operating period, in preparation for cluster analysis.

[0027] The method for obtaining the equipment non-mechanical state coefficients at each moment based on the multidimensional non-mechanical data at each moment within the reference runtime period is as follows:

[0028] (1) For any dimension of non-mechanical data in the multidimensional non-mechanical data at any time during the reference running period, since the current data, voltage data, power supply temperature and ambient humidity exhibit stable fluctuation characteristics under normal conditions, when the non-mechanical data of any dimension is current data or voltage data or power supply temperature or ambient humidity, the device state coefficient of the non-mechanical data of any dimension is obtained according to the discrete characteristics of the non-mechanical data of any dimension and the previous non-mechanical data of the same dimension.

[0029] Specifically, according to a preset time period, the reference runtime is divided into at least two reference periods. If the number of historical reference periods preceding the reference period containing any dimension of non-mechanical data is greater than or equal to a first preset number, then the non-mechanical data belonging to the same dimension as the non-mechanical data in each historical reference period are grouped into a historical data sequence. In this embodiment, the preset time period is set to 6 hours, and the first preset number is set to 5. This is not a limitation and can be set according to the specific implementation scenario. The reciprocal of the coefficient of variation of each historical data sequence is obtained as the stable characteristic value of each historical data sequence. The coefficient of variation is existing technology and will not be elaborated here. The stable characteristic values ​​of each historical data sequence are sorted in descending order, and the average value of a preset proportion of stable characteristic values ​​is obtained as the... The reference stability index for any dimension of non-mechanical data is set to a preset ratio of 60% in this embodiment. This is not limited and can be set according to the specific implementation scenario. If the number of historical reference periods before the reference period of any dimension of non-mechanical data is less than the first preset number, the rated value of any dimension of non-mechanical data is obtained as the reference stability index for any dimension of non-mechanical data. The rated values ​​of voltage data and current data are the rated voltage and rated current of the concentrator. The rated values ​​of power supply temperature and ambient humidity are the actual monitored power supply temperature and ambient humidity after the concentrator starts running. In this embodiment, the rated values ​​of power supply temperature and ambient humidity are set to the power supply temperature and ambient humidity at the 5th moment after the concentrator starts running. This is not limited and can be set according to the specific implementation scenario.

[0030] Using any given moment as the cutoff moment, a time window of a preset size of 20 (i.e., containing 20 moments) is established. This is not limited and can be set according to the specific implementation scenario. Within the time window, the non-mechanical data of any dimension and its sequence of non-mechanical data of the same dimension are combined to form a first data sequence. Within the reference period where the non-mechanical data of any dimension is located, if the number of non-mechanical data of the same dimension preceding the non-mechanical data of any dimension is greater than or equal to a second preset number, then the non-mechanical data of any dimension and its sequence of the same dimension preceding it are combined to form a second data sequence. If the number of non-mechanical data of the same dimension preceding the non-mechanical data of any dimension is less than the second preset number but greater than or equal to a third preset number, then the non-mechanical data of any dimension and its sequence of all preceding non-mechanical data are combined to form a second data sequence. In this embodiment, the second preset number is set to 19 and the third preset number is set to 3. This is not limited and can be set according to the specific implementation scenario.

[0031] Based on the reference stability index of the non-mechanical data in any dimension, the coefficient of variation of the first data sequence, and the coefficient of variation of the second data sequence, the formula for calculating the equipment state coefficient of the non-mechanical data in any dimension is as follows:

[0032]

[0033] in, Equipment status coefficients for any dimension of non-mechanical data; A reference stability index for non-mechanical data in any dimension; The coefficient of variation for the first data sequence; The coefficient of variation for the second data sequence; It is the absolute value symbol; The third preset quantity; m is the number of non-mechanical data of the same dimension preceding the non-mechanical data of any dimension within the reference period where the non-mechanical data of any dimension is located.

[0034] It should be noted that, The stable eigenvalue represents the stability of local changes in non-mechanical data of any dimension. The smaller the value, the closer the stable characteristic value of the first data sequence is to the reference stable characteristic value. In this case, the better the equipment operating status is represented by any dimension of non-mechanical data. The larger it is; This represents the stable eigenvalue of the second data sequence, that is, the stability of local changes in any dimension of non-mechanical data within the reference period. The smaller the value, the closer the stable characteristic value of the second data sequence is to the reference stable characteristic value. In this case, the better the equipment operating status is represented by any dimension of non-mechanical data. The larger it is; when At that time, the data within the reference period of any dimension of non-mechanical data is insufficient for change stability analysis, so only the first data sequence is used to analyze the equipment operating status represented by any dimension of non-mechanical data.

[0035] (2) Since there is a consistent relationship between the internal temperature of the equipment and the load, that is, the internal temperature of the equipment is low when the load is low and the internal temperature of the equipment is high when the load is high, when any dimension of non-mechanical data is the internal temperature of the equipment, the internal temperature of the equipment at each moment in the time window is formed into an internal temperature data sequence, and the equipment load data (i.e., charge data, which is directly obtained through the concentrator) at each moment in the time window is obtained into an equipment load data sequence. Based on the correlation between the internal temperature data sequence and the equipment load data sequence, the non-mechanical state coefficient of the equipment at any moment is obtained.

