An IO module for industrial monitoring system

Abstract

The present invention relates to the field of automatic control technology, specifically to an IO module of an industrial monitoring system, the module including a time series data reconstruction unit, an abnormal monitoring and adaptation unit, a frequency domain analysis and positioning unit and a parameter optimization control unit. The present invention reconstructs a multidimensional phase space through a delayed coordinate mapping method, improves the visualization and analysis capabilities of monitoring dynamics, enhances the analysis and prediction capabilities of behavior patterns in complex industrial environments by identifying attractor structures, improves the recognition and prediction accuracy of abnormal states by the application of random forests and support vector machines, realizes accurate real-time abnormal monitoring, ensures that the system can adaptively adjust the response process and monitoring parameters in real time, improves the flexibility and response speed of the system, deeply analyzes IO signals through frequency domain analysis, and locates abnormal frequency sources, optimizes fault diagnosis and preventive measures, and improves the operating efficiency and production stability of industrial equipment.

Description

Technical Field

[0001] The present invention relates to the field of automatic control technology, and in particular to an IO module of an industrial monitoring system. Background Art

[0002] The field of automatic control technology can operate independently without direct human control, ranging from basic closed-loop feedback control systems (such as temperature controllers) to automated control systems (such as robots and intelligent manufacturing systems). The key components of this technology include sensors for monitoring the environment and system status, controllers for decision-making, and actuators for executing control commands. Automatic control plays a key role in improving production efficiency, reducing labor costs and enhancing product quality. It is widely used in many fields such as industrial automation, aerospace, automobiles and electronic equipment.

[0003] Among them, the IO module in the industrial monitoring system is the key component responsible for data exchange, enabling the system to effectively monitor and control industrial equipment and processes. The module receives signals from sensors in the industrial environment (such as temperature, pressure or flow sensors) and converts the data into digital signals so that the central monitoring system can parse and respond. At the same time, the IO module is also responsible for sending control signals from the monitoring system to various actuators to regulate machine operations or adjust production processes. Its application improves the level of automation in industrial operations and enhances the real-time response capability of the system.

[0004] Traditional technologies rely on preset parameters and rules, lack the ability to quickly adapt to unexpected environmental changes and system states, limit the efficiency of the control system when faced with unknown or rapidly changing operating conditions, and result in untimely or inaccurate monitoring and response. Traditional technology data exchange does not support advanced data processing and analysis, which limits the ability to deeply analyze collected data and fails to fully optimize the operating process, resulting in low production efficiency, high energy consumption, increased maintenance costs and increased production safety risks. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings in the prior art and to propose an industrial monitoring system IO module.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: an industrial monitoring system IO module, the module comprising:

[0007] The time series data reconstruction unit receives the time series IO data from the monitoring device, applies the delayed coordinate mapping method to reconstruct the multidimensional phase space, sets the embedding dimension and delay time, identifies the attractor structure, extracts the key features of the monitoring dynamics, and generates a time series feature set;

[0008] The abnormal monitoring and adaptation unit uses random forest and support vector machine based on the time series feature set to identify and predict abnormal states in the IO data, automatically adjust the response process and monitoring parameters according to the monitoring results, and generate a response adjustment configuration;

[0009] The frequency domain analysis and positioning unit processes the IO signal of the industrial equipment based on the reaction adjustment configuration, utilizes Fourier transform, analyzes the signal frequency distribution, performs spectrum analysis according to the frequency characteristics, identifies and locates the abnormal frequency source, and generates an abnormal source identification record;

[0010] The parameter optimization control unit automatically adjusts the IO parameters of the industrial equipment based on the abnormal source identification record, optimizes the operating efficiency of the industrial equipment through feedback control and adjustment of operating parameters, and generates operating efficiency optimization results.

