Equipment monitoring state evaluation method and system based on multi-agent cooperation

Abstract

The invention relates to the technical field of multi-agent collaborative evaluation, in particular to an equipment monitoring state evaluation method and system based on multi-agent collaboration, and the method comprises the following steps: carrying out the partition processing of a vibration signal, a temperature change signal and a displacement signal based on the monitoring data of equipment operation, carrying out the time sequence analysis of each signal partition, and carrying out the time sequence analysis of each signal partition; and calculating amplitude change amplitude and change rate values in the partitions, integrating and analyzing signal features, and generating a multi-signal distribution feature matrix. According to the method, the identification accuracy of the abnormal signal is optimized through the clarification of the signal linkage intensity, the abnormal signal is accurately identified and analyzed by utilizing the comparison of trend fitting and the equipment operation reference, the detailed degree of state monitoring is improved, the signal analysis result and the task path are tightly coupled in the task execution process, and the task execution efficiency is improved. Through dynamic priority ranking and resource allocation, the resource utilization efficiency and the task execution flexibility are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of multi-agent collaborative evaluation, and particularly relates to an equipment monitoring state evaluation method and system based on multi-agent collaborative cooperation. BACKGROUND

[0002] The technical field of multi-agent collaborative evaluation includes the research, design and application of multi-agent systems. The core content of this field is to achieve the analysis and optimization of complex systems through the coordinated interaction of multiple agents. Multi-agent systems usually include multiple agents with independent perception, decision-making and execution capabilities. These agents collaborate to complete tasks that a single agent cannot independently complete. This technical field is widely used in industrial control, equipment monitoring, task allocation and state evaluation scenarios. It focuses on the exchange of information between agents, task allocation strategies, collaborative decision-making methods and overall system performance optimization. In addition, this field also emphasizes the development of agent collaboration mechanisms to ensure real-time adjustment and effective cooperation in dynamic environments, thereby improving the adaptability and efficiency of the system.

[0003] Among them, the equipment monitoring state evaluation method based on multi-agent collaborative cooperation refers to the use of the collaborative capabilities of multiple agents to achieve real-time monitoring and evaluation of the running state of equipment. This patent subject addresses the state perception, anomaly detection and evaluation problems that may be involved in the running process of equipment. It covers the collaborative data processing method of multiple agents through perception devices to collect equipment running parameters, the evaluation technology of information relevance and state weight between agents, and the collaboration mechanism of agents through data transmission and distributed analysis. The specific method includes collecting multiple running data generated during the operation of the equipment, using agent distributed computing technology to extract specific running parameters, and using the state relevance calculation method of multi-agent collaboration to comprehensively analyze the running state of the equipment, thereby completing the monitoring and evaluation of the running state of the equipment.

[0004] The prior art lacks effective mining of deep correlations among multiple signals in equipment operation state monitoring, usually focusing on feature analysis of independent signals, failing to effectively cover the correlation characteristics between different signals, resulting in the interaction between multiple source signals being ignored. This deficiency is particularly prominent in complex operating states where multiple signals exist simultaneously and interact with each other. In addition, the prior art lacks specific abnormal feature description and correlation analysis means in state anomaly identification, and cannot extract the difference characteristics of the equipment state through multi-dimensional data, making it difficult to locate the specific abnormal cause. During task execution, the priority sorting and resource allocation of the task path are usually based on fixed rule settings, lacking flexibility in dynamic adjustment, and cannot be optimized in real time in a multi-task parallel environment, easily causing resource allocation to be rigid and efficiency to be reduced. At the same time, feedback data is not fully utilized, and the historical data of task execution lacks deep analysis and close combination with resource allocation, resulting in insufficient optimization ability after task completion, limiting the improvement of system adaptability in dynamic environment. SUMMARY

[0005] The purpose of the present application is to solve the shortcomings in the prior art, and a method and system for evaluating the monitoring state of equipment based on multi-agent collaborative cooperation are proposed.

[0006] In order to achieve the above purpose, the present application adopts the following technical scheme: a method for evaluating the monitoring state of equipment based on multi-agent collaborative cooperation, comprising the following steps: S1: Based on the monitoring data of equipment operation, the vibration signal, temperature change signal and displacement signal are processed by partitioning, the time series analysis of signal partitioning is performed, the amplitude change amplitude and change rate are calculated, the signal features are integrated and analyzed, and a multi-signal distribution feature matrix is generated; S2: Based on the multi-signal distribution feature matrix, the time distribution of vibration and displacement signal peak points is analyzed, the temperature signal change rate is compared with the vibration displacement, the correlation degree between signals is calculated and a threshold is set, the signal time trend with high linkage strength is summarized, and an abnormal linkage signal matrix is generated; S3: Based on the abnormal linkage signal matrix, trend fitting analysis is performed on the time series of linkage strength signal points, signals that do not match the equipment operation mode benchmark are extracted, a signal distribution density matrix is established, the distribution results are integrated into a multi-dimensional difference data structure, and an abnormal signal difference matrix is generated; S4: Based on the abnormal signal difference matrix, the task chain nodes are prioritized according to the signal difference, the node execution order and resource distribution of the task path are dynamically adjusted, the execution resources are optimized according to the signal difference density matrix, and a task chain path optimization model is generated; S5: According to the task chain path optimization model, the task execution feedback data is accumulated and analyzed, the resource consumption, the task completion time and the signal change trend in the execution process are calculated, the node state benefit result is filed and integrated, and the task execution resource dynamic distribution result is generated.