[0036] Specifically, the Spearman correlation coefficient between the internal temperature data sequence and the load data sequence of the equipment is obtained. The Spearman correlation coefficient is existing technology and will not be elaborated here. Based on the sum of the Spearman correlation coefficient and the constant 1, the equipment state coefficient of the non-mechanical data of any dimension is obtained.

[0037] In one embodiment, the formula for calculating the equipment state coefficient of the internal temperature of the equipment is:

[0038]

[0039] in, The equipment state coefficient representing the internal temperature of the equipment; This represents the Spearman correlation coefficient between the equipment's internal temperature data sequence and the equipment load data sequence.

[0040] It should be noted that, A higher value indicates a stronger positive correlation between the equipment's internal temperature and its load data. This means the internal temperature generally corresponds to a lower temperature under low load and a higher temperature under high load, indicating better equipment performance. The larger it is.

[0041] (3) Obtain the equipment state coefficient of each dimension of the non-mechanical data in the multidimensional non-mechanical data at any time, and obtain the corresponding cumulative multiplication result as the equipment non-mechanical state coefficient at any time.

[0042] In one embodiment, taking the b-th time point within the reference operating period as an example, the formula for calculating the non-mechanical state coefficient of the equipment at the b-th time point is:

[0043]

[0044] Wherein, FM is the non-mechanical state coefficient of the equipment at time b; For the device state coefficient of the i-th dimension of the multidimensional non-mechanical data, including current data, voltage data, power supply temperature, and ambient humidity, at time b; is the equipment state coefficient of the internal temperature of the equipment in the multidimensional non-mechanical data at time b; I is the total number of current data, voltage data, power supply temperature and ambient humidity in the multidimensional non-mechanical data at time b.

[0045] Furthermore, the method for obtaining the equipment mechanical state coefficients at each moment based on the multi-dimensional mechanical data at each moment within the reference operating period is as follows:

[0046] (1) Since the state of vibration data is obtained by comparing it with the average vibration data benchmark value during normal operation in the past, for example, a failure of the cooling fan bearing may cause high-frequency vibration, while rotor imbalance may cause low-frequency vibration, causing the vibration data to deviate from the benchmark value. Therefore, for the vibration data in the multidimensional mechanical data at any time during the reference operating period, the vibration state coefficient of the vibration data is obtained based on the difference between the vibration data and the historical vibration data.

[0047] Specifically, if the number of historical reference periods before the reference period where the vibration data is located is greater than or equal to a first preset number, then the vibration data in each historical reference period are combined into a historical vibration data sequence, and the reciprocal of the coefficient of variation of each historical vibration data sequence is obtained as the stable characteristic value of each historical vibration data sequence. Since the vibration data usually maintains a stable fluctuation value under normal equipment operation, the average value of the historical vibration data sequence corresponding to the maximum stable characteristic value is obtained as the reference vibration index of the vibration data. If the number of historical reference periods before the reference period where the vibration data is located is less than the first preset number, then the rated value of the vibration data is obtained as the reference vibration index of the vibration data. In this embodiment, the rated value of the vibration data is set to the vibration data at the 5th moment after the concentrator starts running. This is not limited here and can be set according to the specific implementation scenario.

[0048] Obtain the absolute value of the difference between the vibration data and its reference vibration index to obtain the vibration index difference value. Obtain the reciprocal of the sum of the vibration index difference value and the constant 1 to obtain the vibration state coefficient of the vibration data.

[0049] In one embodiment, the formula for calculating the vibration state coefficient of the vibration data is:

[0050]

[0051] in, The vibration state coefficients of the vibration data; V represents the reference vibration index for the vibration data; V represents the vibration data. It is the absolute value symbol.

[0052] It should be noted that, The smaller the value, the closer the current vibration data is to the normal value, and the better the equipment condition indicated by the current vibration data. The larger it is.

[0053] (2) Since the crimping force data of components such as terminals and antenna connectors need to maintain the consistency and stability of crimping force in each part under normal conditions, the crimping force state coefficient of the multidimensional crimping force data is obtained based on the discrete characteristics of the multidimensional crimping force data and its previous multidimensional crimping force data in the multidimensional mechanical data at any time during the reference running period.

[0054] Specifically, if the number of historical reference periods preceding the reference period in which the multidimensional pressure relay data is located is greater than or equal to a first preset number, then the pressure relay data of each dimension within each historical reference period are combined into a historical pressure relay data sequence. The reciprocal of the coefficient of variation of each historical pressure relay data sequence within each historical reference period is obtained as the stable characteristic value of each historical pressure relay data sequence. The mean of the stable characteristic values ​​of each historical pressure relay data sequence within each historical reference period is obtained. Since the multidimensional pressure relay data usually maintains a stable fluctuation value under normal equipment operation, the mean of the maximum stable characteristic value is used as the reference pressure relay index of the multidimensional pressure relay data. If the number of historical reference periods preceding the reference period in which the reference period at any given time is located is less than a first preset number, then the rated value of the pressure relay data is obtained as the reference pressure relay index of the multidimensional pressure relay data. In this embodiment, the rated value of the multidimensional pressure relay data is set to the mean of the multidimensional pressure relay data at the 5th time after the concentrator starts operating. This is not limited here and can be set according to the specific implementation scenario.