[0011] The present invention is improved in that the steps of setting the embedding dimension and the delay time are specifically as follows:

[0012] Receive time series IO data from monitoring equipment, use filtering algorithms to remove noise and outliers, and obtain processed time series data sets;

[0013] Based on the time series data set, the phase space reconstruction technique is used to estimate the embedding dimension and delay time using the formula:

[0014]

[0015] and

[0016]

[0017] Get the initial embedding dimension and delay time ,in, is the sampling frequency, which determines the time interval for data acquisition. is the autocorrelation function, Represents the principal component analysis eigenvalues, Represents the principal component analysis eigenvalues, is the total number of features;

[0018] The preliminary embedding dimension and delay time are optimized to reduce the prediction error, using the formula:

[0019]

[0020] and

[0021]

[0022] Get the optimized embedding dimension and delay time ,in, and is the partial derivative of the error, which represents the rate of change of the error with respect to the embedding dimension and delay time, and is a coefficient that adjusts the embedding dimension and delay time.

[0023] The present invention is improved in that the step of extracting the key features of the monitoring dynamics is specifically as follows:

[0024] Based on the multidimensional phase space, the geometric distribution characteristics of high-dimensional data are used to calculate the Euclidean distance between multiple points through the distance matrix to identify the attractor structure, using the formula:

[0025]

[0026] The attractor structure is obtained, where Represents data points and The distance in the reconstructed phase space, Representative The data point in The value of the dimension, Representative The data point in The value of the dimension, Representative The weight of the dimension, is the total number of dimensions of the reconstructed phase space;

[0027] Based on the attractor structure, the key dynamic monitoring features are calculated using the formula:

[0028]

[0029] Get the key dynamic features, where is the Lyapunov exponent, is the limit symbol, Represents a point in time, represents the sampling time interval, is the natural logarithm function, is the dynamic equation For status The partial derivative of is the scaling parameter, It's in time Monitoring status.

[0030] The present invention is improved in that the abnormal state identification and prediction steps are specifically as follows:

[0031] Based on the time series feature set, key feature integration is performed using the formula:

[0032]

[0033] Get the integrated feature vector ,in, is the total number of features, is the feature weight, is the feature mean, Indicates eigenvalues, is the parameter that adjusts the nonlinear effects;

[0034] Based on the integrated feature vector, random forest and support vector machine are used for training and anomaly prediction, using the formula:

[0035]

[0036] Get the predicted output , represents the probability of abnormal state, where yes The number of elements in , is the weight parameter, representing The weight of is the first eigenvector after integration elements;

[0037] Based on the predicted output results, the abnormal state is evaluated and the judgment formula is adopted:

[0038]

[0039] Identified abnormal state ,in, is the predicted output result, is the threshold used to determine whether to trigger an exception.

[0040] The present invention is improved in that the step of obtaining the reaction adjustment configuration is specifically as follows:

[0041] The abnormal state monitoring results are analyzed, and the probability and potential impact of each prediction are combined, using the formula:

[0042]

[0043] Get quantitative impact assessment results ,in, It is The probability of a predicted result, is the probability mean, It is The weight of the result, The number of prediction results;

[0044] Based on the quantitative impact assessment results, the response process and monitoring parameters are adjusted using the formula:

[0045]

[0046] Get response adjustment configuration ,in, is the adjustment coefficient, which is used to control the adjustment range. is the nonlinear characteristic parameter of the regulation response, It is the quantitative impact assessment result.

[0047] The present invention is improved in that the processing steps of the industrial equipment IO signal are specifically as follows:

[0048] Based on the reaction adjustment configuration, using sensors to collect IO signals from industrial equipment interfaces, including current and voltage signal data, to obtain time domain signals;

[0049] Based on the time domain signal, Fourier transform is applied to frequency domain conversion using the formula:

[0050]

[0051] Get the signal represented in the frequency domain ,in, is the total number of time points, It's time point The signal strength of the original IO signal at the time point The measured value of is the angular frequency, representing the variable in the frequency domain,

[0052] Based on the signal represented in the frequency domain, the key frequency components and amplitudes in the signal are determined, and the signal energy distribution is calculated using the formula:

[0053]

[0054] Get the signal frequency distribution analysis results ,in, is the signal represented in the frequency domain The mold length.

[0055] The present invention is improved in that the step of obtaining the abnormal source identification record is specifically as follows:

[0056] The frequency feature is identified, the energy value in the frequency distribution is calculated and compared with the threshold, using the formula:

[0057]

[0058] Get the abnormal frequency set ,in, is the signal frequency distribution analysis result, is the signal represented in the frequency domain The module length, is the threshold value;

[0059] Based on the abnormal frequency set, source analysis is performed to capture patterns that deviate from the norm, using the formula:

[0060]

[0061] Get the abnormal source identification record ,in, is an analysis function used to compare the abnormal frequency with the equipment specifications. is the set of abnormal frequencies, It is the equipment specification parameter.