[0007] As a further scheme of the application, the multi-signal distribution feature matrix includes a signal time sequence matrix, a signal amplitude change matrix, and a signal change rate matrix, the abnormal linkage signal matrix includes a signal linkage intensity matrix, a signal time correlation matrix, and a signal change trend matrix, the abnormal signal difference matrix is specifically a signal deviation degree matrix, a signal distribution density matrix, and a signal difference feature matrix, the task chain path optimization model includes a node priority distribution matrix, a resource dynamic distribution matrix, and a task path sorting matrix, and the task execution resource dynamic distribution result includes a resource use matrix, a node benefit distribution matrix, and an execution time distribution matrix.

[0008] As a further scheme of the application, based on the monitoring data of equipment operation, the vibration signal, the temperature change signal and the displacement signal are processed in partitions, time sequence analysis is performed on each signal partition, the amplitude change amplitude and the change rate value in the partition are calculated, the signal features are integrated and analyzed, and the specific steps of generating the multi-signal distribution feature matrix are as follows: S101: Based on the monitoring data of equipment operation, the vibration signal, the temperature change signal and the displacement signal are partitioned according to the signal type, the time sequence information of the signal partition data is extracted one by one, the instantaneous state and the time change trend of the amplitude data are summarized, the correlation degree characteristic value of the time sequence data with the signal state change is selected, and the signal partition time sequence feature set is obtained; S102: Based on the signal partition time sequence feature set, the continuity of the signal partition data is analyzed, the ratio of the amplitude change range to the time interval of each group of time sequence feature data is compared, the amplitude change range is summarized according to the continuity feature, the change rate reference data table is established for the time interval data, the relationship characteristic value of the signal amplitude and the time change rate is calculated, and the signal amplitude and rate feature set is obtained; S103: Based on the signal amplitude and rate feature set, the signal interrelation of the partition feature data of the vibration signal, the temperature change signal and the displacement signal is classified, the distribution features of the partition signals are integrated, the amplitude and rate feature data matrix of different signals are summarized and adjusted, and the multi-signal distribution feature matrix is generated.

[0009] As a further scheme of the application, based on the multi-signal distribution feature matrix, the peak points of the vibration and displacement signals are analyzed in time distribution, the change rate of the temperature signal is compared with the vibration displacement, the correlation degree between different signals is calculated, a threshold is set, the signal time trend with high linkage intensity is summarized, and the specific steps of generating the abnormal linkage signal matrix are as follows: S201: Based on the multi-signal distribution feature matrix, extract the peak point information from the vibration signal and displacement signal, organize the signal peak point data in chronological order, analyze the distribution characteristics of the peak point time interval, classify and organize the peak point data according to the time interval, and obtain the peak point time distribution feature set. S202: Based on the peak point time distribution feature set, extract the temperature signal change rate data, compare the time distribution characteristics of the peak points of the vibration signal and the displacement signal, calculate the correlation strength value between the temperature signal rate data and the peak points, compare the data results to generate a signal correlation feature parameter table, and obtain the signal correlation feature matrix. S203: Based on the signal correlation feature matrix, set the signal linkage threshold range, filter the time points where the signal linkage intensity meets the threshold, organize the linkage time trend data of vibration signal, displacement signal and temperature signal, establish the signal linkage relationship based on the filtering results, and generate an abnormal linkage signal matrix.

[0010] As a further aspect of the present invention, the formula for calculating the correlation strength value is specifically as follows: ; In the formula, This represents the correlation strength between the rate of change of the temperature signal and the peak point. Representing the The rate of change of the temperature signal at any given time, This represents the corresponding peak point time distribution characteristic value. and These represent the rate of change of the temperature signal and the mean of the time distribution of the peak point, respectively. This represents the total number of data points.

[0011] As a further aspect of the present invention, the following steps are taken to generate an abnormal signal difference matrix: using the aforementioned abnormal linkage signal matrix, trend fitting analysis is performed on the time series of linkage intensity signal points to extract signals that do not match the equipment operation mode baseline; a signal distribution density matrix is ​​established based on the time series; and the distribution results are integrated into a multi-dimensional difference data structure. S301: Based on the abnormal linkage signal matrix, extract the time series data of linkage intensity signal points, perform trend fitting processing on the time series data, compare the fitting results with the equipment operation mode benchmark value point by point, filter out the parts of the signal point time distribution that do not match the benchmark, and generate an abnormal signal time series feature set. S302: Based on the abnormal signal time series feature set, extract the time axis distribution data of the signal points, divide the time series according to the time period, calculate the signal distribution density value in the interval, summarize the signal density distribution characteristics in each time period, and generate a signal distribution density matrix. S303: Based on the signal distribution density matrix, perform multi-dimensional classification on the difference data of signal distribution density in each time interval, classify and integrate the signals according to the time interval difference data, organize the multi-dimensional feature data and establish a complete correlation structure to generate an abnormal signal difference matrix.

[0012] As a further aspect of the present invention, the formula for calculating the signal distribution density is as follows: ; in, Represents signal distribution density, Representing the The number of signals within a time period, This represents the average number of signals across all time periods. Represents the total number of time intervals. This represents the duration of each time period.