[0055] The multidimensional pressure relay data within the time window is divided into multiple pressure relay data sequences according to dimensions, with one dimension corresponding to one pressure relay data sequence. The reciprocal of the coefficient of variation of each pressure relay data sequence is obtained as the stable feature value of each pressure relay data sequence. The average value of the stable feature value of each pressure relay data sequence is obtained to obtain the stability coefficient. The absolute value of the difference between the reference pressure relay index and the stability coefficient is obtained to obtain the stable difference value. The reciprocal of the sum of the stable difference value and the constant 1 is obtained to obtain the first pressure relay state value.

[0056] Obtain the reciprocal of the mean difference of the multidimensional pressure force data to get the second pressure force state value, and obtain the mean between the first pressure force state value and the second pressure force state value to get the pressure force state coefficient of the multidimensional pressure force data.

[0057] In one embodiment, the formula for calculating the pressure state coefficient of the multidimensional pressure pressure data is as follows:

[0058]

[0059] in, For multidimensional pressure bonding force data, the pressure bonding force state coefficient is used. A reference pressure force index for multi-dimensional pressure force data; The first in the multidimensional pressure relay data Stable characteristic values ​​of a pressure relay data sequence; L is the number of dimensions of the multidimensional pressure relay data; The mean difference of the multidimensional pressure relay data; It is the absolute value symbol.

[0060] It should be noted that, This is the first pressure state value. The smaller the value, the closer the change in crimping force data at each part is to the reference crimping force index, and the more consistent it is with the characteristic that the crimping force data of components such as terminals and antenna connectors should maintain the stability of crimping force changes at each part under normal conditions. The larger it is; This is the second pressure state value. The smaller the value, the more consistent the crimping force data is across all parts, which better reflects the requirement for consistent crimping force across components such as terminals and antenna connectors under normal conditions. The larger it is.

[0061] (3) Obtain the product of the vibration state coefficient of the vibration data and the pressure state coefficient of the multidimensional pressure force data to obtain the mechanical state coefficient of the equipment at any time.

[0062] In one embodiment, taking the b-th time point within the reference operating period as an example, the formula for calculating the mechanical state coefficient of the equipment at the b-th time point is:

[0063]

[0064] Where ME is the mechanical state coefficient of the equipment at time b; The vibration state coefficient is the vibration data at time b. is the pressure state coefficient of the multidimensional pressure pressure data at time b.

[0065] Thus, the non-mechanical state coefficients and mechanical state coefficients of the equipment at each moment within the reference operating period are obtained.

[0066] Step S103: Divide the reference running period into at least two time periods. For any time period, construct a two-dimensional feature value for each moment by combining the equipment mechanical state coefficient and the equipment non-mechanical state coefficient. Cluster the two-dimensional feature values ​​for each moment to obtain a multi-dimensional clustering result index for any time period. Obtain the multi-dimensional clustering result index for each time period. Based on the differences in the multi-dimensional clustering result index for each time period, obtain the attenuation coefficient and attenuation window value.

[0067] After obtaining the equipment non-mechanical state coefficients and equipment mechanical state coefficients at each moment within the reference operating period, the equipment non-mechanical state coefficients and equipment mechanical state coefficients at each moment within the reference operating period are combined to form a two-dimensional feature value for each moment, which is used for cluster analysis.

[0068] Because the data exhibits relatively consistent behavior when the concentrator is in normal operation, the data under normal operation is mostly in a similar state. When using the DBSCAN algorithm for clustering, a large amount of data under normal operation will be concentrated in the same cluster. If a fault occurs, the fault will mainly manifest as mechanical data fault, non-mechanical data fault, or simultaneous mechanical and non-mechanical data fault. Therefore, when using the DBSCAN algorithm for clustering, the cluster positions are generally in relatively fixed areas. For example, in the image obtained from the clustering results, normal data, due to its higher coefficients for both states, is often located in the upper right corner of the image, while data with simultaneous mechanical and non-mechanical data faults, due to its lower coefficients for both states, is often located in the lower left corner of the image. Mechanical data faults or non-mechanical data faults, due to one state coefficient being lower, are generally located in the upper left or lower right corner depending on the setting of the horizontal and vertical coordinates.

[0069] However, since the multidimensional mechanical and non-mechanical data at each moment are related to the actual power consumption, the reference stability index, reference vibration index, and reference pressure force index obtained in step S102 may change. When these indexes change, the corresponding non-mechanical state coefficients or mechanical state coefficients of the equipment may also change, which may cause the cluster positions in the image obtained from the clustering results to shift. Moreover, as the equipment ages, the distribution of some data may change. Therefore, old data can affect the position of clusters in the clustering results, which may affect the classification results of new data, leading to inaccurate monitoring results of equipment operating status.