[0062] The present invention is improved in that the steps of obtaining the operation performance optimization result are specifically as follows:

[0063] Based on the abnormal source identification record, the IO parameters of the industrial equipment are adjusted using the formula:

[0064]

[0065] Get the adjusted parameters ,in, Indicates the original parameters of the device. It is a proposal for adjustment ratio based on anomaly identification analysis. is the adjustment sensitivity coefficient;

[0066] Based on the adjusted parameters, the impact on the equipment performance is monitored using the formula:

[0067]

[0068] Get adjustment feedback results ,in, is the feedback gain, and They are the target and current equipment performance indicators;

[0069] The adjusted parameters and the adjustment feedback results are integrated to obtain an operation performance optimization result.

[0070] Compared with the prior art, the advantages and positive effects of the present invention are:

[0071] In the present invention, the multidimensional phase space is reconstructed by the delayed coordinate mapping method, which improves the visualization and analysis capabilities of monitoring dynamics. By identifying the attractor structure, the analysis and prediction capabilities of behavior patterns in complex industrial environments are enhanced. The application of random forests and support vector machines improves the recognition and prediction accuracy of abnormal states, realizes accurate real-time abnormal monitoring, ensures that the system can adaptively adjust the response process and monitoring parameters in real time, improves the flexibility and response speed of the system, deeply analyzes the IO signal through frequency domain analysis, and locates the abnormal frequency source, optimizes fault diagnosis and prevention measures, and improves the operating efficiency and production stability of industrial equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 is a system flow chart of the present invention;

[0073] Figure 2 A flow chart for setting the embedding dimension and delay time in the present invention;

[0074] Figure 3 A flowchart for extracting key features of monitoring dynamics in the present invention;

[0075] Figure 4 This is a flow chart for identifying and predicting abnormal conditions in the present invention;

[0076] Figure 5 A flowchart for obtaining the reaction adjustment configuration in the present invention;

[0077] Figure 6 This is a processing flow chart of the IO signal of the industrial equipment in the present invention;

[0078] Figure 7 This is a flowchart for obtaining abnormal source identification records in the present invention;

[0079] Figure 8 The figure is a flow chart for obtaining the operating performance optimization results in the present invention. DETAILED DESCRIPTION

[0080] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0081] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0082] Example

[0083] See also Figure 1 The present invention provides a technical solution: an industrial monitoring system IO module includes:

[0084] The time series data reconstruction unit receives the time series IO data from the monitoring device, applies the delayed coordinate mapping method to reconstruct the multidimensional phase space, sets the embedding dimension and delay time, identifies the attractor structure, extracts the key features of the monitoring dynamics, and generates a time series feature set;

[0085] The abnormal monitoring and adaptation unit uses random forest and support vector machine based on the time series feature set to identify and predict abnormal conditions in IO data, automatically adjust the response process and monitoring parameters according to the monitoring results, and generate response adjustment configuration;

[0086] The frequency domain analysis and positioning unit uses Fourier transform based on the reaction adjustment configuration to process the IO signals of industrial equipment, analyze the signal frequency distribution, and perform spectrum analysis based on the frequency characteristics to identify and locate the abnormal frequency source and generate the abnormal source identification record;

[0087] The parameter optimization control unit automatically adjusts the IO parameters of industrial equipment based on abnormal source identification records, eliminates the impact of the production process through feedback control and adjustment of operating parameters, optimizes the operating efficiency of industrial equipment, and generates operating efficiency optimization results.

[0088] The time series feature set includes response curves, behavior pattern indicators, and state change frequency. The response adjustment configuration includes update lists, parameter adjustment records, and warning level settings. The abnormal source identification records include peak frequency positions, abnormal vibration sources, and signal distortion areas. The operation efficiency optimization results include energy consumption reduction ratios, production throughput optimization results, and response time optimization results.