[0013] As a further aspect of the present invention, based on the abnormal signal difference matrix, the nodes in the task chain are prioritized according to signal differences, the execution order and resource distribution of the task path are dynamically adjusted, and the execution resources are optimized according to the signal difference density matrix. The specific steps for generating a task chain path optimization model are as follows: S401: Based on the abnormal signal difference matrix, extract the signal difference data of the nodes in the task chain, sort each node according to the difference value, determine the logical relationship between the difference value and the sorting result for each node, generate a priority data table according to the sorting weight, and obtain the node priority sorting table. S402: Based on the node priority sorting table, adjust the execution order of nodes in the task chain in sequence, calculate the node resource requirement value according to the adjusted order, optimize resource allocation by comparing with the global resource distribution rules of the task chain, and obtain the task path adjustment structure; S403: Based on the task path adjustment structure, the resource optimization results are adjusted a second time by combining the signal difference density matrix. The rationality of the execution order of the path nodes is verified according to the optimization results. The resource and path adjustment data are integrated to reconstruct the task chain path structure and generate a task chain path optimization model.

[0014] As a further aspect of the present invention, based on the task chain path optimization model, the feedback data after task execution is cumulatively analyzed, and the revenue results of all node states are archived and integrated by calculating resource consumption, task completion time, and signal change trends during execution, to generate dynamic distribution results of task execution resources. The specific steps are as follows: S501: Based on the task chain path optimization model, summarize the feedback data after task execution, extract data on resource consumption, task completion time and signal change trend, organize and archive different types of data node by node, integrate the archived content according to the task chain node distribution rules, and obtain the task execution feedback dataset. S502: Based on the task execution feedback dataset, calculate and compare the resource consumption and task completion time in the order of nodes, classify the node status benefit parameters item by item in combination with the signal change trend data, organize the node status classification results according to the benefit parameter data, and obtain the node status benefit result set. S503: Based on the node status benefit result set, integrate the node status benefit results with the task chain path data, optimize the resource distribution of the benefit results in combination with resource allocation requirements, readjust the node resource configuration according to the optimized data, and generate dynamic distribution results of task execution resources.

[0015] An equipment monitoring and status assessment system based on multi-agent collaborative cooperation includes: The signal partitioning module, based on the monitoring data of equipment operation, partitions vibration signals, temperature change signals and displacement signals, performs time series analysis on each signal partition, calculates the amplitude and rate of change within the partition, and generates a multi-signal distribution feature matrix. Based on the multi-signal distribution feature matrix, the linkage analysis module performs time distribution analysis on the peak points of vibration and displacement signals, compares the rate of change of temperature signal with vibration displacement, calculates the correlation between different signals, sets thresholds, and generates an abnormal linkage signal matrix. The anomaly detection module uses the anomaly linkage signal matrix to perform trend fitting analysis on the time series of linkage intensity signal points, extracts signals that do not match the equipment operation mode benchmark, integrates the distribution results into a multi-dimensional difference data structure, and generates an anomaly signal difference matrix. Based on the abnormal signal difference matrix, the task optimization module prioritizes the nodes in the task chain according to their signal differences, dynamically adjusts the node execution order and resource distribution of the task path, and generates a task chain path optimization model. The resource regulation module performs cumulative analysis on the feedback data after task execution based on the task chain path optimization model. By calculating resource consumption, task completion time, and signal change trends during execution, it generates dynamic distribution results of task execution resources.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, the accuracy of abnormal signal identification is optimized by clarifying the signal linkage strength. By comparing trend fitting with the equipment operating benchmark, abnormal signals are accurately identified and analyzed, improving the detail of status monitoring. During task execution, the signal analysis results are closely coupled with the task path. Through dynamic priority sorting and resource allocation, resource utilization efficiency and task execution flexibility are significantly improved. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0018] Figure 1 This is a schematic diagram of the steps of the present invention; Figure 2 This is a flowchart of steps S1 of the present invention; Figure 3 This is a flowchart of steps S2 of the present invention; Figure 4 This is a flowchart of steps S3 of the present invention; Figure 5 This is a flowchart of step S4 of the present invention; Figure 6 This is a flowchart of steps S5 of the present invention; Figure 7 This is a system module diagram of the present invention. Detailed Implementation

[0019] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0020] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0021] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0022] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0023] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0024] Please see Figure 1 A method for assessing the status of equipment monitoring based on multi-agent collaborative cooperation includes the following steps: S1: Based on the monitoring data of equipment operation, vibration signals, temperature change signals and displacement signals are processed by partition, time series analysis is performed on the signal partitions, the amplitude change magnitude and change rate are calculated, the signal characteristics are integrated and analyzed, and a multi-signal distribution feature matrix is ​​generated. S2: Based on the multi-signal distribution feature matrix, the time distribution analysis of the peak points of vibration and displacement signals is performed. The temperature signal change rate is compared with the vibration displacement. The correlation between signals is calculated and a threshold is set. The time trend of signals with high linkage intensity is summarized, and an abnormal linkage signal matrix is ​​generated. S3: Based on the abnormal linkage signal matrix, perform trend fitting analysis on the time series of linkage intensity signal points, extract signals that do not match the equipment operation mode benchmark, establish a signal distribution density matrix, integrate the distribution results into a multi-dimensional difference data structure, and generate an abnormal signal difference matrix. S4: Based on the abnormal signal difference matrix, prioritize the task chain nodes according to the signal differences, dynamically adjust the node execution order and resource distribution of the task path, optimize the execution resources according to the signal difference density matrix, and generate a task chain path optimization model. S5: Based on the task chain path optimization model, perform cumulative analysis on task execution feedback data, calculate resource consumption, task completion time and signal change trends during execution, archive and integrate node status benefit results, and generate dynamic distribution results of task execution resources.

[0025] The multi-signal distribution feature matrix includes the signal time series matrix, the signal amplitude change matrix, and the signal change rate matrix. The abnormal linkage signal matrix includes the signal linkage strength matrix, the signal time correlation matrix, and the signal change trend matrix. The abnormal signal difference matrix specifically includes the signal deviation matrix, the signal distribution density matrix, and the signal difference feature matrix. The task chain path optimization model includes the node priority distribution matrix, the resource dynamic distribution matrix, and the task path ranking matrix. The task execution resource dynamic distribution results include the resource usage matrix, the node revenue distribution matrix, and the execution time distribution matrix.