[0070] Therefore, in this embodiment, the reference runtime is divided into at least two time periods. For any given time period, DBSCAN clustering is performed on the two-dimensional feature values ​​at each moment. DBSCAN clustering is an existing technology and will not be elaborated here. A clustering result graph is obtained after clustering the two-dimensional feature values ​​at each moment within the given time period. The horizontal axis of the clustering result graph represents the equipment mechanical state coefficient at each moment, and the vertical axis represents the equipment non-mechanical state coefficient at each moment. The number of clusters, the coordinates of the center data point of each cluster, and the proportion of discrete data points in the clustering result graph are obtained to form a multidimensional clustering result index for any given time period. Similarly, the multidimensional clustering result index for each time period is obtained. Based on the differences in the multidimensional clustering result index for each time period, an attenuation coefficient and an attenuation window value are obtained. These are used to remove influencing old data when monitoring the equipment operating status, reducing the impact of old data on the equipment operating status monitoring results.

[0071] To illustrate, the reference runtime period is divided into at least two time periods. Assuming the reference runtime period is 3 months (90 days), it is divided into 9 cycles, each cycle being 10 days. The first cycle within the reference runtime period is the first time period, the first two cycles are the second time period, the first three cycles are the third time period, and so on, resulting in 9 time periods. The division of cycles within the reference runtime period needs to be set based on the magnitude of changes in equipment operating data. If the electricity consumption behavior of the users of the concentrator varies greatly, such as in residential areas, a shorter time period can be set, such as 30 minutes, 1 hour, or 2 hours. If the electricity consumption behavior of the users of the concentrator varies little, such as when it is used for the continuous operation of an industrial machine for a long time, and the electricity consumption is relatively consistent at different times, a longer time period can be set, such as 24 hours, 36 hours, or 72 hours. There are no restrictions here; the setting can be based on the specific implementation scenario.

[0072] Furthermore, based on the differences in the multidimensional clustering results indicators for each time period, the decay coefficient and decay window value are obtained using the following method:

[0073] (1) Obtain the variation coefficient of each time period within the reference runtime period.

[0074] After clustering the two-dimensional feature values ​​at each moment within each time period, the position of each cluster in the clustering result graph changes as the total number of moments within each time period changes, and the number of relatively discrete data points also changes. Therefore, for any time period within the reference runtime (excluding the first time period), the formula for calculating the change coefficient of the cluster center data point in the clustering result graph of that any time period, by comparing the cluster center data point of each cluster center data point with the cluster center data point of that any time period, is as follows:

[0075]

[0076] in, is the coefficient of change for any given time period; C is the number of clusters in the multidimensional clustering result index for any given time period; This refers to the number of clusters in the multidimensional clustering results index of the previous time period for any given time period; P represents the distance between the center data point of the k-th cluster and the center data point of the compared cluster in the clustering result graph for any time period; P represents the proportion of discrete data points in the multidimensional clustering result index for any time period. The percentage of discrete data points in the multidimensional clustering results index of the previous time period for any given time period; Let z be the maximum value function; z is a preset constant, which is set to 0.000000001 in this embodiment to prevent... and A value of 0 will affect the calculation results; however, there is no restriction here, and it can be set according to the specific implementation scenario. It is the absolute value symbol.

[0077] It should be noted that, The smaller the value, the smaller the change in the number of clusters between any given time period and the previous time period. The smaller it is; This represents the mean offset of the coordinates of the cluster center data points in the multidimensional clustering results index for any given time period compared to the previous time period. The smaller the value, the smaller the offset of the coordinates of the cluster center data points in the clustering results of any given time period compared to the previous time period. The smaller it is; This represents the difference in the proportion of discrete data points in the clustering results of any given time period compared to the previous time period. The smaller the value, the smaller the change in the number of discrete data points in the clustering results between any given time period and the previous time period. The smaller it is.

[0078] Similarly, obtain the variation coefficient for each time period other than the first time period.

[0079] The method for obtaining the comparison cluster center data points of each cluster center data point in the clustering result graph for any given time period is as follows:

[0080] For any cluster in the clustering result graph of any time period, the cluster corresponding to the previous time period of any time period is recorded as a historical cluster, and the data points in each historical cluster are recorded as the historical data points corresponding to each historical cluster.

[0081] If there is at least one historical data point in any of the clusters, then count the number of historical data points corresponding to each historical cluster in any of the clusters, and take the center data point of the historical cluster containing the most historical data points as the center data point of the comparison cluster of any cluster.

[0082] If no historical data point exists in any of the clusters, then the comparison cluster center point of the center data point of any cluster is set as the origin of the clustering result graph for any time period.

[0083] (2) Obtain the first half of all time periods in the reference runtime period and record it as the target time period (for example, if there are 10 time periods in the reference runtime period, obtain the first 5 time periods and record them as the target time period; if there are 11 time periods in the reference runtime period, obtain the first 6 time periods and record them as the target time period), and obtain the turning coefficient of each target time period.