[0089] See also Figure 2 , the specific steps for setting the embedding dimension and delay time are:

[0090] Receive time series IO data from monitoring equipment, use filtering algorithms to remove noise and outliers, and obtain processed time series data sets;

[0091] Based on the time series data set, the phase space reconstruction technology is used to estimate the embedding dimension and delay time using the formula:

[0092]

[0093] and

[0094]

[0095] Get the initial embedding dimension and delay time ,in, is the sampling frequency, which determines the time interval for data acquisition. is the autocorrelation function, Represents the principal component analysis eigenvalues, Represents the principal component analysis eigenvalues, is the total number of features;

[0096] The initial embedding dimension and delay time are optimized to reduce the prediction error, using the formula:

[0097]

[0098] and

[0099]

[0100] Get the optimized embedding dimension and delay time ,in, and is the partial derivative of the error, which represents the rate of change of the error with respect to the embedding dimension and delay time, and is a coefficient that adjusts the embedding dimension and delay time.

[0101] Assume 100Hz, that is, 100 data points are collected per second;

[0102] ACF is the autocorrelation function, which measures the relationship between a data point and its delay The correlation between versions captures the maximum autocorrelation function value;

[0103] Suppose there is a set of data points In time , calculate the value of each The autocorrelation function value is found in ACF at 2 seconds Reaching the maximum value of 0.9, then:

[0104]

[0105] It indicates that the optimal time delay is 0.02 s for reconstructing the phase space.

[0106] Assume it is 10;

[0107] Capture the smallest , so that the front The cumulative contribution of the eigenvalues ​​exceeds 90%;

[0108] Assume that the eigenvalue is , then the sum , calculate the cumulative contribution of the first three eigenvalues:

[0109]

[0110] therefore, , that is, the embedding dimension is 3.

[0111] and Assume 0.05;

[0112] and Assume 0.1 and 0.2.

[0113] According to the above assumptions, we have:

[0114]

[0115]

[0116] This means that the optimized embedding dimension is slightly reduced, while the latency is halved, which improves the accuracy and efficiency of the model.

[0117] See also Figure 3 ,The steps for extracting the key features of monitoring dynamics are as follows:

[0118] Based on the multidimensional phase space, the geometric distribution characteristics of high-dimensional data are used to calculate the Euclidean distance between multiple points through the distance matrix to identify the attractor structure, using the formula:

[0119]

[0120] The attractor structure is obtained, where Represents data points and The distance in the reconstructed phase space, Representative The data point in The value of the dimension, Representative The data point in The value of the dimension, Representative The weight of the dimension, is the total number of dimensions of the reconstructed phase space;

[0121] Based on the attractor structure, the key dynamic monitoring features are calculated using the formula:

[0122]

[0123] Get the key dynamic features, where is the Lyapunov exponent, is the limit symbol, Represents a point in time, represents the sampling time interval, is the natural logarithm function, is the dynamic equation For status The partial derivative of is the scaling parameter, It's in time Monitoring status.

[0124] Suppose there are two three-dimensional data points and ;

[0125] Weight .

[0126] The calculation process is as follows:

[0127]

[0128]

[0129]

[0130]

[0131] Distance value indicates point and The distance after considering the weights of each dimension.

[0132] Assume that the dynamic equation of the system ,exist The derivative of ;

[0133] Pick 1.

[0134] The calculation is as follows:

[0135]

[0136]

[0137]

[0138]

[0139] The numerical value indicates that the system is The points show strong chaotic characteristics.

[0140] See also Figure 4 , the specific steps of abnormal state identification and prediction are:

[0141] Based on the time series feature set, key features are integrated using the formula:

[0142]

[0143] Get the integrated feature vector ,in, is the total number of features, is the feature weight, is the feature mean, Indicates eigenvalues, is the parameter that adjusts the nonlinear effects;

[0144] Based on the integrated feature vector, random forest and support vector machine are used for training and anomaly prediction, using the formula:

[0145]

[0146] Get the predicted output , represents the probability of abnormal state, where yes The number of elements in , is the weight parameter, representing The weight of is the first eigenvector after integration elements;

[0147] Based on the predicted output results, the abnormal state is evaluated and the judgment formula is adopted:

[0148]

[0149] Identified abnormal state ,in, is the predicted output result, is the threshold used to determine whether to trigger an exception.