[0026] Please see Figure 2 The specific steps of S1 are as follows: S101: Based on the monitoring data of equipment operation, vibration signals, temperature change signals and displacement signals are divided into partitions according to signal type. The time sequence information of the signal partition data is extracted one by one. The instantaneous state and time-time change trend of amplitude data are summarized. The correlation feature values ​​of the time sequence data with the signal state change are screened out to obtain the signal partition time sequence feature set. Data preprocessing algorithms are used to denoise and normalize the monitoring data, eliminating the impact of outliers on feature analysis. The data is partitioned according to signal type. For the dataset of each signal partition, a time series decomposition method is used to split the signal sequence into trend, seasonal, and random noise components. Fast Fourier Transform (FFT) is used to perform frequency domain analysis on the signal components to extract the main frequency components and amplitude features. Short-Time Fourier Transform (STFT) is used to obtain time-frequency distribution information. The sliding window method is combined to extract time-series features. Based on the set window size, the mean, variance, skewness, kurtosis, and other statistical characteristics of the data within the sliding window are calculated to summarize the instantaneous state of the amplitude data. The autoregressive moving average (ARMA) model is used to analyze the time-varying trend of the amplitude data. Feature values ​​related to signal state changes are screened from the time-series data, and key indicators such as abrupt change points, trend change rate, and periodic features are selected to finally obtain the time-series feature set of the signal partition.

[0027] S102: Based on the signal partition time series feature set, through the continuity analysis of the data of each signal partition, compare the ratio of the amplitude change range to the time interval of each group of time series feature data, summarize the amplitude change range according to the continuity characteristics, establish a change rate reference data table for the time interval data, calculate the relationship feature value between the signal amplitude and the time change rate, and obtain the signal amplitude and rate feature set. Based on the time-series feature set of signal partitions, the continuity of data in each signal partition is analyzed according to the formula. ; Calculate the characteristic value of the rate of change of signal amplitude.

[0028] In the formula, The characteristic value representing the rate of change of signal amplitude. Representing the The signal amplitude at time , Represents the corresponding point in time. This represents the total number of signal data points.

[0029] For continuous analysis of signal data, the first step is to obtain the signal amplitude sequence. and the corresponding time series Calculate the change between adjacent amplitudes. And accumulate them to calculate the time interval between adjacent time points. The values ​​are then accumulated, and finally the characteristic value of the signal amplitude change rate is calculated using a formula. .

[0030] Specific calculation example: Set the vibration signal zone data as follows ; Calculate the magnitude change of adjacent data: Calculate adjacent time intervals: ; Substitute into the formula: ; The results show that the average amplitude change rate of this signal partition is 0.3, which characterizes the temporal trend of the signal and provides a reference for subsequent signal correlation analysis.

[0031] S103: Based on the signal amplitude and velocity feature set, the feature classification of the inter-signal correlation of vibration signal, temperature change signal and displacement signal partition feature data is performed, the distribution features of partition signals are integrated, and the amplitude and velocity feature data matrix between different signals is summarized and adjusted to generate a multi-signal distribution feature matrix. The signals are categorized by type and frequency distribution. Principal component analysis (PCA) is used to reduce dimensionality and extract the main signal features. A feature matrix is ​​constructed based on the distribution characteristics of the signals. The correlation coefficient between different signals is calculated using covariance analysis. By combining the amplitude and rate feature data matrices of each signal, a mapping relationship of signal feature distribution is established. Dynamic time warping (DTW) is used to align and match the time series patterns of different signals. The optimal matching path is found on the signal time axis to measure the correlation between signals. Finally, a multi-signal distribution feature matrix is ​​generated.

[0032] Please see Figure 3 The specific steps of S2 are as follows: S201: Based on the multi-signal distribution feature matrix, extract the peak point information in vibration and displacement signals, organize the peak point data in chronological order, analyze the distribution characteristics of peak point time intervals, classify and organize the peak point data according to the time intervals, and obtain the peak point time distribution feature set. First, time-domain analysis was performed on the vibration and displacement signals. Local maxima were detected using the sliding window technique, and the significance of signal peaks was enhanced using the three-point smoothing method. The time index corresponding to each peak point was calculated, and the peak point data were organized in chronological order. The time intervals between adjacent peak points were statistically analyzed, and a histogram of time interval distribution was constructed. The distribution characteristics of peak point time intervals were analyzed using the normal distribution fitting method. Hierarchical cluster analysis was performed on the time interval data, and the time intervals were classified and organized according to different cluster centers. The time distribution characteristics of peak points in each category were extracted, and finally, the time distribution feature set of peak points was obtained.

[0033] S202: Based on the peak point time distribution feature set, extract the temperature signal change rate data, compare the time distribution characteristics of the peak points of vibration signal and displacement signal, calculate the correlation strength value between temperature signal rate data and peak points, compare the data results to generate a signal correlation feature parameter table, and obtain the signal correlation feature matrix. The specific formula for calculating the correlation strength value is as follows: ; In the formula, This represents the correlation strength between the rate of change of the temperature signal and the peak point. Representing the The rate of change of the temperature signal at any given time, This represents the corresponding peak point time distribution characteristic value. and These represent the rate of change of the temperature signal and the mean of the time distribution of the peak point, respectively. This represents the total number of data points.