[0084] For any target time period, obtain the absolute value of the difference between the change coefficient of the target time period and the change coefficient of the previous target time period to obtain the change coefficient difference value. Obtain the difference between the constant 1 and the reciprocal of the change coefficient difference value to obtain the first turning point index.

[0085] The second turning point index is obtained by taking the reciprocal of the difference between the length of any target time period and the length of the first time period within the reference running period.

[0086] The product of the first turning point index and the second turning point index is obtained to obtain the turning point coefficient of any target time period.

[0087] In one implementation, taking the t-th target time period as an example, the formula for calculating the transition coefficient of the t-th target time period is:

[0088]

[0089] in, Let be the turning point coefficient for the t-th target time period; Let be the coefficient of change for the t-th target time period; is the change coefficient of the previous target time period of the t-th target time period; T is the difference in time length between the t-th target time period and the first time period within the reference running period; It is the absolute value symbol.

[0090] It should be noted that, As the first turning point indicator, This represents the difference in the coefficient of change between the t-th target time period and its previous target time period. The larger the value of T, the greater the impact of all newly added data from the t-th target time period compared to the previous target time period on the cluster distribution in the clustering results. Conversely, the smaller the value of T, the greater the impact of adding time periods of length T on the cluster distribution in the clustering results compared to the first time period. The larger it is.

[0091] Similarly, obtain the turning point coefficient for each target time period other than the first target time period.

[0092] (3) Obtain the attenuation coefficient.

[0093] All the change coefficients are combined into a change coefficient sequence. The larger the turning point coefficient, the greater the degree of change in the clustering results of the corresponding target time period. In other words, the more influential old data needs to be removed. Therefore, the largest turning point coefficient in the change coefficient sequence is identified as the obvious turning point of the change coefficient sequence. That is, within the target time period corresponding to the largest turning point coefficient, old data needs to be removed.

[0094] Therefore, the difference between the length of the target time period corresponding to the maximum turning point coefficient and the length of the first time period within the reference running period is obtained to get the time length difference. The ratio of the time length difference to the length of the reference running period is then obtained to get the attenuation coefficient.

[0095] (4) Obtain the attenuation window value.

[0096] The target time period corresponding to the maximum turning coefficient is denoted as the turning time period. For any data point in the clustering result graph of the turning time period, the distance between the data point and its cluster center is obtained in the clustering result graph of the turning time period and denoted as the first distance of the data point.

[0097] If the clustering result diagram of the previous target time period of the transition period contains any of the data points, then the distance between any of the data points and their cluster centers is obtained in the clustering result diagram of the previous target time period of the transition period, and recorded as the second distance of any of the data points. If the clustering result diagram of the previous target time period of the transition period does not contain any of the data points, then the second distance of any of the data points is set to 0.

[0098] Obtain the first distance and the second distance of each data point in the clustering result graph of the transition time period, obtain the absolute value of the difference between the first distance and the second distance of each data point, obtain the distance difference of each data point, form a distance difference sequence, perform DBSCAN clustering on the distance difference sequence to obtain at least one cluster, wherein the cluster containing the most distance differences is recorded as the largest cluster. The data points in the largest cluster are similar in distance to the cluster centers of the two clustering results, that is, the change of the cluster center has little impact on such data points, but has a greater impact on other data points not in the largest cluster. Therefore, data points not in the largest cluster should be deleted. So, the data points corresponding to each distance difference in the cluster containing the most distance differences are recorded as target data points.

[0099] For any cluster among all clusters corresponding to the transition time period and its previous time period, obtain the proportion of target data points in any cluster, obtain the average proportion of target data points in each cluster, and obtain the attenuation window value.

[0100] Thus, the attenuation coefficient and attenuation window value are obtained.

[0101] Step S104: Based on the attenuation coefficient and the attenuation window value, obtain the clustering reference time period for each moment in the preset future operating cycle after the reference operating period. For any moment in the preset future operating cycle, obtain the two-dimensional feature values ​​of the any moment and each moment in the clustering reference time period and perform clustering to obtain the operating status monitoring result of the concentrator at any moment.

[0102] After obtaining the attenuation coefficient and attenuation window value, the product of the reference runtime length and the attenuation coefficient is obtained to get the new runtime length. The product of the number of moments contained in the first time period within the reference runtime and the attenuation window value is obtained to get the old runtime length. That is, the previous target time period of the transition time period is recorded as the initial time period. Starting from the first moment of the transition time period, every time a new time period passes, the time period of the old time period needs to be removed from the initial time period to obtain a new initial time period.

[0103] For example, suppose the initial time period is 1 hour, the new time period is 10 minutes, and the old time period is 5 minutes. Starting from the first moment after the initial time period, every 10 minutes, the first 5 minutes need to be removed from the initial time period to obtain a new initial time period.

[0104] Furthermore, according to the above-mentioned method for obtaining the new initial time period, for any moment within the preset future operating cycle, the target time length from the next moment of the initial time period to any moment is obtained. The ratio of the target time length to the new time length is rounded down to obtain the target multiple. The product of the target multiple and the old time length is obtained to obtain the time length to be deleted. The time period to be deleted is constructed with the first moment of the initial time period as the starting moment. The time length of the time period to be deleted is the time length to be deleted.