[0150] Assume the following data:

[0151] ;

[0152] ;

[0153] ;

[0154] .

[0155] Calculation process:

[0156]

[0157]

[0158]

[0159]

[0160] result Represents the value of the integrated feature vector, which is used for model prediction.

[0161] Assumptions ;

[0162] .

[0163] Calculation process:

[0164]

[0165]

[0166]

[0167] result The probability of indicating an abnormal state is 79.4%, based on the simplified model prediction of a single feature.

[0168] Assumptions is 0.7, and the judgment formula is used to judge. Since 0.794>0.7, then The output is abnormal and the current status is recognized as abnormal.

[0169] See also Figure 5 , the specific steps for obtaining the reaction adjustment configuration are:

[0170] The abnormal state monitoring results are analyzed, combined with the probability and potential impact of each prediction, using the formula:

[0171]

[0172] Get quantitative impact assessment results ,in, It is The probability of a predicted result, is the probability mean, It is The weight of the result, The number of prediction results;

[0173] Based on the quantitative impact assessment results, the response process and monitoring parameters are adjusted using the formula:

[0174]

[0175] Get response adjustment configuration ,in, is the adjustment coefficient, which is used to control the adjustment range. is the nonlinear characteristic parameter of the regulation response, It is the quantitative impact assessment result.

[0176] Suppose there are three probabilities of prediction results ;

[0177] The corresponding weight .

[0178] Calculate the average of the probabilities :

[0179]

[0180] Then calculate :

[0181]

[0182]

[0183]

[0184]

[0185] result Quantifies the overall deviation of the forecast results from the mean. The weight reflects the importance of each result. Values ​​indicate larger deviations.

[0186] Assumptions and , using the previously calculated .

[0187] calculate:

[0188]

[0189]

[0190]

[0191]

[0192] result Represents the intensity of adjustments that the system should make based on the deviation of the predicted results.

[0193] See also Figure 6 , the specific steps for processing IO signals of industrial equipment are:

[0194] Based on the reaction adjustment configuration, sensors are used to collect IO signals from industrial equipment interfaces, including current and voltage signal data, to obtain time domain signals;

[0195] Based on the time domain signal, apply Fourier transform to frequency domain conversion, using the formula:

[0196]

[0197] Get the signal represented in the frequency domain ,in, is the total number of time points, It's time point The signal strength of the original IO signal at the time point The measured value of is the angular frequency, representing the variable in the frequency domain,

[0198] Based on the signal represented in the frequency domain, the key frequency components and amplitudes in the signal are determined, and the signal energy distribution is calculated using the formula:

[0199]

[0200] Get the signal frequency distribution analysis results ,in, is the signal represented in the frequency domain The mold length.

[0201] Suppose there is a signal , including data at three time points:

[0202]

[0203] Total time points .

[0204] Calculate frequency of :

[0205]

[0206]

[0207]

[0208]

[0209]

[0210]

[0211] Indicates how to convert a time domain signal into its frequency domain representation at a specific frequency. Describes the complex amplitude of this frequency component.

[0212] Using the above values, calculate:

[0213]

[0214]

[0215]

[0216]

[0217] result It indicates the energy of the component with a frequency of 1 in the signal. A higher energy value indicates that this frequency component is more significant in the signal and is suspected to be a key frequency or anomaly source.

[0218] express The modulus length is a measure of the signal strength in the frequency domain and is calculated as:

[0219]

[0220] in and Represent the real and imaginary parts of a complex number respectively.

[0221] See also Figure 7 ,The specific steps for obtaining abnormal source identification records are:

[0222] To identify the frequency features, calculate the energy value in the frequency distribution and compare it with the threshold, using the formula:

[0223]

[0224] Get the abnormal frequency set ,in, is the result of signal frequency distribution analysis, indicating the frequency The energy is the signal represented in the frequency domain The modulus length indicates the frequency The amplitude of is the threshold value;

[0225] Based on the abnormal frequency set, source analysis is performed to capture patterns that deviate from the norm, using the formula:

[0226]

[0227] Get the abnormal source identification record ,in, is an analysis function used to compare the abnormal frequency with the equipment specifications. is the set of abnormal frequencies, It is the equipment specification parameter.