[0034] For the correlation analysis between the rate of change of temperature signal and the time distribution of peak points, the first step is to obtain the rate of change sequence of the temperature signal. and peak point time distribution series Calculate the mean of each data point. and Calculate the product of the deviations of each data point from the mean, and then calculate the square root of the sum of their squares. Finally, substitute these squares into the formula to calculate the signal correlation strength value. .

[0035] Specific calculation example: Set the temperature signal change rate data as ; Calculate the mean: ; Calculate the product of deviations and the sum of squares of deviations: ; ; Calculate the correlation strength value: ; The results indicate that the rate of change of the temperature signal is highly correlated with the time distribution of the peak point, providing a basis for the generation of signal-related characteristic parameters.

[0036] S203: Based on the signal correlation feature matrix, set the signal linkage threshold range, filter the time points where the signal linkage intensity meets the threshold, organize the linkage time trend data of vibration signal, displacement signal and temperature signal, establish the signal linkage relationship with the filtering results, and generate an abnormal linkage signal matrix. The linkage conditions of vibration, displacement, and temperature signals are determined, and time points where the signal linkage intensity meets the threshold are selected. First, time series data between signals are extracted, and time point data of different signal types are normalized to establish a signal linkage matching model. The time offset between different signals is calculated using the Dynamic Time Warping (DTW) method. Based on the offset, signal data that meets the linkage threshold is selected, and time series pattern recognition is performed on the selected time points. The Support Vector Machine (SVM) classification method is used to classify and organize the signal linkage time trend data. The signal linkage weight matrix is ​​calculated based on the selection results, and the time points of abnormal linkage signals are extracted. Cluster analysis is performed on the distribution of abnormal linkage signals to generate an abnormal linkage signal matrix.

[0037] Please see Figure 4 The specific steps of S3 are as follows: S301: Based on the abnormal linkage signal matrix, extract the time series data of linkage intensity signal points, perform trend fitting processing on the time series data, compare the fitting results with the equipment operation mode baseline value point by point, filter out the parts of the signal point time distribution that do not match the baseline, and generate an abnormal signal time series feature set. The signal variation trend is fitted using a multinomial regression method, the fitting error of the signal point time series is calculated, the trend deviation of each signal point is calculated using a moving window, the fitted trend is compared with the equipment operation mode baseline value, a time deviation curve is constructed, the deviation value is statistically analyzed, the parts of the signal point time distribution that do not match the baseline value are screened out, the deviation amplitude of each signal point is calculated, an abnormal signal deviation index is established, abnormal signal points with deviations exceeding a set threshold are screened out, the time distribution characteristics of abnormal signal points are extracted, and finally an abnormal signal time series feature set is generated.

[0038] S302: Based on the time series feature set of abnormal signals, extract the time axis distribution data of signal points, divide the time series according to time periods, calculate the distribution density value of the signal in the interval, summarize the signal density distribution characteristics in each time period, and generate a signal distribution density matrix. The specific formula for calculating signal distribution density is as follows: ; in, Represents signal distribution density, Representing the The number of signals within a time period, This represents the average number of signals across all time periods. Represents the total number of time intervals. This represents the duration of each time period.

[0039] This formula is used to calculate the signal distribution density, where This represents the average density of the signal over a time interval. The time axis distribution data of the signal is collected through monitoring equipment, and the signal density is calculated based on piecewise statistical data. The parameters are explained below:

[0040] Representing the The number of signals recorded within a time period is counted by the signal monitoring system.

[0041] The average number of signals across all time periods is calculated as follows: ; This value is obtained by calculating the ratio of the total number of signals across all time periods to the number of time periods.

[0042] This represents the total number of time intervals, which are divided into time periods according to the timeline, with each time period having a fixed length.

[0043] This represents the length of each time period, and is usually set to a fixed time window, such as 1 second, 5 seconds, or 10 seconds.

[0044] Specific calculation example: Assuming the monitoring process is divided into... A time period, the length of each time period The number of signals recorded by the monitoring system per second is as follows: ; Calculate the mean: ; Calculate the sum of absolute deviations: ; ; Calculate the signal distribution density: ; Analysis of calculation results: The results indicate that a signal distribution density of 0.6 means that an average of 0.6 signals deviate from the overall mean per second. This density can be used to assess the distribution of signals and further for signal anomaly analysis or optimization of data distribution strategies.

[0045] S303: Based on the signal distribution density matrix, perform multi-dimensional classification of the difference data of signal distribution density in each time interval, classify and integrate the signals according to the difference data of time intervals, organize multi-dimensional feature data and establish a complete correlation structure to generate an abnormal signal difference matrix; Statistical features of signal distribution in different time periods are extracted, the mean and variance of each signal type are calculated, the density changes of different signal types on the time axis are analyzed, principal component analysis (PCA) is performed on the time density data to extract the main density feature vectors, signals are classified according to time period differences, K-means clustering is used to classify signal density, the time distribution overlap of each type of signal is calculated, multi-dimensional feature data are organized, a time density classification mapping matrix is ​​established, correlation patterns of different signal types are extracted, features of the time series are fused, the time overlap rate of abnormal signal types is calculated, an abnormal signal classification structure is established, and finally an abnormal signal difference matrix is ​​generated.

[0046] Please see Figure 5 The specific steps of S4 are as follows: S401: Based on the abnormal signal difference matrix, extract the signal difference data of the nodes in the task chain, sort each node according to the difference value, determine the logical relationship between the difference value and the sorting result for each node, generate a priority data table according to the sorting weight, and obtain the node priority sorting table. Based on the abnormal signal difference matrix, signal difference data of nodes in the task chain are extracted, and each node is prioritized and sorted according to the formula. ; In the formula, Representing the The priority of each node, Representative node In the Data across different dimensions, Representing the Weights of each difference dimension This represents the number of dimensions of difference.