[0105] Starting from the first moment of the initial time period, the historical time period before any moment is obtained, and the time period to be deleted is removed from the historical time period to obtain the clustering reference time period for any moment.

[0106] Following the method for obtaining the two-dimensional feature values ​​at each moment within the aforementioned reference runtime period, the two-dimensional feature values ​​of any moment and each moment within the clustering reference time period are obtained and DBSCAN clustering is performed to obtain the monitoring results of the concentrator's operating status at any moment. DBSCAN clustering is an existing technology and will not be elaborated here. If the monitoring results of the operating status are normal, it indicates that the current concentrator equipment is operating normally. If the monitoring results of the operating status are abnormal, it indicates that the current concentrator equipment may be malfunctioning, and staff should be notified to conduct inspection and maintenance.

[0107] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A concentrator intelligent fusion terminal with equipment operation status monitoring function, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it performs the following steps: Acquire multidimensional mechanical data and multidimensional non-mechanical data at every moment during the operation of the concentrator; Obtain the reference runtime of the concentrator, and based on the multidimensional non-mechanical data at each moment within the reference runtime, obtain the equipment non-mechanical state coefficient at each moment. Based on the multidimensional mechanical data at each moment within the reference runtime, obtain the equipment mechanical state coefficient at each moment. The reference runtime is divided into at least two time periods. For any given time period, the mechanical state coefficient and non-mechanical state coefficient of the equipment at each moment within that time period are combined to form a two-dimensional feature value for that moment. The two-dimensional feature values ​​at each moment are clustered to obtain a multi-dimensional clustering result index for that given time period. The multi-dimensional clustering result index for each time period is obtained. Based on the differences in the multi-dimensional clustering result index for each time period, the attenuation coefficient and attenuation window value are obtained. Based on the attenuation coefficient and the attenuation window value, obtain the clustering reference time period for each moment in the preset future operating cycle after the reference operating period. For any moment in the preset future operating cycle, obtain the two-dimensional feature values ​​of the any moment and each moment in the clustering reference time period and perform clustering to obtain the monitoring result of the concentrator's operating status at any moment. The step of obtaining the decay coefficient and decay window value based on the differences in the multidimensional clustering results indicators for each time period includes: For any time period within the reference runtime (excluding the first time period), obtain the comparison cluster center data points of each cluster center data point in the clustering result graph for that any time period, and calculate the change coefficient for that any time period: ， in, is the coefficient of change for any given time period; C is the number of clusters in the multidimensional clustering result index for any given time period; This refers to the number of clusters in the multidimensional clustering results index of the previous time period for any given time period; P represents the distance between the center data point of the k-th cluster and the center data point of the compared cluster in the clustering result graph for any time period; P represents the proportion of discrete data points in the multidimensional clustering result index for any time period. The percentage of discrete data points in the multidimensional clustering results index of the previous time period for any given time period; The function is the maximum value; z is a preset constant. It is the absolute value symbol; Within all time periods, at least two target time periods are obtained. For any target time period, the absolute value of the difference between the change coefficient of the target time period and the change coefficient of the previous target time period is obtained to obtain the change coefficient difference. The difference between the constant 1 and the reciprocal of the change coefficient difference is obtained to obtain the first turning point index. The second turning point index is obtained by taking the reciprocal of the difference between the length of any target time period and the length of the first time period within the reference running period. The product of the first turning point index and the second turning point index is obtained to obtain the turning point coefficient of any target time period. Obtain the change coefficient for each time period other than the first time period, obtain the turning point coefficient for each target time period other than the first target time period, and obtain the attenuation coefficient and attenuation window value based on the change coefficient for each time period other than the first time period and the turning point coefficient for each target time period other than the first target time period.

2. The intelligent fusion terminal for concentrators with equipment operation status monitoring function according to claim 1, characterized in that, The step of obtaining the equipment non-mechanical state coefficient at each moment based on the multidimensional non-mechanical data at each moment within the reference runtime period includes: Multidimensional non-mechanical data includes current data, voltage data, power supply temperature, ambient humidity, and internal equipment temperature; For any dimension of non-mechanical data in the multidimensional non-mechanical data at any moment within the reference running period, a time window of a preset size is established with the time at any moment as the cutoff time. When the non-mechanical data of any dimension is current data, voltage data, power supply temperature, or ambient humidity, the device state coefficient of the non-mechanical data of any dimension is obtained based on the discrete characteristics of the non-mechanical data of any dimension and the previous non-mechanical data of the same dimension. When the non-mechanical data of any dimension is the internal temperature of the equipment, the internal temperature of the equipment at each moment within the time window is formed into an internal temperature data sequence. The equipment load data at each moment within the time window is obtained and formed into an equipment load data sequence. The Spearman correlation coefficient between the internal temperature data sequence and the equipment load data sequence is obtained. Based on the sum of the Spearman correlation coefficient and the constant 1, the equipment state coefficient of the non-mechanical data of any dimension is obtained. Obtain the equipment state coefficient of each dimension of the non-mechanical data in the multidimensional non-mechanical data at any given time, and obtain the corresponding cumulative multiplication result as the equipment non-mechanical state coefficient at any given time.