[0228] Assume that the amplitudes of the three frequencies obtained by Fourier transform are as follows:

[0229]

[0230]

[0231]

[0232] Setting Thresholds .

[0233] Calculate the energy at each frequency:

[0234]

[0235]

[0236]

[0237] Compare the energy of each frequency to the threshold:

[0238] ——Frequency 1 is the abnormal frequency;

[0239] ——Frequency 2 is an abnormal frequency;

[0240] ——Frequency 3 is an abnormal frequency.

[0241] therefore, The frequencies are identified as abnormal.

[0242] Assumptions Corresponding to the different operating modes of the equipment, by analyzing historical fault records and equipment specifications, it is determined that frequencies 1 and 3 match the known equipment failure modes, and frequency 2 is related to new or unknown problems. The analysis helps to accurately locate the source of the problem and provide key information for maintenance.

[0243] See also Figure 8 , the specific steps for obtaining the operation performance optimization results are:

[0244] Based on the abnormal source identification records, adjust the IO parameters of industrial equipment using the formula:

[0245]

[0246] Get the adjusted parameters ,in, Indicates the original parameters of the device. It is a proposal for adjustment ratio based on anomaly identification analysis. is the adjustment sensitivity coefficient;

[0247] Based on the adjusted parameters, monitor the impact on the equipment performance using the formula:

[0248]

[0249] Get adjustment feedback results ,in, is the feedback gain, and They are the target and current equipment performance indicators;

[0250] Integrate the adjusted parameters and adjustment feedback results to obtain the operation efficiency optimization results.

[0251] Assume the following data:

[0252] ;

[0253] ;

[0254] .

[0255] Calculation process:

[0256]

[0257]

[0258]

[0259]

[0260] This shows that the adjusted parameter value is 96, which is 4% lower than the original value, which is in line with the adjustment measures recommended after the abnormal source is identified.

[0261] Assumptions:

[0262] ;

[0263] ;

[0264] .

[0265] Calculation process:

[0266]

[0267]

[0268]

[0269]

[0270] Indicates adjustment response , indicating that based on the current performance gap, adjustments should be intensified to reach the target performance.

[0271] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. An industrial monitoring system IO module, characterized in that: The modules include: The time series data reconstruction unit receives the time series IO data from the monitoring device, applies the delayed coordinate mapping method to reconstruct the multidimensional phase space, sets the embedding dimension and delay time, identifies the attractor structure, extracts the key features of the monitoring dynamics, and generates a time series feature set; The abnormal monitoring and adaptation unit uses random forest and support vector machine based on the time series feature set to identify and predict abnormal states in the IO data, automatically adjust the response process and monitoring parameters according to the monitoring results, and generate a response adjustment configuration; The frequency domain analysis and positioning unit processes the IO signal of the industrial equipment based on the reaction adjustment configuration, utilizes Fourier transform, analyzes the signal frequency distribution, performs spectrum analysis according to the frequency characteristics, identifies and locates the abnormal frequency source, and generates an abnormal source identification record; The parameter optimization control unit automatically adjusts the IO parameters of the industrial equipment based on the abnormal source identification record, optimizes the operating efficiency of the industrial equipment through feedback control and adjustment of operating parameters, and generates an operating efficiency optimization result; The steps of identifying and predicting the abnormal state are specifically as follows: Based on the time series feature set, key feature integration is performed using the formula: Get the integrated feature vector ,in, is the total number of features, is the feature weight, is the feature mean, Indicates eigenvalues, is the parameter that adjusts the nonlinear effects; Based on the integrated feature vector, random forest and support vector machine are used for training and anomaly prediction, using the formula: Get the predicted output , represents the probability of abnormal state, where yes The number of elements in , is the weight parameter, representing The weight of is the first eigenvector after integration elements; Based on the predicted output results, the abnormal state is evaluated and the judgment formula is adopted: Identified abnormal state ,in, is the predicted output result, is the threshold used to determine whether an exception is triggered; The steps for obtaining the reaction adjustment configuration are specifically as follows: The abnormal state monitoring results are analyzed, and the probability and potential impact of each prediction are combined, using the formula: Get quantitative impact assessment results ,in, It is The probability of a predicted result, is the probability mean, It is The weight of the result, is the number of prediction results; Based on the quantitative impact assessment results, the response process and monitoring parameters are adjusted using the formula: Get response adjustment configuration ,in, is the adjustment coefficient, which is used to control the adjustment range. is the nonlinear characteristic parameter of the regulation response, It is the quantitative impact assessment result.