[0047] For priority evaluation of each node, the data of each node in different difference dimensions is first extracted based on the abnormal signal difference matrix. Set the weights for each dimension. And calculate the weighted average to obtain the priority of each node. .

[0048] Specific example: Assume there are three difference dimensions, with weights as follows: The difference data for node A is Calculate the priority of node A: ; The result indicates that node A has a priority of 0.68, reflecting the importance of comprehensive evaluation based on its signal difference data.

[0049] S402: Based on the node priority sorting table, adjust the execution order of nodes in the task chain in turn, calculate the node resource requirements according to the adjusted order, optimize resource allocation by comparing with the global resource distribution rules of the task chain, and obtain the task path adjustment structure; First, the execution order of the nodes is rearranged, and the resource requirements of each node are calculated according to the new order. The resource allocation of the entire task chain is optimized using a linear programming method. The resource allocation ratio is optimized based on the resource requirements of each node and the total resource limit. The resource allocation model is used to evaluate the resource utilization efficiency of each node to ensure the fairness and efficiency of resource allocation. After adjustment, the resource requirements of the task path are recalculated, and the differences in resource allocation before and after optimization are compared to obtain the adjusted task path structure.

[0050] S403: Based on the task path adjustment structure, the resource optimization results are adjusted a second time by combining the signal difference density matrix. The rationality of the execution order of the path nodes is verified according to the optimization results. The resource and path adjustment data are integrated to reconstruct the task chain path structure and generate a task chain path optimization model. First, the resource usage of each node is simulated using an algorithm to evaluate the rationality of the adjusted resource configuration. Based on the optimization results, the execution order of the path nodes is verified to be optimal. Through simulated execution tests, the smoothness of task execution and the balance of resource usage are examined. The optimized resource configuration and path data are integrated to reconstruct the task chain path structure. Graph theory analysis methods are applied to ensure the optimization of the path structure and generate a task chain path optimization model. This model aims to improve task execution efficiency and reduce resource waste.

[0051] Please see Figure 6 The specific steps of S5 are as follows: S501: Based on the task chain path optimization model, it summarizes the feedback data after task execution, extracts data on resource consumption, task completion time and signal change trends, organizes and archives different types of data node by node, integrates archived content according to the task chain node distribution rules, and obtains task execution feedback dataset. For resource consumption data, the resource usage of each task node is calculated and grouped and archived according to resource type. For task completion time data, the execution time of each node is calculated and a task chain execution time series is established. For signal change trend data, the temporal characteristics of signals during task execution are extracted and classified and analyzed in conjunction with the node execution status. Finally, based on the task chain node distribution rules, the archived data is integrated to construct a task execution feedback dataset.

[0052] S502: Based on the task execution feedback dataset, the resource consumption and task completion time are calculated and compared in the order of nodes. Combined with the signal change trend data, the node status benefit parameters are classified item by item. Based on the benefit parameter data, the node status classification results are sorted out to obtain the node status benefit result set. Based on the task execution feedback dataset, resource consumption and task completion time are calculated and compared sequentially according to the node order, following the formula. ; In the formula, The parameter representing the node's state and revenue. Representing the Resource consumption of each task node Representing the The task completion time for each task node. Representing the Signal change trend parameters for each task node This represents the total number of nodes in the task chain.

[0053] The calculation of node status revenue is based on the task execution feedback dataset. First, the resource consumption data of each task node is obtained. Get task completion time Extract signal change trend parameters The difference between resource consumption and task time is calculated, and a weighted reward value is calculated based on signal change trends. Finally, the overall average reward of the task chain is calculated. .

[0054] Specific calculation example: A task chain is defined as containing five task nodes, with the following parameters: resource consumption data, task completion time, and signal change trend. ; Calculate the revenue contribution of each node: ; ; Calculate node state revenue parameters: ; The results show that the average node revenue of the task chain is 22.76, which reflects the execution efficiency and resource consumption of the task nodes, providing a basis for task optimization.

[0055] S503: Based on the node status revenue result set, integrate the node status revenue result with the task chain path data, optimize the resource distribution of the revenue result in combination with resource allocation requirements, readjust the node resource configuration according to the optimized data, and generate dynamic distribution results of task execution resources. Based on changes in resource consumption, task completion time, and signal trend parameters, resource distribution is optimized for the revenue of each task node. First, the resource usage ratio of each node in the task chain is calculated and compared with the total global resources to establish a resource allocation model. For nodes with low revenue, their resource configuration schemes are adjusted, and the impact of resource changes on task execution is analyzed. Based on the optimization results, the total resources in the task chain are redistributed, the resource usage ratio of each task node is adjusted, and the dynamic resource distribution scheme for task execution is optimized, ultimately generating the dynamic resource distribution result for task execution.