3. A concentrator intelligent fusion terminal with equipment operation status monitoring function according to claim 2, characterized in that, The step of obtaining the equipment state coefficient of the non-mechanical data in any dimension based on the discrete characteristics of the non-mechanical data in any dimension and the previous non-mechanical data in the same dimension includes: According to a preset time period, the reference runtime is divided into at least two reference periods. If the number of historical reference periods before the reference period where the non-mechanical data of any dimension is located is greater than or equal to a first preset number, then the non-mechanical data belonging to the same dimension as the non-mechanical data of any dimension within each historical reference period is formed into a historical data sequence. The reciprocal of the coefficient of variation of each historical data sequence is obtained as the stable feature value of each historical data sequence. The stable feature values ​​of each historical data sequence are sorted in descending order, and the average value of the stable feature values ​​of a preset proportion is obtained as the reference stability index of the non-mechanical data of any dimension. If the number of historical reference periods before the reference period where the non-mechanical data of any dimension is located is less than a first preset number, then the rated value of the non-mechanical data of any dimension is obtained as the reference stability index of the non-mechanical data of any dimension. Within the time window, the non-mechanical data of any dimension and the sequence of non-mechanical data of the same dimension are combined to form a first data sequence. Within the reference period where the non-mechanical data of any dimension is located, if the number of non-mechanical data of the same dimension preceding the non-mechanical data of any dimension is greater than or equal to a second preset number, then the non-mechanical data of any dimension and the second preset number of non-mechanical data of the same dimension preceding it are combined to form a second data sequence. If the number of non-mechanical data of the same dimension preceding the non-mechanical data of any dimension is less than a second preset number but greater than or equal to a third preset number, then the non-mechanical data of any dimension and all the non-mechanical data preceding it are combined to form a second data sequence. Based on the reference stability index of the non-mechanical data in any dimension, the coefficient of variation of the first data sequence, and the coefficient of variation of the second data sequence, calculate the equipment state coefficient of the non-mechanical data in any dimension: ， in, Equipment status coefficients for any dimension of non-mechanical data; A reference stability index for non-mechanical data in any dimension; The coefficient of variation for the first data sequence; The coefficient of variation for the second data sequence; It is the absolute value symbol; The third preset quantity; m is the number of non-mechanical data of the same dimension preceding the non-mechanical data of any dimension within the reference period where the non-mechanical data of any dimension is located.

4. A concentrator intelligent fusion terminal with equipment operation status monitoring function according to claim 3, characterized in that, The step of obtaining the equipment mechanical state coefficient at each moment based on the multi-dimensional mechanical data at each moment within the reference operating period includes: Multidimensional mechanical data includes vibration data and multidimensional pressure force data; For vibration data in multidimensional mechanical data at any moment within the reference running period, if the number of historical reference periods before the reference period where the vibration data is located is greater than or equal to a first preset number, then the vibration data in each historical reference period are combined into a historical vibration data sequence, and the reciprocal of the coefficient of variation of each historical vibration data sequence is obtained as the stable characteristic value of each historical vibration data sequence. The average value of the historical vibration data sequence corresponding to the maximum stable characteristic value is obtained as the reference vibration index of the vibration data. If the number of historical reference periods before the reference period where the vibration data is located is less than the first preset number, then the rated value of the vibration data is obtained as the reference vibration index of the vibration data. Obtain the absolute value of the difference between the vibration data and its reference vibration index to obtain the vibration index difference value. Obtain the reciprocal of the sum of the vibration index difference value and the constant 1 to obtain the vibration state coefficient of the vibration data. For the multidimensional pressure force data in the multidimensional mechanical data at any moment within the reference running period, the pressure force state coefficient of the multidimensional pressure force data is obtained based on the discrete characteristics of the multidimensional pressure force data and its previous multidimensional pressure force data. The mechanical state coefficient of the equipment at any given time is obtained by multiplying the vibration state coefficient of the vibration data with the pressure state coefficient of the multidimensional pressure force data.

5. A concentrator intelligent fusion terminal with equipment operation status monitoring function according to claim 4, characterized in that, The step of obtaining the pressure state coefficient of the multidimensional pressure data based on the discrete characteristics of the multidimensional pressure data and its previous multidimensional pressure data includes: If the number of historical reference periods preceding the reference period in which the multidimensional pressure relay data is located is greater than or equal to a first preset number, then the pressure relay data of each dimension in each historical reference period are combined into a historical pressure relay data sequence. The reciprocal of the coefficient of variation of each historical pressure relay data sequence in each historical reference period is obtained as the stable feature value of each historical pressure relay data sequence. The mean of the stable feature values ​​of each historical pressure relay data sequence in each historical reference period is obtained, and the mean of the maximum stable feature value is used as the reference pressure relay index of the multidimensional pressure relay data. If the number of historical reference periods preceding the reference period in which the reference period at any moment is located is less than the first preset number, then the rated value of the pressure relay data is obtained as the reference pressure relay index of the multidimensional pressure relay data. The multidimensional pressure relay data within the time window is divided into multiple pressure relay data sequences according to dimensions, with one dimension corresponding to one pressure relay data sequence. The reciprocal of the coefficient of variation of each pressure relay data sequence is obtained as the stable feature value of each pressure relay data sequence. The average value of the stable feature value of each pressure relay data sequence is obtained to obtain the stability coefficient. The absolute value of the difference between the reference pressure relay index and the stability coefficient is obtained to obtain the stable difference value. The reciprocal of the sum of the stable difference value and the constant 1 is obtained to obtain the first pressure relay state value. Obtain the reciprocal of the mean difference of the multidimensional pressure force data to get the second pressure force state value, and obtain the mean between the first pressure force state value and the second pressure force state value to get the pressure force state coefficient of the multidimensional pressure force data.