2. The industrial monitoring system IO module according to claim 1, characterized in that: The steps for setting the embedding dimension and delay time are specifically as follows: Receive time series IO data from monitoring equipment, use filtering algorithms to remove noise and outliers, and obtain processed time series data sets; Based on the time series data set, the phase space reconstruction technique is used to estimate the embedding dimension and delay time using the formula: and Get the initial embedding dimension and delay time ,in, is the sampling frequency, which determines the time interval for data acquisition. is the autocorrelation function, Represents the principal component analysis eigenvalues, Represents the principal component analysis eigenvalues, is the total number of features; The preliminary embedding dimension and delay time are optimized to reduce the prediction error, using the formula: and Get the optimized embedding dimension and delay time ,in, and is the partial derivative of the error, which represents the rate of change of the error with respect to the embedding dimension and delay time, and is a coefficient that adjusts the embedding dimension and delay time.

3. The industrial monitoring system IO module according to claim 1, characterized in that: The steps for extracting the key features of the monitoring dynamics are specifically as follows: Based on the multidimensional phase space, the geometric distribution characteristics of high-dimensional data are used to calculate the Euclidean distance between multiple points through the distance matrix to identify the attractor structure, using the formula: The attractor structure is obtained, where Represents data points and The distance in the reconstructed phase space, Representative The data point in The value of the dimension, Representative The data point in The value of the dimension, Representative The weight of the dimension, is the total number of dimensions of the reconstructed phase space; Based on the attractor structure, the key dynamic monitoring features are calculated using the formula: Get the key dynamic features, where is the Lyapunov exponent, is the limit symbol, Represents a point in time, represents the sampling time interval, is the natural logarithm function, is the dynamic equation For status The partial derivative of is the scaling parameter, It's in time Monitoring status.

4. The industrial monitoring system IO module according to claim 1, characterized in that: The processing steps of the industrial equipment IO signal are specifically as follows: Based on the reaction adjustment configuration, using sensors to collect IO signals from industrial equipment interfaces, including current and voltage signal data, to obtain time domain signals; Based on the time domain signal, Fourier transform is applied to frequency domain conversion using the formula: Get the signal represented in the frequency domain ,in, is the total number of time points, It's time point The signal strength of the original IO signal at the time point The measured value of is the angular frequency, representing the variable in the frequency domain, is an imaginary unit used to represent the phase portion of a signal; Based on the signal represented in the frequency domain, the key frequency components and amplitudes in the signal are determined, and the signal energy distribution is calculated using the formula: Get the signal frequency distribution analysis results ,in, is the signal represented in the frequency domain The mold length.

5. The industrial monitoring system IO module according to claim 1, characterized in that: The steps for obtaining the abnormal source identification record are specifically as follows: The frequency feature is identified, the energy value in the frequency distribution is calculated and compared with the threshold, using the formula: Get the abnormal frequency set ,in, is the signal frequency distribution analysis result, is the signal represented in the frequency domain The module length, is the threshold value; Based on the abnormal frequency set, source analysis is performed to capture patterns that deviate from the norm, using the formula: Get the abnormal source identification record ,in, is an analysis function used to compare the abnormal frequency with the equipment specifications. is the set of abnormal frequencies, It is the equipment specification parameter.

6. The industrial monitoring system IO module according to claim 1, characterized in that: The steps for obtaining the operation performance optimization result are specifically as follows: Based on the abnormal source identification record, the IO parameters of the industrial equipment are adjusted using the formula: Get the adjusted parameters ,in, Indicates the original parameters of the device. It is a proposal for adjustment ratio based on anomaly identification analysis. is the adjustment sensitivity coefficient; Based on the adjusted parameters, the impact on the equipment performance is monitored using the formula: Get adjustment feedback results ,in, is the feedback gain, and They are the target and current equipment performance indicators; The adjusted parameters and the adjustment feedback results are integrated to obtain an operation performance optimization result.

Images