[0056] Please see Figure 7 An equipment monitoring and status assessment system based on multi-agent collaborative cooperation includes: The signal partitioning module, based on the monitoring data of equipment operation, partitions vibration signals, temperature change signals and displacement signals, performs time series analysis on each signal partition, calculates the amplitude and rate of change within the partition, and generates a multi-signal distribution feature matrix. The linkage analysis module is based on a multi-signal distribution feature matrix. It performs time distribution analysis on the peak points of vibration and displacement signals, compares the rate of change of temperature signals with vibration displacement, calculates the correlation between different signals, sets thresholds, and generates an abnormal linkage signal matrix. The anomaly detection module uses the anomaly linkage signal matrix to perform trend fitting analysis on the time series of linkage intensity signal points, extracts signals that do not match the equipment operation mode benchmark, integrates the distribution results into a multi-dimensional difference data structure, and generates an anomaly signal difference matrix. The task optimization module prioritizes nodes in the task chain based on the abnormal signal difference matrix, dynamically adjusts the node execution order and resource distribution of the task path, and generates a task chain path optimization model. The resource regulation module performs cumulative analysis on the feedback data after task execution based on the task chain path optimization model. By calculating resource consumption, task completion time, and signal change trends during execution, it generates dynamic distribution results of task execution resources.

[0057] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An equipment monitoring state evaluation method based on multi-agent collaborative cooperation, characterized in that, The method comprises the following steps: S1: based on the monitoring data of equipment operation, the vibration signal, the temperature change signal and the displacement signal are processed in partitions, time series analysis is performed on the signal partitions, the amplitude change amplitude and the change rate are calculated, the signal characteristics are integrated and analyzed, and a multi-signal distribution characteristic matrix is generated; S2: based on the multi-signal distribution characteristic matrix, time distribution analysis is performed on the vibration and displacement signal peak points, the temperature signal change rate is compared with the vibration displacement, the correlation degree between signals is calculated and a threshold is set, the signal time trend with high linkage strength is summarized, and an abnormal linkage signal matrix is generated; S3: based on the abnormal linkage signal matrix, trend fitting analysis is performed on the linkage strength signal point time series, signals that do not match the equipment operation mode benchmark are extracted, a signal distribution density matrix is established, the distribution results are integrated into a multi-dimensional difference data structure, and an abnormal signal difference matrix is generated; S4: based on the abnormal signal difference matrix, the task chain nodes are prioritized according to the signal difference, the node execution order and resource distribution of the task path are dynamically adjusted, the execution resources are optimized according to the signal difference density matrix, and a task chain path optimization model is generated; S5: according to the task chain path optimization model, cumulative analysis is performed on the task execution feedback data, the resource consumption, the task completion time and the signal change trend in the execution process are calculated, the node state benefit results are archived and integrated, and a task execution resource dynamic distribution result is generated.

2. The method of claim 1, wherein, The multi-signal distribution characteristic matrix includes a signal time series matrix, a signal amplitude change matrix and a signal change rate matrix. The abnormal linkage signal matrix includes a signal linkage strength matrix, a signal time correlation matrix and a signal change trend matrix. The abnormal signal difference matrix is specifically a signal deviation degree matrix, a signal distribution density matrix and a signal difference characteristic matrix. The task chain path optimization model includes a node priority distribution matrix, a resource dynamic distribution matrix and a task path sorting matrix. The task execution resource dynamic distribution result includes a resource use matrix, a node benefit distribution matrix and an execution time distribution matrix. 3.The equipment monitoring state evaluation method based on multi-agent collaborative cooperation according to claim 1, characterized in that, Based on the monitoring data of equipment operation, the vibration signal, the temperature change signal and the displacement signal are processed in partitions, time series analysis is performed on each signal partition, the amplitude change amplitude and the change rate value in the partition are calculated, the signal characteristics are integrated and analyzed, and the specific steps of generating a multi-signal distribution characteristic matrix are as follows: S101: based on the monitoring data of equipment operation, the vibration signal, the temperature change signal and the displacement signal are partitioned according to the signal type, the time sequence information of the signal partition data is extracted one by one, the instantaneous state and the time change trend of the amplitude data are summarized, the correlation degree characteristic value of the time sequence data with the signal state change is selected, and a signal partition time sequence feature set is obtained; S102: Based on the signal partition time sequence feature set, the amplitude change range and the time interval ratio of each group of time sequence feature data are compared through the continuity analysis of the signal partition data, the amplitude change range is summarized according to the continuity feature, the change rate reference data table is established for the time interval data, the relationship feature value of the signal amplitude and the time change rate is calculated, and the signal amplitude and the rate feature set are obtained; S103: Based on the signal amplitude and rate feature set, the partition feature data of the vibration signal, the temperature change signal and the displacement signal are classified according to the correlation between signals, the distribution characteristics of the partition signals are integrated, the amplitude and rate feature data matrices of different signals are summarized and adjusted, and a multi-signal distribution feature matrix is generated.

4. The method of claim 1, wherein, Based on the multi-signal distribution feature matrix, the time distribution of the peak points of the vibration and displacement signals is analyzed, the change rate of the temperature signal is compared with the vibration displacement, the correlation degree between different signals is calculated, a threshold is set, the signal time trend with high linkage strength is summarized, and the specific steps of generating an abnormal linkage signal matrix are as follows: S201: Based on the multi-signal distribution feature matrix, the peak point information in the vibration signal and the displacement signal is extracted, the signal peak point data is arranged in time sequence, the distribution characteristics of the peak point time interval are analyzed, the peak point data is classified and arranged according to the time interval, and a peak point time distribution feature set is obtained; S202: Based on the peak point time distribution feature set, the temperature signal change rate data is extracted, the time distribution characteristics of the peak points of the vibration signal and the displacement signal are compared, the correlation strength value between the temperature signal rate data and the peak points is calculated, a signal correlation feature parameter table is generated by comparing the data results, and a signal correlation feature matrix is obtained; S203: Based on the signal correlation feature matrix, a signal linkage threshold range is set, time points with signal linkage strength meeting the threshold are screened, linkage time trend data of the vibration signal, the displacement signal and the temperature signal are arranged, a signal linkage relationship is established combined with the screening results, and an abnormal linkage signal matrix is generated.