6. A concentrator intelligent fusion terminal with equipment operation status monitoring function according to claim 1, characterized in that, The process involves combining the equipment mechanical state coefficient and the equipment non-mechanical state coefficient at each moment within any given time period to form a two-dimensional feature value for that moment. Then, clustering these two-dimensional feature values ​​yields a multi-dimensional clustering result index for any given time period, including: A clustering result image is obtained after clustering the two-dimensional feature values ​​at each moment within any time period. The horizontal axis of the clustering result image is the mechanical state coefficient of the equipment at each moment, and the vertical axis is the non-mechanical state coefficient of the equipment at each moment. The number of clusters, the coordinates of the center data point of each cluster, and the proportion of discrete data points in the clustering result image are obtained to form a multidimensional clustering result index for any time period.

7. A concentrator intelligent fusion terminal with equipment operation status monitoring function according to claim 1, characterized in that, The step of obtaining the comparison cluster center data points of each cluster center data point in the clustering result graph for any given time period includes: For any cluster in the clustering result graph of any time period, the cluster corresponding to the previous time period of any time period is recorded as a historical cluster, and the data points in each historical cluster are recorded as the historical data points corresponding to each historical cluster. If there is at least one historical data point in any of the clusters, then count the number of historical data points corresponding to each historical cluster in any of the clusters, and take the center data point of the historical cluster containing the most historical data points as the center data point of the comparison cluster of any cluster. If no historical data point exists in any of the clusters, then the comparison cluster center point of the center data point of any cluster is set as the origin of the clustering result graph for any time period.

8. A concentrator intelligent fusion terminal with equipment operation status monitoring function according to claim 1, characterized in that, The step of obtaining the attenuation coefficient and attenuation window value based on the change coefficient of each time period other than the first time period, and the transition coefficient of each target time period other than the first target time period, includes: Obtain the transition coefficient for each target time period except the first target time period; obtain the difference between the length of the target time period corresponding to the maximum transition coefficient and the length of the first time period within the reference running period; obtain the length difference; obtain the ratio of the length difference to the length of the reference running period; obtain the attenuation coefficient. The target time period corresponding to the maximum turning coefficient is denoted as the turning time period. For any data point in the clustering result graph of the turning time period, the distance between the data point and its cluster center is obtained in the clustering result graph of the turning time period and denoted as the first distance of the data point. If the clustering result diagram of the previous target time period of the transition period contains any of the data points, then the distance between any of the data points and their cluster centers is obtained in the clustering result diagram of the previous target time period of the transition period, and recorded as the second distance of any of the data points. If the clustering result diagram of the previous target time period of the transition period does not contain any of the data points, then the second distance of any of the data points is set to 0. Obtain the first distance and the second distance of each data point in the clustering result graph of the turning point time period, obtain the absolute value of the difference between the first distance and the second distance of each data point, obtain the distance difference of each data point, form a distance difference sequence, perform clustering on the distance difference sequence to obtain at least one cluster, and record the data point corresponding to each distance difference in the cluster containing the most distance differences as the target data point. For any cluster among all clusters corresponding to the transition time period and its previous time period, obtain the proportion of target data points in any cluster, obtain the average proportion of target data points in each cluster, and obtain the attenuation window value.

9. A concentrator intelligent fusion terminal with equipment operation status monitoring function according to claim 7, characterized in that, The step of obtaining the clustering reference time period for each moment in a preset future running cycle after the reference running period based on the attenuation coefficient and the attenuation window value includes: The new time length is obtained by multiplying the length of the reference runtime segment by the attenuation coefficient, and the old time length is obtained by multiplying the number of moments contained in the first time segment within the reference runtime segment by the attenuation window value. The previous target time period of the transition period is recorded as the initial time period. For any moment within the preset future operating cycle, the target time length from the next moment of the initial time period to any moment is obtained. The ratio of the target time length to the new time length is rounded down to obtain the target multiple. The product of the target multiple and the old time length is obtained to obtain the time length to be deleted. The time period to be deleted is constructed with the first moment of the initial time period as the starting moment. The time length of the time period to be deleted is the time length to be deleted. Starting from the first moment of the initial time period, the historical time period before any moment is obtained, and the time period to be deleted is removed from the historical time period to obtain the clustering reference time period for any moment.

Images