5. The method of claim 4, wherein, The correlation strength value calculation formula is ； wherein, represents the correlation strength value between the temperature signal change rate data and the peak point, represents the temperature signal change rate at the time point, represents the temperature signal change rate at the time point, represents the corresponding peak point time distribution characteristic value, and respectively represent the mean values of the temperature signal change rate and the peak point time distribution, represents the total number of data points.

6. The method of claim 1, wherein, Using the abnormal linkage signal matrix, trend fitting analysis is performed on the time sequence of the linkage strength signal points, signals that do not match the equipment operation mode benchmark are extracted, a distribution density matrix of the signals is established based on the time sequence, the distribution results are integrated into a multi-dimensional difference data structure, and the specific steps of generating an abnormal signal difference matrix are as follows: S301: Based on the abnormal linkage signal matrix, the time sequence data of the linkage strength signal points are extracted, trend fitting processing is performed on the time sequence data, the fitting results and the equipment operation mode benchmark value are compared point by point, the parts with signal point time distribution not matching the benchmark are screened, and an abnormal signal time sequence feature set is generated; S302: Based on the abnormal signal time sequence feature set, the time axis distribution data of the signal points are extracted, the time sequence is divided according to the time period, the distribution density value of the signal in the interval is calculated, the signal density distribution characteristics in each time period are summarized, and a signal distribution density matrix is generated; S303: Based on the signal distribution density matrix, the difference data of signal distribution density in each time interval is classified in multiple dimensions, the signals are classified and integrated according to the time period difference data, the multi-dimensional feature data is arranged and the complete association structure is established, and an abnormal signal difference matrix is generated.

7. The method of claim 6, wherein, The signal distribution density calculation formula is specifically: ； wherein, representing the signal distribution density, representing the number of signals in the first time period, representing the average number of signals over all time periods, representing the total number of time intervals, representing the length of each time period. 8.The method of claim 1, wherein, Based on the abnormal signal difference matrix, the nodes in the task chain are prioritized according to the signal difference, the node execution order and resource distribution of the task path are dynamically adjusted, and the execution resources are optimized according to the signal difference density matrix, and the specific steps of generating the task chain path optimization model are: S401: Based on the abnormal signal difference matrix, the signal difference data of the nodes in the task chain is extracted, and each node is prioritized according to the difference value, the logical relationship of the difference value relative to the sorting result is judged node by node, the priority data table is generated according to the sorting weight, and the node priority sorting table is obtained; S402: Based on the node priority sorting table, the execution order of the nodes in the task chain is adjusted in turn, the node resource requirement value is calculated according to the adjusted order, the resource allocation is optimized by comparing the global resource distribution rule of the task chain, and the task path adjustment structure is obtained; S403: Based on the task path adjustment structure, the resource optimization result is adjusted again in combination with the signal difference density matrix, the execution order rationality of the path nodes is verified according to the optimization result, the resource and path adjustment data are integrated to reconstruct the task chain path structure, and the task chain path optimization model is generated. 9.The equipment monitoring state evaluation method based on multi-agent collaboration according to claim 1, wherein, According to the task chain path optimization model, the feedback data after task execution is accumulated and analyzed, the benefits of all node states are archived and integrated by calculating the resource consumption, task completion time and signal change trend, and the specific steps of generating the task execution resource dynamic distribution result are: S501: Based on the task chain path optimization model, the feedback data after task execution is summarized, the data of resource consumption, task completion time and signal change trend are extracted, and the data is arranged and archived node by node for different types of data, and the archived content is integrated according to the task chain node distribution rule, and the task execution feedback data set is obtained; S502: Based on the task execution feedback data set, the resource consumption and task completion time are calculated and compared according to the node order, the node state benefit parameters are classified item by item in combination with the signal change trend data, the node state classification results are arranged according to the benefit parameter data, and the node state benefit result set is obtained; S503: Based on the node state benefit result set, the node state benefit result and the task chain path data are integrated, the benefit result is optimized in combination with the resource allocation demand, the node resource configuration is adjusted again according to the optimized data, and the task execution resource dynamic distribution result is generated.

10. An equipment monitoring state evaluation system based on multi-agent collaborative cooperation, characterized by, The equipment monitoring state evaluation method based on multi-agent collaborative cooperation according to any one of claims 1-9, the system comprises: The signal partition module partitions and processes the vibration signal, the temperature change signal and the displacement signal based on the monitoring data of the equipment operation, performs time series analysis on each signal partition, calculates the amplitude change range and the change rate value in the partition, and generates a multi-signal distribution feature matrix; The linkage analysis module performs time distribution analysis on the peak points of the vibration and displacement signals based on the multi-signal distribution feature matrix, compares the change rate of the temperature signal with the vibration and displacement, calculates the correlation degree between different signals, sets a threshold, and generates an abnormal linkage signal matrix; The abnormal detection module performs trend fitting analysis on the time series of the linkage intensity signal points using the abnormal linkage signal matrix, extracts signals that do not match the equipment operation mode benchmark, integrates the distribution results into a multi-dimensional difference data structure, and generates an abnormal signal difference matrix; The task optimization module prioritizes the nodes in the task chain according to the signal difference based on the abnormal signal difference matrix, dynamically adjusts the node execution order and resource distribution of the task path, and generates a task chain path optimization model; The resource regulation module accumulates and analyzes the feedback data after task execution according to the task chain path optimization model, calculates the resource consumption, task completion time and signal change trend during execution, and generates a task execution resource dynamic distribution result.

Images